Perception, Cognition, and Aesthetics [1 ed.] 9781138615939, 9780429462658

Table of contents :
Cover
Half Title
Series
Title
Copyright
Contents
Acknowledgments
Introduction
Section I Perception
1 Disjunctivism and the Internal: A Problem for McDowell’s Epistemological Disjunctivism?
2 Xenophanes’s Figs and Honey: An Essay about a Program of Philosophy of Perception
3 A Neurophilosophical Approach to Perception
4 Smelling Molecular Structure
5 “Hierarchical Bokeh” Theory of Attention
6 Perceiving Live Improvisation in the Performing Arts
Section II Cognition
7 Consciousness and Content in Perception
8 Perceptual Capacities
9 Thinking Differently About Thought
10 Immediate and Reflective Senses
11 The Unity of Unconsciousness
12 Phenomenal Experience and the Thesis of Revelation
Section III Perception and Cognition in Aesthetics
13 Would You Buy Absence Art?
14 Penetrating Beauty: Knowledge, Culture and Context in Aesthetic Perception
15 Gist Experience
16 How Do I Know When I Am Dancing?
17a Interview with James Cohan
17b Interview with Leonel Moura
Notes on Contributors
Index

Citation preview

Perception, Cognition and Aesthetics

This volume addresses key questions related to how content in thought is derived from perceptual experience. It includes chapters that focus on single issues on perception and cognition, as well as others that relate these issues to an important social construct that involves both perceptual experience and cognitive activities: aesthetics. While the volume includes many diverse views, several prominent themes unite the individual essays: a challenge to the notion of the discrete and non-temporal unit of perception, a challenge to the traditional divide between perception and cognition and a challenge to the traditional divide between unconscious and conscious intentionality. Additionally, the chapters discuss the content of perceptual experience, the value of traditional notions of content, disjunctivism, adverbialism and phenomenal experience. The final section of essays dealing with perception and cognition in aesthetics features work in experimental aesthetics and unique perspectives from artists and gallerists working outside of philosophy. Perception, Cognition and Aesthetics is a timely volume that offers a range of unique perspectives on debates in philosophy of mind surrounding perception and cognition. It will also appeal to scholars working in aesthetics and art theory who are interested in the ways these debates influence our understanding of art. Dena Shottenkirk is assistant professor in the Philosophy Department at Brooklyn College. Before receiving her PhD from the Graduate Center (CUNY), Shottenkirk had received an MFA in painting from Rutgers University and was employed as an art critic at both Artforum and Art in America magazines. Manuel Curado is professor at the University of Minho (Portugal). He’s mainly interested in intellectual history, the history of Jewish and Portuguese medicine, and the history of the representations of mental life, the Philosophy of Mind debate and the history of Psychiatry until the nineteenth century. Steven S. Gouveia is a PhD candidate in Philosophy of Mind at the University of Minho under the supervision of Professor Manuel Curado (University of Minho) and Professor Georg Northoff (University of Ottawa), funded by the Science and Technology Portuguese Foundation. More info: https://stevensgouveia.weebly.com/.

Routledge Studies in Contemporary Philosophy

Questions of Practice in Philosophy and Social Theory Edited by Anders Buch and Theodore R. Schatzki The Phenomenology of Real and Virtual Places Edited by Erik Malcolm Champion Context, Truth, and Objectivity Essays on Radical Contextualism Edited by Eduardo Marchesan and David Zapero Good Thinking A Knowledge First Virtue Epistemology Christoph Kelp Epiphenomenal Mind An Integrated Outlook on Sensations, Beliefs, and Pleasure William S. Robinson The Meanings of Violence From Critical Theory to Biopolitics Edited by Gavin Rae and Emma Ingala A Defense of Simulated Experience New Noble Lies Mark Silcox The Act and Object of Judgment Historical and Philosophical Perspectives Edited by Brian Ball and Christoph Schuringa Perception, Cognition and Aesthetics Edited by Dena Shottenkirk, Manuel Curado and Steven S. Gouveia For more information about this series, please visit: www.routledge.com/ Routledge-Studies-in-Contemporary-Philosophy/book-series/SE0720

Perception, Cognition and Aesthetics Edited by Dena Shottenkirk, Manuel Curado and Steven S. Gouveia

First published 2019 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2019 Taylor & Francis The right of the editors to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Names: Shottenkirk, Dena, editor. Title: Perception, cognition, and aesthetics / edited by Dena Shottenkirk, Manuel Curado, and Steven S. Gouveia. Description: 1 [edition]. | New York : Taylor & Francis, 2019. | Series: Routledge studies in contemporary philosophy ; 119 | Includes bibliographical references and index. Identifiers: LCCN 2019005354 | ISBN 9781138615939 (hardback) Subjects: LCSH: Aesthetics. | Perception (Philosophy) | Cognition. Classification: LCC BH39 .P3546 2019 | DDC 111/.85—dc23 LC record available at https://lccn.loc.gov/2019005354 ISBN: 978-1-138-61593-9 (hbk) ISBN: 978-0-429-46265-8 (ebk) Typeset in Sabon by Apex CoVantage, LLC

Contents

Acknowledgmentsviii Introduction

1

DENA SHOTTENKIRK, MANUEL CURADO, AND STEVEN S. GOUVEIA

SECTION I

Perception7 1 Disjunctivism and the Internal: A Problem for McDowell’s Epistemological Disjunctivism?

9

DAVIDE DALLA ROSA AND FEDERICO SANGUINETTI

2 Xenophanes’s Figs and Honey: An Essay about a Program of Philosophy of Perception

25

MANUEL CURADO

3 A Neurophilosophical Approach to Perception

41

STEVEN S. GOUVEIA AND GEORG NORTHOFF

4 Smelling Molecular Structure

64

BENJAMIN D. YOUNG

5 “Hierarchical Bokeh” Theory of Attention

85

ANATOLY NICHVOLODA

6 Perceiving Live Improvisation in the Performing Arts AILI BRESNAHAN

106

vi  Contents SECTION II

Cognition121   7 Consciousness and Content in Perception

123

BILL BREWER

  8 Perceptual Capacities

137

SUSANNA SCHELLENBERG

  9 Thinking Differently About Thought

170

NICHOLAS GEORGALIS

10 Immediate and Reflective Senses

187

ANGELA MENDELOVICI

11 The Unity of Unconsciousness

209

TIM CRANE

12 Phenomenal Experience and the Thesis of Revelation

227

MICHELLE LIU

SECTION III

Perception and Cognition in Aesthetics253 13 Would You Buy Absence Art?

255

ANYA FARENNIKOVA

14 Penetrating Beauty: Knowledge, Culture and Context in Aesthetic Perception

279

JESSE PRINZ

15 Gist Experience

302

DENA SHOTTENKIRK

16 How Do I Know When I Am Dancing? ROMAIN BIGÉ

319

Contents vii 17a Interview with James Cohan 

333

DENA SHOTTENKIRK

17b Interview with Leonel Moura 

340

STEVEN S. GOUVEIA

Notes on Contributors344 Index351

Acknowledgments

The editors would like to take a moment to thank all the authors for their valuable contributions. We especially would like to thank Diana Neiva for her support. Finally, we would like to thank Andrew Weckenmann for believing in this project.

Introduction Dena Shottenkirk, Manuel Curado, and Steven S. Gouveia

This edited volume not only addresses the usual question in philosophy of mind of how content in thought is derived from perceptual experience but also has many distinguished writers questioning the traditional distinction between the structure of perceptual phenomenon and that of cognition. That traditional distinction saw perception as passively obtained and providing the content for an actively operating cognition, but the last several years have seen various challenges to this paradigm, including challenges to the general notion of content. The conversation around the interface between perception and cognition has become widely published in journals and monographs during the last few years, and the importance of this edited volume is in the bringing together of many of the major voices in this conversation in the philosophy of perception and adding to them the voices of newer and younger thinkers, resulting in a compilation that properly addresses the main obstacle still present in the discourse: the relationship between perception and thought. But it is not only the timeliness of the issue that makes this volume distinct; it is also its embracing of aesthetics under the larger umbrella of perception and thought. Much recent work in aesthetics has pulled aesthetics out from the sidelines of Platonic beauty or Human taste, and has instead insisted on aesthetics’ rightful discourse amid the concerns of research into perceptual experience. This is true both within the analytic tradition as well as within the continental. In the former, research in aesthetics has been addressed most often by empirically minded philosophers who, building upon scientific research in various fields, see art as an important instance of perceptual experience. The latter tradition, e.g., continental philosophy, addresses the concerns of phenomenological research into perceptual experience. Thus, within both traditions, there has been a shift toward taking aesthetics as an integral part of perception theory. We agree with this trend and thus have given aesthetics equal weight in the examination of how thought is derived from perceptual experience. It is this derivation that is the crucial junction in individual and social human experience: the dovetailing of perceptual experience and cognitive

2  Dena Shottenkirk, et al. operations, or to phrase it in a slightly different fashion, the complex knitting together of our sensory absorption of the external world and our cognitive organization of those attentional experiences. This relationship is addressed by all the contributors, whether they have mainly chosen to focus on the single issue of perception or the single issue of thought, or whether they have instead chosen to write about the related concerns of perception and thought in aesthetics. The volume is therefore divided into three categories: perception, cognition and aesthetics. The first section gives varying perspectives on how we stand to the phenomenal world. Accordingly, issues of intentionality, attention, realism and content are dealt with. In the first section, “Perception,” Davide Dalla Rosa and Federico Sanguinetti offer a global picture of McDowell’s disjunctivism about perceptual experience which exhibits some possible problems regarding the internalistic features inscribed in McDowell’s position. Overthrowing the traditional passive model of perception in a slightly different way, Manuel Curado gives us a notion of perceptual content that is relational and context-driven, wherein he reexamines the preSocratic Xenophanes of Colophon’s writing fragment that questions whether the fig would have tasted sweeter if the eater had not already tasted honey. The relational instability of the content “sweetness” has far-reaching theoretical consequences that could be generalized. That instability is the distinctive sign of all phenomenic content. Curado sees in the Xenophanes’s fragment a pioneering way of looking at the metaphysical foundations of perception and eventually recommends that the Aristotelian category of relation should move to the central place where the category of substance remains for centuries. Also arguing against naïve realism, and concerned with how we stand to perceptual experience while wanting to preserve mind-independent physical objects, Steven S. Gouveia and Georg Northoff use empirical data from contemporary neuroscience to argue for a “spatiotemporal” model, as opposed to both a specific direct realist (naïve realism) account and, generally, the indirect realist accounts, explaining how the spatiotemporal model gives us the most plausible account of perception by providing for a bi-directionality between the perceptual facilities and the cognitive ones. Benjamin Young also examines a non-visual perceptual facility in his analysis of smells as intentional objects; relying on empirical data from the sciences, Young examines various theories and evaluates their ability to explain both smells as distal entities and our experience of them as intentional objects. Anatoly Nichvoloda develops his own outlook against the traditional view of perception as passively obtained and presents a novel theory explaining the mechanism behind information processing called the

Introduction 3 Bayes-Shannon system. A  theory of attention in relation to top-down and bottom-up perception, Nichvoloda’s chapter gives a view of perception consistent with a complex feedback model that enables perception to function as a temporally accruing constituent of cognition. Aili Bresnahan examines the relation between perception and cognition by taking us into the experience of dance, and develops a theory of perception that uses dance to illustrate the fact that there are cognitive processes that occur before conscious awareness of the perceptual experience itself. An examination of top-down influences that provides an understanding of both performance behavior and audience participation at the level of cortical circuitry, the chapter also demonstrates the centrality of the aesthetic experience in perception theory. The second section, “Cognition,” sees philosophers grappling with issues of intentional or representational states, capacities that are employed in understanding and distinctions between such things as conscious and unconscious mental states or between individual thoughts and the linguistic entities that are designated for them. Bill Brewer argues against the Content View by arguing that perceptual consciousness of the world extends beyond the range of conscious attention that delineates perceptual representational content. So content cannot capture all there is to conscious perception. Susanna Schellenberg reexamines the Aristotelian notion of mental capacities in cognitive systems, drawing on the fact that contemporary scientific research routinely employs the notion of capacity in analyses of cognitive functions. By providing an analysis of capacities (which is distinguishable from functionalism) within their neural base conditions, she is able to give accounts of capacities that are both context-driven and those that are not, and to explain the difference between possessing and exercising a capacity. Nick Georgalis develops a distinction between thought-tokens (e.g., individual agents’ thoughts) and thought, where the latter is attributed to a linguistic sentence with standard contents; it is a many–one relationship. By drawing attention to the social nature of linguistic attribution, Georgalis draws the general discussion toward the pertinent social role of language within the exercise of cognition. In a switch back to the individual’s cognitions, Angela Mendelovici examines the individual’s varying awarenesses of objective referents in, on the one hand, immediate sense experience, and, on the other, reflective senses, e.g., those things of which we are aware after reflection. Arguing against understanding the distinction between immediate and reflective senses using an analysis of cognition in terms of functionalism or tracking relations, she instead explains it in terms of phenomenal intentionality theory.

4  Dena Shottenkirk, et al. Tim Crane argues for the distinction between unconscious and conscious intentionality, focusing on the argument that the structure and contents of these states are not, as is commonly thought, the same. Concerning himself with states that represent the world, his argument is against the view that unconscious states can be modeled on propositional content. Michelle Liu’s chapter is an examination of revelation through the work of David Lewis and his view on the folk-psychological concept of qualia. Looking from within the physicalist-dualist debate, Liu points out that revelation is in tension with physicalism. She does this by carefully spelling out how revelation is to be understood and how it is incompatible with standard versions of physicalism which take qualia to have physical or functional essences. The final section, “Perception and Cognition in Aesthetics,” brings in several authors who are, in a few instances, both philosophers and art practitioners, and in a couple of instances, people only involved directly in the art world, e.g., the views of people actually “on the ground.” Here, too, we see a challenge to the traditional divide between perception and cognition, and here, too, questions of content and concepts are acute. The connection between phenomenal awareness and the objects of which we are phenomenally aware, how it is that we become aware of the objects and events in front of us, whether or not perceptual processes are fallible and to what degree our cognitive structures inhibit and inform our perceptual processes are all questions that are, to varying degrees, addressed by the writers. Anya Farennikova delves into the world of absence in art and examines instances of contemporary art wherein a part of the artwork is actually absent from viewing but is nonetheless essential to conceptually understanding the work. The reference to the absent thing makes us experience the art itself differently from those instances where we see something as actually there because it is a contradictory experience of seeing and notseeing, and therefore it necessarily brings with it a sense of longing. Jesse Prinz gives us a related view by arguing against the traditional notion of perception (in this case aesthetic) that is driven entirely by external stimuli in conjunction with our bodily processes. We don’t just see what is in front of us. Instead, he provides a top-down view of art that depends on seeing as a function of our cultural and historical situation; we can’t assume that we “understand” art that is from a very different place and time. Education and connoisseurship are essential for the aesthetic process. Narrowing down the focus, Dena Shottenkirk looks at the role of gist, e.g., the first few hundred milliseconds of perception, in the aesthetic experience. Relying on empirical data provided by psychologists and neuroscientists, Shottenkirk examines how whole-scene perception controls subsequent aesthetic examinations, and how understanding a

Introduction 5 work of art is dependent on viewers’ initial gist experience. Also an artist, Shottenkirk uses the example of Diego Velázquez’s Portrait of a Man to illustrate the theory. Switching back to dance, Romain Bigé, who is both a dancer and a philosopher, also focuses on the small slivers within temporal experience and asks the unusual question, “How do I know when I am dancing?” By pulling us into the minute moments that construct the barrier between ordinary movement and dancing, he directs our attention to the phenomenology of the act itself. Concluding this section and the book are the two interviews from “people on the ground”: American gallerist James Cohan and Portuguese artist Leonel Moura. James Cohan, with galleries in New York and Shanghai, is interviewed by Dena Shottenkirk. Leonel Moura is interviewed by Steven S. Gouveia. These two interviews give us correlating views by people who have thought deeply and very hard about what it is they do for a living and how their activities in the art world gain meaning and are absorbed by others. Together, all of these chapters form a body of voices articulating how we see and sensorially experience, how we think about what we see and experience, and how it is that it has meaning and value for us. These are the core issues in philosophy. But this is not just an academic book: It is engaged philosophy at its best, working alongside the sciences, art and the real world. It is possible that these themes will disappear in the distant future. Researchers in many scientific fields have the centuries-old experience of disappearing concepts and problems; philosophers tend to appreciate the company of their problems, preferring to transfigure rather than replace them. In any case, the themes discussed here are important in themselves. Nothing that exists in the world is devoid of goodness, but it is fair to recognize that perception, cognition and art have worth beyond themselves, as if in some way they were keys to the mystery of reality. All the minutes of intellectual work devoted to them provide a perhaps disproportionate effect. This effect is called understanding. Finally, and not without great emphasis, we want to thank all those who have done this work. To the philosophers who wrote the chapters: We are utterly indebted to each of you for those many hours of hard labor. (Not to mention, of course, also the emails and housekeeping chores involved!) We want to also thank the editors at Routledge, in particular Andrew Weckenmann, without whom this project could not have advanced. We are grateful.

Section I

Perception

1 Disjunctivism and the Internal A Problem for McDowell’s Epistemological Disjunctivism? Davide Dalla Rosa and Federico Sanguinetti

“[T]ruth at once reveals itself and also what is false, because truth is made clear through truth, that is through itself, and through it also is falsity made clear; but falsity is never revealed and made manifest through itself. So that anyone who is in possession of the truth cannot doubt that he possesses it, while one who is sunk in falsity or in error can well suppose that he has got at the truth; just as someone who is dreaming can well think that he is awake, but one who is actually awake can never think that he is dreaming.” Spinoza, Short Treatise, Chap. 15

1. Introduction This chapter will explore the relation between several key theoretical features characterizing McDowell’s epistemological disjunctivism. More specifically, we will focus here on the relations among the notions of appearance, warrant, perceptual capacity and fallibility. The aim of the chapter is twofold. First, we would like to provide a faithful reconstruction of McDowell’s disjunctivism and thus clarify the way some constitutive concepts which shape his views on perceptual experience relate to one another. Furthermore, our reconstruction will also highlight a possible difficulty that stems from McDowell’s attempt to accommodate the theoretical ambitions of his philosophical enterprise. The chapter proceeds as follows:

We shall begin with an analysis of McDowell’s disjunctivism, starting from the notion of appearance (Section 2). We will first show the link between the notion of appearance and the notion of warrant. This analysis is conducted from a third-person perspective, i.e., from the perspective of an observer considering the epistemic situations of two subjects who undergo, respectively, a veridical and a falsidical experience.    We reconstruct McDowell’s characterization of the perceptual capacity and its exercises, highlighting the link between the notion of

10  Davide Dalla Rosa and Federico Sanguinetti warrant and the notion of fallibility of perception (Section 3). This analysis is conducted from a first-person perspective, i.e., from the internal perspective of a subject who is in a sensible, conscious mental state. In the final part of this section, we point out a possible problem stemming from McDowell’s characterization of the first-person perspective when it is considered against the general background of his theoretical ambitions and aims. More specifically, we aim in the final part of this section and in the Conclusion (Section 4) to interrogate the success of McDowell’s attempt to reconcile, on the one hand, the structural fallibility of our perceptual capacity and, on the other hand, the capacity of rational subjects to know, from a firstperson perspective, that they possess indefeasible warrants for suitably related perceptual-based beliefs.

2. Appearances Our aim in this section will be to clarify the concept of “appearance,” which plays a key role in McDowell’s epistemological disjunctivism. According to McDowell’s definition, appearances are “mental state types.” From a third-person perspective, these mental state types can be expressed as follows: “it appears to one that p.” These mental states are considered by McDowell to be a common element between veridical and falsidical experiences. More exactly, he defines them in terms of a “common factor” between these two different kinds of experience. Because of their status as a “common factor,” they can account for the indistinguishability between veridical and non-veridical experiences and, more generally, for the fallibility of our perceptual capacity. That is a common kind that is relevant to the explanatory purposes of epistemology; it enters into making it intelligible that experiences of the “bad” kind can be indiscriminable from experiences of the “good” kind even though their epistemic significance is different. (McDowell, 2013b: 263) However, the epistemic significance is a feature which determines a strong distinction between two disjoint kinds of appearances. Appearances in veridical experiences are epistemologically different from the “mere” appearances found in falsidical experiences. This separation lies at the core of McDowell’s disjunctivism. If appearances, in sum, can be treated as a common factor, they cannot, in any case, figure as the Highest Common Factor between these two kinds of experiences. As McDowell clearly states, veridical and falsidical experiences share no common mental state type which can be relevant from an epistemological point of view. The only common mental state is the appearance,

Disjunctivism and the Internal 11 the appearing to one that p—which does not have any epistemological relevance per se. The separation between these two kinds of perceptual experience can be explained as follows: • When a perceiving subject undergoes a veridical experience, what she perceives is not just a mere appearance. In these cases, in fact, the subject is presented with a “piece of environment”, such as a physical object and its properties (shape, color and so on). This particular kind of appearance is object-dependent—it corresponds to a Fregean de re sense. In this veridical case, the situation of a perceiving subject who sees a red cube in front of her can be described as follows: S sees a red cube in front of her. • In non-veridical cases, a perceiving subject has mere appearances. The content of her experience is not dependent on an external object and there is no object which is a proper part of her perceptual content. The mere appearance is in fact object-independent and it does not correspond to a Fregean de re sense. A  subject S can claim to see a red cube in front of her, even though there is no actual red cube there. In this case, S’s experience can be expressed as follows: It merely seems to S that there is a red cube in front of her. McDowell’s externalism about appearances in veridical experience is strictly tied up with the way in which he characterizes the notion of warrant, i.e., the justification of our perceptual judgments. Let’s keep thinking from a third-person perspective. If a subject S is in a suitable perceptual relation with her environment and asserts that she sees a red cube in front of her, then the appearance of a red cube (the latter constituting the content of the corresponding appearance) is a warrant for the truth of S’s judgment. If a subject S has a mere appearance of something, she is not presented with a state of affairs and this appearance cannot be a warrant for a perceptual judgment which corresponds to her perceptual experience. That is to say: If S asserts that she sees a red cube in front of her even though there is no red cube in front of her, the absence of the object of perception determines the falsity of her judgement. Therefore, the obtaining of the two different kinds of appearance determines either the presence or the absence of the warrant for the corresponding perceptual judgment. Considering this explanation, it seems that there is room to cast doubt upon this thesis. Since McDowell holds that there are two disjoint kinds of appearances, it seems reasonable to expect a constructive analysis (from an epistemological point of view) of what happens in the case that a subject has a “mere” appearance of something. However, McDowell denies that further clarification should be given about the concept of “mere” appearance, aside from the assumption that “mere” appearances can be indistinguishable from the appearance in veridical experiences. He

12  Davide Dalla Rosa and Federico Sanguinetti claims that cognitive psychology (not epistemology!) is the science which is entrusted with the task of giving an exact description of the so-called object-independent appearances. Still, even if we may feel as though McDowell ought to have given a more detailed analysis of mere appearances, this explanatory gap does not actually cause any real damage to McDowell’s position (at least from a third-person perspective). Our analysis has shown that McDowell’s position is, up to this point, consistent and convincing. With that said, we are not so sure that the same can be said of his position when speaking from a first-person point of view. We will investigate this possibility in the second part of this chapter.

3.  The Perceptual Capacity The epistemologically relevant capacity that puts the subject in relation with the world is the perceptual capacity. The exercises of the perceptual capacity are disjointed and correspond to two distinct kinds of appearances (seeing that and seeming that). However, both exercises are exercises of the same perceptual capacity—and this is a sort of common factor between defective and non-defective exercises. As we shall see, defective and non-defective exercises of the perceptual capacity share common features.

3.1 The Disjunction Between the Exercises of the Perceptual Capacity. Relationality and Self-Consciousness The exercises of the perceptual capacity are disjunct and are characterized by a strong typological difference. There are two key features that determine this typological difference. The first feature is relationality. Relationality implies that perceptual capacity, in non-defective exercises (= seeings), puts our surroundings in view for us. In defective exercises (= seemings), in which the perceptual capacity does not do what it ought to, the perceptual capacity fails to put our surroundings into view for us. Therefore, defective exercises are non-relational.1 From the point of view of the epistemic subject, there is also a second difference. In non-defective exercises of the perceptual capacity (= seeings) the perceiving subject is in a position to know that she is in a suitable perceptual relation with the world. The specific reflective accessibility to perceptual content in cases of seeing puts the subject in the position to know that she is perceiving the world as it is—and, therefore, to rule out the possibility that things are different from how she perceives them. McDowell

Disjunctivism and the Internal 13 calls this reflective access to the content in a non-defective exercise of ­perception: Self-consciousness. In defective cases (seemings), the epistemic subject is not in a position to know that she is not perceiving the world as it is—and, therefore, to rule out the possibility that things are different from how she perceives them. In these cases, the subject is not self-conscious to the extent that she does not know what is happening to her.2

3.2 Commonalities Between the Exercises of the Perceptual Capacity. Contentfulness and Reflectiveness Although the exercises of the perceptual capacity are treated by McDowell as typologically distinct, they nevertheless have two common features. The first feature is contentfulness. Within McDowell’s conceptualist framework, contentfulness implies that sensible experience is not a blind given.3 The second feature is reflectiveness. Both in cases of seeing and in cases of seeming, the epistemic subject has a reflective access to the content of experience.4 Despite the strong disjunction between seeings and seemings, these commonalities (contentfulness and reflectiveness) between the two kinds of exercises contribute to accounting for cases of indistinguishability between cases of seeming and cases of seeing. In cases of seeming, as noted above, a subject is not in a position to know that she is in a “bad case,” since the presentational content of her experience is phenomenologically indistinguishable from the presentational content in cases of seeing. However, in both cases the subject can introspect and conceptually articulate the content which she is presented with in her sensible experience.

3.3 Why Does McDowell Argue in This Way? The Defense of the Indefeasible Warrants McDowell’s proposal, although it may look counterintuitive at first sight, is sophisticated and aims to make different theoretical demands compatible. In particular, it aims to reconcile the following two theses: • The perceptual capacity, in its non-defective exercises, is a capacity for knowledge to the extent that it provides indefeasible warrants to the subject, who is in the position to know that she possesses an indefeasible warrant and, therefore, she is in the position to know that she is not wrong about how things are in the world. In this case, perception is contentful, relational and reflectively accessible through a genuine act of self-consciousness.

14  Davide Dalla Rosa and Federico Sanguinetti • Our perceptual capacity is fallible. This means that there are some cases in which it does not do what it ought to. And in the defective cases, sometimes, we are not in the position to know that we are misperceiving the world. In these cases, sensible experience is contentful but not relational and the reflective access to the presentational content does not constitute a genuine act of self-consciousness. McDowell’s central claim is that the fallibility of our perceptual capacity—i.e., the fact that it can be exercised defectively—does not exclude that, in non-defective exercises, we are in direct contact with the world and we know that we are in such a situation. McDowell thus works to block any skeptical attempt to claim that the indistinguishability at the level of content in the defective and non-defective cases can prevent our perceptual capacity from being relational and being able to provide us with indefeasible warrants.5

3.4  Possible Problems? On the basis of his disjunctivism, McDowell seems to argue that cases of seeing have a certain kind of epistemic significance that, if we understand him well, should be part of the warrant-constituting character of experience for the subject. This warrant is externalistically conceived but must be available to self-consciousness—that is, it must have a certain kind of epistemic relevance from the internal perspective of the subject as well. This represents the internalistic component of McDowell’s proposal, a component that plays a decisive role in his theory of justification. “If an experience could by itself guarantee that things in the environment are a certain way, the subject would have to be in a position to know, in her self-consciousness in enjoying the experience, that it had that epistemic significance.” (McDowell, 2013a: 149) This means that McDowell not only claims that there is an objective asymmetry in the epistemic significance of the two ways of presenting (seeing and seeming). He also seems to claim that there is a discriminative asymmetry between the two ways of presenting—and that this asymmetry has a cognitive significance for the first-person perspective. In his Marquette Lectures “Perception as a Capacity for Knowledge,” McDowell quotes the following passage from Sebastian Rödl’s book “Self-Consciousness” to express this point: Had I been fooled, I would not have known that I was. I would not have been able to tell my situation apart from one in which I am not fooled. [. . .] When I know that p as I perceive it to be the case, then

Disjunctivism and the Internal 15 I know that I perceive that p. Thus, I am in a position to distinguish my situation from any possible situation in which I would be fooled, for, in any such situation, I would not perceive that p, while in the given situation I do. (Rödl, 2007: 158 apud McDowell, 2011: 44) This passage seems to imply that, although McDowell agrees that our perceptual capacity is fallible and that, when we misperceive the world, we do not distinguish our experience from a veridical experience, he claims that, on some occasions—“locally,” so to speak—we are in a position in which we can recognize that we are conclusively and indefeasibly warranted in a determinate perception-based belief. The indistinguishability between “bad case” and “good case” is not bi-univocal, but it applies only to the “bad cases.” Therefore, [i]f a state of affairs consisting in there being something red and rectangular in front of one is present to one in an experience, one is thereby in a position to know that that is how it is with one and thus to recognize that one is conclusively warranted in believing there is something red and rectangular in front of one. (McDowell, 2013a: 150–151) If this is right, McDowell’s position implies an asymmetrical reflective accessibility to the discrimination between “good case” and “bad case” in a determinate token of sensible experience.6 It must be noted that this position does not fall into a definition of perceptual capacity as an infallible capacity tout-court.7 On the contrary, it leaves room for fallibility to the extent that, in the “bad case,” it is the very same perceptual capacity which is exercised in a defective way. Fallibility, however, is not considered as a threat for the fact that, provided that I am in a “good case,” I am able to recognize the peculiar epistemic significance of my experience and therefore, I am able to exclude the possibility that I am in a “bad case.”8 From the mere possibility of misleading experiences does not follow that the capacity cannot be correctly described as a capacity to get into positions in which one knows that it is through one’s perceptual state that one knows something about the environment—and so knows that one’s perceptual state conclusively warrants one in the belief in question. (McDowell, 2011: 42) Nevertheless, in our opinion, this strong asymmetry which characterizes my capacity of recognizing from a first-person perspective whether or not I  am in a good case is not convincing. We do not see how those experiences which we take to be accurate and factive can have the peculiar epistemic significance McDowell attributes them: i.e., an epistemic

16  Davide Dalla Rosa and Federico Sanguinetti significance which can be recognized de facto from the first-person perspective and which allows us to infallibly exclude the possibility that they are not misleading experiences. Provided that I am in a bad case, the way in which things appear to me does not put me in a position to know that I am in a bad case. However, the way of presenting which characterizes genuine experiences, given the phenomenological indistinguishability at the level of content, also does not seem to present the relevant and introspectively accessible features which should allow me to recognize that I am in a good case.9 In other words, the typological difference (assumed from an objective, third-person perspective) between the two ways according to which the subject is presented with appearances (= seemings and seeings, or objectindependent appearances and object-dependent appearances) does not have a phenomenological correlate which is reflectively available to the first-person perspective and which should put de facto the subject who is in a good case in the position to know that she is in a good case. Given that McDowell seems to deny the presence of specific reflectively accessible phenomenological elements10 that justify the asymmetric discrimination in the good case from a first-person perspective, the claim of a transparently different epistemic significance of veridical experience for the first-person perspective seems to be dogmatic.11 However, not only does it seem that the asymmetric discrimination from the first-person perspective is not grounded on theoretical elements that are able to justify it, but it is also threatened by the very conception of fallibility McDowell endorses.12 McDowell repeatedly states that he does not intend to conceive of fallibility as a property of exercises, but of capacities:13 “fallibility is a property of capacities, or perhaps of cognitive subjects as possessors of capacities” (Dennis, 2014: 4011). McDowell’s conception of fallibility as a property of the perceptual capacity seems to imply that fallibility applies across the board, r­ egardless of the particular cases in which the perceptual capacity is exercised. Since McDowell intends to block the skeptical consequences of the a­ rgument from illusion, the conception of fallibility as a property of the ­perceptual capacity could (if one accepts our claim that de facto accessibility to asymmetric discrimination is problematic) lead McDowell’s theory to a weaker result, namely: A subject which is in a determinate sensible experience is at least in principle in a position to know that she is in a suitable perceptual relation with his environment.14 Nevertheless, this result seems to be available to McDowell only at the cost of renouncing the internalist elements of his proposal. For a subject to be aware that, on some occasions, she can in principle (therefore from a third-person perspective) be in the position to know that things cannot be different from how she perceives them never tells her when she de facto is in such a good case. This is because—as stated above—according to McDowell there seems to be no phenomenological elements that are

Disjunctivism and the Internal 17 both accessible and relevant for the discrimination of the good case. On this reading, the strong discriminative asymmetry (de facto and accessible from a first-person perspective) is not present: For this reason, a subject can always recognize the possibility that she might be in a bad case even when she is in a good case.15 In fact, this reading merely affirms that, in the good cases, when a subject affirms that she knows that things are thus and so, things effectively are thus and so—and provide her with an indefeasible warrant for a suitably related belief. However, the subject would never know, at the first person, when she is in such a position in which she knows how things are in the world.16 For this reason, it seems to be possible to conclude that the subject never knows, from a first-person perspective and in the single case, to be indefeasibly and conclusively warranted in believing something about the world on perceptual grounds. Even if she knows that she can be in principle conclusively and indefeasibly warranted in claiming to know something on perceptual grounds (this is the “external” standpoint of the third-person perspective and it does not make reference to the single case), she never knows from a first-person perspective when she is de facto in that position in a determinate single case. The fallibility of the perceptual capacity can always bring her to doubt whether she is in “bad case,” even though she is in fact in a “good case.”17 In order to support his position, McDowell claims that (if we are not in an experimental scenario, in which environmental conditions are artificially manipulated) we have a subject who in ninety-nine cases in a hundred is in, and knows she is in, a perceptual state that provides conclusive warrant for a belief about the colour of something. In one case in a hundred, the fallibility of her capacity kicks in and she is wrong in taking herself to be in that position. But that does not require us to give up saying that she is in that position in the other cases. (McDowell, 2011: 52) McDowell seems, therefore, to suggest that (outside of artificial experimental scenarios) the low rate of failure of our perceptual capacity does not constitute a good reason to doubt that, in the overwhelming majority of cases, a subject is conclusively warranted in claiming to be in the position to know (also from a first-person perspective) how things are in the world. Moreover, he endorses an even stronger claim: Even if fallibility could be quantified and the result of the quantification would be more that 1%, this very fact would not threaten his position. When we imagine that fallibility might be quantified, we leave the sheer fact of fallibility, as before, posing no threat to the following idea: When all goes well in the exercise of a fallible perceptual

18  Davide Dalla Rosa and Federico Sanguinetti capacity, its possessor is in a position that conclusively warrants her in believing something. (McDowell, 2011: 52) In our opinion, this conception of fallibility is incompatible with the goals McDowell aims to achieve.18 The problem of fallibility is not quantitative, but qualitative. McDowell’s concession to pragmatic considerations (something similar to saying: “99% of the time you perceive the world as it actually is and you know it!”) seems to underestimate the problem of fallibility. The “qualitative” dimension of the problem is that fallibility, as a property of perceptual capacity, seems to be valid across the board with respect to the particular exercises of said capacity. Yet, fallibility, which is assumed de jure in order to explain the possibility of misleading experiences, is completely excluded from the single acts of veridical perception—the latter are described as conclusively warranting the subject in her corresponding beliefs and, hence, putting her in a position to know, in her self-consciousness, that she perceives the world as it is and to exclude that she might be wrong. However, due to the lack of any elements that could allow me to recognize (in a single case, from the first-person perspective) that I am not misperceiving the world, the sheer fact of fallibility, even if it is quantified in a very small percentage, does not put me in any position to decide whether, in a determinate occasion, I am in a good case or in a bad case. The mere theoretical possibility of the “bad cases” (namely, the fallibility of our perceptual capacity) undermines our capacity to recognize, from a first-person perspective, that we are in a suitable relation with our environment and that there is no possibility that things could be otherwise than as we perceive them.

4. Conclusion If our analysis is correct, it should follow that McDowell’s disjunctivism succeeds in making room for a position according to which, from a thirdperson perspective, it is possible to ascribe to a rational subject a perceptual state that put her in direct contact with the world. Furthermore, this contact represents an indefeasible warrant for her corresponding belief, which as a result is conclusively justified.19 However, as noted earlier, McDowell’s aim is to confer upon the indefeasible warrant an epistemic significance that is reflectively available from a first-person perspective. This peculiar epistemic significance should justify an asymmetric and mono-directional conception of the self-conscious discriminability (from the first-person perspective) between a “good case” and “bad case.” This represents the internalistic dimension of the objective disjunction between veridical and misleading

Disjunctivism and the Internal 19 experiences. McDowell needs this internalistic dimension in order to satisfy his ambition of giving an account of cases in which the subject conclusively knows, from a first-person perspective, that things in the world cannot be otherwise than as she perceives them—and that conclusive knowledge is a self-conscious one, so that the subject conclusively knows that she is not in a bad case.20 However, the conceptual resources displayed by McDowell do not seem to be enough to defend such strong theoretical claims. McDowell’s ambition is to provide a convincing account of fallibility as a property attached to capacities—therefore applying in principle to all exercises of a capacity. At the same time, he seems to claim that we must consider some exercises as completely immune to fallibility in order to block the skeptical consequences of the argument from illusion—he therefore seems to claim that there are cases in which we cannot doubt, because we are perfectly self-conscious that things in the world cannot be otherwise than as we perceive them. Within McDowell’s theoretical framework, this ambition does not seem to be adequately justified. On the contrary, by understanding fallibility as a property attached to capacities and by excluding phenomenologically relevant elements which could support the claim of a discrimination in the single (“good”) case, McDowell seems to provide the skeptic with the conceptual resources which allow him to find a breach in the “impregnable fortress” represented by cases of veridical and self-conscious perception, in which we are supposed to know, from the first-person perspective, to be indefeasibly warranted with respect to a determinate perceptual belief.21 In order to appreciate this point, it is interesting to notice that Leddington’s (2018) and Pritchard’s (2009, 2012)22 attempts to improve McDowell’s views, although they go in different directions, originate precisely from this problematic situation. Leddington’s solution aims to defend the infallibility of the subject in cases of veridical perception by virtue of the disjunction of the perceptual capacity from a sort of pseudo-capacity that would be responsible for misleading experiences. In an analogous way, Pritchard seems to introduce the notion of favoring ground with the aim of accounting for fallibility in the single case, without having to claim that we have a self-conscious accessibility to the asymmetric discrimination of the cases in which we are indefeasibly warranted. What emerges from our analysis is that the common root of Leddington’s dissatisfaction with McDowell’s view and Pritchard’s dissatisfaction with the formulation of McDowell’s view can be recognized in the incompatibility of a de jure conception of fallibility (as a property attached to capacities) with the self-conscious capacity to exclude de facto (in the single case) that she is misperceiving the world. Leddington adopts his solution with the aim to avoid the propagation of the fallibility de jure to the totality of its exercises de facto. Pritchard, on the contrary, introduces the notion of merely favoring (= defeasible) grounds de facto in order to

20  Davide Dalla Rosa and Federico Sanguinetti avoid falling back into a theory in which the subject (in the good cases) would be able to infallibly exclude (from first-person perspective) that she is misperceiving the world. Nevertheless, Leddington’s and Pritchard’s positions do not seem to be able to accommodate the theoretical ambitions found in McDowell’s proposal. On the one hand, Pritchard does not seem to offer an account that envisages cases in which a subject infallibly knows de facto, in her self-consciousness (= from the first-person perspective), that she has indefeasible warrants for a determinate perceptual-based belief. On the other hand, Leddington’s suggestion to split perceptual capacity (as it is conceived by McDowell) into two different capacities does not introduce new elements which are able to account for a self-conscious reflective access, in the single case, to the discrimination of the “good case” from the “bad case.” The aim of our chapter has been to highlight a problematic issue in McDowell’s proposal. Nevertheless, we do not aim at prescribing a therapy able to solve the problem which we take to plague McDowell’s position. However, we would be satisfied if we have at least succeeded in offering a convincing analysis of McDowell’s proposal and of the difficulties which it runs into. Moreover, we hope to have shed light on a specific theoretical tension between the concepts of warrant and fallibility, tension which gives rise to Pritchard’s and Leddington’s attempts to improve on McDowell’s views. A positive account is not only beyond the limits of this chapter, but also something of which we have no pretensions to accomplish.23

Notes 1. It might be objected that illusions, unlike hallucinations, are experiences in which the subject is in a relation with an object, but the object in question is not correctly perceived in its qualities (e.g., a red cube is experienced as being blue). The fact that McDowell understands the relationality of perception in terms of relation to states of affairs, rather than relation to objects, seems to justify the characterization of illusions as non-relational experiences. As highlighted by Byrne and Logue (2008), in particular p. 60, the status of illusions is a controversial issue among disjunctivists. 2. McDowell’s position seems to be in line with Williamson (2000), chapter 8, in regard to the asymmetry of the self-conscious access to our evidence. We are not sure whether Williamson’s “relaxed” conception of self-consciousness and the externalistic features of Williamsons’s proposal are consistent with the strong claims implied by McDowell’s conception of self-consciousness and with the internalistic features of his position. 3. On this point, McDowell (2013a), 144, explicitly distances himself from Travis—see also p. 144, fn. 2, for critical references to other authors. 4. It is important to stress that self-consciousness and reflectiveness are not identical properties. In this context, self-consciousness is the subject’s capacity to know that she is in a suitable perceptual relation to her environment— self-conscious obtains only in the “good case.” Reflectiveness is the subject’s

Disjunctivism and the Internal 21 introspective capacity to conceptually articulate the presentational content of experience—and obtains both in the “good case” and in the “bad case.” 5. See McDowell (2011, 2013a). 6. Pritchard (2009), 467–479, and Pritchard (2012) suggest a weaker form of disjunctivism, to the extent that he appeals to a conception of warrants as merely “favouring” (therefore: defeasible), without their being conclusive. This is meant to avoid the skeptical consequences of the arguments McDowell intends to block without appealing to indefeasible warrants. In our view, Pritchard’s proposal, which avoids the necessity of a strong discrimination of the indefeasible warrants, is too weak to satisfy McDowell’s ambitions—on this point, see Dennis (2014). 7. McDowell rejects both the thesis according to which the perceptual capacity would be an infallible capacity tout court (i.e., a capacity which is not liable to defective exercises) and the thesis according to which our perceptual capacity would always allow us to infallibly discriminate between “good cases” and “bad cases”—on this point, see McDowell (2011: 40–41) 8. It must be stressed that McDowell does not think that our capacity to know that we are in a “good case” is a separate capacity, distinct from our perceptual capacity. Rather, it is part of our perceptual capacity—more exactly, it is part of our perceptual capacity when correctly exercised. See McDowell (2013a: 153). This seems to be intended, on the one hand, to block a possible regress in the conception of self-consciousness, and, on the other hand, to avoid the conception of self-consciousness on Cartesian lines qua a sort of internal infallible eye. 9. M.G.F. Martin seems to argue in favor of a phenomenological difference between a “good case” and a “bad case.” However, we are inclined to think that McDowell is not committed to this thesis. 10. See McDowell (2008: 381) For an excellent discussion of this point and of the possible reply by McDowell, see Soteriou (2016: 126ff). 11. Given the fact that McDowell seems to deny his commitment to any phenomenological aspect relevant to the discrimination, the commonalities between veridical and misleading experiences (= contentfulness and reflectiveness) contribute to muddling the waters—and this because of their function of contributing to a convincing account of the indiscriminability of the “bad case” from the “good case.” 12. This problem seems to be correctly highlighted by Dennis (2014). In general, Dennis’s paper seems to identify the problem at the center of our critical reconstruction and, in particular, it seems to correctly highlight the difference between McDowell’s and Pritchard’s positions. Nevertheless, although Dennis’s analysis of the problem of infallibility is interesting, he seems not to articulate this problem in a fully convincing way (in particular, in his paper does not emerge a correct distinction between the perspectives of the thirdperson and of the first-person—this is, for us, a decisive point). 13. This results from the controversy between Burge and McDowell about the characterization of McDowell’s disjunctivism—see Burge (2003, 2005, 2011), McDowell (2011, 2013b). On this point, see Dennis (2014: 4009). 14. See McDowell (1982: 390), fn. 37: “[F]or some purposes the notion of being in a position to know something is more interesting than the notion of actually knowing it.” What we consider as problematic is the step from merely “being in a position to know” and “knowing, from the first-person perspective to have indefeasible grounds for a determinate belief.”

22  Davide Dalla Rosa and Federico Sanguinetti 15. Following Austin, McDowell seems to hold that we must have a determinate ground to doubt a particular perceptual situation—he therefore denies that we must take into account a sort of generalized and abstract doubt. Indefeasible warrants do not depend on the exclusion of all the error-possibilities. See McDowell (2011: 49) and McDowell (2013a: 153–154). 16. Soteriou (2016: 134–135) reconstructs a similar line of objection—together with a possible reply in McDowellian spirit. McDowell’s reply could be, again, to resist the argument which, from the fallibility of perception and of the capacity to tell how things internally are with one, denies the possibility to get into positions in which one can indefeasibly know how things are both externally (= in the world) and internally (= in one’s conscious life). This is exactly the point we find problematic. In our view, Soteriou’s attempt to present a plausible reply on McDowellian lines makes the mistake of overlapping “indefeasibly knowing (on a first-person perspective)” and “being in a position to know (on a third person perspective).” 17. Again, this possibility seems to be considered by McDowell, even though McDowell seems to restrict this possibility to motivated doubt—see Soteriou (2016: 140ff). However, this is not intended by McDowell to be a threat to his position. 18. This point is examined by Leddington (2018). In order to save the infallibility of the perceptual capacity in its exercises, Leddington suggests extending disjunctivism to capacities. Defective cases are not operations of the same capacity that perceptually puts us in relation with the world. Rather, they are operations of a sort of “pseudo-capacity” which is disjunct from the former. We will not discuss whether or not Leddington’s corrective proposal runs the risk of simply pushing the problem of fallibility one level further—from the exercises of a capacity to capacities. What is important to stress is that it does not seem to offer a solution to the problem at the center of our discussion. According to our reconstruction, in McDowell’s framework the epistemic subject does not know whether a determinate exercise of her perceptual capacity is defective or not. In a similar vein, in Leddington’s framework, a subject would not know whether a determinate exercise is an exercise of her perceptual capacity or of her “pseudo-capacity.” Pritchard (see above, fn. 6) seems to try to correct McDowell’s proposal by trying to solve the difficulties concerning the level of the single exercise, in a single case, of the perceptual capacity. Leddington seems to aim to correct McDowell by trying to solve the difficulties concerning the level of the capacity in general. 19. See Soteriou (2016: 126) and Soteriou (2016: 131–132). 20. It seems problematic that McDowell tends to conceive of the internalistic component of justification exactly on the model of an externalistic account— McDowell’s picture of the internal is exactly modeled on his picture of the external. This is functional to his proposal, because—as we have argued— from a third-person perspective his reconstruction is plausible. However, the internalistic standpoint from the first-person perspective seems to have features that make it peculiar and that do not seem to allow a mere transposition of the third-person perspective onto the first-person perspective. To see this, it can be useful to think of the analogy proposed by McDowell between the perceptual capacity and the capacity of an expert golfer to sink eightfoot putts. This analogy shows how the internalist element that McDowell let enter through the front door (by claiming that we have a self-conscious access to the discrimination of the “good case” from the first-person perspective) leaves in disguise through the back door. As a matter of fact, the golfer does not only perfectly know when he can sink an eight-foot putt. He also perfectly knows when he does not sink the putt—if we hold to McDowell’s analogy, this implies that a subject who is in a “bad case” is in a position

Disjunctivism and the Internal 23 to know to be in a “bad case.” Using Martin’s words, this would mean “to attribute to responsible subjects’ potential infallibility about the course of their experiences”—Martin (2004), 51. This is incompatible with McDowell’s position—see above, fn. 7. 21. Fallibility is always excluded from the veridical perceptual case (non-defective exercises of the perceptual capacity are beyond the possibility of failure. But how is it possible to demonstrate that there is no room, from a first-person perspective, to doubt that a determined perceptual experience happening to be true could have not been as such? 22. See above, fn. 18 and 6. 23. A first attempt to develop the argument we defend in this chapter can be found in our “Appearances, indefeasible warrants e fallibility. Sul disgiuntivismo epistemologico di John McDowell,” Rivista di Filosofia 108 (1), 61–90. We thank Lucia Oliveri for comments on previous drafts.

References Austin, J. L. (1979) “Other Minds” in Philosophical Papers (ed. J. O. Urmson and G. J. Warnock), 3rd ed., Oxford: Oxford University Press, 76–116. Burge, T. (2003) “Perceptual Entitlement” in Philosophy and Phenomenological Research, 67 (3), 503–548. Burge, T. (2005) “Disjunctivism and Perceptual Psychology” in Philosophical Topics, 33 (1), 1–78. Burge, T. (2011) “Disjunctivism Again” in Philosophical Explorations: An International Journal for the Philosophy of Mind and Action, 14 (1), 43–80. Byrne, A., Logue, H. (2008) “Either/Or” in Disjunctivism: Perception, Action, Knowledge (ed. A. Haddock and F. Macpherson), Oxford: Oxford University Press, 57–94. Dennis, P. (2014) “Criteria for Indefeasible Knowledge: John McDowell and ‘epistemological disjunctivism’ ” in Synthese, 191, 4099–4113. Leddington, J. (2018) “Fallibility for Infallibilists” in In the Light of Experience: New Essays on Perception and Reasons (ed. J. Gersel, R. Thybo Jensen, M. S. Thaning, and S. Overgaard), Oxford: Oxford University Press, 162–184. Martin, M. G. F. (2004) “The Limits of Self-Awareness” in Philosophical Studies, 120, 37–89. McDowell, J. (1982) “Criteria, Defeasibility, and Knowledge” in Proceedings of the British Academy, 68, 455–479. McDowell, J. (2008) “The Disjunctive Conception of Experience as Material for a Transcendental Argument” in Disjunctivism: Perception, Action, Knowledge (ed. A. Haddock and F. Macpherson), Oxford: Oxford University Press, 376–389. McDowell, J. (2010) “Tyler Burge on Disjunctivism” in Philosophical Explorations, 13 (3), 243–255. McDowell, J. (2011) Perception as a Capacity for Knowledge, Milwaukee: Marquette University Press. McDowell, J. (2013a) “Perceptual Experience: Both Relational and Contentful” in European Journal of Philosophy, 21 (1), 144–157. McDowell, J. (2013b) “Tyler Burge on Disjunctivism (II)” in Philosophical Explorations, 16 (3), 259–279. Pritchard, D. (2009) “Wright contra McDowell on Perceptual Knowledge and Scepticism” in Synthese, 171, 467–479.

24  Davide Dalla Rosa and Federico Sanguinetti Pritchard, D. (2012) Epistemological Disjunctivism, Oxford: Oxford University Press. Rödl, S. (2007) Self-Consciousness, Cambridge, MA: Harvard University Press. Soteriou, M. (2016) Disjunctivism, London and New York: Routledge. Williamson, T. (2000) Knowledge and its Limits, Oxford: Oxford Blackwell.

2 Xenophanes’s Figs and Honey An Essay about a Program of Philosophy of Perception Manuel Curado

1.  A Little Everyday Experience The Diels-Kranz B38 fragment of Xenophanes of Colophon famously related the sweetness caused by figs to the sweetness caused by honey. Men would find the figs much sweeter if they had not tasted the honey. The phrase that Herodian has fixed and that has crossed the centuries is the following one: εἰ μὴ χλωρὸν ἔφυσε θεὸς μέλι, πολλόν ἔφασκον γύλσσονα σῦκα πέλεσθαι. [ei mê chlôròn éphyse theòs méli, pollón éfaskon gúlssova sýka pélesthai]. Several translations have been proposed. Kathleen Freeman proposes “If God had not created yellow honey, they would say that figs were far sweeter” (1947: 24); W.K.C. Guthrie, in the same vein, proposes “If God had not made yellow honey, men would think figs were much sweeter” (1962: 401); Kirk and Raven propose “If god had not made yellow honey, men would consider figs far sweeter” (1971: 180); James H. Lesher translates “If god had not made yellow honey, they would think that figs were much sweeter” (1992: 41); Sir Karl Popper, in one of the last texts to which he dedicated himself in his life and which left incomplete, proposes the translation “If God never had chosen to make the light-yellow honey,/Many a man would think of figs as being much sweeter” (1998: 45); Daniel W. Graham makes the following proposal: “If God had not created yellow honey, they would say figs are much sweeter” (2010: 127); more recently, Shaul Tor proposes “Had it not been the case that god makes . . . yellow honey, they would say that figs are much sweeter” (2017: 122). As can be seen clearly, discounting the normal latitude that separates all the translations, there is agreement in the essential. This text has an unsophisticated simplicity and seems to be confined to a mere testimony of an experience that anyone could have uttered. On personal record, one could say that this person started by eating figs

26  Manuel Curado and then ate honey. When he ate the figs, it seemed to him that they were very sweet. When he had eaten the honey, it seemed to him that it was very sweet and remembering the figs, it seemed to him that they were not as sweet as honey. Xenophanes’s elegant phrase does not address sophisticated intellectual problems or mysterious matters. It all boils down to a modest point of view about everyday life. All people could live full lives in the most complete ignorance of fragment 38 of Xenophanes. It does not appear to be a sapiential aphorism that can guide one’s life, does not seem to be a mantra that can be repeated to calm down emotions, does not seem to be an equation of a complex theoretical problem as well. The thought that is expressed does not seem to be really important in people’s lives. Simply put, the author confines himself to saying something about a person’s experience. What is affirmed may already have been lived and described countless times, and it only seems remarkable because the broken materiality of a fragment of an old thinker with the prestige of an iconoclast like Xenophanes gives it a perhaps disproportionate prominence in relation to its content. In a hurried assessment, the least that can be said is that the testimonial simplicity of this text does not compel anyone’s intelligence. Therefore, we can all continue to live without thinking about the fragment of Xenophanes. Nothing is gained by the useless exercise of thinking about so little. For various reasons, this old reflection of Xenophanes (c. 530 BCE) never came to its final consequences. The status of the pre-Socratic philosophers in the history of philosophy has caused later authors to think habitually about them but not with them. It will be difficult, for example, to take seriously the doctrine of Thales of Miletus which states that water is the constitutive principle of reality. Instead of water, any other philosophical problem, including the figs and honey of Xenophanes and perhaps even the spherical and brilliant being of Parmenides, may be placed. The doctrines of the pre-Socratics are very much worked from the point of view of the history of thought as if they were the preparatory years for the more advanced studies of modern and contemporary thought. They do not seem to have the sophistication to participate equitably in a debate on contemporary philosophical issues. Pre-Socratic thought has to be “aided” by strenuous inquiries with the aim of being properly contextualized, otherwise, either one does not perceive what is at stake in each philosophical thesis, or the little that is actually perceived of each one of them ends up looking inevitably anachronistic. All this contributes to the fact that pre-Socratic thought has ceased to be at the forefront of intellectual debate to become a property of a specific historiographical approach: The history of philosophy. In order to understand by analogy the impoverishment that this entails, it is sufficient to imagine the possible world in which the sacred texts of the great world religions (Christianity, Judaism, Islam, Buddhism, etc.) have become the exclusive subject of academic researchers. The belief and meaning of the existence

Xenophanes’s Figs and Honey 27 of millions of people would have been transformed into the poor version of a literary and documentary content that is scientifically investigated. The same could be said about love, since the ideology of scientism that relentlessly dominates contemporaneity gives itself permission to study with its simple conceptual tools all the complexity of love life. If the sciences dealing with love were to be compared with the literary and religious representations of love, one would see that they are of an embarrassing poverty; obviously the very literary and religious representations of love are, in turn, overwhelming poor compared to the actual experience of love itself. Now with the pre-Socratic thought something similar is happening: Philosophically important subjects are hostage to the academic disciplines of hermeneutics, philosophical historiography or even the history of science. It is not possible to prove that the intelligibility gained by the academic approach of subjects is richer than the original experience of the subjects. In the case of religion, this is evident. The understanding of the religious phenomenon is unbelievably poorer than the rich sense that millions of people derive from religious practice. There is no scientific theory that can rival the role that religions play in people’s lives. The same can be said about the investigations of love. At a time when simulations of reality have reached unprecedented levels of sophistication, this is increasingly difficult to understand. People tend to confuse representation with reality. As far as philosophical debate is concerned, there is only one way of liberating the pre-Socratic philosophy of its current academic masters: To put it into dialogue with contemporary scholars. (It is necessary to take into account, of course, what centuries of interpretation have been able to ascertain about the ancient texts.) In fact, what is ostensibly affirmed in the fragment of Xenophanes could also be affirmed by any person of the twenty-first century. This is the very subject that needs clarification. What truth is expressed about the general order of the world in a statement which, having been originally uttered by Xenophanes, can be affirmed in any other historical epoch? Since fragment 38 refers to taste experiments, all thinking about it is inevitably included in a philosophy of perception. A  small human experience is held hostage by a structured body of intellectual traditions that claims the privilege of being able to pronounce with authority over it. This privilege certainly comes with a price, which cannot be other than the total clarification of the original little experience. In the absence of this clarification, there is no reason why the privilege should be maintained. From 530 BCE to the twenty-first century, perhaps enough years have passed to demand this definitive clarification. One cannot put the small experience of which Xenophanes speaks about in the same category of “philosophical problems” without a known solution, a category that includes momentous problems like the status of being according to Parmenides or the conjecture about the existence of an infinite number of worlds in Anaxagoras’s cosmology, because this

28  Manuel Curado little experience is universal, in the precise sense that no known human being can live without experiencing what is described in fragment 38 (or something very similar). To assert in the twenty-first century that we do not know everything that is at stake in that small experience is indeed demoralizing. Looking deliberately and systematically at all the libraries in the world, where is the ultimate science that explains why a human being feels all the flavor of figs when eating figs, feels the taste of honey when eating honey and when eating honey after eating figs, the sweetness of these figs seem less to him or her than the sweetness of the honey? This embarrassing question has no known answer. It suffices, to understand this point, to recall the hard problem of consciousness: It is not known why sentient beings feel anything at all, since one might think that they could live without feeling something (for a reference formulation of this problem, see Chalmers, 1995 and 1996). The hard problem of consciousness manifests itself only in the initial part of the formulated question, since an explanation should be added for the second part of the question, which deals with the exchange of intensities of mental contents. No one can seriously claim that a single book is known that has a final theory that allows a true answer to this question; if such a book existed, the continued existence of sciences such as psychophysiology, philosophy of mind, philosophy of perception and phenomenology would not be understood. The mere existence over centuries of structured sets of knowledge (concepts, theories, researchers, institutions that promote this knowledge, publications, scientific events, etc.) that deal with this type of subject is in itself sufficient proof that throughout of all this time one has thought wrongly about these questions. Taking an analogy with the history of medicine, no one in the twenty-first century wants to be taken care of with medicine from medieval times, from the Renaissance, from the eighteenth century or even from much of the nineteenth century. People realize that many centuries of medical history have been a huge misunderstanding and that there is no significant continuity (if at all) between Sir William Osler’s and Galen’s medicine, for example. Complicating our understanding of Xenophanes’s small experience is a fossilized way of thinking about these issues. As if they had a stony but anachronistic existence, some matters are thought of in inevitably dated ways. It is not only the content (the concepts, the theories, etc.) but the very form of thought that are concerned; and, of course, what is considered possible and impossible to be thought of. Here is a modest example of this fossilized way of thinking about matters: An instance of intelligence tries to understand through language what is involved in an experience of another instance of intelligence. Simplifying the equation of this problem, the rational module of human intelligence makes the exercise of trying to understand what happens when the person it belongs to experiences eating figs and eating honey. For this exercise, the module

Xenophanes’s Figs and Honey 29 uses the instrument of language, trying to find the most accurate labels to refer to parts of the continuum of experience. To some parts it calls “sweetness of the figs,” to other parts it calls “sweetness of the honey”; to other parts it calls only “figs” and only “honey”; other parts are still called “more” and “less.” If it were possible to contemplate these matters without these conceptual and perspectival constraints, it would perhaps be possible to see that all this is very strange. The rational module pays the expenses of the exercise: It is it that seems to start the activity, it is it that performs the comparisons, it is it that searches for the most appropriate conceptual tools and it is it that fights for the ultimate prize of understanding. The perceptual experience is passive and seems to have the unworthy status of being the matter that awaits patiently the enlightening intervention of rationality, an almost salvific intervention because this charitable approach seems to be the only one that can clarify the residual opacity of experience. To understand the strangeness of this intellectualized way of looking at the apparently modest experience of the sweetness of figs and honey, a way of proceeding which is neither innocent nor spontaneous, since it stems from centuries of intellectual history, suffice it to say that it is difficult to look at this relationship in the opposite direction. Perception does not seem to be able to evaluate on its own terms the module of rationality. The statement “perceiving a thought” has no meaning, nor, indeed, the more limited versions, such as “seeing a thought” or “savoring a thought.” As far as transparency is concerned, the case is not clear. Rational activity is unlikely to be able to reach the essence of itself by explaining, for example, the vast problem of what it means to be rational. All human faculties are concerned only with their contents, not with themselves. The vision sees apples and apple trees but does not see itself. No human has ever seen sight. This is the rule of all human faculties. No other is known. Moreover, reason cannot give reasons explaining why it exists when it might not exist, or why it is as it is, or even explain why the sense of transparency and well-being that results from understanding happens. Sensory experience, in turn, is prototypically diaphanous; however, the transparency or diaphaneity of the experience deserves an explanation which, from Xenophanes to Thomas Reid and J.L. Austin, has been slow to appear. As far as proximity to personal identity is concerned, the case is also difficult to resolve. People feel that they are the ones who are thinking and feel that they are the ones who are feeling. However, at the risk of splitting hair, as the end result of rational activity does not seem to be as immediate as that of perception, there is a slight asymmetry in favor of perception. These constraints diminish the value of any philosophical or rational exercise on perception. This truth is difficult to grasp because, in some way, the history of thought has already made it accepted as a dogma

30  Manuel Curado that the rational module can take care of any other subject, including any other module of the mind. One cannot see the common background that surrounds all the modules. The question of the initiative of rationality, as if it were its spontaneity, is interesting in this respect. Many questions could be asked in this regard. Here are a few. In the vast set of motives that determine the focus of intellectual attention, what is the factor that initiates rational reflection? Does reason give itself the motive that determines it or, differently, need something outside itself to initiate its action (an emotion, a perceptual content, a memory, etc.)? These issues emphasize the fact that there seems to be a constant internal struggle within the human mind. It is not just what modern philosophy has called conflict of faculties. The human mind seems to organize itself in its totality according to an agonistic model. Using expressions that are already commonplace, “reason” opposes “heart” and “sensibility.” Within each of these high-level categories there are also conflicts that might be termed low-level. Xenophanes gives as a matter of fact an example of low-level conflict: The sweetness of the figs is not exactly the same as the sweetness of honey. The fragment that reached the twentyfirst century emphasizes the different intensity of sweetness but could emphasize also the unique trait of the taste of one with respect to the taste of the other. The sweetness of the fig has a unique content that the sweetness of the honey does not have and vice versa. Xenophanes suggests a quantitative approach to the relations between perceptual contents, anticipating what the nineteenth century did about sensations (e.g., Weber-Fechner’s law). Even the sweetness of the figs is not always perceived in the same way. The sweetness of the first fig that is eaten is manifestly different from the sweetness of the hundredth fig that is eaten. Other contextual factors condition the perception of a modest fig and so does the perception of honey. The fragmentation of the human mind into faculties and contents illustrates the agonistic model that in the West was built, a model that has in Xenophanes one of the earliest intellectual manifestations. Different solutions have been proposed over the centuries to ensure the possible and desirable unity of this disconnected set, such as Idealism, Panpsychism and Personal Identity, just to give a few examples. These constraints have to be respected. Others will certainly exist and it is perhaps impossible to get a complete inventory of all. In any case, it is necessary to explore the way in which Xenophanes observation allows to clarify the totality of the perceptual content. It could be said that this is the founding intuition of the philosophy of perception, the little curiosity that started a research program that, despite the intellectual wealth it has already provided, is still far from complete. It is this program that should be made ostensive, seeking to glimpse not only what can positively be developed, but also omissions that may give rise to new understandings about mental life.

Xenophanes’s Figs and Honey 31

2.  A Philosophy of Perception Program What is precisely stated in the text of Xenophanes? What is ostensibly affirmed and what is tacitly assumed constitute a small world that describes what is at stake in perception. It is as if Xenophanes proposed a kind of cartography of perception, an intellectual project with the aim of indicating what is possible to discern in the subtle realm of subjective experience of something and what, in the empty spaces that reside between the contents of that experience, it is not possible to name or to identify, only knowing of its existence by an absence marked by a space. Before one sees explicitly what is present in this fragment of the “founder of Greek Enlightenment,” as Sir Karl Popper (1998: 33) characterized his author, it will not be possible to guarantee in advance that the project is limited only to a philosophy of perception. It may be that the detailed map of a trivial perceptual act amplifies the rapid and ephemeral instant in such a way that one glimpses the fine grain of reality beyond perception. No one can tell for sure. In this anticipatory sense that it is not yet possible to prove, the phenomenological exercise would give rise to an ontology. This would be the ultimate meaning of the perceptual act, since knowing something is only a poor and regional interpretation of the existence of something. Xenophanes seems to have contributed to the tradition of ancient thought that values everything that exists, seeing in the smallest object or in the most fleeting event signs of a wider order of things. Each thing is a summary of all things. This conviction is central to the thinking of later authors. Remember the seeds of the whole (spermata) that are in each thing, in Anaxagoras of Clazomenae (e.g., DK 4, 6, 11, 12), or, much later, in Lucretius, who believed that things clarified one another (De rerum natura, V, 1455–1457). Shaul Tor observes with deep insight that Xenophanes’s fragment reveals a “gnomic nuance” (2017: 122). It is precisely the vast wealth of a gnomic wisdom that must be made explicit. The following aspects of this hidden wisdom in the almost idyllic statement are especially relevant: 1. Good perceptual forms. Fragment 38 mentions and relates honey, figs, color of honey, sweetness of figs and honey, sight of objects called figs and honey, activity of savoring objects such as figs and honey, and creation of the things of the world by a divine entity. The inventory of the conspicuous contents of the fragment has these pregnant or well-formed forms. 2. Good forms that are merely alluded to or accepted tacitly. There are obviously enthymemes of the explicit statement of the fragment (i.e., information not made ostensive in the statement itself and merely alluded to) that it would be necessary to make explicit, namely, the manipulation of fig and honey objects, the universe of persons who have already tasted the fig and honey objects, the universe of people

32  Manuel Curado who have tasted other foods other than figs and honey, the human activity of utilizing for food purposes natural objects such as figs and honey. (One cannot be sure of the right number of elements of the phantasmatic or shadowy set of informations that were not ostensibly placed in the final statement. It is therefore enough to make clear that the above-mentioned are merely indicative. Others interpreters add a moral message to that set, seeing in the fragment a prudential ideal that states that, as Lesher phrases it, “men would have found a life of moderate pleasure perfectly satisfactory if they had not been exposed to more intense pleasures,” 1992: 182.) It is presumed that, in spite of the “creation” of honey by god or by a god, it is a fact that bees produce honey, as fig trees produce figs; for this to happen, specific conditions are required that are not made ostensible. These conditions may be partially listed (fertile land, mild winters, etc.) but not exhausted; in this sense, the fact that it is impossible to make a complete list of these conditions reveals that many of them are unknown, a feature they share with the human beings’ lack of knowledge about the creation of honey by god and, of course, the creation all other natural things. The most subtle of the enthymemes refers to the autonomous existence of these referents, to the mere fact that they exist (regardless of whether they exist with these or with those characteristics). 3. Rough sketches as summaries of reality. The great objects mentioned by linguistic nouns sum up a wider reality that is not expressed in its entirety. The words used by Xenophanes summarize broader realities. Honey is not described in the hive, in a precise quantity, in the manner in which it is transported, in the space and time in which it exists. The boundary between god-made honey, honey produced by bees and honey harvested by humans is not drawn. One does not realize also if what produces in the human beings the flavor of the honey is only the honey, if it is a mixture of honey with something else in the human body or if it is a process that happens in the body of human beings but has nothing to do with honey itself, as if honey were merely the initial cause of a secondary effect called the perception of honey, a cause that could be replaced by some other factor (for, as an enthymematic reason necessarily determines, an initial cause is not present in the final outcome of a process, e.g., a person does not carry the parents who conceived him or her). Figs, on the other hand, are not accurately described: Nothing is said about their variety, about their degree of ripeness or about the precise location where the fruit was separated from the branch of the tree. The human beings referred to in the fragment share these imprecise lines. It is not indicated which are the actual human beings who would consider that figs would be much sweeter if they did not know the taste of honey. The human beings referred to by Xenophanes are the

Xenophanes’s Figs and Honey 33 totality of human beings who were his contemporaries; the totality of human beings who existed in the past, who were his contemporaries and who will exist in the future; part of this set; humans at all possible ages and in all living conditions that human beings can experience, including the times they are asleep? The chromatic category is also not specified. The yellow color of the honey is not compared with anything that allows a safe identification and the color of the figs is not mentioned at all. Sweetness follows the same model because the references that are made to it have a low level of resolution or discrimination of details. As expected, the creation of honey by a god is so vaguely alluded to that it could be confused with a colloquial expression without any profound meaning. Nothing is said about the identity of this god (if any god indeed was in the spirit of Xenophanes), nor about the reason that would have motivated this creation. In short, the fragment describes in a highly imprecise manner a possible event, as if represented with a low level of attention to detail. It is as if it were a “photograph” of reality at a high power of ten, to take as a parallel the famous book of Philip and Phylis Morrison on the powers of ten and the scales of the universe (1982). This “photograph” captures in a grossly imprecise way something that could perhaps be grasped in a less imprecise way. Of course, supposing that the exercise of constructing a complete representation of the original perceptual event is not at all impossible for human bounded rationality, this hypothetical representation might seem unintelligible to humans, more imprecise and even perhaps more grotesque. (One could compare for instance this significant improvement of precision in the final representation with the law of diminishing returns in economics. More investment does not mean automatically more returns and could even mean diminishing returns, or, more dramatically, no return at all.) One does not know what a total representation would be, nor whether such a subject is accessible to human intelligence; however, by promoting some exploratory exercises, one sees that it is possible that this more complete representation would not altogether be recognized by humans as a representation of figs, honey and the human experience of comparing and tasting the two edible objects. One could imagine that a description of this encounter of factors is made at the level of microphysical constituents. How would one identify the protons of the figs, the protons of the honey and the protons of the mouth of the person who eats them, for example? Xenophanes represents with almost literary elegance an abstract relationship and not, surprisingly, a concrete event. He is not describing what a hypothetical individual called, for instance, Callias, would experience when eating figs and honey; however, if Xenophanes were confronted with a representation of a concrete event

34  Manuel Curado that instantiates the abstract relationship from the point of view of cellular metabolism or neuronal processing, it is likely that he would not recognize at all the referent of that preternaturally rich representation. One witnesses all the time this situation: No user of computers has the necessary intelligence in order to follow the binary code; humans do need shortcuts and simplifying strategies to cope with reality. Since these modest examples yield negative results, a hypothetical total representation can only be considered unrecognizable by humans. 4. The counterfactual context of interpretation. The fragment has a conditional structure that proposes a double counterfactual. The antecedent of the conditional proposes a possible world in which an order of things is exactly equal to the order of things that exist in fact in the world with the sole exception that honey does not exist. Everything would be the same except the existence of honey. The consequent of the conditional also draws an alternative scenario. In case honey does not exist, men would consider figs to be sweeter than they consider them to be in a world where honey does exist and has been tasted by men. There is also another enthymematic counterfactual. The statement establishes a relation between honey and figs but presupposes that this type of exercises can be done with other pairs of fruits known to the author, not making it impossible to compare figs with other fruits not known by Xenophanes, or even with other products that are not fruits but, notwithstanding that, could produce sensations of sweetness. 5. Qualia and subjective experience. The combination of several factors in the statement seems to place all of them on the same level; however, this junction has the high aim of highlighting the oscillation latitude of sweetness. In this precise sense, sweetness (especially the sweetness of figs) is salient, that is, stands out in relation to the rest. There is of course a subjective appreciation of other factors; for example, there is no representation of god or divine tasks without the subjectivity of a person shaped by cultural traditions, and in other fragments Xenophanes is eloquent in this regard (e.g., DK B14, B15, B16); the color of honey cannot be apprehended without a subjective appreciation; the very categorization of the objects of the world into entities that belong to the categories of honey, human beings, divine beings and figs. Nevertheless, the whole fragment revolves around sweetness, a reality that cannot be grasped without a subjective appreciation by conscious beings. 6. The empire of the category of relation. The text of the fragment presumes that the factors mentioned above do have autonomy. There is, from the point of view of the philosophical narrator, no possibility of god being confused with humans, existence with non-existence, taste with sight and, of course, honey with figs. As has been seen

Xenophanes’s Figs and Honey 35 previously, this appearance of horizontality is fragile, since factors are called in to create the case in favor of the oscillating nature of the sweetness of figs. It is precisely here that one could find one of the first notable contributions of this fragment for a possible and future program of a philosophy of perception. The autonomy of factors creates the case that each of them is solid and robust. This impression also exists when one tastes a spoonful of honey and eats a single fig. The specific flavor of each of these foods does not seem likely to change. There is something that tastes like honey and there is, differently, something that is experienced as fig. Xenophanes, with a shrewd and slightly sly spirit, could have related the taste of honey and the taste of fig; however, it is not at this level that the mutual play of his ideas unfolds. The Greek bard selects sweetness, a common property to honey, figs and many other foods and products. Of course, a person who does not know the existence of honey could nevertheless experience figs and pronounce on them by saying that they taste sweet to him or her. When that person eats a fig, he or she would say that that fruit is sweet, leaping rapidly from his or her subjective appreciation to the objective order of reality by mentioning a property that would allegedly exist in the thing itself. It is not this accidental or deliberate confusion of properties that should be underlined at the moment. Sweetness itself, whether it be a subjective event or a property of a substance that is itself sweet (this debate hints as a shadow of the fragment that will manifest itself with all its might only in modern thinkers such as John Locke and Thomas Reid), has a latitude in its intensity. Not being a territory, the sweetness that is perceived, seems to have a scope and even a space of its own. The remarkable contribution of Xenophanes is to have noticed that this scope is not fixed. The intensity of the sweetness of the figs changes when compared to the sweetness of the honey. This alteration of intensity erodes the nature of mental content; after all, if the intensity diminishes to a possible nadir, the sweetness will disappear; if it is increased to a possible zenith, it will be transformed into something else. All perception of sweetness, therefore, only exists relative to other realities; it is not an entity with absolute existence in the world. Xenophanes stood at the door of a vast problem. There is no evidence that he has thought of possible developments in his observation of the relational nature of sweetness. However, what applies to sweetness may be extended to all other factors. Fragment 38 equates a vast intellectual program which may include other contents (in addition to specifically perceptive contents); however, it is not known how far it can be extended. What is at issue could be generalized. To begin a possible line of thought, one might ask whether a significantly less intense sweetness of fig (compared with the sweetness not only of honey but of a normal fig) would still be a sweetness of fig, or whether it would be something

36  Manuel Curado else. One possible limit to the generalization lies in perception; after all, the verification that the intensity of sweetness changes when comparing such factors as honey and figs only happens in the mind of a sentient being. However, the intervention of a figure such as the god who can or cannot create honey, whether or not it is a mere way of expressing a possible scenario, extends the field of perception to the field of substances. Xenophanes himself, in the already mentioned fragments on the representation of the divine by various peoples and even, in a surprisingly imaginative and comprehensive way, by various beings other than human beings (namely, oxen, horses and lions), gives examples of the relativity of representations of subjects related to religious belief. There is, therefore, no reason for the core subject of the fragment 38 to be limited to perception. Doing some exploratory exercises in this direction, one can see what might be involved. The identification of honey as honey seems an absolute without relational context. Quite a mistake, of course. Honey’s own physical state, its viscosity that mediates the solid state and the physical state, exists only as a relation. In a blacksmith’s oven with high temperature, the physical state of the honey changes radically as well as its nature. Of course, the same reasoning may apply to some other factors (figs and humans for sure and perhaps even the god who created honey but who had the possibility of not creating honey). It is not known whether it can be applied to factors such as god or as the space that separates the objects or even to the time that structures the rhythm of the event that allows the perception at different moments of the sweetness of honey and figs. From this point of view, perception could be a particular case of a wider metaphysical subject, a subject that is not sweetness, is not subjectivity, is not fruit, is not honey, is not human being and is not the god creator of honey. It is not certain, however, what is the relative order of these matters. Perception is not an irrelevant detail in this line of thought, but it is not clear the level at which the problem could be precisely equated. Of course, despite the slight lack of precision in this detail, the consequences of the relevance of perception must be considered.

3.  The Relation in Place of Substance The widening of the relational dimension of perception to a wider metaphysical scope does not, as has been claimed, have a textual basis in the surviving fragments of Xenophanes’s work. It is irrelevant to start a quasi-forensic process to determine if there is any indication that this has happened or not. That would be a way of not appreciating on its own terms the eloquent and profound lesson on the relational nature of sweetness.

Xenophanes’s Figs and Honey 37 There is, however, one small exception which deserves mention. Xenophanes, in other fragments, attributes to God—in the precise sense of a god who is above the popular representations of the divine, which he criticizes, as has been seen (cf. fragment 23)—the property of feeling. Fragment 24, in particular, represents god as having the ability to see, think and listen, and this ability is superlatively described: The whole nature god is able to see, think and hear. It is of no interest to open here a parallel debate on Xenophanes’s negative and positive theology (for an exercise in these lines, see Lesher, 1992: 105–106). One has not explored a line of thought that imposes itself. It is the following one: If seeing and hearing are sensory activities and if these sensory activities are indeed common to humans and to god, it follows that something similar to what is described in fragment 38 will also happen to god. (It is useless to explain the expected details of the differences between the first and the second; suffice it to say that everything in god is superlative in relation to human beings.) After the moment when god knows that figs are sweet, when he realizes that honey is also sweet, does the sweetness of figs also seem less intense to him? Moreover, is it due to the relativity of the sweetness of figs as it is perceived by god the relativity of the sweetness of figs as perceived by humans? Is the first relativity the cause and foundation of the second? Generalizing this point, is god the foundation of all perceptual relativity, as if it were its deep structure and its maximum limit? The interesting contribution of this conjecture is not to enter into tiring and endless theological and philosophical debates, but to propose a sketch of the idea that the set of perceptual contents of conscious beings would depend on a relational matrix that is not itself perceptive. If the embarrassing relativity of sweetness is common to men and to God, it is possible to conjecture that this sameness could be a sign of a matrix common to all sentient beings. It is not possible to think perception in a free and neutral way. There is an entire intellectual history that enforces its empire. The idea that all perceptions of all sentient beings have a relational matrix, founded or not on a divine entity, is not easily understood (this point is not decisive for the argument), because one tends to think this matter in a substantial way. Sweetness seems to be something (even if it is a subjective thing or, forcing the linguistic expression, a “subjective object,” in the manner of an intention or a dream or a belief) just as figs and honey seem to be things to humans. The corrosive effect of the relationship proposed by fragment 38 runs counter to this substantialist conception. Aristotle gave this conception the theoretical foundation that it needed. One could recall the relatively unimportant position that the category of relation deserves in the set of categories of his Organon (namely, in Categories, 6a36–8b24; see Barnes, 1984: 10–14). Perception, as understood by

38  Manuel Curado Aristotle, is relational: “The following, too and their like, are among relatives: State, condition, perception, knowledge, position” (Categories, 6b1–2). The problem is that no substance is relative, nor the first substances (e.g., man, ox, hand), nor second substances (e.g., human species, bovine species) (Categories, 8a13–20). The ascendant of substance when compared with the category of relation marked the intellectual history of the West (see Categories, 2a10–15). Only by a deliberate act of philosophical imagination can one think about the putative interest and urgency of changing the relative positions of categories in the Aristotelian category table. Fragment 38 is decisive in this respect. It is perhaps the first formulation of an important subject from the relational point of view. It is true that other pre-Socratic authors made intellectual proposals that promoted the relational approach (namely, the Pythagoreans, Heraclitus and Philolaus of Croton), but only in this fragment is it seen clearly how a whole area of studies could be reformulated from the relational point of view. These pre-Socratic exploratory and pioneering exercises were never sufficiently developed. Over the centuries, the subject was being proposed without much consistency, as if everything were limited to philosophical exercises on extreme possibilities. On the side of the classics of the reflection on the human mind, consider the inversion of qualia, advanced by John Locke (e.g., An Essay Concerning Human Understanding, II.XXXII.15), or the theory of sensations as natural signs in Thomas Reid (e.g., An Inquiry into the Human Mind on the Principles of Common Sense, IV.1; Essays on the Intellectual Powers of Man, I.VI.3). On the side of contemporary thought, one could recall the doubt of J.L. Austin about the true flavor of saccharine, since this product makes the tea sweeter but is itself bitter. With a temporal distance of about 25 centuries, Austin formulates a doubt that has exactly the same scope as fragment 38 of Xenophanes: “ ‘What is the real taste of saccharine?’ We dissolve a tablet in a cup of tea and we find that it makes the tea taste sweet; we then take a tablet neat and we find that it tastes bitter. Is it really bitter, or really sweet?” (1962: 66). The Idealistic philosophical tradition could have brought these piecemeal intuitions to an unprecedented development. It is known that this did not happen. Therefore, against what might be expected, a question which was supposed to be as simple as the bones and blood of living beings is incredibly complex, to the point of having no solution. In conclusion, it is important to see that there is an alternative way of thinking about perception and this alternative mode is relational. The fecund intuition of Xenophanes comes at a high price, and here lies one of the reasons why this intuition has not been exploited to its fullest extent. The main reason is perhaps the corrosive effect of a kind of universal acid. Daniel Dennett stressed a similar point comparing the theory

Xenophanes’s Figs and Honey 39 of evolution to a universal acid that erodes everything (1995: 61 sq. and 521 sq.). Likewise, a relational theory of perception is hardly confined to perception itself. As one has seen above, one moves quickly from this localized subject to broader questions related to the set of categories that structure human thought and even to questions connected with an eventual divine foundation of the structure of the relationship. All this seems a prototypical example of a bad method of rational inquiry. Against this methodological wisdom, it is possible to identify many recurrent manifestations of the hope of reaching a relational theory in areas not connected to each other, such as religion, technology and computation (for literary, philosophical and technological representations of the category of relation, see Curado, 2017: 29–44, in Cibercultura: Circum-Navegações em Redes Transculturais de Conhecimento, Arquivos e Pensamento [Cyberculture: Circum-Navigations in Transcultural Networks of Knowledge, Archives and Thought] (eds. Helena Pires, Manuel Curado, Pedro Andrade and Fábio Ribeiro), V. N. Famalicão: Húmus, 29–44. This justified hope (or bad method) is hard to explain. It is possible that this hope points to the end of the need to continue to reflect philosophically on perception. That would be good news: A  philosophical problem would be solved. Perhaps Xenophanes’s pioneering steps in this regard could be followed with confidence.

References Austin, J. L. (1962) Sense and Sensibilia, Reconstructed from the Manuscript Notes by G. J. Warnock, New York: Oxford University Press. Barnes, Jonathan, ed. (1984) The Complete Works of Aristotle: The Revised Oxford Translation, Vol. I, Princeton, NJ: Princeton University Press. Chalmers, David J. (1995) “The Puzzle of Conscious Experience” in Scientific American, 273 (6), 62–68. Chalmers, David J. (1996) The Conscious Mind. In Search of a Fundamental Theory, New York: Oxford University Press. Curado, Manuel (2017) “A Internet de Luz: Ensaio sobre o Luminar do Romance Eumeswil de Ernst Jünger [The Internet of Light: An Essay on the Luminar of the Novel Eumeswil by Ernst Jünger]” in Cibercultura: Circum-Navegações em Redes Transculturais de Conhecimento, Arquivos e Pensamento [Cyberculture: Circum-Navigations in Transcultural Networks of Knowledge, Archives and Thought] (ed. Helena Pires, Manuel Curado, Pedro Andrade and Fábio Ribeiro), V. N. Famalicão: Húmus, 29–44. Dennett, Daniel C. (1995) Darwin’s Dangerous Idea: Evolution and the Meanings of Life, New York: Simon & Schuster. Freeman, Kathleen (1947) Ancilla to the Pre-Socratic Philosophers. A Complete Translation of the Fragments in Diels, Fragmente der Vorsokratiker, Oxford: Basil Blackwell.

40  Manuel Curado Graham, Daniel W. (2010) The Texts of Early Greek Philosophy. The Complete Fragments and Selected Testimonies of the Major Presocratics, Part I, New York: Cambridge University Press. Guthrie, W. K. C. (1962) A History of Greek Philosophy, vol. I, The Earlier Presocratics and the Pythagoreans, Cambridge: Cambridge University Press. Kirk, G. S. and Raven, J. E. (1971) The Presocratic Philosophers: A Critical History with a Selection of Texts, Cambridge: At the University Press. Lesher, J. H. (1992) Xenophanes of Colophon: Fragments, A Text and Translation with a Commentary, Toronto: University of Toronto Press. Morrison, Philip; and Phylis Morrison (1982) Powers of Ten: About the Relative Size of Things in the Universe (Scientific American Library Series), San Francisco: W. H. Freeman. Ostenfeld, Erik (1987) Ancient Greek Psychology and the Modern Mind-Body Debate, Aarhus: Aarhus University Press. Popper, Karl R. (1998) “The Unknown Xenophanes: An Attempt to Establish His Greatness” in The World of Parmenides: Essays on the Presocratic Enlightenment (ed. Arne F. Petersen and Jørgen Mejer), London: Routledge, 33–67. Tor, Shaul (2017) Mortal and Divine in Early Greek Epistemology: A Study of Hesiod, Xenophanes and Parmenides, Cambridge: Cambridge University Press.

3 A Neurophilosophical Approach to Perception Steven S. Gouveia and Georg Northoff

1. Introduction Perception is an important feature of our mental life. Most of the best experiences that we have come with perceptual experience: The taste of a good coffee or a nice pizza. Perceptual experience is also vital for our survival. Since ancient times, perception theory has been dealt with as part of investigations into the nature of reality or in terms of how we access the world. The question about the nature of perception is, however, different from the empirical question: That is, how do our sense organs in general actually work? In that sense, we are asking for the specific mechanisms of, e.g., vision and smell. Even if, in a scenario of epistemic perfection where we know every detail, e.g., the neuropsychological theory of vision and smell, there will always be the question: Why does this process give a reason to justify, for example, the belief that I’m seeing this table right now? In this chapter, we will focus on the importance of one of the most relevant answers to the problem of perception: Namely, Direct Realism (DR). We will characterize the main idea of this approach and the several sub-positions (and which model—sensory, cognitive, or other—of perception they are attached to) underlying the main intuition of DR. We will then demonstrate why this approach is more plausible than the indirect realists’ theories of perception. However, DR suffers nevertheless from some problems, most notably the adoption of an implausible model of perception. Presupposing an empirically more plausible model of perception, we, therefore, sketch a neurophilosophical alternative, a spatiotemporal approach, as we describe it. Theories of Perception I: Direct Realism Perceptional experience is one of the oldest topics of human thought. For example, Plato’s Cave or the Evil Genius from Descartes raise many questions about the validity of the knowledge that our senses get from the world. The problem about the nature of perceptual experience can be

42  Steven S. Gouveia and Georg Northoff then stated as the struggle to merge two somewhat clear statements about perception and about the real possibility of a certain kind of perceptual error, namely, the idea that we have direct access to the world but that sometimes this access is not accurate (cf. Crane, 2005: 241). Why is it so difficult to combine these two ideas? Let’s sketch first, then, the problem of perception in more detail. Our common view about perception is that (a) there are ordinary objects that exist in the world, (b) these objects are not dependent on our perceptual activity (i.e., they are mind-independent) and, sometimes, (c) we are perceptually aware of such mind-independent material objects. So, I’m seeing this table in front of me, right now, and the existence of the table doesn’t depend on my perceptual awareness today, as the same table will still be here in the office in Ottawa the next day if nobody changes its location. The first claim (a) is an ontological statement: It is a claim about what exists and what is real. The second claim (b) is an epistemic statement already committed to a specific theory of perception. For now, we will just explain that the objects that we perceive are mind-independent in the sense that their existence doesn’t depend on our perceptual activity toward them. The inverse idea would be to claim that (some) objects in the world are mind-dependent: That is, for a thing x to be mind-­ dependent, it must be an object of perceptual experience—it must be perceived. This idea is somewhat related to the famous statement from George Berkeley: “[E]sse est percipi”—“To be is to be perceived.” The important consequence, for now, is that these x are considered mental entities of a particular form. The last claim (c) is just a trivial statement that during most of our lifetime—when we are not sleeping or dreaming—we see objects in the world that are not dependent on our existence to be perceived. Figure 3.1 shows a visual representation of what we just claimed:

Figure 3.1  Direct Realism

A Neurophilosophical Approach to Perception 43 Against these ideas about perception, we have several cases that can contradict some of our initial claims. For now, we ask you to observe the following image in Figure 3.2:

Figure 3.2  Müller-Lyer illusion1

Our perceptual experience of the image in Figure 3.3 shows, clearly, that the lines A, B and C are different: A is smaller than C and B is longer than C. We are not drunk, we didn’t take any drugs, we are not dreaming and our visual system is working perfectly. We have, then, to conclude that the direct visual experience that we have must be coherent with our three claims about perception. So, the lines A, B and C are mind-independent and they really exist with such and such features and characteristics. If, however, it is the case that those lines are not so different as our perceptual experience seems to perceive them, what would that mean for our first approach about perception? It would mean that maybe (a), (b) or (c) are wrong.

Figure 3.3  Müller-Lyer illusion

44  Steven S. Gouveia and Georg Northoff As you can see it now, our first perception about the lines was wrong: Factually, all the lines have the same length (A = B = C). However, this goes against our first visual experience of the image. What is wrong in this picture? Note that, however artificial this example is, similar cases of illusion and hallucination happen in several mental diseases. For example, in schizophrenia, people usually have several hallucinations (they can hear, see, smell or feel things no one else does). The most common hallucination is hearing voices in their heads.2 Hallucinations are very real to the person, even though people around them can’t notice any voices. Empirical data show us that this particular kind of auditory hallucination may be triggered by an imbalance between the Default Mode Network3 (DMN) and the Central Executive Network (CEN) and their relationship with the auditory cortex: The DMN seems less connected in the resting state,4 which is strongly connected to the CEN (cf. Northoff, 2015, Ganzetti and Mantini, 2013). So auditory verbal hallucinations (AVH)—the experience of perceiving an external voice without any external stimuli—is a natural case that affects more than two-thirds of patients with schizophrenia. They can also suffer several delusions: Delusions are false beliefs or misinterpretations of a particular event and of their meaning.5 For instance, a person might think someone is trying to control their brain or that someone is stalking them. Moreover, hallucinations and illusions also happen with other senses: For example, synesthesia is a perceptual experience where the activation of one sense leads automatically to an involuntary experience in a second sense (e.g., people who see colors usually associate a certain sound). All the senses might be crossed with each other, making the perception of reality different from that of most people. The general problem is that we are facing a phenomenon that raises many doubts about our common view of perception. Thus, we should conceptually divide the perceptual experience into two main positions: i. Good cases (or veridical experience): When we have a real (nonillusionary) perceptual experience, e.g., where one sees a table and has an experience in which it looks brown to one and it is, in fact, that way. ii. Bad cases (illusory/hallucinatory experience): When we have illusionary/hallucinatory perceptual experience, e.g., here one sees a table and has an experience in which it looks brown to one, when in fact it is black. This division will be important to show some problems with Direct Realism (DR). We can now present, based on what we said, a version of what

A Neurophilosophical Approach to Perception 45 is known in the literature by the “Argument of Illusion/Hallucination” (AIH): (1) The perceptual experience in (ii) consists in an experience that is mind-dependent of the object; (2) The perceptual experience in (i) and (ii) is the same; (3) Therefore, the perceptual experience in (i) also consists of an experience that is mind-dependent of the object. This argument is really important because it denies one of the main features of our common view about perception: Namely, (b) and (c), that is, that we perceive objects that are mind-independent. In order to assure the validity of the argument, (2) must be read in the following sense: (2*) If two perceptual experiences (x and y) are the same, then if x involves the presentation of a mind-dependent object, then y also involves the presentation of a mind-dependent object. We can now detail every premise of the argument: (1) states that in (ii), something must be present for the subject. As we saw above, people with schizophrenia usually have positive symptoms like auditory hallucination (e.g., hearing voices) and they don’t state that they are not hearing anything: There is something that doesn’t have an external cause that is presented in the mind of the person. However, if something is presented in (ii), then it can only be something mind-dependent (as we know, by definition, that that experience doesn’t have an external cause). Therefore, we must conclude (1), that is, in (ii), we are presented with an object that is mind-dependent from the subject. What about (2)? What is the meaning of (i) and (ii) being the same? To argue about that premise, we have to present a few ideas which will lead to the conclusion (2). The first idea is to state that perceptual experience in (i) and (ii) are phenomenological or subjectively indistinguishable for the subject. That is, to see a (real) table or to see a illusionary table is the same subjective experience—that’s why we believe that in the two cases that there is a table in front of us. If the two experiences are indistinguishably (at least for the subject of experience), we can then conclude that they have the same nature. We can, then, combining the two ideas, conclude (2): The perceptual experience in (i) and (ii) have the same nature. Finally, we conclude (3) from (1) and (2). We can now ask: If the AIH is valid (and, let’s assume, for now, sound), what follows for the problem of perception? The more important consequence is that this argument collides with the common picture about

46  Steven S. Gouveia and Georg Northoff perception that we declare above in (a), (b) and (c). Those ideas complete what we call Direct Realism (DR): A theory of perception which main intuitions are (a), (b) and (c). But how can resist DR to the AIH? And how, in detail, does the AIH goes against DR? Finally: With which kind of model of perception is DR attached to? Theories of Perception II: Indirect Realism We can now introduce the main argument against DR. The argument follows:   (4) In both (i) and (ii), there is a common element (quale) presented in your perceptual experience;   (5) Because the quale is qualitatively identical, an analysis of one must be the same as the other;   (6) In both (i) and (ii), we are aware of something (F);   (7) But in (ii), by definition, we cannot have an external/material thing; therefore, it must be a mental entity (or sense-datum or representation);   (8) Because (5), what follows is that we only perceive sense-data;   (9) Following (8), we conclude that we never perceive material/external objects; (10) Therefore, DR is false.6 This second argument show us clearly how the AIH goes against DR. If the argument is valid and sound, we will have to conclude that DR is refuted and that a direct realist approach to perception is mistaken. Maybe we will have to adopt, as the argument seems to suggest, an indirect realist approach to perception. What is, then, different in Indirect Realism? Indirect realism about perception is a theory that denies (b): That is, that the objects that we observe from the world are independent of the subject of experience. As the argument above shows, we have to conclude that we cannot have experience of objects that are mind-independent of the subject but only a subjective impression that is mind-dependent. But what is “indirect” and “realist” on this approach? The “realist” part agrees with the DR: We perceive physical objects which are mind-independent. However (here comes the indirect part), we do so via, or in virtue of, perceiving mind-dependent sense-data. These sense-data are, of course, caused by and represent physical objects of the world (that’s the realist part). Therefore, indirect realism is the theory that claims that we perceive sense-data immediately and physical objects indirectly. Again, the postulation of “sense-data” comes from the idea that we perceive something having some property F and that when we perceive something having that F, then there is something that has this property F.

A Neurophilosophical Approach to Perception 47 However, we saw that (ii) the bad cases show us that the physical object does not have the property F. We have to conclude that what has the property F is something mental, non-material, the sense-data. The concept of sense-datum was introduced by G.E. Moore. (cf. Moore, 1953: 30) Its precise definition is highly controversial, as it is used differently by different philosophers. For example, the main intuition of the concept introduced by Moore was somehow neutral regarding the mind’s dependence or independence of the thing we perceive. Therefore, if we want to be rigorous, for Moore there is no necessary relationship between sense-datum and mind-dependence, as the concept can also make sense regarding a mind-independence model of perception. (cf. Snowdon, 2007) But most of the contemporary authors have defended the notion that sense-datum is necessarily connected with mind-dependence. The big picture that arises from this is that the world itself is a representation, a replica of the real external world, as if the world was something that is present in your television and you are just sitting in the theater, watching what the TV is showing to you (and not the real world). What do you perceive in the TV, then? The content of sensations or appearances, the sense-data, that are mental entities which are the way we perceive them to be as opposed to physical objects that can appear differently from how they really are. Hence, the sense-data only exist while they are being experienced as opposed to physical objects that exist even if no one is experiencing them. Finally, the sense-data are private, as no one else but you can experience your sense-data, as opposed to physical objects that are public (the same object is experienced by different people). Figure 3.4 represents visually this theory of perception:

Figure 3.4  Indirect Realism

48  Steven S. Gouveia and Georg Northoff Is this approach a viable alternative to Direct Realism? One good point against DR  is that it seems hard to understand the causal role of the objects of the external world. How can they affect the subject directly and immediately? We know from the neuropsychological sciences that perception involves a very complex and long causal line of events. If so, it’s impossible that we perceive the world directly, but something else must be mediating the perception and the object (e.g., sense-datum or representation or idea)— that’s the direct entity of perception. Does this really entail the falsehood of DR? We don’t think so. When a direct realist states that external objects are the direct objects of perception, what a direct realist is denying is that there is something else mediating by a priori awareness of the complex causal series that ends up with perception. So of course that perception is, in a narrow sense, indirect: It involves a long series of causal relationships. Nevertheless, it is implausible that perception is indirect in the wider sense of involving some mediator between the external object and its perceptional experience. The argument just confuses how we perceive with what we perceive. The causes are simply certain conditions that makes seeing the object possible. Of course: The object alone is not sufficient for having a perceptual experience—we also need a nervous system working properly (and a body and the interaction with an environment). Nonetheless, even if sense-data were the medium by which the external objects of the world are perceived, then it would follow that sense-data are only the causal intermediary and we don’t really perceive sense-data directly. But sense-data are, by definition, what we perceived. And that’s a difficult paradox to solve. Moreover, Indirect Realism is an implausible response to the problem of perception because the postulation of sense-data requires necessarily a connection with representation. But representationalism simply fails because it is assumed that something in the brain must be interpreting the perceptual data from the world (that is, something must be in the theater watching the movie that provokes the perceptual experiences). What kind of entity can be in the brain? There must be something: Usually, we call this a homunculus. The problem with this kind of entity is that it leads to a second problem: Necessarily, it requires an infinite regression, because we would have to explain how the homunculus1 can make sense of the information of the senses and then explain that it’s because a homunculus2 is interpreting the perceptual information and so on. Due to such infinite regress, this approach seems ontologically implausible. Furthermore, Indirect Realism seems to leave an open door to skepticism about the existence of the external world: If we never have direct experience of the world, how can we be sure that it really exists? One way is to accept this consequence and being an idealist about perception: The only thing that exists are ideas. Of course, this approach is

A Neurophilosophical Approach to Perception 49 too implausible, as it might lead to the postulation of a super-being, or, alternatively, leads to solipsism. The postulation of the sense-datum is also philosophically implausible: They are usually characterized by being private (nobody else can access my sense-data), they are exactly as they appear to my perceptual experience (they cannot be misperceived, we cannot be wrong about them), they are mind-dependent (they exist only when they are being experienced) and they represent things in the external world. All these characteristics are somewhat problematic, as we show below. Finally, as we will show in the second part of the chapter, perception seems to be non-representational.7 We will sketch a detailed framework that will explain how perception really works. What is wrong, then, with the argument against DR? The AIH rests on the claim that when it seems to one that something has some property F, then there is something that has that property F. Is this a good claim? Again, we don’t think so. Following Searle (cf. Searle, 2015: 18–24), the problem with the argument is related to the premise (6): (6) In both (i) and (ii), we are aware of something (F). The problem with (6) is that it suffers from a fallacy: The fallacy of ambiguity. As Anscombe (1965) showed us, there is a vital distinction to be made on the intentional stance: between content and object. That is: If I see a table in front of me, the content is “There is a table in front of me” and the object is the table itself. So perception has content, but this content is not itself the object of perception: The opposite will put us in the “argument of illusion” scenarios. In the bad cases of perception (ii), we just have content, but no object. The content is caused internally by a problem in the nervous system of the person who is having a hallucination—there isn’t any external cause.8 We cannot treat the perceptual experience itself as the object of the perceptual experience in the (ii), as there is simply no object that is causing that experience. Moreover, following Northoff (2018: 269–314), we cannot infer from the phenomenal realm of conscious activity to its ontological basis. This amounts to what can be described as a (11) “phenomenal-ontological fallacy.” Against what we just claimed, some philosophers argue that (5) is a reasonable claim to follow. For example, J. Foster claims that “(. . .) if perception and hallucination have the same subjective character, the simplest and most obvious way of accounting for this would be to suppose that they also have, at the fundamental level of description, the same psychological character through and through (. . .)” (Foster, 2000: 8). Other philosophers, with the same intuition but different approaches, argue for the reasonability of (5), like Martin (1997) or Crane (2005).

50  Steven S. Gouveia and Georg Northoff But even if we assume, for the sake of the argument, that the objects of perception in the bad cases (ii) are the sense-data, it does not follow that they must be also be the objects of perception in the good cases (i). The fallacy can be formalized in the following form: (11) If x and y are phenomenally indistinguishable, then x and y are ontologically indistinguishable. But then again, as we have shown above, we don’t have any plausible reason to assume that the subjective experience should drag necessary consequences for ontology. For instance, a hologram of a table is phenomenally indistinguishable from a real table. But to conclude that, because of that, the two tables are real (or that the two tables are minddependent) would be to commit the (11). We conclude the first part of the chapter stating that Direct Realism is the right approach to follow. However, we have to find out now which sub-position of this wider approach is more plausible. We will do that by analyzing different models of perception and their consequences for the problem of perception.

2. Realism of Perception and Models of the Brain in Neuroscience Naïve Realism and Passive Model of the Brain We will now analyze what kind of model of the brain we need to explain the perceptual experience. In doing so, we will characterize several approaches and show their consequences for the problem of perception. Following the three claims about the common view about perception, we should think about the role that the brain has in this phenomenon. The first model—the naïve model of perception—considers the brain as a passive organ in that its activity is determined only by the external sensory stimuli without any direct or indirect contribution of the brain. This extrinsic model of the brain (Northoff, 2012, 2016a, Raichle, 2015) focuses exclusively on the role that the external world has in the construction of our perception. The main idea is that the mind-independent object is a sufficient condition to account for the total process that is the perceptual experience. In this approach, an active role of the brain is denied and its only function is to receive and process the information received by our sense organs. Moreover, the only function of the brain is to associate the different stimuli with each other. This model of the brain is associated with Naïve Realism, a radical sub-position of DR that asserts that the role of the brain itself, e.g., its intrinsic input to the neural activity underlying perception, is minimal. This position somehow fears that denying this is to succumb to the

A Neurophilosophical Approach to Perception 51 conclusion that (ii) the bad cases seems to infer: The idea that we don’t have access to the external world and that we only have access to ideas that our brains construct. The model of the brain defended by the Naïve approach can be then characterized by the passive receiving and processing of external sensory stimuli. One consequence of this is that the brain’s neural activity is necessarily and sufficiently dependent on the external stimulus which entails that the brain just processes passively the information of the environment. The active participation of the brain is completely denied in this naïve model of perception. Even if we measure some spontaneous activity in the brain, this model will ignore its impact and causal role in the perceptual activity and neural activity in general. (cf. Northoff, 2016a) We can then affirm that this model can be described as defining the brain as a mere stimulus-response-based (or neuro-sensory model of the brain). This approach implies necessarily that the resulting stimulusinduced activity should be the same across different subjects. Is that what the empirical data show us though? We think that this model of the brain is simply implausible to explain the role of the brain in the process of perception. Based on the discovery of neural activity within the brain itself, totally independent of any external stimuli, we will suggest another type of model of the brain: An intrinsic rather than extrinsic model of the brain. (cf Northoff, 2012, 2018, Raichle, 2010) This spontaneous activity seems to be scientifically and philosophically relevant to understanding the role of the brain in perception. The crucial point is that spontaneous activity remains independent of a specific task or stimuli. (cf. Logothetis et al., 2009; Northoff, 2014a) We should note that, however, the difference between the spontaneous and the stimulus-induced activity is not absolute, but rather relative, i.e., related to spatiotemporal differences. (cf. Klein, 2014; Northoff, 2014a) The spontaneous activity that reflects what operationally is described as resting-state activity, is manifested in several neural networks like de DMN (Default Mode Network), the CEN (Central Executive Network), SN (Salience Network) and the SMN (Sensorimotor Network) (cf. Klein, 2014). The relevant point here is that even in regions of the brain that are not reliant on external stimuli, there is neural activity. This will be relevant for the model of the brain that we will suggest below. More importantly, we will show how this naïve model of the brain cannot account for the pre-stimulus neural activity. Duncan et al. (2015) investigated healthy college students (free of mental or psychiatric diseases) with a psychological questionnaire (CTQ) that assessed early childhood trauma. Using fMRI, it measured the spatiotemporal structure of the spontaneous activity, as indexed by entropy.9 The findings can be summed up in the following ideas (cf. Northoff, 2016b): (a) The subjects who scored high for early childhood trauma showed higher degrees of entropy in the perigenual anterior cingulate

52  Steven S. Gouveia and Georg Northoff (PACC) and intrinsic activity in early adulthood—there is a correlation between PACC entropy and their questionnaire score; (b) Using magneto resonance spectroscopy (MRS), they measured the resting-state concentration of glutamate in the same region of the brain, the perigenual anterior cingulate cortex (PACC)—it was observed that there was a direct association between glutamate and early childhood trauma: The higher the former was (as measured with CTQ), the lower the resting-state concentration of glutamate in PACC was found; (c) Duncan et al. (2015) included yet another component in the imaging experiment in fMRI, namely, a task that applied aversive stimuli10 and measured its stimulus-induced activity. Neural activity was measured precisely during the anticipation of aversive stimuli. It was observed that neural activity in the right anterior insula and the motor cortex during the aversive anticipation correlated significantly with all three measures: Early childhood trauma, glutamate concentration and degree of entropy. Those empirical data are important for our naïve model of the brain because we can conclude the following consequences: (d) Events in the world impact and modulate the brain’s intrinsic activity in such a way, i.e., in a statistically based and spatiotemporal way, that they can become traumatic events for the respective person. The spatiotemporal structure of the brain’s intrinsic activity may, therefore, serve as “spatiotemporalized memory” of how the world impacted and modulated the brain and its spontaneous activity’s spatiotemporal structure; (e) This suggests that a particular event in the world can impact and modulate the biochemical features of the brain’s intrinsic activity, its level of glutamate, in such way that the respective event may later be regarded as traumatic—events in the world can also modulate and impact biochemical measures in the brain; (f) The empirical data suggest that the spontaneous activity of the brain is both passively affected by the environment and at the same time actively modifies its brain activity; this amounts to what has been described as rest-stimulus interaction (see Northoff et al., 2010; Huang et al., 2016) and cannot be accounted for by a mere passive model of the brain. We can, therefore, conclude that the spontaneous activity’s spatiotemporal structure is dependent on the prior experience of the respective subject. That’s why the naïve model of the brain fails to predict the interindividual variation with the different subjects showing different degrees that are related to the degree of early childhood trauma. If this first model was the correct one, we should have an inter-­individual similarity between all those factors during the stimulus-induced activity. That’s not the case, as the study showed us, though.

A Neurophilosophical Approach to Perception 53 Moreover, the naïve model of the brain conceives the external stimulus as necessary and sufficient. Hence, the spontaneous activity (e.g., level of glutamate and entropy) should not be accounted for. That’s not what the empirical data show us, again. On the contrary, the findings suggest that the stimulus-induced activity is clearly related with the spontaneous activity (e.g., entropy and glutamate levels). Since this model is based only stimuli-response structure, it cannot predict these findings. Therefore, we should conclude that the naïve model of perception is not plausible. We can summarize the implausibility of our first model with the following argument: (12) A plausible model of perception must be aligned with the empirical data;11 (13) The naïve model of perception is not aligned with the empirical data; (14) Therefore, the naïve model of perception is not a plausible model of perception. The quoted study supports our (13) premise. Moreover, Northoff (2014b, Chapter 8) show us that the preceding resting-state activity predisposes our conscious content during stimulus-induced activity. Again, if the naïve model of perception were the right one, we wouldn’t have interindividual differences and, if even if we did, the model would ignore the active role of the spontaneous activity (pre-stimulus) and its consequence for the behavioral level and for the perceptual content. Finally, this model can’t account for the role of pre-stimulus activity of the brain and its impacts on the stimulus-induced activity. Another fMRI study focused on the auditory cortex (cf. Sadaghinani et al., 2009) and demonstrated that the resting-state activity has a strong influence on the stimulus-induced activity and associated perception of objects and events in the world. The design of the task was the following: Subjects had an auditory detection task hearing broadband noise stimulus in unpredictable intervals of 20–40 ms. Whenever the subject heard the specific sound, they had to press a button (otherwise, they did not have to). The idea was to compare the neural activity preceding hits with the neural activity preceding the instance of the subject when they failed to detect the specific sound. The study found that successful detection of the sound was preceded by a significantly higher pre-stimulus activity in the auditory cortex in comparison to misses. The same data was analyzed by another team (Sadaghiani et al., 2015) that showed that specific neural networks like DMN have more functional connectivity before the detected auditory stimuli. Taken together, these studies (and others: See Northoff et  al., 2010; Northoff, 2014a) show that the resting state exerts a strong impact on

54  Steven S. Gouveia and Georg Northoff the contents of our perception. The resting state’s pre-stimulus activity level seems to be central in determining the contents that we subsequently perceive. (cf. Hohwy, 2013, 2014) The naïve model cannot predict those facts exactly because the characterization of the brain as a static, non-temporal and isolated object is implausible. The main Humean intuition that the stimuli from the external world are necessary and sufficient for our perception of objects is implausible and, therefore, we conclude (14). We should notice that the only thing that we have demonstrated until now is that a specific model of perception—the naïve model of the brain— based on DR is implausible. But DR is still the right road to follow. We need to find, however, the specific model that can properly account for perception. Indirect Realism and Active Model of Brain A second model of perception, different from the naïve realist approach, can be presented. Instead of considering the brain a passive organ with the only task of associating the various stimuli from the world, we can argue (as Kant did, for example) that that is not sufficient to obtain knowledge via perceptual experience. Instead of offering a passive model of the brain, we can give an active role to the brain, considering that its participation in the process of perception and cognition, in general, is also necessary to account for perceptual experience. In this model of perception, we can finally account for the spontaneous activity so empirically relevant. This spontaneous activity is independent of the external stimuli—the role of the brain is now to structure and organize the perceptual data in a coherent unity. There is active participation of the brain to process the sensory stimuli. Therefore, we are in a better position now from the previous model: We can now account for the impact of the spontaneous activity on stimulus-induced activity. We can also explain the inter-variability of the subjects of the study above quoted, something that we couldn’t do with the previous model. In this model, the spontaneous activity directly impacts the stimulusinduced activity: There is a direct relationship, and the first is a necessary condition of the latter. However, the model cannot account for some of the data of the first study, namely, the impact of the encoding of the environmental life events by the spontaneous activity (cf. Northoff, 2016b: 25), and perhaps a more spatiotemporal account is necessary to account for that element— this account will be sketched in the third model proposed. This model seems to account for the impact of the spontaneous activity on stimulus-induced activity. To do so, it must do it via representation:

A Neurophilosophical Approach to Perception 55 The early life events are represented in the brain and they influence the higher cognitive functions like decision-making or memory. More generally, the sensory information of the world is then represented by the brain to form the perceptual content of our experience. The brain, here, has an active contribution: It processes the sensory stimuli and constitutes the mental features. The only way that this model can account the inter-individual variations is via representation. And this is the main error of this model: As we saw above, representationalist models of perception are philosophically problematic and we will show, now, that they are also empirically implausible: (15) A plausible model of perception must be aligned with the empirical data; (16) The representationalist model of perception is not aligned with the empirical data; (17) Therefore, the representationalist model of perception is not a plausible model of perception. Again, there are several meta-philosophical assumptions that will be discussed in further research.12 At this point, what we can claim is: (15) represents the modal intuition that [all] the empirical data should be explained using a representational language. So we just need to find one counterexample—see below—that can be explained theoretically without using the representational language.13 That will show that the necessity of the premise is implausible. How can we support our premise (16)? First of all, we should expect that the early life event to only correlate with the stimulus-induced activity, but not with the spontaneous activity—the level of glutamate and entropy. But the study (Duncan et al., 2015) shows us a direct relationship between those events and the spontaneous activity. Nonetheless, this model remains insufficient when it comes to the encoding of the environmental life by the spontaneous activity. (cf. Northoff, 2016b: 24–25) To encode the environmental life by spontaneous activity, we need a more dynamic and relational model to account for those facts. Moreover, this model can only account for the objects of perception indirectly: That is, the input from the environment is mediated by the brain’s representation of them. The problem is that, as we will show below, this approach ignores the temporal features of the brain that are so relevant. This cognitive (instead of sensory) and representationalist (instead of naïve) model of perception commits also a relevant fallacy: a methodological-ontological fallacy. That is: “[T]he fact that we probe the brain’s neural activity by using our cognitive tasks does not entail or justify the assumption that the brain itself is a cognitive device.”14 (Northoff, 2016c: 10)

56  Steven S. Gouveia and Georg Northoff

Figure 3.5  An example of fMRI research

Consider, for example, the use of functional brain imaging (fMRI). This method represents the activity of the brain through colorful dots and spots as in Figure 3.5.15 To conclude that from the different shades of black, gray and white that the brain’s activity (representation) is itself colorful would be a methodological error. The represented properties are related not by the specific ontology of the brain or its neural activity, but by the specific method used. (cf. Northoff, 2016c: 9) The active model of the brain is an upgrade from the passive model that can accommodate important conceptual and empirical features. However, it fails to do it properly, and a more plausible model is needed to account for perception. Spatiotemporal Realism and Neuro-Ecological Model of Brain Until now, the models of perception proposed have failed to be logically and empirically plausible. What kind of model can accommodate all those facts? We will now propose an alternative that considers the first two models without considering them mutually exclusive but incorporating some of their features. The naïve model of the brain is right about its direct realist position. However, the direction of fit is not between the world and a passive brain but between a world-body-brain relation. The concept of spatiotemporal alignment can account for the directionality needed to explain the perceptual activity. For example, Northoff (2018) show us that there is a coupling process between the brain and the heart and stomach that presuppose a specific directionality between them. That is: A healthy brain tries to align the phase onsets of its spontaneous activity to the heartbeat and, on the other hand, its amplitude is coupled to the phase onset of the stomach. In both cases you can see the directionality: The brain’s spontaneous activity

A Neurophilosophical Approach to Perception 57 aligns and integrates itself to the body (i.e., heart or stomach) as it is also supported by the data on the information transfer from the stomach to the brain. (cf. Northoff, 2018: 201–202) Moreover, the relationship between the world and the brain is not a-temporal as the first two models predicted, but temporal. More importantly, it is bidirectional, because the brain actively conforms to the information of the world. Therefore, we can talk about a world-brain relation in this third model. The inverse relation—a brain-world relation—would put us again in the representationalist model of perception: The directionality is from the brain to the world. But that’s not the case. However, we still have to explain why our third model of perception— the spatiotemporal model of perception—has more explanatory power than the two previous ones. We will argue, with Johnston (2007), that our brain does not represent the world, as the representationalist model of perception suggests, but quite on the contrary, the brain samples from the environment the events or objects. Such sampling is a selective process that is temporal, contrary to the sampling that occurs in the passive model of perception: In this case, what is sampled by our mind would be a passive sampling, a one-to-one correspondence in a spatiotemporally isomorphic way. But we are not arguing for such a passive process: As Johnston suggests, the sampling in our neuro-ecological model of perception is selective. The main difference is to accept, again, that the brain and the body select16 specific parts of the environment while discarding others. In a nutshell: The environment contributes to the objects of perception while the mind/ brain contributes their selection, making all processes dynamic. Again, this is only possible based on a world-brain relation, as distinguished from the indirect relationship that we have with representation. This implies the following conceptual difference: In this spatiotemporal model of perception, representation is substituted by disclosure of the events or objects in the environment. How can we, still, adopt a realist position? Using the concept of “selective sampling” we are providing an explanation that is i) realist, ii) direct and iii) causal and that can account for all those empirical data above quoted. First, the selective sampling presupposes an active contribution of the brain: Against the naïve model of perception, that presupposes a rather passive contribution, here the relationship between the brain and world is not unidirectional (as it is in the naïve model of perception and in the representationalist model of perception), but it is bidirectional: The environment gives the sensory input which the brain relates and, at the same time, it selects the relevant information via sampling. Therefore, the environment contributes to the inputs and the brain with their selection. We can now detail how the sampling works: By sampling the world, the brain conforms its own spontaneous activity to the world and its inputs. The origin of the contents of perception are now “located” in the

58  Steven S. Gouveia and Georg Northoff world itself and, more specifically, in the relationship between world and brain, namely, in the degree to which the brain conforms to the world— that process is statistically based (cf. Northoff, forthcoming: 6). We can also easily explain the (ii) bad cases of perception. Those cases happen exactly because the proper alignment between the brain/mind and the world is missing: The former cannot conform (for several reasons) to the latter. This way, our spatiotemporal model of perception can do something epistemically powerful: It can also explain why the scenarios of hallucination and illusion happen. In this model, we can explain both artificial and natural cases. For example, if a schizophrenic patient is “hearing” a sound that is not present in the environment, we can explain this by saying that this is clearly a case where the relationship between the world and the brain is imbalanced, that is, the brain is not coupling with the world, and the spatiotemporal feature of the brain loses its balance, causing the hallucination. As this process is statistically based, it’s easy to see why this can happen: An injured brain will look in toward itself without considering the input from the environment. Our model is also ontologically powerful: We do not need to postulate another metaphysical category like representation in order to explain both (i) the good cases and (ii) the bad cases of perception. One of the biggest problems of representationalism is to offer more ontological entities that are needed to explain perception. That is: With this model of perception, we need to postulate two entities:   (*) The object of the world, mind-independent; (**)  The representation (or idea or sense-datum) of the object, mind-dependent; Following Ockham’s Razor Principle, when presented with competing hypotheses, one should select the solution with the fewest ontological assumptions. In our second model of perception, if we are aware of a table in front of us, we are directly (**) aware of the idea (**) of this table and indirectly aware of the table (*) itself. This unnecessary duplication is problematic. As Campbell claims: “We seem to be faced with the idea of an absurd duplication of the material world in some other area.” (Campbell, 2002: 156) Considering this, our third model absorbs the lessons of the two previous models: It accepts that the world has influence on the brain (like the passive model), but it also accepts that the brain has influence on the world (like the active model). In this sense, we can recall Plato’s Theaetetus and its “Secret Doctrine of Protagoras”: Perception comprises not an active or passive relation with the world. Quite differently, this relation is a distinct “double-sided” interface between the world and the brain/ mind and vice-versa.

A Neurophilosophical Approach to Perception 59 One major consequence of what we argued for is to hold that cognition is not essentially representational17 (against, e.g., Fodor, 1981 and partially in favor of, e.g., Brooks, 1991). One option is to claim that the representationalist conceptual framework is just a useful theoretical façon de parler18 that can help the theorists of the mind to sketch their theories and guide the empirical research. However, most researchers, when they use the representationalist approach to perception (or cognition in general), they really believe that representation is a real thing: It is part of our ontology. In our view, that is highly implausible. Another option is to try to redefine the notion of representation (for example, Varela et al., 1991; Clark and Grush, 1999 or Gallagher, 2005). Either way, the classic notion of representation should be replaced from current neuroscience and cognitive sciences in general as it is part of a conceptual scheme that is problematic to describe relevant empirical data.

Conclusion In the first part of the chapter, we discussed how the (ii) bad cases of perception influenced our ideas about perception. The main conclusion that we got from that discussion was that indirect theories of perception look problematic and that the direct realist approach seems to be more plausible. In the second part of the chapter, we tried to connect the different models of the brain used in current neuroscience and cognitive science in general with the specific positions regarding the problem of perception. We concluded, based on some empirical data, that the more radical view (the naïve realist position) of DR is problematic because it assumes an implausible model of the brain. The same happened with the indirect realist position: It assumes an implausible model of the brain, but for different reasons than the first model—namely, because this second model can only explain the empirical data by proposing the concept of representation that is problematic. We finally concluded with a third model that seems to incorporate both intuitions of the previous two models. The main goal was to show how a spatiotemporal model of the brain can well account for the perceptual experience in both (i) the good cases and (ii) the bad cases. More detailed research needs to be done in order to sketch in specific how this model works, something that we hope to achieve with a further upcoming investigation.

Notes 1. Our version of the Müller-Lyer illusion. 2. In the case of auditory hallucination, there is solid empirical evidence for increased resting-state activity in the auditory cortex (Northoff, 2014d;

60  Steven S. Gouveia and Georg Northoff Northoff and Quin, 2011). This activity may lead to the constitution of unusually strong predicted inputs which are not impacted by any actual input, giving the internally generated predicted input more influence than the actual input with the result of the subject hearing the auditory hallucinations. So, in general, a schizophrenic’s brain confuses the predicted input with the actual input, ignoring the externally generated input of the world and focusing on the internal predicted input, something that is also empirically supported (Northoff, 2014; Northoff and Quin, 2011). 3. The DMN includes various regions: the anterior and posterior cortical midline structures as the lateral parietal córtex and the hippocampus (cf. Quin and Northoff, 2011). 4. The “resting state” can be defined operationally as the absence of specific stimuli (e.g., during eyes closed). (cf. Northoff and Quin, 2011). 5. A detailed analysis of the concept of delusion can be found in Gonçalves (2018). 6. Recreation of Searle’s formalization (cf. Searle, 2015: 22–23). 7. Nothing new here: A  similar intuition is followed by O’Regan and Noë (2001). 8. In the same line, Johnston claims that “we simply have an intensional context, not a non-existent intentional object” (Johnston, 2004: 174). 9. Entropy is a measure that matches the statistically based degree of complexity (or disorder) of the signal (as measured in fMRI) across time and space within the brain. More concretely, it measures how much the signal at one point in time and space is different from the signal obtained at the subsequent and later points in time. Therefore, one can say that entropy can be regarded as a statistically based measure of the spatiotemporal structure of the brain’s resting-state activity (cf. Northoff, 2016) 10. A short painful tactile stimulus to the index finger was used. 11. There are several methodological and meta-philosophical assumptions here that we cannot go into detail about in this chapter. Several papers are being prepared specifically about this issue. But a first general approach can be found in Northoff, 2014c. 12. Some are already discussed in Northoff, 2014c; Gouveia, 2018. 13. Fore more examples of empirical data explained without representation, see, for example, Orlandi (2010, 2012, 2016) or Hutto (2013). 14. See, for instance, Thagard, 2012 and Churchland, 2012. 15. Example of a fMRI study of the brain. Retrieved from Quin et al., 2018. 16. The specific choice of selection is still unknown and needs further research. 17. Again, this is not new: Some embodied theorists have argued for that for a while now: “Radical embodied cognitive science, very roughly, is the thesis that cognition is to be described in terms of agent-environment dynamics, and not in terms of computation and representation” (Chemero, 2009). Those “environment dynamics” are what we call the spatiotemporal features of the third model here presented. 18. Although others think that classical representation has lost its explanatory power already—at least in the context of a post-cognitivist theory of the mind (cf. Ramsey, 2007).

References Anscombe, Elizabeth (1965) “The Intentionality of Sensation: A  Grammatical Feature” in Analytic Philosophy (ed. Ronald J. Butler), Oxford: Blackwell. Brooks, Rodney (1991) “Intelligence Without Representation” in Artificial Intelligence, 44, 139–160.

A Neurophilosophical Approach to Perception 61 Campbell, John (2002) Reference and Consciousness, Oxford: Oxford University Press. Chemero, Anthony (2009) Radical Embodied Cognitive Science, Cambridge, MA: MIT Press. Churchland, Paul (2012) Plato’s Camera: How the Physical Brain Captures a Landscape of Abstract Universals, Cambridge, MA: MIT Press. Clark, Andy and Grush, Rick (1999) “Towards a Cognitive Robotics” in Adaptive Behavior, 7 (1), 5–16. Crane, Tim (2005) “What Is the Problem of Perception?” in Synthesis Philosophica, 20, 237–264. Duncan, Niall et al. (2015) “Negative Childhood Experiences Alter a Prefrontal-insularmotor Cortical Network in Healthy Adults: A Multimodal rsfMRIfMRI-MRS-dMRI Study” in Human Brain Mapping, 36 (11), 4622–4437. Fodor, Jerry (1981) Representations, Cambridge, MA: MIT Press. Foster, John (2000) The Nature of Perception, Oxford: Oxford University Press. Gallagher, Shaun (2005) How the Body Shapes the Mind, Oxford: Oxford University Press. Ganzetti, Marco and Mantini, Dante (2013) “Functional Connectivity and Oscillatory Neuronal Activity in the Resting Human Brain” in Neuroscience, 240, 297–309. Gonçalves, Jorge (2018) “Why Are Delusions Pathological?” in Schizophrenia and Common Sense Explaining the Relation Between Madness and Social Values (ed. Inês Hipólito, Jorge Gonçalves and João G. Pereira), Cham, SL: Springer. Gouveia, Steven S. (2018) Philosophical Reflections: Art, Mind and Justice, Famalicão: Editora Humus [Portuguese Edition]. Hohwy, Jakob (2013) The Predictive Mind, Oxford: Oxford University Press. Hohwy, Jakob (2014) “The Self-evidencing Brain” in Noûs, 50 (2), 259–285. Huang, Zhen et al. (2016) “Decoupled Temporal Variability and Signal Synchronization of Spontaneous Brain Activity in Loss of Consciousness: An fMRI Study in Anesthesia” in Neuroimage, 124 (Pt A), 693–703. Hutto, Daniel (2013) “Exorcising Action-oriented Representations: Ridding Cognitive Science of its Nazgûl” in Adaptive Behavior, 21 (3), 142–150. Johnston, Mark (2004) “The Obscure Object of Hallucination” in Philosophical Studies, 120 (1–3), 113–183. Johnston, Mark (2007) “Objective Mind and the Objectivity of Our Minds” in Philosophy and Phenomenological Research, 75 (2), 233–268. Klein, Colin (2014) “The Brain at Rest: What It Is Doing and Why That Matters” in Philosophy of Science, 81 (5), 974–985. Logothetis, Nikos et  al. (2009) “How Not to Study Spontaneous activity” in Neuroimage, 45 (4), 1080–1089. Martin, Michael (1997) “The Reality of Appearances” in Thought and Ontology (ed. M. Sainsbury), Milan: Franco Angeli. Moore, G. E. (1953) Some Main Problems of Philosophy, London: George, Allen and Unwin. Northoff, Georg (2012) “Psychoanalysis and the Brain—Why Did Freud Abandon Neuroscience?” in Frontiers of Psychology, 3, 71. Northoff, Georg (2014a) Unlocking the Brain I: Coding, Oxford: Oxford University Press. Northoff, Georg (2014b) Unlocking the Brain II: Consciousness, Oxford: Oxford University Press.

62  Steven S. Gouveia and Georg Northoff Northoff, Georg (2014c) Minding the Brain, New York: Palgrave Macmillan. Northoff, Georg (2014d) “Are Auditory Hallucinations Related to the Brain’s Resting State Activity? A ‘Neurophenomenal Resting State Hypothesis’ ” in Clinical Psychopharmacology and Neuroscience, 12 (3), 189–195. Northoff, Georg (2015) “Is Schizophrenia a Spatiotemporal Disorder of the Brain’s Resting State?” in World Psychiatry, 14 (1), 34–35. Northoff Georg (2016a) “Is the Self a Higher-order of Fundamental Funciton of the Brain? The ‘basis model of self-specifity’ and its Encoding by the Brain’s Spontaneous Activity” in Journal of Cognitive Neuroscience, 7 (1–4), 203–222. Northoff, Georg (2016b) “Neuroscience and Whitehead I: Neuro-ecological Model of Brain” in Axiomathes, 26 (3), 219–252. Northoff, Georg (2016c) “Neuroscience and Whitehead II: Process-Based Ontology of Brain” in Axiomathes, 26 (3), 253–277. Northoff, Georg (2018) The Spontaneous Brain: From the Mind—Body to the World—Brain Problem, Cambridge, MA: MIT Press. Northoff, Georg (forthcoming) “Is Perception Neuro Temporal and Neuro-Ecological?—Whitehead’s Non-traditional Theory of Perception” in Consciousness: Bridging the Gap between Science and Philosophy (ed. Alex Gomez-Marin and Rod Hemsell), Anoka, MN: Process Century Press. Northoff, Georg and Quin, Pengmim (2011) “How Can the Brain’s Resting State Activity Generate Hallucinations? A ‘resting state hypothesis’ of Auditory Verbal Hallucinations” in Schizophrenia Research, 127 (2011), 202–214. Northoff, Georg et al. (2010) “Rest-stimulus Interaction in the Brain: A Review” in Trends in Neuroscience, 33 (6), 277–284. O’Regan, Kevin and Noë, Alva (2001) “A Sensorimotor Account of Vision and Visual Consciousness” in Behavioral and Brain Sciences, 24 (5), 883–917. Orlandi, Nico (2010) “Are Sensory Properties Represented in Perceptual Experience?” in Philosophical Psychology, 23 (6), 721–740. Orlandi, Nico (2012) “Embedded Seeing-as: Multi-stable Visual Perception without Interpretation” in Philosophical Psychology, 25 (4), 1–19. Orlandi, Nico (2016) “Bayesian Perception Is Ecological Perception” in Philosophical Topics, 44 (2), 327–351. Quin, Pengmim and Northoff, Georg (2011) “How Is Our Self Related to Midline Regions and the Default-mode Network?” in NeuroImage, 57 (2011), 1221–1233. Quin, Pengmin et al. (2018) “Vascular-metabolic and GABAergic Inhibitory Correlates of Neural Variability Modulation. A Combined fMRI and PET Study” in Neuroscience, 379, 142–151. Raichle, Markus (2010) “Two Views of Brain Function” in Trends in Cognitive Science, 14 (4), 180–190. Raichle, Markus (2015) “The Restless Brain: How Intrinsic Activity Organizes Brain Function” in Philosophical Transactions of the Royal Society B: Biological Sciences, 370 (1668). Ramsey, William M. (2007) Representation Reconsidered, Cambridge: Cambridge University Press. Sadaghinani, Sepideh et al. (2009) “Distributed and Antagonistic Contributions of Ongoing Activity Fluctuations to Auditory Stimulus Detection” in Journal of Neuroscience, 29 (42).

A Neurophilosophical Approach to Perception 63 Sadaghiani, Sepideh et al. (2015) “Ongoing Dynamics in Large-scale Functional Connectivity Predict Perception” in Proceedings of the National Academy of Sciences of the United States of America, 112 (27), 86, 8463–8468. Searle, John (2015) Seeing Things as They Are, Oxford: Oxford University Press. Snowdon, Paul (2007) “G. E. Moore on Sense-Data and Perception” in Themes From G. E. Moore: New Essays in Epistemology and Ethics (ed. S. Nuccetelli and G. Seay), Gloucestershire: Clarendon Press. Thagard, Paul (2012) Cognitive Science of Science: Explanation, Discovry and Conceptual Change, Cambridge, MA: MIT Press. Varela, Francisco, Thompson, Evan and Rosch, Eleanor (1991) The Embodied Mind, Cambridge, MA: MIT Press.

4 Smelling Molecular Structure Benjamin D. Young

1. Introduction Smells are mysterious entities.1 They seemingly arise inside our nose, providing a private phenomenological experience, yet at the same time we acknowledge that smells are external entities carried upon the wind. We talk about smells as objective entities of shared attention but are not sure of their ontological status. While the chemoreceptive sciences are still researching the mechanisms of sensory transduction in olfaction, minimally there is consensus that we perceive chemical compounds. The exact property of these compounds is still under investigation, as well as how these yield our sensory experience of what things smell like. Nonetheless few would disagree with the statement that using olfaction we perceive chemical compounds composing gaseous clouds—that is, few except philosophers. In the last decade, there has been a rapid development of philosophical theories regarding the nature of smell.2 This chapter assesses the current philosophical debate about smells. The theories to be covered are: Naïve Realism, Odor Theories, Stuff Theories, Non-Objectivist Theories, Process Theory and Molecular Structure Theory.3 The theories will be evaluated based on their explanations of (a) the olfactory quality of a smell, (b) smells as distal entities and (c) our experience of smells as intentional objects,4 which are arguably the three central issues in answering what are smells.5 The chapter concludes with a defense of Molecular Structure Theory that demonstrates its strengths in accounting for each of these three aspects of smell.

2.  Naïve Realism Our pretheoretic conception of smells, predicated upon ordinary language descriptions of olfactory experience, is that smells are the emanations of ordinary objects. According to Naïve Realism, what we smell is the distal ordinary object by means of its effluvia becoming lodged in our nostrils. What we represent as the intentional object of olfactory experience is the ordinary object from which the gaseous odor plume

Smelling Molecular Structure 65 emanates. The olfactory quality of the effluvia is determined by the source object. Lastly, smells can be distal entities that become lodged in our nose. Despite its intuitive nature, the theory does not provide a viable explanation of olfactory quality and the intentional object of olfactory experience. The Naïve Realists’ account of the intentional object of olfaction has been criticized based on the different persistence conditions between ordinary objects and their smells6 and issues concerning illusory content and misrepresentations.7 Additionally, the theory’s explanation of olfactory quality has been criticized based on the naïvety of assuming that all of the chemicals composing the ordinary object are necessary components for generating the olfactory quality of smells.8 Lastly, Naïve Realism faces difficulty in assessing the identity of the distal source of a smell given its dispersal rates and persistence conditions, as well as the ability of the olfactory quality of a smell to inhere in objects other than the ordinary objects from which it emanated.9 Thus, despite its intuitive appeal, the Naïve view is inadequate.

3.  Odor Theories A large number of philosophers who discuss olfaction argue that smells are odors.10 Contemporary Odor Theories are an outgrowth of Plato’s Vapor theory (Timaeus 66d-67). Odor Theories assert the weaker claim that smells are the odorous subset of the gaseous effluvia of ordinary objects. The external objects of olfactory perception are the chemical odor clouds given off by ordinary objects. Odors identified as gaseous chemical clouds are curious perceptual objects. The clouds’ spatiotemporal boundaries are less truncated than those of ordinary objects identified by vision and touch. Since only diffused subparts of the object are required to elicit a smell, Odor Theory has the resources to explain veridical olfactory perception when the ordinary object is not present in the immediate surroundings as well as the possibility of olfactory misrepresentation The three most developed positions within Odor Theory have been proposed by Batty, Lycan and Richardson. Each theory endorses the central tenet that the intentional object of olfactory experience are odors. However, they differ in their account of the external object of olfactory perception and the spatiotemporal nature of odors. Odor Theory is enticing, but it faces difficulties both in generating individuation conditions of odors and in giving an account of what determines olfactory quality, thus for brevity these inadequacies will be left aside in the discussion of each theory.11 3.1  The Abstract Theory of Odors The Abstract Theory is defended by Batty over the course of multiple arguments concerning the veridicality of odor perception,12 the

66  Benjamin D. Young individuation of multiple odors within an array,13 and the multiple properties ­problem.14 The object of olfactory experience is not of an odor which is a non-spatial non-objective property of environment. Thus, the intentional object of olfactory experience is not as of a particular entity composed by a gaseous cloud. Rather, an odor is a property of the environment. To correctly describe the representational nature of our olfactory experiences we must make use of an existential quantification that there exists some smell hereabouts in our environment.15 The Abstract Theory is primarily constructed to handle the intentional object of olfactory experience using our synchronic phenomenological experience of the distal nature of smells. However, Batty’s focus on the phenomenology of synchronic experiences of smells generates the overarching claim that olfaction does not present locatable entities with fixed spatial locations. While the theory concedes that olfactory experiences have spatial aspects to them, the olfactory object does not have spatial properties. However, there are good reasons to doubt that we do not perceive objects with properties in olfaction and moreover that these do not have spatial properties.16 The theory provides a strong explanation of the intentional object of olfactory experience synchronically conceived but does not provide an explanation of the olfactory quality of smells or an account of our distal perception of smells across time. 3.2  Odor Objects and Representational Layering The original odor theory proposed by Lycan claims that odors are not experienced as having spatial properties, such that they are not presented as being at a fixed spatial location.17 But he disagrees that they are properties of the environment. For Lycan, the intentional objects of olfactory experience are odors, which are objective entities composed by a gaseous chemical cloud. The primary object that is represented in intentional olfactory experiences is a non-spatial objective entity. More recently, Lycan has further distanced himself from other forms of Odor Theory by arguing for a layering account of olfactory experience according to which we experience odors, but there can be multiple layers of intentional object representation.18 For example, though we experience odors, we also might attribute the odor’s smell to an ordinary object somewhere in the vicinity. Despite endorsing an objectivist perspective on the olfactory object as external to the perceiver, Lycan’s view has difficulties generating an adequate explanation of olfactory quality and the distal nature of diachronic olfactory experience. Moreover, Lycan denies that the olfactory object as a gaseous cloud presents us with spatial properties of the smell.19 His argument depends upon our agreeing with his account of the synchronic experience of the odorous cloud’s vague boundaries not presenting us phenomenologically with spatial locatedness. However, even if his claims

Smelling Molecular Structure 67 regarding synchronic experiences are granted, the theory will not do justice to our diachronic experience of mereologically complex olfactory objects within an overlapping smellscape. For these situations, Lycan’s layering approach might be of help, since our background knowledge and conceptual olfactory resources might enhance the meager nature of synchronic olfactory experience. Nevertheless, without a fuller account of the olfactory quality of an odor and some manner of how we determine the spatial extent of these distal entities across time, Lycan’s theory only provides us with a partial explanation of the distal nature of smells and the intentional object of olfactory experience. 3.3  Exteroceptive Odorous Objects The Exteroceptive approach straddles the boundary between Naïve Realism and Odor Theory. Richardson argues in defense of the commonsense view that smells are objective entities with spatial properties.20 Her driving claim is that olfaction is an exteroceptive sense used to perceive spatial entities. Smells present us with spatial entities located in exteroceptive space, but the objective and spatial nature of smells depends on the active movement of airflow in the nostrils. The Exteroceptive theory of smell is crafted to account for the intentional object of olfaction as determined by our olfactory phenomenology. Her theory is that we perceive odors as distal entities. Despite being distal external entities, odors are not synchronically presented at a specific distance or direction from the subject. Rather, the olfactory object is experienced as qualitatively being pulled in through the nostril. Enlarging the range of olfactory qualities inherent in the olfactory experience to include the somatosensory sensation of airflow in the nostrils requires expanding the olfactory modality to include everything that goes on inside the nostrils. However, it has been argued elsewhere that if we begin by assuming that smells must have an olfactory quality, then there is good reason for excluding these subsystems as part of the olfactory experience, because they are not necessary for generating olfactory quality.21 Even if we expand the olfactory modality to include motorsensory and trigeminal sensations, it is still far from clear how the olfactory quality of an odor is derived. Moreover, the addition of trigeminal and somatosensory stimulation facilitates the localization of smells synchronically either intranasally22 or within external space.23 Thus, Richardson’s claim that synchronically the olfactory object does not present us with a spatial entity at a distance or direction from us, becomes empirically questionable. The Exteroceptive theory progresses the debate by arguing that the intentional object of experience (diachronically considered) is of a particular entity with qualities including spatial properties, but its explanation of synchronic distal olfactory perception is debatable.

68  Benjamin D. Young

4.  Smell as the Property of Stuff The Stuff Theory of smell proposed by Mizrahi maintains the naïve realist position that smells inhere in the ordinary objects and their effluvia.24 Smells are both identifiable with properties of the ordinary objects from which they emanate and properties of the gaseous plume. To account for the duality of smells, Mizrahi requires that we recognize a new category of stuff within our ontological framework. While particulars are objective individual entities, stuff is considered to be the constituent matter. According to Mizrahi’s account of our phenomenology, we smell both the ordinary object and the gaseous odor, but the only way to capture the dual existential nature of our pretheoretic descriptions of smells is to embrace an ontology of stuff. While it would seem that stuff is identified by our use of mass noun terms and commonsense phenomenology, a closer reading of Mizrahi’s argument is that stuff has an inner structure which explains its dispositional properties. Stuff is the ontological categorical basis required for us to understand the chemical nature of smells. She does not fully elaborate on the ontological nature of stuff, but a closer look at the theorists cited reveals that they are primarily concerned with generating an ontological category that is proprietary to chemistry.2526 Properly understood, Stuff Theory’s account of the duality of smells amounts to the following two claims. First, we can identify that the smell is within the ordinary object, because the ordinary object has a given chemical composition. Second, we experience a gaseous cloud as having a given olfactory quality, because it is composed of a given set of chemicals. Further issues arise when considering that stuff has the property of homeomerosity. However, each sample of the odorous cloud will not necessarily have the same olfactory quality, given that it might have the same concentration levels. Not only does positing stuff not generate any explanatory gains but it also makes matters worse with its claimed mereological simplicity. For example, chemical compounds often have identifiable functional groups that provide a predictable olfactory quality such that the entire stuff-like entity is not mereologically simple.27 A further problem is that in demarcating the edge of a given odor plume one must track the concentration gradient within the stuff. Yet if stuff is wholly present in each one of its component parts, then it seems odd to think that we would have a diffused gradient instead of something being fully present at each instance of the smelly stuff. Mizrahi’s account is structured around the motivation to account for our phenomenological experience of smells both in the air and at the source, yet olfactory quality is not dealt with. The theory does not explain why one kind of stuff has a given smell as opposed to another kind of stuff. However, the theory allows that the distal nature of smells is of properties of objects that are mind-independent and external to the

Smelling Molecular Structure 69 perceiver. Additionally, Mizrahi is able to justify our naïve intuitions that smells are both located in the air and ordinary objects. Moreover, the new ontological category allows us to track the stuff dispersed in the air back to their original sources. Given its own posits, the theory does capture the duality of smell by adverting to some more basic category of chemical composition. But identifying stuff requires evoking chemical structures; the nature of stuff does not seem to be at bottom a commonsense phenomenological category but something more readily identifiable and individuated by chemists. Stuff Theory does not supersede explaining the nature of smell using the molecular structure of chemical compounds, hence the proposed new ontology generates no further epistemic purchase.

5.  Non-Objective Theories of the Olfactory Object In contrast to the previous theories that primarily engaged with naïve realism, an interesting alternative is to jettison the object centric approach. The three theories in this section share the view that the intentional object of olfactory experience should not be conceived of as representing chemical stimuli as objects within the environment. However, each theory offers its own innovative research program concerning the function of olfactory perception and the ontological commitments required to fully account for the nature of smells. 5.1  Smells as Affective Categories The non-objectivist theory proposed by Castro and Seeley claims that the olfactory object is generated by affective categorizations. Olfactory experiences are not about a particular set of stimuli in the environment.28 Rather, olfactory objects are tracked by the olfactory system in accordance with their affective categories for generating ecologically significant behaviors. What we represent as the intentional object of olfactory experience are affective categories of ecological valuable kinds. The affective categories coarsely specify the dimensions of the olfactory stimulants relative to their biological value for the organism. Despite the label, the affective categories do not correspond merely to hedonics or perceived pleasantness, as they are a multidimensional set of categories meant to explain the mind-dependent realization of the olfactory object. Castro’s research on the categorical dimensions of human odor description space using affective responses serves as a major piece of evidence in favor of the theory.29 However, more parsimonious theories for demarcating olfactory quality spaces have been proposed using judgments of just noticeable difference.30 Starting with our consciously reported phenomenology of smells, Castro and Seeley correctly note that we do not report our olfactory experiences as being of chemical structures, yet it is not clear that any

70  Benjamin D. Young objectivist theory endorses such a claim. Additionally, they employ evidence for coarse coding of molecular features in the olfactory system in support of the claim that we do not wholly perceive the molecular structure of chemical compounds. While it was commonly thought that odorants are coded in a coarse manner at the receptor and olfactory bulb, Vincis et  al. recorded robust fine-grained representations of odorants within the glomeruli of the olfactory bulb under more natural conditions using ordinary odorants at varying levels of concentration.31 Moreover, it is unclear how the encoding evidence is meant to bear upon the conscious experience of the intentional object. They offer further support for the theory based on the claim that the primary determinant of odor identity is its property of valence. Drawing on a series of studies, Yeshurun and Sobel claim that valence is the perceptible property that determinates odor identity instead of olfactory quality.32 For instance, there is greater cross-cultural agreement on the categorization and identification of odors in respect of judged properties of pleasantness than olfactory quality.33 Moreover, a computational model of the olfactory object can predict olfactory valence from chemical structure alone.34 However, a recent set of studies indicates that the object of olfactory perception is more likely identified by humans in terms of its olfactory quality.35 Additionally, Kumar created an alternative computation model using descriptors of qualities and not judgments of valence, as well as measures of psychochemical structures to predict olfactory quality.36 The theory attempts to account for the olfactory quality of smells at the conscious level of experience. Olfactory quality is determined by the affective categories that the organism generates in encoding odor stimuli that are instrumental in the guidance of ecological behavior. Given their theoretical starting point of accounting for conscious reports of olfactory experience, they exclude nonconscious perception of olfactory qualities. But this is a highly dubious assumption given that most olfactory experiences occur in the absence of awareness.37 Also, it is arguably the case that conscious awareness is not required for our perception of olfactory quality.38 Additionally, olfactory experiences generate behavior even in the absence of conscious awareness.39 Castro and Seeley do not account for distal nature of smells, as they are primarily concerned with the representational nature of the intentional content of olfactory experience. The intentional objects of olfactory experiences are mind-dependent entities generated from higher-level processing. Their theory is an interesting account of the intentional object of olfactory experience if one is only interested in accounting for conscious phenomenology. 5.2  Quality Space and Olfaction’s Role in Guiding Behavior Keller offers the most comprehensive treatment of the philosophy of olfaction centering around the claim that olfaction is designed for the

Smelling Molecular Structure 71 determination of behavioral output.40 Olfactory perception should not be conceived of as generating accurate representations of external chemical stimuli, because the purpose of olfactory function is the detection of salient entities for behavioral output. His theory is similar to Castro and Seeley’s, yet it differs in its determination of olfactory quality, which is ascertained using a triadic relation of just noticeable difference judgments of olfactory stimuli. Thus, the qualitative character of an odor is generated by its holistic placement in the mental quality space of each subject. Regarding the distal nature of smells, Keller argues that olfactory qualities do not have spatial properties, because we do not have spatial properties presented to us as part of the olfactory quality. In those instances where it seems that olfaction provides us with spatial information, he rightly notes that these might be attributed to chemesthesis. His arguments derive from a review of experimental evidence on olfactory spatial abilities and tracking. However, there are some empirical studies that arguably show that olfactory experience might be spatial. The binaural rivalry between nostrils yields a slightly different smell percept for each nostril,41 yet this does not yield odor localization. The host of chemoreceptive research indicates that humans cannot use intranasal onset to determine olfactory distal perception.42 However, Negoias et al. demonstrated that individuals can be trained to localized odorants’ onset within a nostril synchronically.43 Their study provides a strong reason to think that we do have the capacity for synchronic odor localization, but it requires attention and training. Additionally, in open field environments subjects can localize a distal object using just the olfactory system without trigeminal stimulation. The use of both nostrils did not make a difference for locating the stimulus at two meters or less from the subject, but beyond two meters the use of both nostrils to demarcate the concentration of the plume seems to be of importance in the localization of the smell.44 Keller entertains the claim that the intention object of olfactory experience could be determined in terms of the objects that are phenomenally present in the experience.45 What is phenomenally presented to us in olfactory perception depends upon background knowledge, including our tracking the placement of odorous ordinary objects in the environment, which he finds problematic because the olfactory object will differ between perceivers. Moreover, based on the olfactory system’s overall anatomical connectivity, he argues that tabled possibility of conceiving of the intentional object of olfactory experience as what is phenomenally present in experience misunderstands the function of olfaction, which is designed for detecting chemical stimuli for the guidance of behavior. His quality space theory generates an account of olfactory quality based on triadic subjective judgments of just noticeable differences between stimuli. Olfactory quality is determined in accordance with its placement within the holistic mental quality space. Chemical stimuli do not have a smell on their own independent of the quality space. Additionally, based

72  Benjamin D. Young upon experimental evidence, he carefully argues that smells do not have spatial properties even though we can use them for navigation. As such, smells are not distal entities on his theory—they inhabit spaces as a property of the environment. Olfaction is not meant to represent the environment, nor is it meant to generate intentional representational objects of experience. Rather, olfactory perception is meant to influence behavior based on the hedonic properties of an odor. 5.3  Process Theory of Olfaction The process theory of olfactory perception proposed by Barwich challenges traditional philosophical conceptions of the olfactory object.46 She argues that traditional approaches incorrectly assume that we can explain the nature of olfactory perception in terms of distal entities in the environment that drive perceptual processes within the individual. Rather, the intentional object of olfactory perception requires an entirely different metaphysical framework of processes. By jettisoning the stimulus-driven approach with its emphasis on the independence of perceptual object and the perceiver, the process ontology provides greater explanatory power in accounting for the nuanced nature of olfactory processing. From an epistemic standpoint, the process view is rather appealing. The theory’s strength derives from acknowledging that the stimuli transduction and neural encoding of odor stimuli within the olfactory system require both top-down and bottom-up processing.47 While most philosophical views propose that we smell things in the environment, which we represent as the objects of olfaction, the process view focuses on how anticipatory effects and background knowledge conditions play a role in the determination of olfactory experience.48 Olfactory processing is not a simple feedforward sensory system whereby the stimulus determines the complete nature of its perceived qualities. Ongoing dynamic processing across multiple sensory and cortical levels of the olfactory system play a role in generating how the sensory stimulus is perceived. Background conditions and top-down processing within the olfactory system mold how an odor is constituted from a host of odorants. For instance, naïve animals treat the same chemical stimulus at varying concentration levels as having different olfactory qualities.49 However, organisms learn concentration invariance (i.e., the shift in concentration of the overall odor does not change its identity) by tracking the concentration of the components50 or by learning that the ratio between components remains constant.51 The process view is well argued for and on firm empirical footing, but jettisoning our traditional ontological framework of particulars, objects, states and properties together with their interrelationships in favor of a process ontology is unmotivated. Not only is the notion of what

Smelling Molecular Structure 73 determines a process vague, but it is unclear that a process ontology generates an explanatory purchase beyond that of our current models. The greatest problem is that the notion of a process as derived from the neuroscientific literature is conflated with the notion of a perceptual process phenomenologically individuated. Moreover, the very concept of a process is never adequately defined in a manner that lets us determine how this contrastive notion is meant to enhance our understanding of olfactory perception beyond that of our current ontological commitments. We can reconstruct our current theories in a critical manner to account for processes, yet what needs to be shown is that the new process ontology constructively generates explanatory purchase over and above that of the original model or system.52 Embracing a process ontology requires that we consider processes to be non-reductive dynamic entities that are both productive and that sustain continuity.53 Given this theoretical starting point, it seems fair to ask how the same olfactory processes realize similar if not the same perceptions of olfactory quality. One specific advantage of the objectivist ontology is that by considering reality to be constructed of discrete entities allows for symbolization and quantification, which thereby generate interpersonal predictions and generalizations. Without a clear delineation of what an olfactory process is across individuals, the processes account will not yield generalizability for the stability of interpersonal and intrapersonal judgments of olfactory quality, intensity or hedonic of smells. Not only does this place the theory at a disadvantage but it does not imbue the approach with any explanatory purchase beyond that of the current scholastic accounts. Process theory’s account of olfactory quality, smell’s distal nature and the intentional object of experience show promise. A  smell’s olfactory quality is determined by an interactive process between the stimulus and top-down cognitive and perceptual processes within the olfactory system. How these processes determine olfactory quality will require future clarification based on empirical investigation, which is a strength of any empirically informed theory in philosophy. The process account clearly holds that olfactory quality is not within the stimulus-driven object. The chemical structures do not have a smell, since olfactory qualities are determined by perceiver-relative processes. In contrast with previous theories that the olfactory objects are distal entities, Barwich argues that the olfactory system is not designed to accurately represent external particulars. The olfactory system tracks information that is instrumental in the guidance of ecological behavior. Lastly, the intentional objects of experience, are generated through an interactive process between the perceiver and the chemical stimulus. There is a blurring of the boundaries between the perceived and the perceiver, which should allow for the intentional object of experience to be a process between top-down anticipatory effects and chemical stimuli encoding.

74  Benjamin D. Young

6.  Molecular Structure Theory According to Molecular Structure Theory (MST), the chemical nature of the stimuli is responsible for our experience of smells. Our experiences represent smells as distal environmental entities with spatial properties such that what determines the olfactory quality and the spatial temporal nature of smells is some property of the chemical structures within the gaseous plumes. Carvalho proposes a precursor of molecular structure theory according to which olfactory objects are chemical structures.54 These chemical structures determine the type of odor that we perceive and generate the token representation of the odor that is the olfactory object of intentional experience. He argues that the existential quantification posited by Batty to explain our experience that there is a smell hereabouts as a property of the environment does not adequately capture our experience of smells. The nuanced temporal development of our experience of smell requires the existence of a spatiotemporally extended object to which properties may be predicated. Additionally, he demonstrates, contrary to Burge55 that there are constancy effects in olfactory perception such that a change in the proximal stimulus does not necessarily yield a change in the olfactory quality of the distal stimulus. The ramifications of his view are that earlier odor theories are incorrect both in their denial of olfactory objects as particulars and their withholding spatial properties from olfactory experiences.56 MST improves upon chemical structure theory by focusing upon the spatial nature of the olfactory plume. The olfactory quality of a given smell, the spatial and temporal boundaries of the distal object of smell and the intentional object of olfactory experience are determined by the molecular structure of chemical compounds within odor plumes.57 According to MST, olfactory quality is determined by the molecular structure of chemical compounds. However, the plume plays a role in the determination of olfactory quality and the spatial extent of the olfactory object. The chemical compounds determine the distal nature of token odor plumes, given their concentration gradients. Additionally, the interaction effects between the different kinds of molecular structures within the plume also play a determinate role in generating olfactory quality. The initial statement of the theory was primarily constructed to account for simple odor mixtures and monomolecular odorants,58 but when considering natural environments, we must consider olfactory mixtures composed of multiple types of chemical compounds. Determining the spatial and distal nature of the olfactory object requires accounting for the plume structure of a given odor within a turbulent sea of overlapping chemical currents. The olfactory system binds together odorants of different molecular structures to compose an overall odor object. What was missing from the initial statement of MST was a specification of the

Smelling Molecular Structure 75 intentional object of olfactory experience. What we experience as the objects of olfactory experience are odors within a smellscape.59 We experience a smellscape of complex olfactory objects that can change and shift their properties across time as against a background of other odors. To do so we employ background knowledge to generate the composition of olfactory mixtures in terms of their groupings. One of the strengths of MST is that it can advert to the olfactory system’s capacity to encode the molar ratios between the components of a given olfactory mixture, the concentration rates and ratios between odorants and overall concentration rates of the key components of complex mixtures. As such, MST can account for the persistence conditions, the mereological identity conditions and individuation conditions of an odor in terms of the molecular structure of chemical compounds that compose the complex mixture. According to MST, it is not possible for humans to have veridical perception of olfactory quality in the absence of stimulation, using chemical odorants. Empirically, MST’s claim is supported by experimental manipulation of the olfactory system using electrical stimulation. In some mammals, it has been shown that stimulating the olfactory epithelium electronically yields stimulation of the olfactory bulb and cortex, though it is unclear if this yields perceptible olfactory qualities.60 Additionally, one recent study generated the experience of olfactory quality in epileptic patients using cortical stimulation induced by subdural electrodes, which they equated with olfactory hallucinations.61 However, to the best of my knowledge no study has yet shown that electronic stimulation of the olfactory sensory system in humans generates experience of olfactory quality.62 63 These findings might not constitute definitive evidence for MST, as they are predicated upon the absence of results, but they are predicted by it. Molecular structure is superior to the alternative approaches, because it comprehensively explains the nature of olfactory quality, the distal nature of smells and the intentional content of olfactory experience in a coherent and systematic fashion. To reiterate, according to MST, the quality of a given odor is determined by the molecular structure of chemical compounds in interactions with the olfactory receptor, as well as in composing the gaseous plume. Additionally, we can individuate the distal nature of olfactory objects in a principled manner that allows for their vague boundary conditions. The overlapping plumes of odors that generate the sea of chemical currents enveloping us generates the environmental smellscape that is the intentional object of olfactory experience. Smellscapes are rather odd things to think about given our visiocentric default mode of thinking about our experience of reality. However, when thinking about smells it becomes natural to consider large-scale environments with overlapping

76  Benjamin D. Young currents that inform us of distal entities of ecological and navigation value to us as organisms.

7. Conclusion In less than a decade, the Philosophy of Smell has blossomed and expanded from just a couple of views to a host of different approaches, all aiming at explaining the nature of our olfactory experiences. I have sought to offer an overview of the debate within philosophy regarding the nature of smell by surveying Naïve Realism, Odor Theories, Stuff Theory, NonObjectivist Theories including Process Theory and Molecular Structure Theory. To properly assess the merits of each theory, three separate issues were used as a means of comparison. The three central questions that any theory of smells must handle concern the distal nature of smells, what generates olfactory quality of smells and what is the intentional object of olfactory experience. Naïve Realism was shown to be inadequate on all aspects of the nature of smell. Odor Theories strengths are built upon their criticisms of Naïve Realism and their explanation of the representational structure of the intentional object of olfactory experience. However, they fare poorly with their narrow synchronic distal object of smell and lack of engagement with the nature of the olfactory quality. Accounting for the olfactory quality of smells is also not handled by Stuff Theory. Mizrahi attempted to maintain our naïve realist conception of smells as inhering within the ordinary object and at the same time existing within the effluvia that reaches our nostrils. She offers an alternative ontological category of stuff to better explain the nature of smells experiences. In elucidating the nature of stuff, it was noted that the ontological shift is not required because stuff is equivalent to the chemical structures and posits of chemistry. The explanatory purchase provided by Stuff Theory can be supplanted by MST without abandoning the traditional ontological framework. The non-objectivist theories offer a compelling alternative. Castro and Seeley’s theory that olfactory quality is determined by affective categories did not set out to comprehensively deal with the intentional object of olfactory experience, nor the distal nature of smells. Hence, Barwich’s Process Theory and Keller’s theory offer more comprehensive non-­ objectivist models of the nature of smell. Barwich’s theory presents a promising alternative to considering the nature of olfactory perception to be object oriented and stimulus driven. However, despite the explanatory value in including top-down processing within the determinants of the olfactory experience, the ontological purchase of abandoning our objectoriented approach in favor of an ontology of processes requires further motivation. Keller offered by far the most comprehensive theory of the nature of smell, centering upon the claim that the olfactory system is

Smelling Molecular Structure 77 designed not to accurately represent distal stimuli in the environment but rather to track the ecological affordances of chemical stimuli in the guidance of behavior. Overall, his theory is well crafted and empirically well supported. However, it has been argued that the olfactory system implements an alternative representational format of non-concatenative compositional representations that do not obey the strictures assumed of a representational format assumed by Keller.64 Moreover, the alternative explanation of the olfactory systems’ representational format might challenge his use of cortical connectivity in support of the claim that the olfactory system does not have a representational format that accurately depicts entities in the environment. MST offers the most comprehensive explanation of the olfactory quality of smells, the distal nature of smells and intentional experience of smells, while at the same time being the most parsimonious in terms of its ontological commitments. The olfactory object should not be conceived as merely determined by the chemical structure of odorants. Instead, it is the molecular structure of chemical compounds within odor plumes that determines the olfactory object. MST does not require the abandonment of our ontological commitments to objects with properties. Furthermore, MST is in keeping with the object-centric stimulus-driven research program of the chemosciences. According to the theory, we smell particulars in external space that are not mind-dependent. The quality of these objects is determined in accordance with the molecular structure of the chemical compounds at the micro-level in their interactions with the olfactory receptors together with their macro-chemical properties in composing the odor plume. The distal nature of the olfactory entity can be determined by the chemical structures composing the overall plume. Lastly, the intentional objects of olfactory experiences are smellscapes composed of a multitude of plumes within the turbulent sea of chemicals we inhabit.

Notes 1. The sole focus of the chapter will be upon orthonasal olfaction. 2. The chapter will not offer historical coverage of the philosophy of smell. Brief introductions to earlier accounts of smell can be found in Perkins, Sensing the World; Young, Olfaction; Richardson “Sniffing and smelling” and Batty “Olfactory Objects.” Also, Johansen, “Aristotle on the Sense of Smell” provides a detailed assessment of the differences between Aristotle’s theory in De Anima and De Sensu. Additionally, for an introduction to the commentators’ debate over Plato and Aristotle’s theories of olfaction, see Kemp, “A  Medieval Controversy about Odor.” Additionally, McHugh’s Sandalwood and Carrion offers a historical overview of Indian Philosophy of Smell. Lastly, Wilson and Stevenson’s Learning to Smell and Barwich’s “Bending Molecules or Bending the Rules?” provide nice introductions to the history of olfactory science. 3. Subjectivism about the olfactory object, i.e., the claim that smells are unstructured sensation of odor qualities that are not representational and merely

78  Benjamin D. Young properties of experience internal to subject, will not be considered. For a more detailed discussion and criticisms, see Carvalho “Olfactory Objects.” 4. The Intentional Object question is similar to giving an account of what Pylyshyn in the first chapter of Seeing and Visualizing refers to as the “phenomenal” content of perception. 5. Cf. Young, 2016. 6. Cf. Batty, 2010c. 7. Cf. Lycan, 1996. 8. Cf. Young, 2011, 2016. 9. Cf. Keller, 2017. 10. Cf. Batty, 2007, 2010a, 2010b; Lycan, 1996, 2000, 2014; Matthen, 2007; Smith, 2002; Tye, 2000, 2002; Richardson, 2013. 11. Cf. Young, 2016. 12. Cf. Batty, 2010. 13. Batty, 2010a, 2010b, 2011. 14. Batty, 2010a, 2010b, 2014. 15. Carvalho (2014) offers detailed criticisms that the existential quantification must also include an olfactory object to account for the diachronic nature of olfactory perception of mereologically complex smells to which we predicate temporal properties. 16. Cf. Richardson, 2013; Carvalho, 2014; Mizrahi, 2014; Young, 2016. 17. Cf. Lycan, 1996, 2000, 2014. 18. Cf. Lycan, 2014. 19. For criticisms see Carvalho, 2014; Young, 2016. 20. Cf. Richardson, 2013. 21. Cf. Young, 2017. 22. Cf. Frasnelli et al., 2009, 2010. 23. Cf. von Bekesy, 1964. 24. Cf. Mizrahi, 2014. 25. Soentgen (2008) claims that the phenomenological experience of reality allows us to categorize things not necessarily as particulars, but as groups of stuff. Yet, ideally the notion is supposed to correspond with our ontological posits in chemistry. The general problem with this account is that it is meant to be a commonsense account given our semantic and linguistic usages, yet all the examples that theoretically motivate positing stuff as an ontological category are derived from chemistry. 26. Lewowicz and Lombardi (2013) offer further reasons for why we must recognize the ontology of stuff based on the linguistic structures of our ordinary language, whereby we identify groups of categories as being of a given kind of stuff. However, the primary reason for positing stuff is to account for the nature of chemical reality. According to Lewowicz and Lombardi, we must have a macro-chemical ontology of stuff together with a micro-structural identification of reality at the level of physics. 27. Cf. Klopping, 1971. 28. Cf. Castro and Seeley, 2014. 29. Cf. Castro et al., 2013. 30. Cf. Young et al., 2014; Keller, 2017. 31. Cf. Vincis et al., 2012. 32. Cf. Yeshurun and Sobel, 2010. 33. Cf. Haddad R. Weiss et al., 2008, 2010. 34. Cf. Snitz et al., 2013. 35. Cf. Olofsson et al., 2012; 36. Cf. Kumar et al., 2015.

Smelling Molecular Structure 79 7. Cf. Sela and Sobel, 2010. 3 38. Cf. Young et al., 2014; Keller, 2017. 39. For a short review, see Young, 2014b. 40. Cf. Keller, 2017. 41. Cf. Zhou and Chen, 2009. 42. Cf. Kobal and Hummel, 1998; Radil and Wysocki, 1988; Frasnelli et  al., 2009, 2010; Kleemann et al., 2009. 43. Cf. Negoias et al., 2013. 44. Cf. Welge-Lussen et al., 2014. 45. Cf. Budek and Farkas, 2014. 46. Cf. Barwich, 2014, 2016, 2018. 47. Cf. Barwich, 2014. 48. Cf. Barwich, 2018. 49. Cf. Cleland et al., 2012. 50. Cf. Le Barre et al., 2008; Sinding et al., 2013, 2014. 51. Cf. Cleland et al., 2012; Uchida and Mainen, 2009. 52. Cf. Seibt, 2005. 53. Cf. Seibt, 2005; Simons, 2000. 54. Cf. Carvalho, 2014. 55. Cf. Burge, 2010. 56. A further aspect of Carvalho’s approach that runs counter to previous theories is that the olfactory objects have distal and temporal constancies such that we have an olfactory psychological ability figure-ground segregation. 57. Cf. Young, 2016. 58. Ibid. 59. The idea of smellscapes is certainly not new, as distal olfactory perception has a long history of debate going back to Plato and Aristotle through to their Medieval Commentators. Additionally, Indian Philosophy contains a rich treatment of olfactory navigation and discussions of garden design to elicit a smellscape (for an overview, see McHugh, Sandalwood and Carrion). More recent discussions can be found in Papi, Animal Homing. Papi poetically writes about smells occurring within a distal array as a mosaic of odor patches. Gatty in Finding Your Way on Land or Sea uses similar terminology when discussing seabirds who use a sensory array of olfactory objects for navigating both at sea and on land. However, the contemporary coinage of smellscape must be attributed to Porteous, who introduces the term to capture our ability to navigate and remember olfactory environments. Porteous’s research shows that smells are not randomly distributed; determining the location of smells requires accounting for the odor’s current position given the smells’ odorous concentration gradient and the wind patterns in the environment. Similarly, Roadaway in Sensuous Geographies examines how olfaction allows interactive navigation through an environment of sensory geographies. Moreover, there are long-standing and ongoing research projects devoted to demonstrating that olfactory perception provides navigational accuracy in using odor gradients within a coordinate space to navigate an environment (Wallraff, “Avian olfactory navigation,” “Avian navigation,” “Ratios among atmospheric trace gases together with winds imply exploitable information for bird navigation” and “Do olfactory stimuli provide positional information for home-oriented avian navigation?” 60. Cf. Ishimaru et al., 1997; Xu et al., 1999; Wei et al., 2003. 61. Cf. Kumar et al., 2012. 62. Cf. Uziel, 1973; Straschill et  al., 1983; Sato et  al., 1996; Ishimaru et  al., 1997; Li et al., 2014; Weis et al., 2016.

80  Benjamin D. Young 63. A possible explanation of these negative findings might be that the electronic stimulation is not sufficiently dynamic and distributed, so it might not stimulate the olfactory epithelium and bulb with the same transduction pattern as chemical stimuli. Moreover, there is reason to be hopeful that, with modern technological advances in computational technology, we might be able to build electronic sensors that mimic the computational patterns at the OB (Persaud et al., “Neuromorphic Olfaction”). However, as of now our optimism in future technology does not threaten the claim that olfactory quality is inherent to the molecular structure of chemical compounds. 64. Young, 2015.

References Aristotle (1984a) De Anima. In The Complete Works of Aristotle, Vol. 1 (ed. Jonathan Barnes), Princeton, NJ: Princeton University Press, 641–692. Aristotle (1984b) De Sensu. In The Complete Works of Aristotle, Vol. 1, (ed. Jonathan Barnes), Princeton, NJ: Princeton University Press, 693–713. Barnes, D. C., Hofacer, R. D., Zaman, A. R., Rennaker, R. L., Wilson, D. A. (2008) “Olfactory Perceptual Stability and Discrimination” in Nature Neuroscience, 11: 1378–1380. Barwich, A. (2014) “A  Sense So Rare: Measuring Olfactory Experiences and Making a Case for a Process Perspective on Sensory Perception” in Biological Theory, 9 (3), 258–268. Barwich, A. (2015) “Bending Molecules or Bending the Rules? The Application of Theoretical Models in Fragrance Chemistry” in Perspectives on Science, 23 (4), 443–465. Barwich, A. (2016) “Making Sense of Smell” in The Philosophers’ Magazine, 73: 41–47. Barwich, A. (2018) “Measuring the World: Towards a Process Model of Perception” in Everything Flows: Towards a Processual Philosophy of Biology (ed. D. Nicholson, and J. Dupré), Oxford: Oxford University Press. Batty, C. E. (2007) Lesions in Smelling: Essays on Olfactory Perception, Linguistics and Philosophy, the MIT PhD Dissertation, Cambridge, MA. Batty, C. E. (2010a) “What the Nose Doesn’t Know” in Journal of Consciousness Studies, 17, 10–17. Batty, C. E. (2010b) “Scents and sensibilia” in American Philosophical Quarterly, 47, 103–118. Batty, C. E. (2010c) “A Representational Account of Olfactory Experience” in Canadian Journal of Philosophy, 40, 511–538. Batty, C. E. (2011) “Smelling Lesions” in Philosophical Studies, 153: 161–174. Batty, C. E. (2014) “Olfactory Objects” in Perception and its Modalities (ed. S. Biggs and D. Stokes), Oxford: Oxford University Press. Budek, Tomasz and Farkas, Katalin (2014) “Which Causes of an Experience are also Objects of the Experience?” in Does Perception Have Content? (ed. Berit Brogaard), Oxford: Oxford University Press, 351–370. Burge, T. (2010) Origins of Objectivity, Oxford: Oxford University Press. Carvalho, F. (2014) “Olfactory Objects” in Disputatio, VI (38). Castro, J. B., Ramanathan, A. and Chennubhotla, C. S. (2013) “Categorical Dimensions of Human Odor Descriptor Space Revealed by Non-Negative

Smelling Molecular Structure 81 Matrix Factorization” in PLoS ONE, 8 (9), e73289. https://doi.org/10.1371/ journal.pone.0073289. Castro, Jason B. and Seeley, William P. (2014) “Olfaction, Valuation, and Action: Reorienting Perception” in Frontiers in Psychology, 5. Cleland, T. A., Chen, S. T., Hozer, K. W., Ukatu, H. N., Wong, K. J. and Zheng, F. (2012) “Sequential Mechanisms Underlyingconcentration Invariance in Biological Olfaction” in Frontiers in Neuroengineering, 4, article 21. Frasnelli, J., Ariza, V., Charbonneau, G., Collignon, O. and Lepore, F. (2010) “Localisation of Unilateral Nasal Stimuli across Sensory Systems” in Neuroscience Letters, 478, 102–106. Frasnelli, J., Charbonneau, G., Collignon, O. and Lepore, F. (2009) “Odor Localization and Sniffing” in Chemical Senses, 34 (2), 139–144. Gatty, H. (1983) Finding Your Way on Land or Sea: Reading Nature’s Maps, Brattleboro, VT: Stephen Greene Press. Haddad, R., Lapid, H., Harel, D. and Sobel, N. (2008) “Measuring Smells” in Current Opinion in Neurobiology, 18 (4), 438–444, doi:10.1016/j .conb.2008.09.007. Haddad, R., Medhanie, A., Roth, Y., Harel, D. and Sobel, N. (2010) “Predicting Odor Pleasantness with an Electronic Nose” in PLOS Computational Biology, 6 (4), e1000740, doi:10.1371/journal.pcbi.1000740. Haddad, R., Weiss, T., Khan, R., Nadler, B., Mandairon, N., Bensafi, M., . . . Sobel, N. (2010) “Global Features of Neural Activity in the Olfactory System Form a Parallel Code that Predicts Olfactory Behavior and Perception” in Journal of Neuroscience, 30 (27), 9017–9026, doi:10.1523/JNEUROSCI.0398–10.2010. Henshaw, V. (2014) Urban Smellscapes, New York: Routledge. Ishimaru, T., Shimada, T., Sakumoto, M., Miwa, T., Kimura, Y. and Furukawa, M. (1997) “Olfactory Evoked Potential Produced by Electrical Stimulation of the Human Olfactory Mucosa” in Chemical Senses, 22, 77–81. Johansen, T. K. (1996) “Aristotle on the Sense of Smell” in Phronesis, XLI/I, 1–20. Johansen, T. K. (2006) “What’s New in the De Sensu? The Place of De Sensu in Aristotle’s Psychology” in Common to Body and Soul: Philosophical Approaches to Explaining Living Behaviour in Antiquity (ed. R. King), Berlin: Walter De Gruyter. Kay, L. M., Crk, T. and Thorngate, J. (2005) “A Redefinition of Odor Mixture Quality” in Behavioral Neuroscience, 119 (3), 726–733. Keller, A. (2017) Philosophy of Olfactory Perception, London: Palgrave Macmillan, Kemp, S. (1997) “A Medieval Controversy about Odor” in Journal of the History of the Behavioral Sciences, 33 (3), 211–219. Kleemann, A. M., Albrecht, J., Schopf, V., Haegler, K., Kopietz, R., et al. (2009) “Trigeminal Perception Is Necessary to Localize Odors” in Physiology  & Behavior, 97, 401–405. Klopping, H. L. (1971) “Olfactory Theories and the Odors of Small Molecules” in Journal of Agriculture and Food Chemistry, 19, 999–1004. Kobal, G. and Hummel, T. (1998) “Olfactory and Intranasal Trigeminal Event-related Potentials in Anosmic Patients” in Laryngoscope, 108 (7), 1033–1035.

82  Benjamin D. Young Kumar, G., Juhasz, C., Sood, S. and Asano, E. (2012) “Olfactory Hallucinations Elicited by Electrecial Stimulation via Subdural Electrodes” in Epilepsy and Behavior, 24, 264–268. Kumar, R., Kaur, R., Auffarth, B. and Bhondekar, A. P. (2015) “Understanding the Odour Spaces: A Step towards Solving Olfactory Stimulus-Percept Problem” in PLoS One, 10 (10), e0141263. doi:10.1371/journal.pone.0141263. Le Berre, E., Thomas-Danguin, T., Béno, N., Coureaud, G., Etiévant, P. and Prescott, J. (2008) “Perceptual Processing Strategy and Exposure Influence the Perception of Odor Mixtures” in Chemical Senses, 33, 193–199. Lewowicz, L. and Lombardi, O. (2013) “Stuff versus Individuals” in Foundations of Chemistry, 15 (1), 65–77. Li, A., Gire, D. H., Bozza, T. and Restrepo, D. (2014) “Precise Detection of Direct Glomerular Input Duration by the Olfactory Bulb” in Journal of Neuroscience, 34, 16058–16064. Lycan, W. G. (1996) Consciousness and Experience, Cambridge, MA: MIT Press. Lycan, W. G. (2000) “The Slighting of Smell” in Of Minds and Molecules; New Philosophical Perspectives on Chemistry, Oxford; New York: Oxford University Press. Lycan, W. G. (2014) “The Intentionality of Smell” in Frontiers in Psychology, 5, 436, doi:10.3389/fpsyg.2014.00436. Matthen, M. (2007) Seeing, Doing, and Knowing, Oxford: Oxford University Press. McHugh, J. (2012) Sandalwood and Carrion, Oxford: Oxford University Press. Mizrahi, V. (2014) “Sniff, Smell, and Stuff” in Philosophical Studies, 171 (2), 233–250. Moessnang, C., Finkelmeyer, A., Vossen, A., Schneider, F. and Habel, U., (2011) “Assessing Implicit Odor Localization in Humans using a Cross-modal Spatial Cueing Paradigm” in PloS One, 6, e29614. Negoias, S., Aszmann, O., Cory, I. and Hummel, T. (2013) “Localization of Odors can be Learned” in Chemical Senses, 38, 553–562. Olofsson, J. K., Bowman, N. E., Khatibi, K. and Gottfried, J. A. (2012) “A Timebased Account of the Perception of Odor Objects and Valences” in Psychological Science, 23, 1224–1232. Olofsson, J. K., Bowman, N. E. and Gottfried, J. A. (2013) “High and Low Roads to Odor Valence? A Choice Response-time Study” in Journal of Experimental Psychology: Human Perception and Performance, 39, 1205–1211. Papi, F. (1992) Animal Homing, London: Chapman & Hall. Perkins, M. (1983) Sensing the World, Indianapolis: Hackett Publishing Company. Persaud, K. C., Marco, S. and Gutiérrez-Gálvez, A. (2013) “Neuromorphic Olfaction” in Frontiers in Neuroengineering, Boca Raton, FL: CRC Press/­ Taylor & Francis, Ch. 1. Porteous, J. D. (1985) “Smellscape” in Progress Physical Geography, 9, 356–378. Pylyshyn, Z. (2003) Seeing and Visualizing: It’s Not What You Think, Cambridge, MA: MIT Press. Radil, T. and Wysocki, C. J. (1988) “Spatiotemporal Masking in Pure Olfaction” in Annals of the New York Academy of Sciences, 855, 641–644. Richardson, L. (2013) “Sniffing and Smelling” in Philosophical Studies, 162 (2), 401–419.

Smelling Molecular Structure 83 Roadaway, P. (1994) Sensuous Geographies, London: Routledge. Sato, M., Kodama, N., Sasaki, T. and, Ohta, M. (1996) “Olfactory Evoked Potentials: Experimental and Clinical Studies” in Journal of Neurosurgery, 85, 1122–1126. Seibt, J. (2005) Process Ontology. Metafisica e Ontologia (ed. G. Imaguire), München: Philosophia Verlag. Sela, L. and Sobel, N. (2010) “Human Olfaction: A Constant State of Changeblindness” in Experimental Brain Research, 1–17. Simons, P. (2000) “Continuants and Occurrents” in Aristotelian Society Suppl, 74, 59–75. Sinding, C., Coureaud, G., Chabanet, C., Chambault, A., Béno, N., Dosne, T., et al. (2014) “Perceptual Interactions in Complex Odor Mixtures: The Blending Effect” in Flavour Science: Proceedings from XIII Weurman Flavour Research Symposium (ed. V. Ferreira and R. B. T.-F. S. Lopez), San Diego: Academic Press, 27–31. Sinding, C., Thomas-Danguin, T., Chambault, A., Béno, N., Dosne, T., Chabanet, C., . . . Coureaud, G. (2013) “Rabbit Neonates and Human Adults Perceive a Blending 6-component Odor Mixture in a Comparable Manner” in PLoS One, 8 (1), e53534, doi:10.1371/journal.pone.0053534. Smith, A. D. (2002) The Problem of Perception, Cambridge, MA: Harvard University Press. Snitz, K., Yablonka, A., Weiss, T., Frumin, I., Khan, R. M. and Sobel, N. (2013) “Predicting Odor Perceptual Similarity From Odor Structure” in PLoS Computational Biology, 9 (9), e1003184, doi:10.1371/journal.pcbi.1003184. Soentgen, J. (2008) “Stuff: A Phenomenological Definition” in Stuff: The Nature of Chemical Substances (ed. J. van Brakel and K. Ruthenberg), Wu ̈rzburg: Konigshausen und Neumann. Sobel, N., Khan, R. M., Saltman, A., Sullivan, E. V. and Gabrieli, J. D. (1999) “The World Smells Different to Each Nostril” in Nature, 402, 35. Straschill, M., Stahl, H. and Gorkisch, K. (1983) “Effects of Electrical Stimulation of the Human Olfactory Mucosa” in Applied Neurophysiology, 46, 286–289. Tye, M. (2000) Consciousness, Color, and Content, Cambridge, MA: MIT Press. Tye, M. (2002) “Representationalism and the Transparency of Experience” in Nous, 36, 137–151. Uchida, N. and Mainen, Z. F. (2009) “Odor Concentration Invariance by Chemical Ratio Coding” in Frontiers in Systems Neuroscience, 1, 3. Uziel, A. (1973) “Stimulation of Human Olfactory Neuro-epithelium by Longterm Continuous Electrical Currents” in Journal of Physiology (Paris), 66, 409–422. Vincis, R., Gschwend, O., Bhaukaurally, K., Beroud, J. and Carleton, A. (2012) “Dense Representation of Natural Odorants in the Mouse Olfactory Bulb” in Nature Neuroscience, 15 (4), 537–539, doi:10.1038/nn.3057. von Bekesy, G. (1964) “Olfactory Analogue to Directional Hearing” in Journal of Applied Physiology, 19 (3), 369–373. Wallraff, H. G. (2004) “Avian Olfactory Navigation: Its Empirical Foundation and Conceptual State” in Animal Behavior, 67, 189–204. Wallraff, H. G. (2005) Avian Navigation: Pigeon Homing as a Paradigm, Berlin: Springer.

84  Benjamin D. Young Wallraff, H. G. (2013) “Ratios among Atmospheric Trace Gases Together with Winds Imply Exploitable Information for Bird Navigation: A Model Elucidating Experimental Results” in Biogeosciences Discuss, 10, 12451–12489. Wallraff, H. G. (2014) “Do Olfactory Stimuli Provide Positional Information for Home-oriented Avian Navigation?” in Animal Behavior, 90, e1–e6. Welge-Lüssen, A., Looser, G.-L., Westermann, B. and Hummel, T. (2014) “Olfactory Source Localization in the Open Field using One or Both Nostrils” in Rhinology, 52, 41–47. Wei, Y., Han, D., Cai, Z., Yang, L., Liu, X., Xian, M. and Zhang, X. (2003) “Experimental Study on Olfactory Evoked Potential in Rabbits” in Journal of Clinical Otorhinolaryngology, 17, 618–620. Weis, T., Shushan, S., Ravie, A. et al. (2016) “From Nose to Brain: Un-Sensed Electrical Currents Applied in the Nose Alter Activity in Deep Brain Structures” in Cerebral Cortex, 1–12, doi:10.1093/cercor/bhw222. Wilson, D. A. and Stevenson, R. J. (2006) Learning to Smell, Baltimore, MD: The Johns Hopkins University Press. Xu, C., Ni, D. and Li, F. (1999) “Olfactory Evoked Potentials Produced by Electrical Stimulation of the Olfactory Mucosa” in Zhonghua er bi yan hou ke za zhi, 34, 224–226. Yeshurun, Y. and Sobel, N. (2010) “An Odor is Not Worth a Thousand Words: From Multidimensional Odors to Unidimensional Odor Objects” in Annual Review of Psychology, 61 (1), 219–241. Young, B. D. (2011) Olfaction: Smelling the Content of Consciousness (doctoral dissertation). City University of New York, New York, NY. Young, B. D. (2014) “Smelling Phenomenal” in Frontiers in Psychology, 5, 713. Young, B. D. (2015) “Formative Non-conceptual Content” in Journal of Consciousness Studies, 22, 201–214. Young, B. D. (2016) “Smelling Matter” in Philosophical Psychology, doi:10.108 0/09515089.2015.1126814. Young, B. D. (2017) “Enactivism’s Last Breaths” in Philosophy of Mind: Contemporary Perspectives (ed. M. Curado and S. Gouveia), Newcastle: Cambridge Press. Young, B. D., Keller, A. and Rosenthal, D. M. (2014) “Quality Space Theory in Olfaction” in Frontiers in Psychology, 5, 1, doi:10.3389/fpsyg.2014.00001. Zhou, W. and Chen, D. (2009) “Binaural Rivalry between the Nostrils and in the Cortex” in Current Biology, 19, 1561–1565, doi:10.1016/j.cub.2009.07.052.

5 “Hierarchical Bokeh” Theory of Attention Anatoly Nichvoloda

1.  Attention as Information Processing A common-sense view of attention is that it is a mental action of structuring one’s mental states so that some become more prioritized, while others more peripheral depending on the task that an agent is trying to accomplish. Many philosophers and neuroscientists have suggested that attention is essential for consciousness (Baars, 1988; Crick and Koch, 1990; Evans, 1982; Lycan, 1996; Mole, 2008; Peacocke, 2000; Posner, 1994; Prinz, 2012; Watzl, 2017). In this chapter I  argue for a theory of perceptual attention in the context of an information-processing mechanism that incorporates Shannon Communication System, Bayesian Predictive Processing and elements of Control Theory. The resulting mechanism (outlined in Figure 5.1) is a Predictive Closed Loop Control Circuit with Negative Feedback, which, going forward I’ll call the BayesShannon system: The goal of this system is to control the behavior of a sensory transducer (e.g., mammalian cochlear cell) by correctly predicting/emulating its physical states and, therefore, correctly predicting environmental stimuli associated with the states of the transducer. In this mechanism, a Bayesian neural network functions as Informer (information source one) and also, importantly, Informee (destination of information). The information-processing cycle starts with a Bayesian net producing a certain message as an output and ends with the network receiving a certain message as an input and, based on that, generating new output and starting a new cycle. The Bayesian net also functions as long-term memory for the overall mechanism, that is, probabilities of possible state values of the sensory transducer are encoded as weights or physical connection strengths between the network’s nodes. The first step in perceptual information processing involves formulation of a prediction/message by the Bayesian neural network (information source one) and encoding of the prediction/message (by the Encoding Transmitter) into the efferent signal in terms of suitable variable physical properties of electric current (e.g., amplitude). This

86  Anatoly Nichvoloda

Figure 5.1  Bayes-Shannon system: Perceptual information-processing mechanism based on Shannon Communication System (Shannon, 1949) that combines Bayesian Predictive Processing (Friston, 2010; Clark, 2013a; Hohwy, 2012) and Modal Emulation (Grush, 2004a: 379).

encoded prediction/message anticipates the properties of the sensory transducer’s states that the system expects to register at the transducer within a certain time interval. The signal carrying the encoded prediction toward the sensory transducer is called the “efferent signal,” while the signal carrying the prediction after it has been modulated by the transducer is called the “afferent signal.” The system also retains a copy of the efferent signal (efferent copy), which is sent directly to the Decoding Receiver to be compared with the corresponding afferent signal (the original efferent signal after its interaction with the sensory transducer). Thus, a given prediction encoded into the efferent signal is sent through the information channel between the Encoding Transmitter and Decoding Receiver via the sensory transducer. As the efferent signal goes through the sensory transducer, if the predicted state of the transducer is different from its actual state, the signal is modulated by the transducer/environment (information source two). If the predicted state of the transducer is the same as the actual state, then the signal passes through the transducer unchanged. Once the modulated

“Hierarchical Bokeh” Theory of Attention 87 (afferent) signal carrying the modulated initial prediction arrives at Decoding Receiver it is compared with the efferent copy signal carrying the initial prediction, and certain physical perfect correlations and/or differences/errors are observed between the two signals. These correlations and differences serve as inputs taken up by the Bayesian neural network that performs a negative feedback operation (Prediction Error Minimization) on the signal and issues a new prediction which begins a new cycle. In Nichvoloda (forthcoming) I argue that perfect correlations and differences observed at Decoding Receiver of Bayes-Shannon information-processing system encode the properties of Access and Phenomenal1 consciousness correspondingly. In this chapter, I’ll use the Bayes-Shannon information-processing system to outline the Hierarchical Bokeh theory of attention. 1.1 Endogenous (Top-Down) vs. Exogenous (Bottom-Up) Attention According to Watzl (2011a, 2011b), attention can be endogenous (topdown, motivationally driven) or exogenous (bottom-up, attention grabbing); it can be focal or global; it can be directed at objects, properties or spatial or temporal regions and so on. Watzl (2010) argues that what explains voluntarism about attention is that attention in all its forms (even when it is involuntary) by its nature is actually active. He uses the notion of an activity that is genuinely weaker than the notion of an intentional action (drawing on O’Shaughnessy, 1981) to defend a view on which attention in all its forms is a mental activity while only voluntary attention is an intentional action. Based on Watzl’s distinction between voluntary (intentionally directed, top-down) vs. involuntary (environment directed, bottom-up) attention, I  suggest that we distinguish two kinds of attention within the BayesShannon hierarchical system: Endogenous and exogenous. I suggest that endogenous (top-down) and exogenous (bottom-up) attention engender phenomenal and access consciousness by influencing/manipulating amounts of differences and perfect correlations observed at the Decoding Receiver of the Bayes-Shannon hierarchical system. 1.2 Andy Clark and Jakob Hohwy: Attention Within the Bayesian Prediction Hierarchy In its basic form, the Bayesian Predictive Processing framework approach describes a system that is primarily driven by external stimuli and is supervised “not by someone else nor by itself, but by the statistical regularities of the properties of the environment.” (Hohwy, 2012: 3) Exogenous attention is an automatic bottom-up process that arises when the system’s perceptual processes (driven by the Prediction Error Minimization (PEM)

88  Anatoly Nichvoloda principle) are modulated by the environment. Essentially, the environment “selects” stimuli for the agent to attend to in a sense that highly unexpected events trigger large prediction errors/differences. According to Hohwy (2012: 7), in the normal course of events the “system is helped in the prediction error minimization task by precision processing, which (following Feldman and Friston, 2010) was claimed to map onto endogenous attention such that attention is precision optimization in hierarchical perceptual inference.” According to Clark, the function of top-down attention is to increase the gain on select error units: “Attention [. . .] is simply one means by which certain error-unit responses are given increased weight, hence becoming more apt to drive learning and plasticity and to engage compensatory action” (Clark, 2013a: 10). In this construal, in the act of attending (e.g., an individual deciding herself to attend left) the cue can be conceived as a desired state, for example, that something valuable will be spotted on the left. This would then generate an expectation of precision for that region such that stimuli located there are more likely to be detected. Attention of this sort is endogenous, because it has a volitional aspect since it is the individual who decides to attend and acts on this decision. Attention as precision expectation modulates the system’s prediction error minimization process subject to precisions predicted by the context (including cues and competing stimuli) and does this for prediction errors of different accuracies. According to Hohwy, “[W]ithout attention, the better a stimulus is predicted the more attenuated its associated signal should be. Attention should reverse this attenuation because it strengthens the prediction error” (Hohwy, 2012: 7). 1.3 Parr and Friston: Attention and Working Memory in Bayesian Predictive Processing Hierarchy According to Parr and Friston (2017: 7), “Gain modulation-based accounts suggest that attention involves increasing the gain of a particular sensory channel, enhancing the processing of that sensation.” Parr and Friston (2017: 3) suggest that the hierarchical nature of Bayesian predictive processing can account for some interesting neuroscientific phenomena. For example, representations in higher regions typically evolve over a longer time scale than lower levels, so that slowly evolving representations can provide the context for faster changes in the lower area, influencing the interpretation of a given stimulus. This hierarchical arrangement of temporally increasing the receptive field can explain delay period activity that persists once a stimulus has been removed—a feature strongly associated with working memory function. Additionally, according to Parr and Friston (2017: 3), “as the higherlevel representation accumulates information from previous sensory samples, it informs actions over a longer time scale than the lower level. This

“Hierarchical Bokeh” Theory of Attention 89 might explain why cortical areas considered to be high in the hierarchy, such as the dorsolateral prefrontal cortex, have been heavily implicated in working memory.” Taking this approach, working memory can be conceptualized as a process of evidence accumulation in a temporally structured hierarchy where new information can be combined with the memory to make more precise predictions about the future. Parr and Friston (2017: 17) suggest that working memory is directly involved in attention since “resistance to distracting stimuli occurs when there is ‘attentive’ biasing toward current beliefs, ensuring the posterior beliefs are kept close to prior beliefs.” Importantly, they suggest that all levels of a predictive hierarchy “can be considered types of working memory, as all involve a maintained representation that is updated over time by new sensory data.” 1.4  Prinz: Attention Engenders Consciousness At the core of Jesse Prinz’s theory is the claim that consciousness occurs when and only when an integrated representation of a stimulus’s several properties, as perceived from a particular point of view, is made available to working memory. This process of making a representation available to working memory is what Prinz calls “attention.” The crucial property of the representations that are made available by this attention process is the property of being integrated into a point-of-view-retaining format (what Prinz calls “intermediateness”). While Prinz doesn’t use predictive processing as the framework in which attention and consciousness are to be found, following some researchers (and, apparently, agreeing with the construal of attention by Clark, Hohwy and Friston), he adopts the term “gain control” from electronics that refers to devices that modulate signal strength (Prinz, 2012: 134). Such a device implements attention by controlling perceptual information flow since “[o]f all of the things that occur when we attend, one seems to be an especially good candidate for the essence, or common denominator: A change in information flow” (Prinz, 2012: 92). According to Prinz (2012: 92), psychologists and neuroscientists agree that attention has a very special relationship with working memory, namely, since attention determines what information gets into working memory, it can be considered as a “gatekeeper” to working memory. In general, working memory refers to a capacity to actively manipulate a limited number of items in a conscious workspace. According to Prinz (2012: 92), working memory is a short-term storage capacity postulated by psychologists that plays an important role associated with the victory of a given stimulus in terms of phenomenal pop-out and that allows for “executive control” (Baddeley, 2007; D’Esposito et al., 1999). According to Block (2007: 14), at a neural level, working memory should be distinguished from long-term structural memory since working memory

90  Anatoly Nichvoloda “is coded in the active firing of neurons—and ceases when that neuronal firing ceases—and structural memory depends on changes in the neural hardware itself, for example, change in strength of synapses.” Block notes that working memory, which is active in the sense that it has to be actively maintained, is sometimes misdescribed as “short term” (see, e.g., Atkinson and Shiffrin, 1968) since it lasts as long as active firing lasts. But, according to Block, firing need not last a short time if the subject is actively rehearsing. Operationally, neuroscientists think of working memory in terms of stimulus-specific neural activations that remain during a “delay period” after a stimulus has been removed. Psychologists think of working memory as a place where temporarily stored information is held “online” for “controlled” information processing, that is, processing that is not automatic in the sense of being reflexive, overlearned or driven entirely by the stimulus (Prinz, 2012: 92). According to Carruthers (2015), the working memory system is a specific system that contains and manipulates different kinds of sensory-based representations. This is also the system which produces our inner speech and the stream of visual imagery. Thus, items encoded in working memory become available to language systems for reporting and for reasoning, planning and comparison. According to Prinz, Attended Intermediate Representation (AIR) theory defines consciousness in terms of attention and attention in terms of working memory. “If we can equate attention with the process by which perceptual information becomes available to working memory, then we can say that all examples of attention involve that process and differ only in what attention is modulating and what allows that modulation to take place” (Prinz, 2012: 95).

2. Exogenous and Endogenous Attention in the Bayes-Shannon System The analysis of information processing in the system offers important insights into how two primary functions of working memory are implemented in the system: Working memory playing a role in guiding effortful attention (e.g., Cowan, 1995), while being a place where attended perceptual states get temporarily stored (Knudsen, 2007). I begin by identifying two kinds of working memory that correspond to two kinds of attention: Exogenous and endogenous. Exogenous attention in the context of the Bayes-Shannon perceptual system occurs when the environment “selects” stimuli for the system to attend to in a sense that highly unexpected events produce large prediction errors/differences, which means that large amounts of phenomenal and small amounts of access consciousness would be observed by the system at the Decoding Receiver. In other words, exogenously attended highly surprising stimuli would phenomenologically appear to be louder and brighter than less unexpected stimuli.

“Hierarchical Bokeh” Theory of Attention 91 On the other hand, highly familiar stimuli would give rise to small (or zero) amounts of phenomenal and large amounts of access consciousness observed at the Decoding Receiver. Thus, depending on the properties of the environment (highly variable or highly uniform), in the BayesShannon hierarchical predictive system that implements only Prediction Error Minimization (PEM), such an environmentally driven perceptual process would tend to vacillate between the two following environmentally driven limits. On the one hand, the system could learn to perfectly predict most (or all) inputs of its environment and, therefore, become a partial or fully functional isomorph of a philosophical zombie/superblindsighter where most or all perception is access conscious and there is little to no phenomenal consciousness. On the other hand, when exposed to significant changes or completely new environmental stimuli, the system would register large errors/differences at the Decoding Receiver, thus making most or all perception phenomenally conscious with little to no access consciousness. And while it is true that human beings tend to go phenomenologically dark of highly familiar or repetitive stimuli while being easily startled/surprised by irrelevant bright lights or loud noises, it seems that exogenous (bottom-up) attention only partially describes the phenomenon we usually call attention. In other words, there is a need for the Bayes-Shannon perceptual system to account for the agent-driven top-down (endogenous) attention, that is, an ability to voluntarily bring different aspects of the environment into a suitable balance of conscious awareness. I suggest that endogenous attention in a hierarchical Bayes-Shannon system can be implemented by a mechanism that involves the use of double prevention, that is, when “a mechanism works by removing a cause, preventing a cause, or inhibiting an inhibitor” (Schaffer, 2004). In other words, if another mechanism performs a negative feedback function on the original negative feedback mechanism (in this case, Prediction Error Minimization), then a reversal of PEM function and its effects will take place. Thus, a hierarchical arrangement of negative feedback controllers and double prevention mechanisms working in a coordinated manner can implement a selective Prediction Error Control (PEC) function in the hierarchy and, therefore, implement the function of endogenous attention. The goal of such a double negation mechanism is an agentcontrolled, environmentally appropriate and contextually normalized balance of differences and perfect correlations at the Decoding Receiver and, therefore, appropriate perception of one’s environment in terms of phenomenal and access consciousness. Following Prinz’s formulation of working memory as the function that modulates perceptual representations, in the Bayes-Shannon perceptual system working memory is defined as the function of modulation of perceptual representations either by the environment or by the system’s neural network. Modulation of perceptual representations by the environment

92  Anatoly Nichvoloda via sensory transducers allows the system to implement exogenous attention while modulation of perceptual representations by the system’s neural network implements endogenous attention. In other words, there are two places in the Bayes-Shannon mechanism where perceptual representations carrying information about a predicted state of the sensory transducer can be manipulated by the system or the environment: Neural network (information source 1) and sensory transducer (information source 2). In each case modulation represents a certain computational input-output transformation of perceptual representations implemented by two sources of information: The environment (information transformation/modulation of perceptual representations between the Encoding Transmitter and the Decoding Receiver) and the system’s neural network (information transformation/modulation of perceptual representations between the Decoding Receiver and the Encoding Transmitter). Thus, (as outlined in Figure 5.2) there are two kinds of working memory implemented by the system: Exogenous and endogenous working memory. The overall perceptual information-processing cycle operates in the following manner. The system’s neural network that implements Prediction Error Control (Information source 1) is in a certain state of endogenous working memory that represents a certain model that contains information that the sensory transducer will be in a certain state. This

Figure 5.2 Endogenous attention is implemented by computational transformations of perceptual representations by the system’s neural network (information source 1, endogenous working memory and attention) while exogenous attention is implemented by computational transformations of perceptual representations by the sensory transducer (information source 2, exogenous working memory and attention).

“Hierarchical Bokeh” Theory of Attention 93 information gets encoded into efferent signal and efferent copy by the Encoding Transmitter in terms of spatiotemporal characteristics of the message (particular sequences of amplitudes of electric current). Thus, the neural network performs modulation of input perceptual representations and outputs a certain expectation at the Encoding Transmitter in the form of a given efferent signal. This efferent signal carries the prediction to the sensory transducer, which encodes information about its actual state into the efferent signal, thus transforming it into the corresponding afferent signal. This is the moment when the environment modulates the message or, in other words, when exogenous attention occurs in the system in virtue of exogenous working memory operation of modulation of the perceptual representations from expected into the actual ones. Further, when a given afferent signal and the corresponding efferent copy are compared at the Decoding Receiver, a certain balance of differences and perfect correlations (phenomenal and access consciousness) are observed; this information is made available to the neural network’s current state that implements endogenous working memory. The neural network takes both kinds of correlations as input and, depending on the goals of the higher levels of the system, performs the Prediction Error Control information transformation/modulation operation on perceptual representations with the purpose of issuing a new prediction that corresponds to the system’s perceptual goals. In other words, the neural net (endogenous working memory) is performing the operation of endogenous attention that becomes the first step in the next cycle of information processing, since the resulting state of the network will be encoded into the next prediction that would go through the transducer and result in a next episode of a certain balance of phenomenal and access consciousness observed at the Decoding Receiver. In other words, attention and working memory are related in the following way: Working memory is the current state of transducer/environment and perceptual model in the system’s neural network/agent, while attention is how exactly the current sensory transducer and neural network’s states influence (what effect they have on) the balance of differences (phenomenal consciousness) and perfect correlations (access consciousness) observed at the Decoding Receiver as outlined in Figure 5.3: Thus, on this formulation, perceptual attention consists of endogenous and exogenous kinds of attention that engender a certain balance of phenomenal and access consciousness and make it available/accessible to endogenous working memory after perceptual representations have been modulated by exogenous working memory. Thus, working memory also consists of endogenous and exogenous components, one of which implements the neural network’s modulation of perceptual representations while the other implements the environment’s modulation of perceptual representations.

94  Anatoly Nichvoloda

Figure 5.3 Signal comparison and observation of access and phenomenal consciousness at the Decoding Receiver

2.1. First-Person Experience of Exogenous and Endogenous Attention in the Bayes-Shannon System For the sake of simplicity, in the previous section I discussed the information-processing dynamic in the Bayes-Shannon mechanism that operates on a single sensory transducer. However, the main principles of perceptual information processing outlined above should also apply to significantly more complex systems that deal with large sensory fields or entire sensory modalities. That is, we can think of perception in terms of bit maps where each sensory transducer corresponds to a pixel (Prinz, 2012: 93). Attention in this formulation can be thought of as coming in degrees, meaning that more or fewer of the pixels in a given region of the visual field are accessible at a given time. Fully attending an object would potentially make every cell responding to that object accessible to working memory. But partial attention results in a subset of cells being accessible, while the others are not. This is like viewing an object through a screen door, letting some details in but not all. The fully attended object will therefore be clearer and sharper in contrast. (Prinz, 2012: 93)

“Hierarchical Bokeh” Theory of Attention 95 According to Prinz (2012: 93), this hypothesis might explain the fact that attention can affect the experience of a stimulus in a number of ways, since increases in attention cause greater apparent resolution, contrast and size. (Carrasco et al., 2004; Anton-Erxleben et al., 2007) In the Bayes-Shannon system, endogenous and exogenous attention can modulate perceptual representations in particular ways and thus determine how exactly a given perceptual stimulus will be represented by the system at the Decoding Receiver. In other words, the working memory control of number, intensity, contrast and other parameters of pixels comprising a given perceptual stimulus will determine resolution, focus, brightness, contrast and other parameters of a particular perceptual object experienced in terms of access and phenomenal consciousness by an agent. In the following example I elaborate how attention engenders consciousness in the context of information processing in the Bayes-Shannon system and the effects it has on the first-person experience in visual modality. Let’s imagine that you are hiking in some hilly, grassy terrain and, upon reaching a hilltop with a beautiful vista, you take a look around. The exogenous attention immediately modulates the overall process of Prediction Error Minimization under the current “rolling, grassy hills” perceptual model by drawing your gaze (capturing your attention) to an area that has higher than average phenomenal pop-out (differences) relative to the currently ongoing perceptual expectation model—a shiny blue spot among matte green rolling hills. According to Prinz (2012: 93), “Pop-out occurs when the representation of one stimulus competes with the representations of surrounding stimuli and wins.” Prinz suggests that the pop-out effect is the psychological correlate of attention, and I agree, with a minor elaboration that in the context of the Bayes-Shannon mechanism, it is the psychological correlate of exogenous attention, namely, a shiny blue spot pops out because its perceptual representations are different compared to the expected visual field’s “rolling grassy hills” model. These differences are registered at the Decoding Receiver as differences and made available to endogenous working memory implemented by the neural network associated with the visual modality. At this point, your visual predictive hierarchy selects an appropriate perceptual model of the environment that corresponds to this particular pop-out circumstance and comes up with a perceptual hypothesis/model: “A  lake.” Then, as you want to take a closer look at the lake, your endogenous attention modulates the perceptual representations from the top-down so as to bring the particular aspects of the lake into a contextually relevant detailed perceptual awareness through a certain balance of access and phenomenal consciousness. For example, endogenous attention can attenuate the Prediction Error Minimization process for color and brightness while still trying to completely match/account for its orientation and shape. The result is that you will see the lake, whose

96  Anatoly Nichvoloda color and brightness have more phenomenal “pop” than its orientation and shape. In other words, through endogenous attention the system has provided you with an appropriately balanced amount of phenomenal and access consciousness that implements a contextually adjusted and top-down controlled perception of the requested aspects of the lake. Conversely, when you want to focus your attention on orientation and shape, the system is capable of making these properties of the lake to have more phenomenal “loudness” than its color and brightness. In this case, the system would endogenously emphasize (by attenuating PEM process applied to) orientation and shape aspects of the lake by biasing them from top down while deemphasizing (strengthening PEM process for) color and brightness aspects. In terms of first-person awareness this would feel that the shape and orientation of the lake are more phenomenally vivid compared to its color and brightness. While the spotlight of your endogenous attention is busy balancing different aspects of the lake, bringing them into sharp attentional focus, the rest of your visual field is experienced as a grainy picture of objects represented by variable degrees of clarity and resolution that doesn’t have much detail but gives you a general medium- to low-resolution sketch of the surroundings. In other words, while you are well aware that you are hiking in some hilly, grassy terrain on a pleasant sunny day and not walking down Fifth Avenue in Manhattan in pouring rain, you are not aware of certain particulars of your environment (e.g., five oak trees a quarter of a mile down the hill, a beautiful fluffy cloud slowly moving in from the west and many other features of the environment) as sharply salient perceptual objects because either those features don’t have enough popout value to engage your exogenous attention or you haven’t paid them your endogenous attention. This view agrees with Prinz (2012, 88): “Once a stimulus is presented, it captures attention, either because it pops out or because we are looking for it. In either case, the initial visual response to the stimulus and the attention directed to it are two distinct processes and the latter happens after the former.” That is, usually exogenous (bottom-up) attention precedes endogenous (top-down) attention (as in the above example of a lake pop-out) in the sense that the environment exogenously “selects” stimuli that differ from our current predictive model and then the system engages its endogenous working memory to either bring a given pop-out into a sharp high-resolution perceptual experience and investigate it further or incorporate it into a current general model of the environment, lower its resolution and move on to the next pop-out. However, exogenous (bottom-up) doesn’t always have to precede endogenous (top-down) attention, since one can proactively engage in the search for a given pattern. For example, having discovered the lake, you want to share the finding with your hiking companion who is still climbing the hill. So your visual system’s endogenous attention

“Hierarchical Bokeh” Theory of Attention 97 engages in pattern matching activity for a red jacket that she wears. In other words, according to Prinz (2012: 93), “[Y]ou use a template of the sought object as a filter on the representation of the [environment]; when a match is found, it becomes available to working memory.” In this case, the Bayes-Shannon system’s neural network issues an appropriately biased expectation that when matched with the sensory input will produce a balanced, contextually normalized perceptual representation of your friend in terms of access and phenomenal consciousness. Thus, according to Prinz (2012: 94), “top-down and bottom-up attention result from different control structures, but they make use of the same resource.” In the context of the Bayes-Shannon system, control structures correspond to exogenous and endogenous kinds of working memory that use the same resource, namely, modulation of perceptual representations to produce a certain balance of differences and perfect correlations at Decoding Receiver and, from the first-person point of view, a perceptual experience of your friend in a red jacket once a match has been found. 2.2. “Perceptual Bokeh” Formulation of First-Person Effects of Hierarchical Processing in the Bayes-Shannon System In many respects operation of endogenous attention can be considered as a perceptual version of bokeh effect in video and photography where depth of field is manipulated to achieve a certain aesthetic effect. “Bokeh” is an optical effect of a soft out-of-focus background that one gets when taking a picture or video of an object using a lens with shallow depth of focus. This effect is usually used to produce an aesthetically pleasing effect of focusing on a particular object while blurring the background. Within the depth of field of a particular lens, objects appear clear and sharp whereas out of the depth of field they appear blurry and amorphous (as represented in Figure 5.4): According to the suggested formulation, the first-person perceptual experience of the world resembles a more sophisticated version of the bokeh effect where the perceptual hierarchy’s endogenous attention can control not only the focus, but also other properties of perceived objects, e.g., resolution, contrast, saturation, etc. By using PEC, endogenous attention functions as an ability to control levels, resolution, focus and other aspects of a perceptual hierarchy in Bayes-Shannon system that can bring any given stimulus (or a group of stimuli) into sharp attentional focus by precisely balancing a contextually appropriate proportion of access and phenomenal consciousness on the appropriate perceptual level(s) to achieve an attentional effect similar to different ranges of bokeh. Some might object to the conception of endogenous attention as producing a bokeh effect in first-person experiences within the Bayes-­ Shannon hierarchical system by saying that our eyes actually have certain

Figure 5.4 The top picture has a very strong bokeh effect produced by a lens with a shallow depth of field. The bottom picture has a very mild bokeh effect produced by a lens with a deep depth of field. Bokeh effect is controlled by the aperture of a lens: The opening that controls how much collimated light gets to a camera’s sensor. The idea of “perceptual bokeh” is that endogenous attention can control the properties of perception so that from the first-person point of view perception could range from the top to the bottom picture. Source: Photo by Basil Morin/CC BY

“Hierarchical Bokeh” Theory of Attention 99 purely mechanical and optical parts that can account for the perceptual features just described. In other words, the phenomenon of bokeh-like focus of endogenous attention in visual modality could be the product of the structure of human eyes that have pupils that serve the function of aperture and muscles that can adjust it. In response, I  would point to similar attentional capacities in other sensory modalities that don’t have any aperture-like functionality that can be implicated in the function of focusing attention. For example, during wine tasting, olfaction and gustation one can focus on subtle flavors of a wine being sampled (blackberry, violets, chocolate, tannins, etc.) and bring them out while suppressing the rest or treating them as general contextually appropriate yet blurry olfactory and/or gustatory background. According to Shomstein and Yantis (2006), the bokeh-type endogenous attention effect is also present in auditory modality. The important point for the discussion that follows is that perceptual predictive control of depth, resolution, etc. that organizes sensory stimuli on many hierarchical levels into perceptual models is information-processing activity, which accounts not only for representational but also cognitive first-person effects of perception. Since endogenous attention modulates perceptual representations in terms of modulating information flow within the Bayes-Shannon hierarchical system it controls not only perceptual representational, but also perceptual cognitive focus of the system. The system accomplishes this by arranging a multitude of individual sensory properties that comprise an object into a cognitively salient hierarchical predictive perceptual model of the object while excluding irrelevant sensory representations of the environment. Along with predictive processing idea about brain p ­ redicting/ modeling one’s environment, the notion of attentional perceptual model of an object is based on Prinz’s view that “we attend to an external stimulus only by attentionally modulating a representation of that stimulus, not by merely attending to the location in which it is presented” (Prinz, 2012: 88) and that “attention is often selective and it seems to make items of experience available to working memory by separating one thing from another” (Prinz, 2012: 243). In other words, the notion of an attentional perceptual model of an object from the third point of view of computational predictive processing hierarchy corresponds to the ability to consistently predict/track object(s) in the environment with contextually appropriate referential and behavioral success. Thus, features of the environment that are in the spotlight of endogenous attention not only have higher resolution, sharper focus, etc., but also higher level of perceptual cognitive saliency. Correspondingly, features of the environment that are out of focus of endogenous attention in, e.g., visual modality, would appear not only as having lower resolution and focus but also as being cognitively less distinct to a variable degree. That is, the lower the resolution of hierarchical representation/processing of the object, the less clear it is processed as an object cognitively.2

100  Anatoly Nichvoloda A “perceptual bokeh” effect can be implemented by the Bayesian Predictive Processing hierarchy in a straightforward way: The system selects appropriate hierarchical representational level(s) and then increases resolution, contrast and other parameters of a certain visual area that corresponds to an object of endogenous attention while simultaneously lowering the resolution, contrast and other variables of other levels and areas around the object(s) of attention. Thus, attended objects would appear sharp and distinct, while nearby objects would appear significantly less distinct to the point of blurring into the general environmental background. The “perceptual bokeh” formulation of endogenous attention agrees with recent experimental findings on attention, namely, that attentional field has what is known as a “Mexican hat” shape. That is, focusing attention on a particular spatial location in our visual environment generates a center-surround modulation where signals and their processing in the center of our attentional spotlight are enhanced, while the immediate vicinity is suppressed. According to Müller et al. (2005), while the perceptual enhancement is often assumed to fall off monotonically with distance from the attended location, there is behavioral and electrophysiological evidence for a suppressive zone in the direct vicinity around the spotlight of attention that together form a “Mexican hat” profile of cortical responsiveness with an excitatory center and an inhibitory circle around it. The physiological correlate of these enhancements is a gain increase of neurons with receptive fields that overlap the attended location, similar to the sensory effect of increasing the salience of a given stimulus (Müller et al., 2005). Importantly, the more difficult the task: (1) the higher the attentional boost, (2) the narrower the field of enhanced view and (3) the narrower the area of suppression. From the first point of view, a difficult task would look more like the top picture in Figure 5.3 where just the stapler is in sharp perceptual focus. An important factor that can make a given task difficult is lack of time for the perceptual hierarchy to construct a sufficiently complex model of an object in the environment since higher levels of the model evolve over a time scale that is slower than that of the level below. To illustrate this point, the time spent reading each word in this sentence is shorter than that for the sentence as a whole. Thus the states representing individual words would be represented at a lower hierarchical level than the states representing sentences (themselves lower than those representing a page, and so on). (Par and Friston, 2017: 7) Conversely, an easier task would make the enhanced field of perception wider, and this can allow the system to focus on multiple objects/models with the same hierarchical depth at the same time. From the first point of view, this would correspond to the bottom picture of the Figure 5.3,

“Hierarchical Bokeh” Theory of Attention 101 where besides the stapler, there are other objects with similar hierarchical depth of perception. That is, endogenous attention can produce a range of effects from maximally biasing processing of information toward the center location and constructing a highly detailed representation of a single object to widening the perceptual field and letting multiple objects have approximately the same perceptual saliency that is more diffused than that of a single object: In focal cases, a single object representation is accessible, and in diffuse attention, wide swaths of the visual field may be accessible. We can access small details, multiple objects, or regions of space; we can even attend focally to one object while diffusely attending to one or more items in the background. Diffuseness here involves both a spread of attention over a spatial region and a reduction in the degree of attention. (Prinz, 2012: 95)

3. Phenomenal Consciousness Overflows Cognitive Access or Cognitive Access Overflows Phenomenal Consciousness? It Depends I suggest that the “perceptual bokeh” formulation of endogenous attention within the Bayes-Shannon hierarchical perceptual system can shed some new light on the debate about whether perceptual phenomenology is rich or poor. Ned Block uses George Sperling’s (1960) experiments to argue that perceptual consciousness is richer (i.e., has a higher capacity) than cognitive access. That is, when observing a complex scene, we are conscious of more than we can report or think about. Sperling (1960) showed subjects an array of letters for a brief period. Although the subjects thought they could see all or almost all of the letters, they could report only 3–4 of them from the whole matrix. However, they could also report 3–4 items from any row that was cued after stimulus offset, suggesting that subjects had a persisting image of more letters than they could report. According to Block’s overflow argument, all or almost all of the 12 items are consciously represented, perhaps fragmentarily but well enough to distinguish among the 26 letters of the alphabet. However, only 3–4 of these items can be cognitively accessed, indicating a larger capacity in conscious phenomenology than in cognitive access. Importantly, the overflow argument does not claim that any of the items in the array are cognitively inaccessible, but rather that necessarily most are unaccessed. (Block, 2011: 1)

102  Anatoly Nichvoloda Some theorists have used “change blindness” phenomenon to argue that the overflow argument is wrong. Unattended features of two similar pictures can differ without the perceivers noticing the difference, despite what seems to be clear perception of both pictures.3 According to Block (2011, 1), there are two competing interpretations of change blindness phenomenon: “inattentional blindness” and “inattentional inaccessibility.” In the first interpretation, supported by Tye (2010), Rensink et al. (1997), O’Regan and Noe (2001), Simons and Rensink (2005) and Noe (2004), “[O]ne does not notice the difference because one simply does not consciously see the specific aspect of the scene that constitutes the difference (for instance, the item that is present in one picture, absent in the other).” (Block, 2011: 2) In the second interpretation, supported by Block (1997, 2001, 2007), Lamme (2003, 2004, 2006) and Dretske (2004, 2007), “[O]ne normally consciously sees the item that constitutes the differences but fails to categorize or conceptualize it in a way that allows for comparison.” (Block, 2011: 2) While it is not my goal here to give an exhaustive treatment of the debate (see, e.g., Block [2011] for an extensive overview), I suggest that the “perceptual bokeh” formulation of endogenous attention within the Bayes-Shannon perceptual predictive hierarchy offers an interesting way to accommodate experimental results and expand on Block’s view about it. To begin with, the hierarchical approach would take an issue with Block’s formulation of cognitive access as an all or nothing phenomenon. That is, a hierarchical structure of perception allows for cognitive access that comes in degrees. Thus, the ability to cognize and report a letter can range from not being able to distinguish a letter from, e.g., a face or a pseudo-letter, to effortless recognition and report of a given letter. In the context of Sperling experiments, the subjects’ endogenous attentional hierarchical bokeh effect brings certain areas and levels into sharp focus (thus making cognizing and reporting of letters in that area an easy task) while suppressing areas and levels outside of endogenous attention (thus making cognizing and report of letters in those areas a difficult task). Importantly, while experimental subjects have trouble perfectly identifying letters that are out of the spotlight of endogenous attention, they still can cognize and report them as letters vs. faces or houses. In other words, the level of cognitive access in the suppressed area is lower than in the attended area, and I suggest that this effect is responsible for the results we see in Sperling type experiments. Thus, this formulation largely agrees with Block, Lamme and Dretske’s interpretation while noting that the unattended conscious area (the fuzzy suppressed part) overflows the level of cognitive access brought to this area as the result of the “hierarchical perceptual bokeh” effect of endogenous attention. Further, it appears that cognitive access can overflow phenomenal consciousness when, for example, the same attended phenomenal

“Hierarchical Bokeh” Theory of Attention 103 appearance can be construed in different conceptual terms, as in the rabbit-duck illusion, when the same phenomenal sensory level representation gives rise to alternating episodes of seeing/cognizing it as a rabbit or as a duck on higher levels of the perceptual hierarchy. Also, different perceptual systems can bring different cognitive resources to the same stimuli, for example, when two people are looking at the same array of letters and one of them happens to be a font designer while the other knows nothing about fonts. The person who knows more about fonts would be able to cognize and report many more features of a given set of letters (font, line width, height, etc.) than a regular person. In either of these cases we see that a hierarchical system’s ability to cognize a given set of phenomenal representations varies depending on what hierarchical representational abilities the system brings to a particular episode of perception. Neither Block and his supporters nor their opponents in this debate consider this possibility, however, from the point of view of the hierarchical system capable of endogenous attention “hierarchical perceptual bokeh” phenomenon seems to be a natural feature of perception.

Notes 1. According to Block (2001), Phenomenal consciousness is a “pretheoretic” notion characterized by mental states that there is something it is like for the subject to have an experience. The paradigm cases of p-conscious states are sensations, states with sensory or qualitative character. Access consciousness involves states that can broadcast for free use in reasoning and for direct “rational” control of action. Access conscious states are “globally accessible” and influence behavior by flexibly interacting with the beliefs, desires and goals of a creature. 2. In the context of this view of perception, it would be natural to suggest that the phenomenon of bokeh has a pleasing aesthetic effect because it mimics our first-person perceptual experience of focusing on objects in the environment not only in terms of visual depth of field but also cognitive depth. That is, a highly focused top picture of a stapler in Figure 5.3 with a lot of bokeh lessens the cognitive load that the visual system has to exert to perceive the stapler since the picture itself has done a lot of the work by making the stapler the only salient object there. Conversely, when perceiving the same stapler in the bottom picture that has very minimal bokeh in Figure 5.3, the visual hierarchy has to work harder since it has to attentionally “carve” the stapler from a busy environment with many objects of a similar level of perceptual and cognitive saliency (similar size, resolution, category of objects, etc.). 3. An excellent overview of change blindness phenomenon with several examples can be found on Rensink’s webpage: https://www2.psych.ubc.ca/~rensink/ flicker/index.html.

References Anton-Erxleben, K., Henrich, C. and Treue, S. (2007) “Attention Changes Perceived Size of Moving Visual Patterns” in Journal of Vision, 7 (11), 1–9.

104  Anatoly Nichvoloda Atkinson, R., Shiffrin, R. (1968) “Human Memory: A Proposed System and Its Control Processes” in Psychology of Learning and Motivation (ed. Kenneth W. Spence and Janet Taylor Spence), Vol. 2, New York: Academic Press, 89–195. Baars, B. (1988) A Cognitive Theory of Consciousness, Cambridge: Cambridge University Press. Baddeley, A. (2007) Working Memory, Thought, and Action, Oxford: Oxford University Press. Block, N. (1997) “On a Confusion about a Function of Consciousness” in The Nature of Consciousness (ed. N. Block, O. Flanagan and G. Guzeldere), Cambridge, MA: MIT Press. Block, N. (2001) “Paradox and Cross Purposes in Recent Work on Consciousness” in Cognition, 79, 197–220. Block, N. (2007) “Consciousness, Accessibility, and the Mesh between Psychology and Neuroscience” in Behavior and Brain Sciences, 30, 481–548. Block, N. (2011) “Perceptual Consciousness Overflows Cognitive Access” in Trends in Cognitive Sciences, 12, 567–575. Carrasco, M., Ling, S. and Read, S. (2004) “Attention Alters Appearance” in Nature: Neuroscience, 7 (3), 308–313. Carruthers, P. (2015) The Centered Mind: What the Science of Working Memory Shows Us About the Nature of Human Thought, Oxford: Oxford University Press. Clark, A. (2013a). “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science” in Behavioral and Brain Sciences, 36 (3). Cowan, N. (1995) Attention and Memory: An Integrated Framework, New York, NY: Oxford University Press. Crick, F. and Koch, C. (1990) “Towards a Neurobiological Theory of Consciousness” in Seminars Neurosciences, 2, 263–275. D’Esposito, M., Postle, B., Ballard, D., Lease, J. (1999) “Maintenance versus Manipulation of Information Held in Working Memory: An Event-Related fMRI Study” in Brain and Cognition, 41 (1), 66–86. Dretske, F. (2004) “Change Blindness” in Philosophical Studies, 120, 1–18. Dretske, F. (2007) “What Change Blindness Teaches about Consciousness” in Philosophical Perspectives, 21, 215–230. Evans, G. (1982) The Varieties of Reference (ed. John McDowell), Oxford: Oxford University Press. Feldman, H. and Friston, K. (2010) “Attention, Uncertainty and Free Energy” in Frontiers in Human Neuroscience, 4, 215. Hohwy, J. (2012) “Attention and Conscious Perception in the Hypothesis Testing Brain” in Frontiers in Psychology, 3, 96. Knudsen, E. (2007) “Fundamental Components of Attention” in Annual Review of Neuroscience, 30, 57–78. Lamme, V. (2003) “Why Visual Attention and Awareness Are Different” in Trends Cognitive Science, 7, 12–18. Lamme, V. (2004) “Separate Neural Definitions of Visual Consciousness and Visual Attention: A  Case for Phenomenal Awareness” in Neural Networks, 17, 861–872. Lamme, V. (2006) “Towards a True Neural Stance on Consciousness” in Trends Cognitive Science, 10, 494–501.

“Hierarchical Bokeh” Theory of Attention 105 Lycan, W. (1996) Consciousness and Experience, Cambridge, MA: MIT Press/ Bradford Books. Mole, C. (2008) “Attention and Consciousness” in Journal of Consciousness Studies, 15 (4), 86–104. Müller, N., Mollenhauer, M., Rösler, A. and Kleinschmidt, A. (2005) “The Attentional Field has a Mexican Hat Distribution” in Vision Research, 45, 1129–1137. Nichvoloda (forthcoming) “Quantum Uncertainty Reduction Theory of Attended Phenomenal and Access Consciousness” in Journal of Consciousness Studies. Noe, A. (2004) Action in Perception, Cambridge, MA: MIT Press. O’Regan, J. and Noe, A. (2001) “A Sensorimotor Approach to Vision and Visual Consciousness” in Behavior Brain Science, 24, 883–975. O’Shaughnessy, B. (1981) The Will: A  Dual Aspect Theory, Cambridge/New York: Cambridge University Press. Parr, T. and Friston, K. (2017) “Working Memory, Attention, and Salience in Active Inference” in Nature: Scientific Reports, 7. Peacocke, C. (2000) “Conscious Attitudes, Attention, and Self-knowledge” in Knowing Our Own Minds (ed. C. Wright, B. Smith and C. Macdonald), Oxford: Oxford University Press. Posner, M. (1994) “Attention: The Mechanisms of Consciousness” in Proceedings of the National Academy of Sciences, 91 (16), 7398–7403. Prinz, J. (2012) The Conscious Brain, Oxford: Oxford University Press. Rensink, R. A. et al. (1997) “To See or Not to See: The Need for Attention to Perceive Changes in Scenes” in Psychology Science, 8, 368–373. Schaffer, J. (2004) “Causes Need Not Be Physically Connected to their Effects: The Case for Negative Causation” in Contemporary Debates in Philosophy of Science (ed. C. Hitchcock), Oxford: Blackwell, 197–216. Shomstein, S. and Yantis, S. (2006) “Parietal Cortex Mediates Voluntary Control of Spatial and Nonspatial Auditory Attention” in Journal of Neuroscience, 26, 435–439. Simons, D. and Rensink, R. (2005) “Change Blindness: Past, Present and Future” in Trends Cognitive Science, 9, 16–20. Sperling, G. (1960) “The Information Available in Brief Visual Presentations” in Psychological Monographs, 74, 1–29. Tye, M. (2010) “Attention, Seeing and Change Blindness” in Philosophical Issues, 20. Watzl, S. (2010) The Significance of Attention, PhD Thesis, Columbia University. Watzl, S. (2011a) “The Philosophical Significance of Attention” in Philosophy Compass, 6 (11), 722–733. Watzl, S. (2011b) “The Nature of Attention” in Philosophy Compass, 6 (11), 842–853.

6 Perceiving Live Improvisation in the Performing Arts Aili Bresnahan

1. Introduction This chapter will explore the ways that improvisational performances before a live audience by professional-level actors, musicians and dancers (who I accept as experts in their respective domains) take place at both cognitive and sub-cognitive levels in ways that are relevant for understanding perception and appreciation of the performing arts. First, evidence from cognitive science will be used to show that we often process visual stimulus at the same time or slightly later than when we prepare the body for action. What this means is that improvising, as in a dance or a music jam session or a scene in theater, may involve physical responses that occur before we are conscious of the event to which we are responding. Second, this chapter will demonstrate how understanding these cognitive processes of the performers can help audience members to pinpoint why live improvisational performances have aesthetic value. The properties that create aesthetic value include the aesthetic qualities of immediacy and “being in the now,” as well as other qualities that provide emotional and intellectual pleasure and satisfaction. Next, this understanding of the performative process of improvisation by audience members will serve to introduce a consideration of the extent to which critical appreciation involves the enrichment and supplementation of perceptual experience with interpretive practice. Like the improvising performing artist, the audience member, too, has cognitive processes that occur before conscious articulation of what they have perceived. It is only after a self-reflective and conscious process of considering what they have perceived and experienced in live performance that a critical appreciator can provide a critical analysis of the performance. This means that evaluative judgments of live improvisation in the arts, like the improvisatory decisions that are made by the performers in the performances that they are judging, are not made at the purely perceptual level. Let us begin by a discussion of how perception and proprioception work together in the motor action of performing artists engaged in live improvisation.

Perceiving Live Improvisation 107

2.  Perception and Proprioception in Motor Action One of the ways that our bodies prepare for motor action is via proprioception, which can be roughly defined as our internal sense of where our bodies are located and positioned in space (cf. Fridland, 2011: 523). For purposes of this chapter I will use the term proprioception, following James Gibson, to describe internal experiences of the body that may rely on but that are not solely limited to experiences of the environment outside of the body (cf. Gibson, 2002: 77–78). As such, proprioception is not identical to perception, which is more commonly defined as limited to experiences of the outside environment (Gibson 2002: 78).1 Proprioception, however, usually accompanies perception in motor action, as will be described more fully below. We can either be consciously aware of our proprioception or it can operate at a level below our conscious awareness. This is true of perception as well. For example, we can choose to focus our vision on a flowering tree, in an attitude where we consciously hone in on the tree, or we can just orient ourselves toward the world in a way that includes the tree as part of our sensory field at large. Maurice Merleau-Ponty provides this example: If I  wish to enclose myself in one of my senses and, for example, I  project myself entirely into my eyes and abandon myself to the blue of the sky, soon I  am no longer aware of gazing and, at just the moment I wanted to give myself over to vision entirely, the sky ceases to be a “visual perception” in order to become my current world. Sensory experience is unstable and wholly unknown to natural perception, which is accomplished with our entire body all at once and opens onto an inter-sensory world. Like the experience of the sensible quality, the experience of isolated “senses” takes place only within an abnormal attitude. (Merleau-Ponty, 2014: 234) “Seeing,” then, can be cognitive awareness of seeing something specific, although it need not be in every case. At the heart of all perception, for Merleau-Ponty and for many philosophers of mind and cognitive scientists in his wake, is motor action—a body oriented and interacting with the world, with the agentive person one who is not separable from their body (as the captain of a ship is separate from the ship) but a necessary part of this body—a moving-thinking-doing live organism (we are more like a whale, perhaps, than like a caption of a ship) (cf. e.g., Noë, 2004). We can think of proprioception in a similar vein. British neuroscientist Karl J. Friston gives this account of how perception and proprioception work together in motor action: In motor systems error signals [incoming signals that do not match our predictions of what we will encounter in the environment]

108  Aili Bresnahan self-suppress, not through neuronally mediated effects, but by eliciting movements that change bottom-up proprioceptive and sensory input. This unifying perspective on perception and action suggests that action is both perceived and caused by its perception. (Clark, 2013: 186) A dancer, for example, can be consciously aware of taking no more than three small turns toward the edge of the stage to avoid falling over the edge, a decision that would entail conscious processing. During the course of these turns, however, they will also be using unconscious proprioception to maintain and fine-tune their balance. And in the case of noticing the edge of the stage it is usually the case (except perhaps in the case of the vision-impaired dancer or if the dance takes place in utter darkness) that their vision aids their proprioception in avoiding danger.2 An additional wrinkle is that after many, many performances of focusing on three small edge-avoiding turns, the turns become so habitual that part of their operation becomes unconscious, even if the dancer is still directing some of what occurs on the conscious level.3 This happens because, as Benjamin Libet and his colleagues discovered in the 1980s, there is a neuroscientific basis for the idea that we plan and start to execute movements unconsciously before we are aware that we have “decided” to do them (Libet, 1983).4 Neuroscience has thus corroborated the earlier account of intentional motor action given by Maurice Merleau-Ponty nearly 40 years earlier. “When I motion to my friend to approach,” Merleau-Ponty explains, my intention is not a thought that I  could have produced within myself in advance, nor do I perceive the signal in my body. I signal across the world; I signal over there, where my friend is. The distance that separates us and his consent or refusal are immediately read in my gesture. There is not first a perception followed by a movement, the perception and the movement form a system that is a modified whole. If, for example, I realize that my friend does not want to obey, and if I thereby modify my gesture, we do not have here two distinct conscious acts. Rather, I see my partner’s resistance, and my impatient gesture emerges from this situation, without any interposed thought. (Merleau-Ponty, 2014: 113) Shaun Gallagher, a “4e” philosopher of mind (one who believes that the mind is embodied, extended, enactive and embedded), whose work builds on that of Merleau-Ponty, has a similar account of how motor action is prepared via pre-conscious awareness, pointing out that “we’ve known for a long time that anticipatory processes are hugely important

Perceiving Live Improvisation 109 for perception and action (Gallagher, 2014: 240)”5 He notes that work in predictive coding accounts for this in neuronal terms: There is general agreement that the process is a hierarchical one involving synaptic inhibition based on an empirical prior—­something that depends on the organism’s previous experience and context-­sensitive learning. This means that specific neural networks, currently in a particular state because the organism has previously encountered a particular stimulus or environment, or has a particular history, determine ongoing processes that lead to top-down synaptic inhibition (a modulation of connections) anticipating further processing consistent with prior processing. Such inhibitory patterns constitute a prediction which is then matched against ongoing sensory input. If there is a mismatch, i.e., if the new stimulus generates a different firing pattern than the one anticipated, prediction errors are sent back up the line and the system adjusts dynamically back and forth until there is a relatively good fit. So, on the predictive coding model, this or something like this is going on when we perceive the world. If that’s the case, if these are the kinds of things that are happening in the brain, it’s not clear that we need to think of it as a kind of inference rather than a kind of dynamic adjustment process in which the brain, as part of and along with the larger organism, settles into the right kind of attunement with the environment—an environment that is physical but also social and cultural. (Gallagher, 2014: 241) An example from driving can be used here to flesh out this idea a bit more. We can drive a familiar route home, for example, without necessarily focusing consciously on where and how to take each turn. Indeed, we may hit the brakes when a car in front of us stops unexpectedly (to avoid hitting a child, say) and only realize after we have stopped what has occurred.6 This is particularly likely to occur where the actions require split-second processing that does not allow time for conscious decisionmaking and planning. Someone who has to make a conscious decision to hit the brakes or wait to recognize that the car in front of them has done so before hitting the brakes is far more likely to get into an accident. This is why experience in common car-related motor actions is necessary to drive a car safely and why experienced motorists get into fewer accidents than do novices. Indeed, highly skilled motor actors (who may not have just experience but excellence in the domain at issue—such as professional racecar drivers or the performing artists at issue here) can rely on unconscious neural processes that regulate simpler physical tasks while also focusing on more complex, higher-level tasks (such as expressive nuances in artistic performance) because these latter tasks require

110  Aili Bresnahan working memory and direction.7 This is where live improvisation in professional-level performing arts comes in.8 One common misconception about improvisation is that it is entirely spontaneous ad hoc activity. In fact, nothing could be further from the truth.9 A professional-level artist who is making and doing something for audience appreciation, whether there are aspects of their activity that are newly improvised or not (those aspects that make the performance or part of it “improvisation”), embodies a complex set of bodily, perceptual and cognitive motor schemas for action that have been trained along the pathways necessary for expertise in their art(s). These “schemas” include the body schema, “the prereflective, automatic and sensorimotor ‘know how’ that the body possesses in familiar action” (Merritt, 2015: 98). (cf. Merleau-Ponty, 2014: 100–103) This “body,” following MerleauPonty, is not a puppet-like object to be manipulated by a thinking brain but “the power for a certain world.” (Merleau-Ponty, 2014: 113) Antonio Damasio’s view that a live organism’s behavior is a multilevelled operation that involves several concurrent biological processes assists this account by explaining how it is that we can both know that we are acting, as in the course of improvising and yet not cognize every part of our performance. “It may be helpful to think of the behavior of an organism as the performance of an orchestral piece,” Damasio points out, “whose score is being invented as it goes along” (Damasio, 1999: 87). Joshua Bergamin’s discussion of the experience of performing music corroborates this idea. Jazz soloists, he points out, have active decisions that “presuppose a ‘smooth, almost automatic’ use of their instrument” (Bergamin, 2017: 410). He further notes that “expert musicians cope smoothly in performances while their minds focus on different thoughts that may or may not be relevant to their playing, with one informant describing the experience as like having ‘two tracks running’ ” (Ibid.). Thus, an actor, musician, or dancer who is improvising, whether an entirely new piece or a section or portion of a performance, can rely on many of the trained tricks of the trade—such as projecting one’s voice or executing a step or playing a note—in an unconscious way while focusing on new embellishments (cf. Limb and Brain, 2008: 3; Altenmüller, 2003). These embellishments may include where to add a humorous aside while reading set lines in a play, for example, a musical rhythm change for expressive purposes, or an extra moment of suspended “hold” in a dance jump. Any actions that become habitual parts of a performer’s unconscious repertoire, that become automatic motor schemas or programs, will rely less on vision and other sensory perception accordingly (Doyon et al., 2009). An artist can also become an expert on improvising within an artistic performance domain due to habitual rehearsal and practice in similar, even if not identical, performance contexts because they can rely on modes of performance they have employed before. In this case, such an

Perceiving Live Improvisation 111 artist has a motor schema (a body-brain-performance-environment habit of action in a live performance context) that I  think of as an artistic schema because of the importance of the artistic context of this training (Crease, 1994: 191). As established by Bruineberg and Rietveld, social and cultural influences such as artistic practices do, indeed, affect us as part of the environmental affordances upon which our “brain as an intrinsically active and unstable organizing system” interacts (Bruineberg and Rietveld, 2014). They note, for example, that our interactions with the environment are often skilled developments of habits and practices that are part of our customary ecological niches, which include sociocultural affordances and choices of “only the relevant affordances in a given situation” (Bruineberg and Rietveld, 2014). All of this is consistent with the account provided by Merleau-Ponty, for whom the body is both a vehicle of understanding and a sense-giving and dynamic organism that gives meaning to both natural and cultural objects (Merleau-Ponty, 2014: 286). A stand-up comic, for example, might develop a stylistic way of dealing with hecklers—finding something to make fun of about their clothing, their inebriation, or something else about them—even if each specific heckler address is particular to that situation. For example, “Oh, you don’t like that, Idaho (where someone is wearing an Idaho t-shirt)? Would you prefer to watch the potatoes grow back home?” A jazz pianist might find when improvising with high-volume brass musicians that they want to simplify the accompaniment and lower the volume to highlight the brass (or take the opposite tack). And a ballet dancer who is dancing a series of “farewell” performances before retirement might decide to add different short excerpts from their past repertoire into their reverences (bows at the end) each night to delight those fans who will recognize the reference. These motor actions all take account of cultural and social artistic contexts in how they operate. We turn now to how understanding the cognitive motor processes involved in improvisation in the performing arts, including perception and proprioception as part of the development of trained motor and artistic schemas as described above, can help to enrich audience appreciation.

3. Understanding Performers’ Cognitive Processes Assists Audience Appreciation Audience members who appreciate live improvisation often do so because of the feeling of excitement and online action that happens when neither the audience nor the performers themselves know precisely what will happen next, although all (especially the performers) have some idea.10 The feeling, as an audience member, is a communal one of sharing a live experience with both the performers and other audience members. The fact that at least some aspects of what happens on stage are a surprise to

112  Aili Bresnahan the performers as well as the audience enhances the excitement of anticipation and suspense. It adds vicarious delight in the performers’ discovery of what they have created at the same time that the audience is experiencing pleasure in their own experiencing of discovering the surprise. Thus, there is joy when comedic performers, as in Saturday Night Live, break “the fourth wall” and laugh at their own improvisational antics as well as that of other performers. We enjoy the performers’ enjoyment of each other’s performances while enjoying the performances ourselves, in part, perhaps, because we recognize the performers as expert appreciators of exceptional achievement.11 Erum Naqvi notes that appreciation of trained improvisational achievement is the aesthetic ethos of the radif tradition in Iranian music. She explains: Achieving improvisational creativity within the stringent structural parameters of the radif is considered the hallmark of artistry in this tradition. Traditional instruction involves one-to-one transmission by rote through observation and bodily mimicry over the course of several years. This process is thought to enable an intuitive absorption of the radif, taken as the structuring framework that implicitly guides improvisation. Under the traditional ethos, apprentices become musicians when they find themselves able to spontaneously reproduce and rearrange motifs in a manner consistent with the radif’s structure without needing to actively think about the structuring of a performance. (Naqvi, 2018: 64) This example attests to both the aesthetic value of achievement and the account of entrained expert motor skills described in Section 2. There is a sort of cognitive, appreciative pleasure—appreciating that something difficult has occurred—but it is also attunement to a feature of performance that also allows for the sort of fine perceptual discriminations that increase pleasure in the perceptions themselves. This is not unlike being able to appreciate notes of floral, of oak and of wood in a wine where before one could only taste fruit. It is not just the fact that one can discriminate tastes that creates the pleasure, but now there is a variety of experiential pleasures where before there was only one. This is how cognition and perception merge to create enhanced perceptual aesthetic pleasure. Gary Peters notes, in his second book on the philosophy of improvisation, that Jimi Hendrix’s famous guitar solos were exceptional precisely because of his great achievement as an improvisor within those solos (Peters, 2017: 177). Indeed, he records the pleasure of being a participant in the aesthetic experience that such improvisation provides by using the example of witnessing one of Hendrix’s guitar solos at the Woburn Pop

Perceiving Live Improvisation 113 Festival on July 6, 1968. He explains that Hendrix arrived late, without his band, and proceeded to play an (ostensibly) improvised guitar solo: at this moment Jimi simply stood, solitary and stationary to one side of the stage in semi-shadow. He neither said nor did anything except play. In truth, he didn’t even play; he simply allowed his guitar to resonate and emit a slowly increasing howl of such despairing intensity that the druggy love-in that had been so happily unfolding all day in the Bedfordshire sunshine suddenly became as trivial as its tinkling bells and faux Buddhist chanting. My memory tells me this continued for at least thirty minutes, thirty minutes when (in philosophical retrospect) the Kantian concept of the sublime as the experience of the pleasure and pain associated with witnessing that which exceeds all experience, was itself experienced. (Peters, 2017: 171–172) Here Peters reports feeling something directly—an aesthetic experience (indeed, the sublime is a traditional aesthetic quality)—that was itself pleasurable. The guitar solo Peters describes above also lent itself to his own appreciative contemplation in the form of the philosophical account he provided in addition to this case study in his book. This is a sort of intellectual pleasure and good in connection with aesthetic and artistic pleasure. This experience of arts improvisation contains emotional pleasure as well. Jenefer Robinson has pointed out that “judgement theorists” of emotion find that experiencing an emotion like joy can come from cognitive judgments like the idea that this performance is an outstanding achievement, or sublime, or “beyond words!” (Robinson, 2005: 8–9). An audience also might find that they experience a catharsis of the emotions of pity or fear during Sophocles’s Oedipus Rex that is triggered (at least in part) by sharing the realization with Oedipus that he has killed his father and slept with his mother despite his extreme efforts to avoid doing both.12 If the actor’s expressions of grief are improvised in whole or in part, we may feel the shock and surprise even more keenly, as the actor to some extent feels the grief afresh as we do. Robinson has also provided evidence to show that emotional responses that include motor action, like jumping or screaming out of surprise or fear, for example, can occur before we are consciously aware of them. She calls these “non-cognitive affective appraisals” (Robinson, 2005: 58). Her list of theorists who hold that physiological changes are a central part of emotion (which is not motor action but related to motor action) includes Aristotle (who discusses the “boiling of the blood” and “hot stuff about the heart” in De Anima) and William James (who correlates certain physiological changes with emotion in his Principles of

114  Aili Bresnahan Psychology) (Robinson, 2005: 28–29). Robinson specifically endorses James’s idea here, holding that “indeed we don’t call a response emotional unless it is a physiological response that has been produced in a certain way” (Robinson, 2005: 57–58). Further, Robinson shows throughout her book how literature, music and the other arts can trigger these non-cognitive affective appraisals. This shows how audience responses are similar (though not identical) to performers’ in the sense that they include both conscious and unconscious perceptual and other processes at the same time. In dance, understanding the cognitive processes of the performers also serves to aid the research being done by dance philosophers who study kinesthetic and empathetic responses to dance as part of understanding the nature of dance appreciation. In short, kinesthetic responses are those that involve felt, bodily experiences of watching dance. These include visceral responses such as feeling the vibrations of the dancers’ jumps and feeling one’s pulse quicken when a dancer makes a terrifying plunge from a great height. When responses include feeling, in an imaginative if not literal way, certain movements witnessed onstage in one’s own muscles, and the emotional experience of what we imagine performing that movement would be like, then our kinesthetic response becomes empathetic as well. Empathetic responses include awareness that we can “get” and affirm through our own experience and imagination what it is that we understand the dancers are undergoing.13 Finally, they can include a sense of anticipation for what will come next. (Reason and Reynolds, 2001: 60, 64) The main thesis of this chapter—that understanding the scope and nature of perceptual and cognitive processes in live improvisation in the performing arts can assist us in understanding this phenomenon in philosophically relevant ways—has now been elucidated. We turn now to the sort of appreciation of live improvisation in the performing arts that is either entirely or mostly cognitive—what I have termed “critical appreciation” elsewhere. (Bresnahan, 2017) This is relevant for purposes of this chapter because it is a type of intellectual appreciation that builds upon, but is not limited to, perceptual and kinesthetic experience.

4. Critical Appreciation Involves Both Perceptual Experience and Interpretive Practice As discussed in Section 3, the audience’s experience of a live improvisation event in the performing arts is similar to the performing artist’s experience in the sense that they have cognitive, emotional, kinesthetic and empathetic responses that occur before conscious articulation of what they have perceived. At the purely perceptual level, what I have elsewhere called “innocent eye appreciation,” the audience member perceives in a way that is “innocent” in the sense that biological perception can operate as it does in animals (cows or pigeons for Danto) or others without

Perceiving Live Improvisation 115 cultural, art-historical or semiotic understanding of the artwork or practice at issue (Bresnahan, 2017: 39–44). At the base level of perception, an audience member has no developed expertise in taking this appreciation to the deeper level of cognitive and art-historical reflective understanding. Expert interpretation, like expert performance, requires training and practice in the domain at issue—it requires its own entrainment and habit-forming practice in order to truly “see” all that is there to be seen. Of course, there is no actual person who exists only at the level of homo sapiens, the term Joseph Margolis uses to denote a human person at the level of biology only. All human beings exist in culture, which is why they are “persons” rather than merely homo sapiens (Margolis, 1995: 190; 2010: 10).14 The issue is therefore not one of distinguishing between cultured and uncultured persons (no one is uncultured who lives in human society) but whether or not the culture in which the person lives has provided training and enculturation in the performing arts practice at issue. If not, properties of the performance may be there that the audience member simply does not “perceive” because they cannot appreciate them. To hear a sound, for example, is not to hear the sound as music; something with semiotic meaning in culture. Indeed, Naqvi is correct that if one isn’t trained in Iranian classical music, for example, one would not be familiar with the structure of the radif (a “canon of motifs,” a “theoretical model” or “a musical grammar of sorts”) and thus would not be able to detect the improvisational creativity within it (Naqvi, 2018: 64). Here too it is not difficult to find analogues in the arts of dance and theater as well. In dance, an untrained audience member might see a dancer spinning around, but they would not understand or appreciate the significance of Odile’s 32 fouetté turns. And in Shakespeare, much of the comedy—what is there to understand as funny—is lost for anyone who does not understand the double entendres, as when Hamlet tells Ophelia “get thee to a nunnery,” where “nunnery” (in Shakespeare’s time) meant both convent and bordello.

5. Conclusion In conclusion, perception of live improvisation in the performing arts is not a matter of pure perception, by which I mean the idea that it could exist in isolation from motor, physiological, emotional, kinesthetic, empathetic, interpretive and appreciative modes of accessing it. This is true because both improvising performers and appreciating audience members are complex live organisms with many levels of information processing who are situated in a cultural environment that includes multiple affordances for action and uptake that occur prior to perception. This accounts for the all-encompassing richness and live vitality that experiences of the performing arts provide. It is no wonder, then, that we value and seek out such experiences.

116  Aili Bresnahan

Notes 1. Cf. Fridland, 2011. 2. See discussion of perceptual adaptation and the role of proprioception and vision in Noe͏̈, Action in Perception, 93–95, and 100. See also Bence Nanay’s discussion of action-oriented perception in “Perception, Action, and Identification in the Theater,” (2009) 246–249. 3. For an interesting debate over to what extent expert or “esoteric” rather than everyday or “ubiquitous” motor action is “automatic” (Dreyfus, 2007), marked by conscious directed thought (Montero, 2010) or “spontaneous,” which can include “instructional nudges” (his own view), see Bergamin (2017: 403–412). 4. Patrick Haggard and Martin Eimer have corroborated this research in “On the Relation Between Brain Potentials and the Awareness of Voluntary Movements” (1999: 128–133), as have Michel Desmurget and Angela Sirigu in “A Parietal-Premotor Network for Movement Intention and Motor Awareness” (2009: 411–419). 5. For a corroboratory account of “action-oriented predictive processing,” see Andy Clark, “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science” (2013: 185–186). 6. In “The Case for Proprioception,” 527, Fridland provides a similar example from Fred Dretske of the sort of minimal consciousness required by driving. 7. See Juan Pablo Bermúdez’s “attention-control” account of skillful motor actions as relying on both automatic and reflective processes in a dual-­ levelled way in “Do We Reflect While Performing Skillful Actions? Automaticity, Control, and the Perils of Distraction” 7 (2017: 900–901). See also Bergamin, “Being-in-The-Flow,” 410. Cf. Barbara Gail Montero’s discussion of what she calls “the just-do-it principle” versus her choice of “cognitionin-action,” Thought in Action: Expertise and the Conscious Mind, 32–40. 8. See Aili Bresnahan, “Improvisation in the Arts,” 2015: 573 for improvisation in the arts in general and 576–577 for an account of improvisation in the performing arts. 9. For a further discussion of this claim, see Bresnahan, “Improvisation in the Arts,” 574. 10. See Foster, “Taken by Surprise: Improvisation in Dance and Mind,” for an account of how this works in dance, particularly at p. 4. See also Bresnahan, “Improvisation in the Arts” for more evidence of this live improvisation phenomenon in music and theater. Cf. Crease, 1994: 183, who discusses performances as valued for their own sake as well as because they are enthralling. 11. Reason and Reynolds’s research has shown that a major component of the audience response to watching dance is what they have termed, “admiration of virtuosity.” Reason Matthew and Reynolds, 2001: 58. 12. See Aristotle’s Poetics for his view of catharsis in tragedy and Politics for his view of catharsis in the experience of music. 13. For more on kinesthetic and empathetic responses in general, see Reason and Reynolds, “Kinesthesia, Empathy, and Related Pleasures,” 49–75. See also Conroy, “Responding Bodily,” 203–10, and Carroll and Seeley, “Kinesthetic Understanding and Appreciation in Dance,” 177–186. 14. For more on Margolis’s theory of the human person as it connects to interpretation in the arts, see also his books: Interpretation Radical but Not Unruly: The New Puzzle of the Arts and History, Selves and Other Texts: The Case for Cultural Realism and What, After All, Is a Work of Art?

Perceiving Live Improvisation 117

References Altenmüller, Eckhart (2003) “Focal Dystonia: Advances in Brain Imaging and Understanding of Fine Motor Control in Musicians” in Hand Clinics, 19 (3), 523–538. Bergamin, Joshua A. (2017) “Being-in-the-Flow: Expert Coping as Beyond Both Thought and Automaticity” in Phenomenology and the Cognitive Sciences, 16, 403–424. Bermúdez, Juan Pablo (2017) “Do We Reflect While Performing Skillful Actions? Automaticity, Control, and the Perils of Distraction” in Philosophical Psychology, 30 (7), 896–924. Bresnahan, Aili (2015) “Improvisation in the Arts” in Philosophy Compass, 9 (10), 573–582. Bresnahan, Aili (2017) “Dance Appreciation: The View from the Audience” in Aesthetics: A  Reader in the Philosophy of the Arts, 4th edition (ed. David Goldblatt, Lee Brown, and Stephanie Patridge), New York: Routledge. Bruineberg, Jelle and Rietveld, Erik (2014) “Self-organization, Free Energy Minimization, and Optimal Grip on a Field of Affordances” in Frontiers in Human Neuroscience, https://doi.org/10.3389/fnhum.2014.00599, accessed May  23, 2018. Carroll, Noël and Seeley, William P. (2013) “Kinesthetic Understanding and Appreciation in Dance” in The Journal of Aesthetics and Art Criticism, 71 (2), 177–186. Clark, Andy (2013) “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science” in Behavioral and Brain Sciences, 36 (3), 181–253. Conroy, Renee M. (2013) “Responding Bodily” in The Journal of Aesthetics and Art Criticism, 71 (2), 203–210. Crease, Robert P. (1994) “The Improvisational Problem” in Man and World, 27, 181–193. Damasio, Antonio (1999) The Feeling of What Happens: Body and Emotion in the Making of Consciousness, San Diego: Harcourt, Inc. Danto, Arthur C. (1992) “Animals as Art Historians: Reflections on the Innocent Eye” in Beyond the Brillo Box: The Visual Arts in Post-Historical Perspective, Berkeley: University of California Press. Danto, Arthur C. (2001) “The Pigeon Within Us All: A Reply to Three Critics” in The Journal of Aesthetics and Art Criticism, 59 (1), 39–44. Desmurget, Michel and Sirigu, Angela (2009) “A Parietal-Premotor Network for Movement Intention and Motor Awareness” in Trends in Cognitive Sciences, 13 (10), 411–419. Doyon, Julien et al. (2009) “Contributions of the Basal Ganglia and Functionally Related Brain Structures to Motor Learning” in Behavioural Brain Research, 199, 61–75. Dreyfus, Hubert L. (2007) “The Return of the Myth of the Mental” in Inquiry, 50 (4), 352–365. Foster, Susan Leigh (2003) “Taken by Surprise: Improvisation in Dance and Mind” in Taken by Surprise: A  Dance Improvisation Reader, Middletown, CT: Wesleyan University Press.

118  Aili Bresnahan Fridland, Ellen (2011) “The Case for Proprioception” in Phenomenology and the Cognitive Sciences, 10 (4), 521–540. Friston, Karl J. (2003) “Learning and Inference in the Brain” in Neural Networks, 16 (9), 1325–1352. Gallagher, Shaun. (2014) “Making Enactivism Even More Embodied” in AVANT, V (2), 232–247. Gibson, James J. (1966) The Senses Considered as Perceptual Systems, Boston: Houghton Mifflin. Gibson, James J. (2002) “A Theory of Direct Visual Perception” in Vision and Mind: Selected Readings in the Philosophy of Perception (ed. Alva Noë and Evan Thompson), Cambridge, MA: MIT Press. Haggard, Patrick and Eimer, Martin (1999) “On the Relation Between Brain Potentials and the Awareness of Voluntary Movements” in Experimental Brain Research, 126, 128–133. Høffding, Simon (2014) “What Is Skilled Coping?” in Journal of Consciousness Studies, 21 (9), 49–73. Libet, Benjamin et al. (1983) “Time of Conscious Intention to Act in Relation to Onset of Cerebral Activity (Readiness-Potential): The Unconscious Initiation of a Freely Voluntary Act” in Brain, 106, 623–642. Limb, Charles J. and Brain, Allen R. (2008) “Neural Substrates of Spontaneous Musical Performance: An fMRI Study of Jazz Improvisation” in PLoS ONE, 3 (2), 1–9, https://journals.plos.org/plosone/article/file?id=10.1371/journal. pone.0001679&type=printable, accessed September 1, 2018. Margolis, Joseph (1995a) Historied Thought, Constructed World: A Conceptual Primer for the Turn of the Millennium, Berkeley: University of California Press. Margolis, Joseph (1995b) Interpretation Radical but Not Unruly: The New Puzzle of the Arts and History, Berkeley: University of California Press. Margolis, Joseph (1999) What, After All, Is a Work of Art, University Park, PA: The Pennsylvania State University Press. Margolis, Joseph (2001) Selves and Other Texts: The Case for Cultural Realism, University Park, Pennsylvania: The Pennsylvania State University Press. Margolis, Joseph (2010) The Cultural Space of the Arts and the Infelicities of Reductionism, New York: Columbia University Press. Merritt, Michele (2015) “Thinking-is-Moving: Dance, Agency, and a Radically Enactive Mind” in Phenomenology and Cognitive Science, 14, 95–110. Merleau-Ponty, Maurice (2014) Phenomenology of Perception (trans. Donald A. Landes), London and New York: Routledge. Montero, Barbara Gail (2010) “Does Bodily Awareness Interfere with Highly Skilled Movement?” Inquiry, 53 (2), 105–122. Montero, Barbara Gail (2016) Thought in Action: Expertise and the Conscious Mind, Oxford: Oxford University Press. Nanay, Bence (2009) “Perception, Action, and Identification in the Theater” in Staging Philosophy: Intersections of Theater, Performance and Philosophy (ed. David Krasner and David Z. Saltz), Ann Arbor: The University of Michigan Press. Naqvi, Erum (2018) “Musical Aesthetics in Iran” in The Philosophers’ Magazine, 82, (3rd Quarter): 63–67. Noë, Alva (2004) Action in Perception, Cambridge, MA: MIT Press.

Perceiving Live Improvisation 119 Obhi, Sukhvinder S. and Haggard, Patrick (2004) “Free Will and Free Won’t: Motor Activity in the Brain Precedes Our Awareness of the Intention to Move, So Sow is it that We Perceive Control?” in American Scientist, 92 (4), 358–365. Peters, Gary (2017) Improvising Improvisation: From Out of Philosophy, Music, Dance, and Literature, Chicago: University of Chicago Press. Pike, Alfred (1974) “A Phenomenology of Jazz” in Journal of Jazz Studies, 2 (1), 88–94. Reason, Matthew and Reynolds, Dee (2001) “Kinesthesia, Empathy, and Related Pleasures: An Inquiry into Audience Experiences of Watching Dance” in Dance Research Journal, 42 (2), 49–75. Robinson, Jenefer (2005) Deeper Than Reason: Emotion and its Role in Literature, Music, and Art, Oxford: Clarendon Press.

Section II

Cognition

7 Consciousness and Content in Perception1 Bill Brewer

Introduction Normal perception involves conscious experience of the world. What I call the Content View (CV) attempts to account for this in terms of the representational content of perception (Brewer, 2011: esp. ch. 4). I offer a new argument against this view in this chapter.2 Ascription of personal-level content, either conceptual or nonconceptual, depends on the idea that determinate predicational information is conveyed to the subject. This determinate predication depends upon the exercise of certain personal-level capacities for categorization and discrimination. Exercise of such personal-level capacities depends in turn upon conscious selective attention. Yet conscious visual acquaintance with the world is the prior ground for the possibility of any such conscious selective attention. Acquaintance obtains throughout the visual field: Where conscious attention is not actually directed as well as where it is. So acquaintance does not depend upon conscious selective attention. Thus, acquaintance is not sufficient for the exercise of the relevant personal-level capacities. Exercise of these capacities is nevertheless necessary for personal-level content. Therefore, visual acquaintance cannot be understood in terms of perceptual content: Basic conscious experience of the world is not a matter of anything like the predication involved in perceptual content. It is rather the relational ground for the possibility of such predication. Clearly every move in this argument needs clarification and defense. I offer this in Section 1. In Section 2, I consider the implications for the Content View. Section 3 concludes.

1.  The Key Argument My Key Argument (KA) may be set out in abbreviated form as follows. (1) P-Level Content → P-Level Predication (2) P-Level Predication → P-Level Capacities (3) P-Level Capacities → Conscious Selective Attention

124  Bill Brewer (4) Not-(Acquaintance → Conscious Selective Attention) (5) So Not-(Acquaintance → P-level Capacities) (6) So Not-(Acquaintance → P-level Content) This section offers elucidation of its constituent notions and motivation for its substantive premises. (1) P-Level Content is the proposition that: A subject, S, is in a conscious perceptual state with personal-level representational content. This means that S’s perceptual experience represents it as being the case that certain elements of the world around her are some more or less determinate way.34 Now her perceptual experience is a personal-level representational state, rather than one deployed only subpersonally. So this more or less determinate predication is conveyed in some way to her. This is P-Level Predication, the proposition that: There is a more or less determinate way that certain worldly elements are experientially represented as being to S herself. Thus: P-Level Content → P-Level Predication (2) It follows from P-Level Predication that S is in receipt of more or less determinate predicational information, or misinformation, about the world around her. Now the predication involved here is in principle indefinitely variable in what might be called its fineness of grain. More precisely, any specific case will involve the ascription of a more or less determinate property to an object, where this will be one of a range of mutually exclusive and jointly exhaustive such determinates of a given determinable; and that determinable may in turn be partitioned into distinct such ranges of determinates in indefinitely many ways. Each of these ranges will contain a determinate, or multiple determinates, whose extensions overlap with that of the determinate ascribed in the initial case in question. Ascriptions of these alternatives would be predications at a different fineness of grain. A highly simplified example in the case of color may help to clarify the point. Suppose we begin with a content ascribing a blue color to an object, b. Candidate more or less determinate predications here might be any of those illustrated in the diagram below, as follows: (a) b is blue as opposed to green or purple; (b) b is dark blue as opposed to light blue; (c) b is medium blue as opposed to navy, Duke, light or sky blue, (d) b is B5, as opposed to any of the other fine Bn shades.5 The point generalizes quite straightforwardly to other basic predications made on the basis of visual perception, such as direction, distance, size, orientation, shape, texture and so on. Thus, the ascription to the subject, S, of more or less determinate predicational information, or misinformation, about the world around her requires

Consciousness and Content in Perception 125 | | B1

B Dark Navy B2

– – | B3

– Blue Duke B4

L – | B5

– Medium B6

– | | B7

U Light Sky B8

– – | B9

– Blue Light B10

E – | B11

| | B12

a principled account of where in this indefinite variety we are to locate the correct fineness of grain involved. The argument for premise (3) is that this principled account of the more or less determinate predication involved in perceptual content is to be given in terms of the specific capacities for categorization and discrimination that are exercised in her experience conveying that information. Furthermore, since the predication involved here is at the personal level, the capacities in question must also be personal-level capacities: Capacities exercised by S herself. The predication involved in personallevel perceptual content conveys more or less determinate information (or misinformation) of a specific fineness of grain to the subject herself. Her receipt of information of that fineness of grain rather than any other is determined by her exercise of the relevant specific personal-level capacities for categorization and discrimination rather than any other such capacities. So, from the assumption of P-Level Predication, we may derive P-Level Capacities, the proposition that: S exercises certain specific personal-level capacities for categorization or discrimination in her perceptual experience: Those that account for the specific fineness of grain of the more or less determinate predicational worldly information or misinformation that is conveyed to her. Again, some examples may help to fix and motivate the principle further. Let us begin with the conceptual paradigm for representational content. Conceiving of S’s perceptual experience as a conscious state with conceptual representational content, that a is F, say, commits the theorist to regarding S as actually deploying the relevant predicational concept, “F,” in her experience, regardless of whether or not she endorses the full content in judgment. A certain worldly element is conceptually categorized as F in her experience. This is an exercise of the very conceptual capacity that is equally involved in her judgment that a1 is F, a2 is F, a3 is not F and so on, as opposed to the distinct capacity conceptually to categorize such things as F’, F’’ and so on, for other candidate predications at a different fineness of grain.6 Thus, a version of the claim in question here is that, in the conceptual case, a principled account of the fineness of grain of the more or less determinate predicational information that is conveyed to the subject in her perceptual experience is given by the specific capacities for conceptual categorization that she exercises in that perception.7 I contend that the general principle extends equally to the nonconceptual case. In terms resonant of Peacocke’s related Discrimination Principle (1988: 468), this may be formulated as follows.

126  Bill Brewer For each content that may be assigned to a representational state of a subject, there is an adequately individuating account of what makes it the case that her state has that content rather than any other. Premise (2) of (KA) elaborates this principle further in two ways. First, it emphasizes that the more or less determinate predication involved in conscious perceptual content is personal-level predication, accessible in some way to the subject herself. From this it derives the qualification that the relevant adequately individuating accounts are equally to be given in personallevel terms. Second, it insists that such personal level individuating accounts make essential reference to certain capacities that she exercises in her perceptual experience. In the nonconceptual case, these are less demanding than any explicit conceptual categorization; but, in order to provide a principled specification of the level of fineness of grain of the predication involved, they must constitute the actual discrimination of a more or less determinate worldly element in such a way as to sustain relevant sorting or other differential behavior in relation to elements alike in that way, rather than in some similar way at a different fineness of grain. Thus, a second version of the claim in question here is that, in the nonconceptual case, a principled account of the fineness of grain of the more or less determinate predicational information that is conveyed to the subject in her perceptual experience is given by the specific capacities for discrimination that she exercises in her perception that sustain her sorting and other differential behavior in relation to worldly elements alike at the relevant fineness of grain. Finally, it may be illuminating to comment on the place of considerations of causal antecedence in content determination. Although these surely make a significant contribution to the determination of perceptual content, they fail to engage directly with the central issue of concern here, which is the principles constraining our ascription of more or less determinate fineness of grain in the predicational information, or misinformation, about the world that is conveyed to the subject in perception. In this connection, I claim, we have to appeal in addition to the personal-level capacities that are actually exercised in her receipt of precisely that fineness of grain of worldly information, or misinformation. Thus: P-Level Predication → P-Level Capacities (3) Recall that P-Level Capacities is the proposition that S exercises certain specific personal-level capacities for categorization or discrimination in her perceptual experience: Those that account for the specific fineness of grain of the more or less determinate predicational worldly information or misinformation that is conveyed to her. Premise (3) asserts that her exercise of those capacities depends upon her attention to the elements of the world around her that this predicational information concerns, consciously selecting them precisely for the further processing that constitutes her exercise of the capacities in question.

Consciousness and Content in Perception 127 Categorizing and discriminating require sustained attention to the specific environmental elements concerned that set the norms for like categorization and discrimination in future judgment and sorting or other differential behavior. Just those elements have to be selected from the full range of alternatives available in order to sustain the capacities in question and control and coordinate the subject’s behavior in actually realizing them. This is a necessary consequence of the processing limitations that govern our cognitive lives and shape our conscious personal-level perspectives upon the world around us. Furthermore, since the personal-level capacities for categorization and discrimination involved are exercised with understanding on the part of the subject herself, in the sense required for her grasp of the perceptual contents in question and their provision of a rational basis for subsequent thought and action, the attentional selection involved here involves conscious attention.8 So, P-Level Capacities imply Conscious Selective Attention, the proposition that: S is consciously attending selectively to certain specific elements of her environment: Those that are the focus of her exercise of the relevant personal level capacities. This premise is highly consonant with the assumptions shaping empirical work on visual attention over the last 50  years and with the main results that have been obtained.9 The basic assumption lying behind the design and interpretation of experimental work on perceptual attention in cognitive psychology is that attention involves a form of subpersonal selection, or filtering, of information for further processing that is essential to deal with the capacity limitations of the brain and with the bodily limitations on simultaneous and sequential action. Wu’s reading of the central results of this empirical work is that this subpersonal processing filter is governed at the personal level by the conscious selection of specific environmental elements as targets to guide and control performance on relevant tasks. His general conclusion is that attention is selection for (mental or physical) action. In any case, the vast body of empirical work and associated psychological theorizing about perceptual attention confirm the central point here, that attention is necessary for the processing in relation to consciously selected elements that constitutes the exercise of personal-level capacities for categorization and discrimination. Thus: P-Level Capacities → Conscious Selective Attention (4) Acquaintance is the proposition that: S is acquainted with certain elements of the world around her.

128  Bill Brewer That is to say, in the current context, that S is visually conscious of those elements. In other words, as I use the notion here, S is acquainted with all and only the worldly elements that are present in her conscious visual field. Thus, acquaintance obtains in relation to elements of the world around the subject that are not consciously selected in attention as well as to those that are. For conscious acquaintance is precisely what delineates the range of alternatives that are available for possible attentional selection. S may therefore be acquainted with certain elements of her environment and yet actually be directing her conscious selective attention elsewhere or otherwise: Not to these particular elements of acquaintance. Indeed, given the capacity limitations noted above, both in terms of neural processing and in terms of bodily action, there will always be elements of the world around her within her acquaintance that are not currently targets of conscious attentional selection: What is often called the “background” of attention. So it is not the case that Acquaintance in relation to certain specific elements of the world around S implies Conscious Selective Attention in relation to those very same worldly elements. Again, the central idea here can be elaborated in relation to relevant empirical work on attention.10 Relations between consciousness and attention are complex and controversial. For present purposes I  steer clear of their most controversial aspects and rely only on claims that may be agreed to on all sides. First, although there is compelling evidence of selective attention in the absence of conscious acquaintance, in cases of blindsight—see, for example, Kentridge et al. (1999, 2004)11— and even in normal subjects under specially contrived circumstances in which they are not conscious of “objects” present in rapidly changing stimuli ­(Norman et al., 2013), we are concerned here only with cases of attention to worldly elements of conscious acquaintance. For the topic is precisely how to understand the nature of normal conscious perceptual experience. Of course the account given must be and is consistent with all of these results, but the phenomenon in question is distinct from blindsight and other cases of “perception” without conscious awareness in precisely this regard: It is a person’s conscious experience of the world around her.12 Second, although attention may be necessary for some forms of consciousness, for example, of “very small or very unexpected changes” (Mole, 2008: 92), there will in general be many elements of the world around the perceiver that she is acquainted with—that are present in her visual field—but that are not objects of selective visual attention.13 This is simply to repeat the point that attention is genuinely selective. Some among the worldly elements of acquaintance get into conscious attention at any time; the remainder do not and yet remain as part of its consciously present background—against which the consciously attentionally selected elements are foregrounded—in ways that elements entirely outside the subject’s visual field are not. Conscious attentional selection is only of a subset of the worldly elements with which she is consciously acquainted in perception.

Consciousness and Content in Perception 129 Thus: Not-(Acquaintance → Conscious Selective Attention) (5) Not-(Acquaintance → P-level Capacities) follows from (3) and (4) (6) Not-(Acquaintance → P-level Content) follows from (1), (2) and (5). Thus, S may be acquainted with various elements of the world around her in spite of not being in a conscious perceptual state with personallevel representational content concerning those elements.14

2.  Against the Content View According to what I call the Content View (CV) (2011: esp. ch. 4), the most fundamental account of our normal perceptual relation with the physical world is to be given in terms of the complete representational contents of perceptual experience. In particular, I focus here on the idea that the correct theoretical articulation of the way things are for the subject, S, when she sees the world around her, begins, as its foundation, with the claim that she is in a conscious perceptual state with personal-level representational content. The remaining substance of the theory is taken up with questions about the type of representational content in question, the range of contents of that type that are eligible contents for perceptual experience itself, the necessary conditions upon and consequences of S being in a perceptual state with a specific content of that type, the relations between the contents of S’s perception and her conscious condition at the time, the role of such representational contents of perceptual experience in her perceptual knowledge of the world around her and so on. This approach can seem almost inevitable, and important arguments from Schellenberg (2011) and Siegel (2010: ch. 2) attempt to make that case. My goal in the current section is to deploy KA, as elaborated in Section 2, against CV and in diagnosis of the error in Schellenberg’s argument for CV.15 The crucial first half of Schellenberg’s Master Argument (MA) runs as follows (2011: 719): (P1) If a subject is perceptually related to the world (and not suffering from blindsight, etc.), then she is aware of the world. (P2) If a subject is aware of the world, then the world seems a certain way to her. (P3) If the world seems a certain way to her, then she has an experience with content C, where C corresponds to the way the world seems to her. (Con) If a subject is perceptually related to the world (and not suffering from blindsight, etc.), then she has an experience with content C, where C corresponds to the way the world seems to her. I accept (P1) as a straightforward elaboration of the kind of perceptual relatedness that is in question: Conscious perceptual experience, or

130  Bill Brewer awareness, of the world. So the crucial issue for MA concerns the relation between the awareness that a subject, S, has of the world around her in perception and her having an experience with content C. In order to focus the discussion and to pinpoint the error in MA that most clearly illustrates my basic objection to CV, I  adopt a maximally sympathetic interpretation of (P2) on which this is also perfectly acceptable. It begins with the Moorean insight that the most explicit description of the conscious character of perceptual experience involves a description of the worldly elements themselves that the subject is apparently aware of in that experience (Moore, 1903: 450; see also Strawson, 1979: 45–46). This description constitutes a plausible understanding of the specific way the world seems to her. Thus, we have (P2). If a subject is aware of the world, then the world seems a certain way to her: The way given in the description of the worldly elements that she is apparently aware of. Next, it is important to distinguish a weak from a strong reading of Schellenberg’s conclusion, (Con). (Conw) If a subject is perceptually related to the world (and not suffering from blindsight, etc.), then she has an experience with content C, where C corresponds to certain aspects of the way the world seems to her. (Cons) If a subject is perceptually related to the world (and not suffering from blindsight, etc.), then she has an experience with content C, where C corresponds to the entirety of the way the world seems to her. According to (Conw), whenever S is perceptually related to the world around her, certain aspects of the way the world seems to her may be articulated in terms of her being in a conscious perceptual state with personal-level representational content. There are legitimate grounds for concern about this claim.16 But so far as KA is concerned, it may yet be true. In this context, its truth depends on that of the claim that whenever S is perceptually related to the world, some of its elements at least are selected, either endogenously or exogenously, by her conscious attention. If the latter claim is correct and the attention involved selects for further processing constituting the exercise of personal-level capacities of categorization and discrimination associated with the more or less determinate predication involved in content C, then there is no obstacle so far as KA is concerned to the idea that that experiential content corresponds with certain aspects of the way the world seems to her. (Cons) is refuted by my Key Argument earlier, though. For, according to premise (4) of KA, whenever S is perceptually related to the world, there will be elements of her environment that she is consciously acquainted with—that must therefore be included in the description of what she is apparently aware of that constitutes the way things seem to her—that are nevertheless not attentionally selected. Given premises (1)(3) of KA, then, these aspects of her awareness cannot correspond with

Consciousness and Content in Perception 131 any perceptual experiential content. So, although S is perceptually related to the world and may have an experience with content C that may indeed capture certain aspects of the way the world seems to her, C cannot possibly correspond to the entirety of the way the world seems to her. I take it that CV is committed to (Cons). For its definitive insight is that the way the world seems to S in perception is precisely a representational matter analogous to the way the world “seems” to her according to the “world view” consisting of her empirical beliefs. I also read Schellenberg’s argument as intended to establish this version of her conclusion. This is strongly suggested in her explicit discussion and motivation of her premises (P2) and (P3) (2011: Section 2.1). In any case, I stipulate CV in this form as my own target for critical evaluation and I assume for the remainder of the discussion that this is also the preferred conclusion of Schellenberg’s MA. So (P3) should be correspondingly strengthened as follows. (P3s) If the world seems a certain way to her, then she has an experience with content C, where C corresponds to the entirety of way the world seems to her. Then KA offers a straightforward diagnosis of the error in MA. Given our understanding of the way the world seems to S in perception, (P3s) is simply false. The notion of content fails to capture the entirety of the way things seem to the subject. For, as I explained above, there will always be unattended worldly elements of acquaintance that are essential to the way things seem to her but do not show up in the more or less determinate predicational information, or misinformation, conveyed by her perceptual content.17 Suppose in objection that we idealize the perceiver.18 Perhaps ordinary perceivers like S cannot process all of the information contributing to their conscious visual perspective on the world because of various limitations. Consider instead a perceiver, S’, not facing the same processing limitations in an otherwise identical perceptual situation to S.19 So far as KA is concerned, there may be a content, C, that corresponds to the entirety of the way the world seems to S’. If so, then should we not also admit that C is the entirety of the way the world seems to S, in the sense that C contains all the information that S could access in some relevant sense? If this is the point of CV, then the objection from KA fails. In reply, though, I claim that, in the absence of the actual exercise of specific capacities for categorization or discrimination, there is no determinacy in the predicational fineness of grain of the information that would be conveyed to the subject were some such capacities exercised. For this depends precisely on the specifics of the capacities in question, and there are, as we have already seen, indefinitely many alternatives available in connection with the various worldly elements that are present in S’s conscious visual perspective. It is the specific capacities for categorization and discrimination that are exercised in attention that determine the predicational fineness of

132  Bill Brewer grain of S’s personal-level perceptual contents. In their absence, there is no saying what exactly C would be were S’ to exercise some such capacities or others in S’s situation.20 It may also be objected that my discussion has been unfair to MA in adopting the Moorean reading of the way the world seems to S. This notion should instead have been more narrowly construed in the way suggested by Schellenberg’s seems-content link, very crudely, that “the content of experience corresponds [entirely] to the way the world seems to the experiencing subject” (2011: 723). Certainly (P3s) is then secured by this definition of the way the world seems to S, and (Cons) immediately follows. But this is really just a case of rearranging the deck chairs on the Titanic. For the strength of the CV that results—how much, if any, of S’s awareness of the world in perception is accounted for by her having an experience with content C—is precisely the strength of the CV that is fed into the argument at (P2). And, as we have already seen, KA demonstrates that the entirety of S’s awareness of the world cannot possibly correspond with the way the world seems to her in this narrower sense. So, we are back to the point that KA refutes the most natural paradigm of CV that is my explicit target here. Schellenberg’s MA attempts to conjoin innocent versions of both (P2) and (P3) in defense of CV, but KA establishes that this cannot be done. KA therefore constitutes a quite general argument against the central CV contention that the correct theoretical articulation of the way things are for the subject, S, when she perceives the world around her, begins, as its foundation, with the claim that she is in a conscious perceptual state with personal-level representational content. Furthermore, my discussion of MA strongly suggests that this CV contention effectively involves a fatal equivocation on the way things seem to the subject in perception, between how things are consciously for the subject in relation to the world in perception, on the one hand, and what more or less determinate predicational information is actually conveyed to the subject as a result of this along with her categorizing or discriminating attentional selection among the worldly elements of her conscious acquaintance, on the other.

3. Conclusion There are aspects of our conscious perceptual relation with the world around us that cannot be captured in terms of the representational content of perceptual experience. I  myself subscribe instead to an account of these by appeal to a fundamental layer of perceptual consciousness consisting in the subject’s standing in a relation of conscious acquaintance with particular mind-independent elements of the world around her (2011: esp. ch. 5). In any case, the principled obstacles to a wholly representational account of perceptual consciousness set out here undermine the prospects of any orthodox Content View of perception.21

Consciousness and Content in Perception 133

Notes 1. This chapter is an authorized republishing from my “Consciousness and Content in Perception,” published in Philosophical Perspectives, 31, Philosophy of Mind, 2017, doi: 10.1111/phpe.12091. 2. As in earlier discussions, I focus throughout on visual perception. 3. I intend the most general notion of representational content and certainly mean to include the various forms of conceptual and nonconceptual content that have been invoked in philosophical theories of perceptual experience. Influential illustrations of the idea and its variety can be found in Anscombe (1965), Armstrong (1968), Peacocke (1983, 1992), McDowell (1994), Tye (1995, 2000), Dretske (1995), Byrne (2001), Siegel (2010), Pautz (2010) and Schellenberg (2010, 2011). 4. Here and throughout I use the term “element” to cover objects, features and regions of the world around the perceiver. 5. The reality is of course far more complex, with colors more accurately arranged in a 3D solid with the blues occupying a segment of the kind that football players might eat of an orange at half-time. See Munsell (1907) for the original conception and Palmer (1999, ch. 3) for a comprehensive review of the major current issues. This additional complexity only adds to the force of my point, with the various candidate predications occupying multiple containing and intersecting solid 3D regions within the blue segment. 6. The idea that a subject exercising genuinely conceptual categorization in any particular case is equally capable of further applications of the same concept in different cases is one conjunct of Evans’s famous Generality Constraint that I  accept as a plausible constraint on genuinely conceptual content ascription (Evans, 1982, ch. 4ff). See also his (1985) for the importance of judgments of not-F-ness, as opposed simply to the withholding of judgments of F-ness. For further discussion, see Campbell (1986), Peacocke (1992, ch. 2) and Heck (2000). For dissent, see Travis (1994). 7. A claim along somewhat similar lines is Davidson’s that belief depends on language, “for the only access to the fine structure and individuation of beliefs is through the sentences speakers and interpreters of speakers use to express and describe beliefs” (1986: 315). The Davidsonian thesis is vulnerable to objection, though, for introducing inessential epistemic conditions into the metaphysics of mind, by insisting that conditions on determinate belief acquisition should in some way be recognizable or knowable. 8. See Campbell (2002) and Smithies (2011) for further development of this idea. They propose different accounts of the most fundamental distinctive cognitive role of conscious attention. Both apply in my view to the personal-level categorization and discrimination characteristic of normal perceptual content. 9. See Wu (2014) for a detailed discussion of the major paradigms and findings of this work and an extended discussion of their philosophical implications. What follows is derived directly from his discussion (esp. chs. 1 and 3). 10. See Mole (2008, 2014) and Phillips (2016, in press, forthcoming) for helpful reviews and discussion of the most important results here. 11. Although see Mole (2008: §4) for an interesting argument that the evidence may not be conclusive. This is based upon a distinction between the cue stimulus, of which the subject is not conscious, but which is also claimed not to be an object of his attention, and the cued region of the blind hemifield, to which the subject is indeed attending, yet of which it is also claimed that he is conscious. This last claim, that the blindsighted subject may nevertheless be consciously oriented towards regions of his blind hemifield rests in turn upon an analogy with sighted subjects’ purported conscious orientation to the whole space around them, including regions to which they are not currently

134  Bill Brewer perceptually receptive—the region directly behind them, say. Whatever the merits of this idea of conscious orientation to unperceived regions of space, it does not obviously entail perceptual acquaintance with those regions as I understand that notion here. So for present purposes, I accept Kentridge et al.’s compelling evidence of attention without consciousness in the relevant sense, in certain abnormal perceptual situations. For an illuminating discussion of our conscious orientation to the whole space around us, see O’Shaughnessy (1980, vol. II, chs. 7 and 8), Ayers (1993, vol. I, pt. iii) and Martin (1992, 1999). 12. For skepticism about the very idea of genuine perception without conscious awareness, see Phillips (2016, in press, forthcoming). 13. Mack and Rock (1998) dispute this claim, insisting that there is “no conscious perception without attention” (p. ix), but they are able to do so only by question-beggingly defining attention as “the process that brings a stimulus to consciousness” (p. 25). Mole persuasively argues that their experimental results fail to establish their conclusion without begging the question in this way (2008, §3). Inattentional blindness phenomena (Mack and Rock, 1998; Simons and Chabris, 1999, 2007; Block, 2007) may be interpreted in line with Mack and Rock’s thesis, but this interpretation is by no means obligatory and is in fact both unpopular and unmotivated. 14. See Amijee (2013) for compelling interpretation and elaboration of a similar line of argument from Russell’s (1912) theory of knowledge: “[A]cquaintance with an object is not alone sufficient for knowledge of any truths about it [nor, as I argue here, for any perceptual content concerning it]; attention is also necessary” (Amijee, 2013, p. 1184). 15. I believe that the application to Siegel is relatively straightforward, but I do not establish this here. 16. See, in particular, Campbell (2002, esp. ch. 6), Travis (2004), Brewer (2011, esp. ch. 4) and Campbell and Cassam (2014). 17. Although I  doubt that this is her intention, it is interesting to note that there is a point at which Schellenberg herself hints at something like this picture, in her suggestion that Chisholm’s (1957: 50–53) noncomparative looks provide a foundation for the comparative looks that perhaps constitute the predications of perceptual content (2011: 722). It would be fruitful to consider further this suggestion that noncomparative looks correspond with unattended elements of acquaintance, whereas comparatives necessarily involve the attention required to sustain more or less determinate predication. 18. Thanks to Matt Parrott for this line of objection. 19. I ignore for the sake of the argument concerns there may be about the very coherence of a perceiver like S’. If the genuine selectivity of conscious attention is essential to the nature of any visual experience, then the objection does not get off the ground. 20. See Stazicker (2011) for a closely related argument for the constitutive dependence of certain aspects of conscious vision on conscious attention that exploits the indeterminacy of visual consciousness in the absence of conscious attention. 21. Many thanks to Dominic Alford-Duguid, Matt Parrott and Matt Soteriou for their written comments on earlier drafts of this material. I’m also grateful for discussions with John Campbell, Imogen Dickie, Naomi Eilan and James Stazicker.

Consciousness and Content in Perception 135

References Amijee, F. (2013) “The Role of Attention in Russell’s Theory of Knowledge” in British Journal for the History of Philosophy, 21, 1175–1193. Anscombe, G. E. M. (1965) “The Intentionality of Sensation” in Analytical Philosophy: First Series (ed. R. J. Butler), Oxford: Blackwell. Armstrong, D. M. (1968) A Materialist Theory of the Mind, London: Routledge. Ayers, M. (1993) Locke, London: Routledge. Berkeley, G. (1975) “A  Treatise Concerning the Principles of Human Knowledge” in George Berkeley: Philosophical Works (ed. M. R. Ayers), London: Everyman. Block, N. (2007) “Consciousness, Accessibility, and the Mesh between Psychology and Neuroscience” in Brain and Behavioural Sciences, 30, 481–548. Brewer, B. (2011) Perception and Its Objects, Oxford: Oxford University Press. Byrne, A. (2001) “Intentionalism Defended” in Philosophical Review, 110, 199–240. Campbell, J. (1986) “Conceptual Structure” in Meaning and Interpretation (ed. Charles Travis), Oxford: Blackwell. Campbell, J. (2002) Reference and Consciousness, Oxford: Oxford University Press. Campbell, J and Cassam, Q. (2014) Berkeley’s Puzzle, Oxford: Oxford University Press. Chisholm, R. (1957) Perceiving: A Philosophical Study, Ithaca, NY: Cornell University Press. Davidson, D. (1986) “A Coherence Theory of Truth and Knowledge” in Truth and Interpretation (ed. E. LePore), Oxford: Blackwell. Dretske, F. (1995) Naturalizing the Mind, Cambridge, MA: MIT Press. Evans, G. (1982) The Varieties of Reference, Oxford: Oxford University Press. Evans, G. (1985) “Identity and Predication” in His Collected Papers, Oxford: Oxford University Press. Heck, R. (2000) “Nonconceptual Content and the ‘Space of Reasons’ ” in Philosophical Review, 109: 483–523. Kentridge, R. W., Heywood, C. A. and Weiskrantz, L. (1999) “Attention Without Awareness in Blindsight” in Proceedings of the Royal Society of London Series B: Biological Sciences, 266, 1805–1811. Kentridge, R. W., Heywood, C. A. and Weiskrantz, L. (2004) “Spatial Attention Speeds Discrimination Without Awareness in Blindsight” in Neuropsychologia, 42, 831–835. Mack, A. and Rock, I. (1998) Inattentional Blindness, Cambridge, MA: MIT Press. Martin, M. (1992) “Sight and Touch” in The Contents of Experience (ed. Tim Crane), Cambridge: Cambridge University Press. Martin, M. (1999) “Sense Modalities and Spatial Properties” in Spatial Representation (ed. Eilan, R. McCarthy and B. Brewer), Oxford: Oxford University Press. McDowell, J. (1994) Mind and World, Cambridge, MA: Harvard University Press. Mole, C. (2008) “Attention and Consciousness” in Journal of Consciousness Studies, 15, 86–104.

136  Bill Brewer Mole, C. (2014) “Attention to Unseen Objects” in Journal of Consciousness Studies, 21, 41–56. Moore, G. E. (1903) “The Refutation of Idealism” in Mind, 48, 433–453. Munsell, A. H. (1907) Colour Notation, Boston: Geo. H. Ellis Co. Norman, L. J, Heywood, C. A. and Kentridge, R. W. (2013) “Object-Based Attention Without Awareness” in Psychological Science, 24, 836–843. O’Shaughnessy, B. (1980) The Will, Cambridge: Cambridge University Press. Palmer, S. (1999) Vision Science, Cambridge, MA: MIT Press. Pautz, A. (2010) “Why Explain Experience in Terms of Content?” in Perceiving the World (ed. B. Nanay), Oxford: Oxford University Press. Peacocke, C. (1983) Sense and Content, Oxford: Oxford University Press. Peacocke, C. (1988) “The Limits of Intelligibility: A  Post-Verificationist Proposal” in Philosophical Review, 97, 463–496. Peacocke, C. (1992) A Study of Concepts, Cambridge, MA: MIT Press. Phillips, I. (2016) “Consciousness and Criterion: On Block’s Case for Unconscious Seeing” in Philosophy and Phenomenological Research, 93, 419–451. Phillips, I. (2017) “Does Unconscious Perception Really Exist?” in Neuroscience of Consciousness, 3(1). Phillips, I. (2018) “Unconscious Perception Reconsidered” in Analytic Philosophy, 59(4). Russell, B. (1912) The Problems of Philosophy, London: William and Norgate. Schellenberg, S. (2010) “The Particularity and Phenomenology of Perceptual Experience” in Philosophical Studies, 149, 19–48. Schellenberg, S. (2011) “Perceptual Content Defended” in Nous, 45, 714–750. Siegel, S. (2010) The Contents of Visual Experience, Oxford: Oxford University Press. Simons, D. and Chabris, C. (1999) “Gorillas in our Midst: Sustained Inattentional Blindness for Dynamic Events” in Perception, 4, 147–155. Smithies, D. (2011) “Attention is Rational-Access Consciousness” in Attention (ed. C. Mole, D. Smithies and W. Wu), Oxford: Oxford University Press. Stazicker, J. (2011) “Attention, Visual Consciousness, and Indeterminacy” in Mind and Language, 26, 156–184. Strawson, P. F. (1979) “Perception and its Objects” in Perception and Identity: Essays Presented to A. J. Ayer (ed. G. F. MacDonald), London: MacMillan. Travis, C. (1994) “On Constraints of Generality” in Proceedings of the Aristotelian Society, 94, 165–188. Travis, C. (2004) “The Silence of the Senses” in Mind, 113, 59–94. Tye, M. (1995) Ten Problems of Consciousness, Cambridge, MA: MIT Press. Tye, M. (2000) Consciousness, Colour, and Content, Cambridge, MA: MIT Press. Wu, W. (2014) Attention, London: Routledge.

8 Perceptual Capacities Susanna Schellenberg

When we perceive, we employ perceptual capacities by means of which we discriminate particulars in our environment. Seeing the red shade of an apple involves, for example, employing one’s capacity to discriminate red from other colors. More generally, we can say that to be a perceiver is to possess certain capacities, to perceive is to employ those capacities and employing perceptual capacities constitutes perceptual states. What are perceptual capacities? A  perceptual capacity is a kind of discriminatory, selective capacity that we employ in perception, hallucination or illusion. It is a low-level mental capacity that functions to differentiate, single out and in some cases classify mind-independent particulars of a specific type—for example, to discriminate and single out instances of red from instances of blue. While discriminating particulars can include classification, it does not require it. To say that perceptual capacities are low-level is not to say that they are subpersonal but rather that they are cognitively less high-level than concepts (at least on most philosophical accounts of concepts). Perceptual capacities come in many varieties: There are perceptual capacities to discriminate luminance, motion, quantities, size, pitch, tone and distances, to name just a few. Some capacities are more basic than others. Some stand in complex hierarchical structures. Some are always employed jointly with other capacities. Drawing on work in cognitive psychology, neuroscience and developmental psychology, this chapter provides an analysis of perceptual capacities. It includes the following key elements: Function of a Perceptual Capacity: The function of a perceptual capacity Cα is to discriminate and single out mind-independent particulars α1, α2, α3, . . . αn, that is, particulars of a specific type. Individuation Condition: A perceptual capacity Cα is individuated by the mind-independent particulars α1, α2, α3, . . . αn that the perceptual capacity functions to single out. Possession Condition: A subject S possesses a perceptual capacity Cα if and only if the following counterfactual is true of S: S would be in a position to discriminate and single out a particular α1, where α1 is any

138  Susanna Schellenberg particular of the type that Cα functions to discriminate and single out, if S were perceptually related to α1, (i) assuming S is perceptually capable (awake, alert etc.), (ii) assuming no finking, masking or other exotic case obtains and (iii) where S being perceptually related to α1 means that (a) the situational features are such that α1 is perceivable by S (good lighting conditions, etc.), (b) S has the relevant sensory apparatus that allows her to gain information about α1 and (c) S is spatially and temporally related to α1 such that S is in a position to gain information about α1 via her sensory apparatus. Fallibility Condition: If a subject S employs a capacity Cα, Cα can either fulfill its function or fail to fulfill its function, such that there is no difference at the level of employing Cα but only a difference at the level of fulfilling its function. The function of Cα is fulfilled if by employing Cα a relevant particular is singled out. The function of Cα fails to be fulfilled if by employing Cα no relevant particular is singled out. Asymmetry Condition: The employment of a perceptual capacity Cα in cases in which Cα fulfills its function is metaphysically more basic than the employment of Cα in cases in which Cα fails to fulfill its function. Repeatability Condition: A necessary condition for Cα to be a perceptual capacity is that Cα is repeatable. Physical Base Condition: If a subject S is employing a perceptual capacity Cα, then there is a physical base of employing Cα that is constituted by physical processes, events and structures (such as the neural activity) of S. Informational Base Condition: If a subject S is employing a perceptual capacity Cα, then there is an informational base of employing Cα that is constituted by the subpersonal psychological mechanism (information processing, computations and other subpersonal functional states, events and processes) of S. I will provide an asymmetric counterfactual analysis of perceptual capacities that is built around these eight conditions. But first it will be helpful to give a brief history of the notion of capacity in cognitive science and philosophy and to lay out the benefits of analyzing the mind in terms of mental capacities.

1. Why Analyze the Mind in Terms of Mental Capacities? The notion of a capacity is deeply entrenched in psychology and the brain sciences. Driven by the idea that a cognitive system has the capacity it does in virtue of its internal components and their organization, it is standard to appeal to capacities in cognitive psychology.1 Critical in the advent of the notion of capacity in cognitive psychology was Chomsky’s distinction between competence and performance, where a competence is a cognitive capacity and a performance is generated by employing a competence. In the case of language, a competence is a tacit grasp of the

Perceptual Capacities 139 structural properties of a language and the performance is the production of utterances (Chomsky, 1995). In contrast to the centrality of capacities in psychology and the brain sciences, questions about mental capacities have been neglected in recent philosophical work.2 This is surprising given their importance in the history of philosophy, in the work of Aristotle and Kant, in particular. Until the beginning of the twentieth century, capacities and related concepts such as abilities, skills, powers and categories featured prominently in philosophical and scientific work on perception. Indeed, it was standard to analyze the mind in terms of capacities. With the linguistic turn, the norms changed and it became standard to analyze the mind in terms of representational content instead. No doubt the linguistic turn brought with it much clarity and precision. However, in sidelining capacities, a great deal was lost. The good news is that we are not forced to choose between analyzing the mind in terms of capacities and analyzing it in terms of representational content. Indeed, I  will argue that employing mental capacities constitutes the representational content of mental states. The main benefit of invoking capacities in an account of the mind is that it allows for an elegant counterfactual analysis of mental states: It allows us to analyze mental states on three distinct yet interrelated levels: 1. A first level of analysis pertains to the function of mental capacities. 2. A second level of analysis pertains to the mental capacities employed, irrespective of the context in which they are employed. 3. A third level of analysis pertains to the mental capacities employed, taking into account the context in which they are employed. On the first level, we focus on the function of perceptual capacities, which is to discriminate and single out particulars of a specific type. A perceptual capacity has this function even if it is employed while failing to fulfill its function, as is the case in hallucination and illusion. Even in such a case, the capacity functions to discriminate and single out particulars of a specific type. Moreover, a perceptual capacity has this function even if it is more often than not employed while failing to fulfill its function. On the second level of analysis, we focus on what is in common between mental states in which the same perceptual capacities are employed. On this level, it is irrelevant whether or not a perceptual capacity is employed such that it fulfills its function. As I argue elsewhere, in perceptions, hallucinations and illusions with the same phenomenal character, the same perceptual capacities are employed. (Schellenberg, 2018b) So, on this second level of analysis, perceptions, hallucinations and illusions with the same phenomenal character are on a par. On the third level of analysis, we focus on the fact that perceptual capacities are employed in a specific environment, whereby a particular

140  Susanna Schellenberg is either successfully singled out or the experiencing subject fails to single out a particular. In contrast to the second level, it matters, on this third level, whether or not a capacity is employed such that its function is fulfilled. So, on this level, perceptions differ from hallucinations and illusions. This is the level of analysis on which we determine the token content of the relevant experiential state.

2.  The Function of Perceptual Capacities Perceptual capacities function to discriminate and single out particulars. More precisely: Function of a Perceptual Capacity: The function of a perceptual capacity Cα is to discriminate and single out mind-independent particulars α1, α2, α3, . . . αn, that is, particulars of a specific type. A particular, as understood here, is a mind-independent object, event or property-instance. I use the notion of “singling out” rather than “referring” so as to remain neutral on whether perceptual capacities are conceptual or nonconceptual. While referring has been argued to require conceptual capacities, singling out particulars requires no such capacities. Singling out a particular can be understood as a proto-conceptual analogue of referring to a particular. Non-rational animals and infants as young as four months old can perceptually single out particulars in their environment, yet on at least some notions of “reference” they do not have the capacity to refer. Moreover, on many views of reference, referring to a particular presupposes that the relevant subject is in a mental state with content.3 While I  will show that analyzing perceptual states as constituted by employing perceptual capacities entails that those perceptual states have representational content, we can remain neutral for now on whether perceptual experience has content. Thus, I use the term “singling out” so as not to presuppose a representational view. The notion of function in play is a notion of natural function. It is natural in that it is independent of interpretation. So what function a capacity has is not relative to an interpreter.4 There are many different kinds of perceptual capacities. There are perceptual capacities that function to discriminate and single out objects of a specific type. Others function to discriminate and single out property instances of a specific type. Still others function to discriminate and single out events of a specific type. Natural functions can be given an etiological analysis; we can, however, work with the notion of a natural function while rejecting such an analysis. As I  will argue, that is what we should do. According to etiological theories, something has a certain function because of what it is selected and adapted for (Ayala, 1970, Wright, 1973, Millikan, 1989, Neander, 1991).5 Consider the heart’s function to pump blood.

Perceptual Capacities 141 The etiological theory explains this function by pointing to the fact that hearts were selected for pumping blood. While this is no doubt how it causally came about that hearts have the function to pump blood, the fact that hearts have this function is neutral on how they came to have it. Even if hearts came to have their function by some other means, they could still have the function to pump blood. More generally, we should distinguish what function something has from how it came to have that function. What is crucial for an analysis of capacities is what function they have, not how they came to have it. In contrast to etiological theories, the view developed here is neutral on how mental capacities came to have their function. No doubt we have the perceptual capacities that we do due to our phylogenetic and ontogenetic background. The point is that we can analyze the function of those capacities without appealing to how we came to have them. Indeed, there is no sense in which the phylogenetic or ontogenetic history of a subject is relevant for determining the function of her capacities. A subject who discriminates and singles out particulars in her environment via an implant can have perceptual capacities with the very same function as a subject who has those capacities due to her phylogenetic and ontogenetic background. While most mental capacities happen to have their function due to natural selection or some other natural process, nothing in the account developed here hinges on the matter. For this reason, the account of mental states developed here does not face well-known problems of etiological theories of mental content. It does not, for example, face the problem of how to account for complex capacities, the possession of which cannot be explained in terms of natural selection, adaptation or meme selection. Moreover, by contrast to etiological accounts, it does not face Davidson’s Swampman objection (Davidson, 1987: 443–444). Swampman is a creature that by astounding coincidence came into existence through a collision of particles caused by a lightning bolt. At the very same moment, Donald Davidson is struck by a lightning bolt and tragically dies. Swampman is a physical duplicate of Davidson, but his history is radically different. He did not partake in any evolutionary history and there are no phylogenetic, ontogenetic or other etiological ways to explain his mental states. For this reason, etiological accounts of function are forced to say that Swampman’s component parts do not have any functions. But according to capacitism, the view developed in the course of this book, a function is in no way dependent on the history of the subject employing the relevant capacity. Therefore, capacitism posits that Swampman possesses all the capacities that Davidson possessed shortly before being struck by lightning. Indeed, I  argue that neither the content nor the epistemic force of a mental state depends on the history or reliability of employing the capacities that constitute that mental state. Since capacitism holds that the function of perceptual capacities is independent of the history of the subject employing those

142  Susanna Schellenberg capacities, the view posits that Swampman not only has mental states with content but also mental states with epistemic force.6 A perceptual capacity has a certain function irrespective of whether it fulfills its function in any particular context of employment. To explain why, it is helpful to distinguish capacities from their employment. While a capacity is a kind of mental tool, the employment of a capacity is a mental activity. Consider Sam, who possesses the perceptual capacity that functions to discriminate and single out red particulars. Just as Sam’s heart has the function to pump blood but may fail to pump blood, so Sam may employ her capacity while failing to single out any red particular. In such a case, the capacity failed to fulfill its function because the target of employing the capacity is not present: No red particular was discriminated and singled out. A few clarifications are in order before we move on to developing the individuation conditions of perceptual capacities. First, for ƒ to be a natural function does not imply that ƒ is a biological function. While biological functions are natural functions, not all natural functions are biological functions. After all, a computer can have a natural function, but it does not have a biological function. Second, it is crucial that the function of a perceptual capacity is not just a matter of discriminating particulars but is also a matter of singling them out. Due to this, perceiving an instance of red is distinct from perceiving an instance of blue. Both cases may involve discriminating red from blue, but in the former case an instance of red is singled out, while in the latter case an instance of blue is singled out. So the capacities employed are distinct and the perceptual states constituted by employing those capacities differ. Third, while capacitism is compatible with functionalism, it does not commit one to functionalism. Functionalism individuates mental states not with regard to their internal constitution or their relation to the environment but on the basis of their function in the cognitive system of which they are a part (e.g., Lewis, 1966, Block, 1978). Capacitism individuates mental states on the basis of mental capacities and mindindependent particulars: Mental states are constituted by the mental capacities employed and the particulars (if any) thereby singled out. The function of those capacities is not understood in terms of the role those capacities play in the cognitive system of which they are a part. Thus, capacitism does not entail functionalism. There may, however, be good reasons to integrate capacitism in a functionalist view of the mind. 2.1.  Material Discrimination When we perceptually discriminate α from β, we discriminate an actual, mind-independent particular α to which we are perceptually related from a distinct actual, mind-independent particular β to which we are similarly

Perceptual Capacities 143 perceptually related. Let’s call this kind of discrimination material discrimination. When I  speak of discrimination without further qualification, I mean always material discrimination. Discriminating between two particulars in this sense does not require attending to both particulars. It requires only registering their differences. Consider Dylan, who is walking through thick foliage. It is unclear how she could be perceptually aware of, say, a leaf without registering how it differs in at least one respect from its surround. More generally, it is unclear how one could be perceptually aware of a particular without registering how it differs in at least one respect from its surround. The basic level of employing perceptual capacities is to discriminate one particular from another, where this discrimination is understood as registering their differences.7 Material discrimination is distinct from any notion of discrimination understood in terms of carving out possibility space. On such notions, to discriminate α is to discriminate α from other possible ways α could be. In particular, material discrimination is to be distinguished from the notion of discrimination in relevant alternative views of knowledge (Austin, 1946, Dretske, 1969, 1981, Goldman, 1976), contextualism and pragmatic encroachment accounts (Hawthorne, 2003, Stanley, 2005, DeRose, 2009), as well as contrastivism (Schaffer, 2005). Subtleties aside, such views have it that to know that an object o has property F (in some circumstance), one must be able to rule out some relevant alternatives, that is, certain relevant situations in which o has, say, property G rather than F. On this notion of discrimination, to discriminate a property F that an object o instantiates is to discriminate F from relevant alternative ways o could be. As Pritchard puts it: In the perceptual case at least, to be able to rule out an alternative is to be able to make the relevant discriminations between the target object and the object at issue in the alternative—e.g., to be able to discriminate between goldfinches and woodpeckers. (Pritchard, 2010: 246) On such relevant alternative views of knowledge, discrimination is necessary for knowledge: To know, one must discriminate the way things are from relevant other ways they might be. The notion of discrimination is a matter of modal appreciation.8 Material discrimination is distinct from discriminating relevant alternatives in two ways. First, material discrimination is a matter of noticing differences between actual, mind-independent particulars to which one is perceptually related rather than appreciating relevant alternatives. Second, material discrimination need not be cognitive (and typically is not), while any kind of modal appreciation and modal theorizing falls squarely in the cognitive realm. It is standard to distinguish perception and cognition. Perception is a kind of mental faculty that we share with

144  Susanna Schellenberg non-rational animals. While human perception might be rife with topdown effects, there is no reason to think that modal appreciation is constitutive of perception. Material discrimination is distinct not only from appreciating relevant alternatives but also from detecting differences between mental states via introspection. It has been argued that two phenomenal states M1 and M2 differ if and only if their subject can introspectively tell them apart. (Shoemaker, 1994) I am not denying that we can discriminate between phenomenal states in this way. We can call this introspective discrimination. The important point here is that in perceiving our environment, we discriminate between external, mind-independent particulars rather than mental states or aspects of mental states. According to capacitism, discriminating such particulars constitutes perceptual states and indeed phenomenal character and so is more basic than introspective discrimination.

3.  The Individuation Condition Perceptual capacities are individuated by the external, mind-independent particulars that they function to single out. Individuation Condition: A perceptual capacity Cα is individuated by the mind-independent particulars α1, α2, α3,  .  .  .  αn that the perceptual capacity functions to single out. Given that perceptual capacities are individuated externally, the perceptual capacity that functions to single out instances of red differs from the perceptual capacity that functions to single out instances of scarlet or vermilion. There will be a perceptual capacity to discriminate and single out instances of red, a distinct perceptual capacity to discriminate and single out instances of scarlet and yet another perceptual capacity to discriminate and single out instances of vermilion. So perceptual capacities can be more or less fine-grained, and we can single out the very same particular with capacities that are more or less fine-grained. Suppose you see a field of flowers that are shades of red and yellow. You can employ your capacity to discriminate between red and yellow and thus be aware of a field of red and yellow flowers. Alternatively, you can employ your capacity to discriminate between crimson, scarlet and vermilion and between lemon, mustard and ocher and thus be aware of the colors in front of you in a more fine-grained way. The external, mind-independent property instances that we can perceive do not just include instances of intrinsic properties, such as intrinsic shapes, colors, sounds, smells, textures and the like.9 We always perceive from a perspective. As a consequence, we perceive under situational features, that is, features such as the lighting conditions, color context, the acoustic conditions and our location in relation to the particulars perceived. Thus, when we perceive a circular coin from different angles, there

Perceptual Capacities 145 is a respect in which the coin looks circular throughout but also a respect in which the coin’s appearance changes. Likewise, when we perceive two trees of the same size located at different distances from us, there is a respect in which they look the same size, but also a respect in which they appear different. (Peacocke, 1983) Perception has both an invariant aspect—an aspect that remains stable across changes in perspective—and a variant aspect—an aspect that changes depending on one’s perspective. How should we account for the variant aspect of perception? One option is to understand the variant aspect in terms of situationdependent properties.10 A situation-dependent property is an external, mind-independent property that is determined by an intrinsic property and relevant situational features (e.g., the perceiver’s location relative to the perceived intrinsic property, the lighting conditions, acoustic conditions, etc.). Situation-dependent properties are exclusively sensitive to and ontologically dependent on intrinsic properties and situational features. Any perceiver occupying the same location would, ceteris paribus, be presented with the same situation-dependent property. As with intrinsic properties, perceivers differ, however, with regard to which situation-dependent properties are perceptually available to them, and they differ in how they represent and are aware of situation-dependent properties. If this is right, then the external, mind-independent property instances that we can perceive include situation-dependent properties in addition to intrinsic properties. The boundaries of the set of particulars that a capacity functions to single out is set by the world. It is not set by what a perceiver takes her perceptual capacity to function to single out. So the boundaries of my capacity to discriminate and single out squares is set by squares, not by what I take to be squares. If in perception I take something to be a square that is not in fact a square, I employ my perceptual capacity to discriminate and single out squares baselessly, while failing to single out a square. Thus, I presuppose a strong form of realism. As we have seen, perceptual capacities are with regard to their individuation conditions analyzed in terms of mere relations to the world and so without any appeal to mental entities, be they, states, capacities or events. In this respect, capacitism builds on causal views of mental states. (Kripke, 1972’; Putnam, 1975; Burge, 1979; Devitt, 1981) However, as I will argue shortly, with regard to their possession conditions, the situation is more complex: The possession of at least some perceptual capacities requires possessing other perceptual capacities.

4.  The Possession Condition Perceptual capacities cannot be analyzed independently of analyzing their possession conditions. To possess a perceptual capacity is to be in a position to discriminate and single out the external, mind-independent particulars that the capacity functions to single out when perceptually related to such particulars and some further conditions hold. So if we

146  Susanna Schellenberg possess such a capacity, then—assuming no exotic case obtains—the following counterfactual holds: If we were perceptually related to a particular that the capacity functions to single out, then we would be in a position to discriminate and single out that particular. More precisely: Possession Condition: A subject S possesses a perceptual capacity Cα if and only if the following counterfactual is true of S: S would be in a position to discriminate and single out a particular α1, where α1 is any particular of the type that Cα functions to discriminate and single out, if S were perceptually related to α1, i. assuming S is perceptually capable (awake, alert, etc.), ii. assuming no finking, masking or other exotic case obtains, and iii. where S being perceptually related to α1 means that (a) the situational features are such that α1 is perceivable by S (good lighting conditions, etc.), (b) S has the relevant sensory apparatus that allows her to gain information about α1 and (c) S is spatially and temporally related to α1 such that S is in a position to gain information about α1 via her sensory apparatus. The condition requires only that a subject be in a position to discriminate and single out a particular of the type that Cα functions to single out when perceptually related to one and not that she in fact do so. The reason for this is that even if the subject is perceptually related to a relevant particular, she might for a variety of reasons fail to single out the particular, perhaps because she does not notice the particular due to her attention being directed elsewhere. It will be helpful to specify each qualification of what it is to be perceptually related to a particular. The qualification that the subject is perceptually capable rules out cases in which the subject is not at that particular moment able to employ her perceptual capacity (perhaps because she is intoxicated or sleepy), even though she is generally capable of doing so. The qualification that no finking, masking or other exotic cases obtain rules out cases in which the subject mysteriously loses her capacity from one moment to the next. The inference from a claim about perceptual capacities to a counterfactual fails in such cases.11 However, all the standard ways of fixing the disposition-to-counterfactual inference can be exploited for the capacity-to-counterfactual inference (see Lewis, 1997). Finding a formulation of the capacity-to-counterfactual inference that is indefeasible in light of all possible finking, masking and similarly exotic cases would be a project of its own. Therefore, I will here work on the independently plausible assumption that no such exotic cases obtain. The first specification of what it means to be perceptually related rules out cases in which the subject is causally related to a relevant particular α1, but it is, for example, too dark or too noisy for her to perceive

Perceptual Capacities 147 the particular. The second specification rules out cases in which the relevant subject does not have the sensory apparatus to perceive α1, perhaps because her sensory organs are damaged. The third specification rules out cases in which the subject is causally related to a relevant particular α1, but not in a way that allows her to gain information about α1 via her sensory organs—perhaps because α1 is so close to her eyes that she cannot properly make it out or so far away that she is unable to discriminate it from its surround. Successfully employing a perceptual capacity to discriminate and single out particulars of a type requires being differentially sensitive to particulars of that type in one’s environment. However, the counterfactual analysis of perceptual capacities entails that one could possess a perceptual capacity despite not being able at that very moment to respond differentially to the relevant particulars. If one is sufficiently intoxicated, one might not be able to respond differentially to much at all. In such states, one nonetheless possesses perceptual capacities. Moreover, if one does not have the relevant sensory apparatus or one’s sensory apparatus is impaired, one cannot be perceptually related to particulars that the perceptual capacity functions to discriminate and single out. In those cases too, one nevertheless can possess perceptual capacities. One will just not be in a position to employ them while fulfilling their function without being appropriately connected to a sensory apparatus. In short, while successfully employing a perceptual capacity requires being differentially sensitive to particulars of the relevant type in one’s environment, possessing a perceptual capacity is not subject to this requirement. There are several close alternatives to the counterfactual analysis provided. A  conditional could, for example, be formulated in terms of a “might” or a “could.” If the conditional were formulated in terms of a “might” or a “could,” the link between possessing a perceptual capacity and successfully employing it would be too weak to entail a constitutive relation between the perceptual capacities employed and the perceptual states thereby constituted.12 Therefore, it is crucial that the conditional is formulated in terms of a “would.” Now one might wonder what the connection is between possessing specific capacities and possessing closely related capacities. One might wonder, for example, whether there could be a perceiver who possesses only the capacity to discriminate red from other colors without possessing any perceptual capacities to discriminate and single out other colors. More radically, can there be a perceiver who possesses only one perceptual capacity? In response, there is empirical evidence that possession of at least some perceptual capacities comes in clusters. For example, if one is able to discriminate angles from straight lines, one will also be able to discriminate curves from straight lines. And if one possesses the capacity to discriminate, for example, red from blue and single out red, one will

148  Susanna Schellenberg also possess the capacity to discriminate blue from red and single out blue.13 4.1.  Possessing a Capacity vs. Employing a Capacity What is the relation between possessing a capacity and employing it? It has been argued that one cannot claim one possesses a capacity if one has never employed it successfully (Aristotle, De Generatione Animalium 2.3, 736b21–6 & 4.1, 766a5–10).14 It has been argued, moreover, that if one employs a capacity without it fulfilling its function, then one does not claim to possess the capacity at that moment (Millar, 2008). Aristotle attributes a view that is even more restrictive to the Megarians: There are some—such as the Megarians—who say that something is capable only when it is acting, and when it is not acting it is not capable. For example, someone who is not building is not capable of building, but someone who is building is capable when he is building; and likewise, too in other cases. It is not hard to see the absurd consequences of this. (Metaphysics, Book: 1046b) On the Megarian view, one can possess a capacity only when one is successfully employing it. Against all these views, I am arguing that we can possess a capacity even if we never employ it. Possessing a capacity is thus metaphysically more fundamental than employing a capacity: A subject cannot employ a capacity that she does not possess, but she can possess a capacity without ever employing it. Aristotle’s distinction between first and second potentiality of capacities and first and second actuality of capacities is helpful here (De Anima II.5: 417a22–417a30). We can distinguish between an English speaker’s innate capacity to speak a language (first potentiality), her capacity to speak English when she is sleeping (second potentiality) and her capacity to speak English when she is speaking English (second actuality). If one has first potentiality of a capacity, one is the kind of being that could possess that capacity. If one has second potentiality of a capacity, one possesses that capacity (Aristotle also calls this the first actuality of a capacity). If one manifests the second actuality of a capacity, one employs the capacity successfully. A necessary condition for possessing a capacity is to be the kind of being who could possess that capacity and to meet some further constraint, such as being in an environment in which one has the opportunity to come to possess the capacity. Aristotle expresses this idea when he maintains that first potentiality is prior to second potentiality (or first actuality). A necessary condition for employing a capacity is to possess that capacity and to meet some further constraint, such as being in a

Perceptual Capacities 149 suitable environment. Aristotle expresses this idea when he maintains that first actuality is prior to second actuality.15

5.  The Fallibility Condition So far, we have analyzed perceptual capacities in light of their function to discriminate and single out particulars in perception. What happens when we fail to single out what we purport to single out, such as in cases of hallucination and illusion? I argue that perceptual capacities are fallible in that the very same perceptual capacity can be employed in perception, hallucination and illusion. Fallibility Condition: If a subject S employs a capacity Cα, Cα can either fulfill its function or fail to fulfill its function, such that there is no difference at the level of employing Cα but only a difference at the level of fulfilling its function. The function of Cα is fulfilled if by employing Cα a relevant particular is singled out. The function of Cα fails to be fulfilled if by employing Cα no relevant particular is singled out. The relevant alternative to understanding capacities as fallible is to understand them as infallible. Millar among others understands perceptual capacities (including recognitional capacities) in this way: If I had judged falsely that the plants in the plot were azaleas, I would not have exercised the recognitional ability in question. The general point here is that the notion of the exercise of a recognitional ability is a success notion. (Millar, 2008: 333) If capacities are understood as infallible, then one cannot employ a capacity if one does not succeed in fulfilling its function. I will not here argue against infallibilist views of capacities but will focus rather on why we should understand perceptual capacities as fallible. By way of analogy, consider that if we possess a concept, then we can employ it even if we fail to refer. After all, if we say “That’s a horse,” pointing to where in fact there is no horse, we are arguably using the very same concept horse that we would use if we were successfully pointing at a horse. The difference between the former and the latter case is simply that in the former, but not the latter, we fail to refer. The failure occurs at the level of reference. There is no failure at the level of employing the concept. If that is right, then there is no reason to think that the two cases differ with regard to employing the concept horse. The very same thing can be said of perceptual capacities. If we possess a perceptual capacity, then we can employ it even if we are not accurately perceiving. One could be prompted to employ a perceptual capacity due to non-standard circumstances: Unusual brain stimulations or misleading

150  Susanna Schellenberg distal inputs, for example. Given that capacities are determined by functional relations between the perceiver and her environment and not by individual token responses, we can employ a capacity even if a relevant particular is not present. If this is right, then like concepts, perceptual capacities are fallible. If we employ a concept but fail to refer, the concept employed remains empty. Analogously, if we employ a perceptual capacity but fail to single out a particular, the capacity is employed baselessly. It is employed baselessly in the sense that the usual target of discrimination and selection— an external, mind-independent particular—is absent. Let’s consider some examples. In the paradigmatic case of hallucination, it seems to us that there is an object where in fact there is no such object. Consider Kim when she hallucinates a white cup. She employs her capacity to discriminate and single out an object of a certain type. Moreover, she employs her capacity to discriminate and single out white from other colors along with capacities to single out various other property instances: Luminance, shapes, textures and so on. Since she is hallucinating and so not perceptually related to a white cup, all these capacities are employed baselessly. In the paradigmatic case of illusion, it seems to us that an object has a property that it does not in fact instantiate. A subject who is suffering an illusion is not perceptually related to at least one particular that she purports to single out. Say she sees an object that instantiates property π, but given misleading circumstances, it seems to her (falsely) to be instantiating property ρ. In such a case, she employs her capacity to discriminate and single out an instance of ρ. But given that there is no ρ-instance present, she employs that capacity while failing to single out any particular. In the typical case, she will be employing several other capacities successfully. But insofar as she is suffering an illusion, she employs at least one capacity baselessly. Now in perception the particulars between which we discriminate are mind-independent particulars in our environment. This invites the question: What do we discriminate between when we employ perceptual capacities baselessly? In response: When we employ a capacity baselessly, we are not discriminating any mind-independent particulars. Indeed, we are not discriminating any particulars. We are employing a mental tool without that mental tool fulfilling its function. The important point for present purposes is that the fact that the mental tool is not fulfilling its function does not imply that we are not employing the mental tool. 5.1. The Dependence of Perceptual Capacities on Mind-Independent Particulars I have argued that while perceptual capacities are individuated by the particulars they function to single out, they can nonetheless be employed

Perceptual Capacities 151 baselessly. This invites the question of whether perceptual capacities are dependent on the particulars they function to single out.16 There are at least three different ways of understanding this question, each of which requires its own response. One way of understanding it is as a question about possessing capacities. Could a subject possess a perceptual capacity, even though she has never been perceptually related to a particular of the kind that the capacity functions to single out? In response: Yes. After all, the capacity could be innate. The perceiver may have been unlucky and never been perceptually related to a relevant particular. So, despite possessing the capacity, the perceiver will never have had a chance to employ her capacity to successfully single out a relevant particular. A second way of understanding the question is as a question about employing capacities. Could a perceptual capacity be employed even if the relevant particular is not present? In response: Yes. As noted, a perceptual capacity could be employed in the absence of any relevant particular. This occurs in cases of hallucination and illusion. A third way of understanding the question is as an existence question. Could a perceptual capacity exist that functions to single out a kind of particular that does not exist and has never existed? In response: No. Any perceptual capacities must be grounded in perception in the sense that any perceptual capacity must have been employed successfully by someone, somewhere. If that is right, then a perceptual capacity could not exist if no particular of the kind exists that the capacity functions to single out.17 In sum, while perceptual capacities are individuated by the particulars they function to single out, they are dependent on particulars only in the following sense: A  perceptual capacity could not exist if no particular that it functions to discriminate and single out exists or ever has existed.

6.  The Asymmetry Condition While perceptual capacities are fallible and employable in perception, illusion and hallucination alike, there is an asymmetry between employing a capacity in perception and employing that same capacity in hallucination or illusion. Asymmetry Condition: The employment of a perceptual capacity Cα in cases in which Cα fulfills its function is metaphysically more basic than the employment of Cα in cases in which Cα fails to fulfill its function. The reason for this asymmetry is that it is the function of a perceptual capacity to discriminate and single out particulars. It is not its function to fail to single out particulars. This is the case even if a perceptual capacity is more often than not employed unsuccessfully. As a consequence, there is both an explanatory and a metaphysical primacy of the employment of a perceptual capacity in perception over its employment in hallucination or illusion.

152  Susanna Schellenberg There is an explanatory primacy of employing a perceptual capacity in perception over its employment in hallucination or illusion since one can give an analysis of the capacity employed in hallucination or illusion only by appealing to its role in perception. Consider again Kim when she suffers a hallucination of a white cup on a desk. Even though she fails to single out anything white, she is in a phenomenal state that is as of an instance of white in virtue of employing the capacity to discriminate and single out white from other colors. She would single out an instance of white were she perceptually related to a white cup—assuming again that no finking, masking or other exotic case obtains. After all, she is employing a perceptual capacity the very function of which is to differentiate white from other colors and to single out white in her environment. In this sense, we need to refer to what Kim would discriminate between and what she would single out in perception to explain the role of the capacities she employs in hallucination. Licensing this explanatory primacy, there is a metaphysical primacy of employing a perceptual capacity in perception over its employment in hallucination or illusion. There is such a metaphysical primacy since a perceptual capacity functions to do what it does in perception, namely, discriminate and single out particulars. It does not function to do what it does in hallucination or illusion, namely, fail to discriminate and single out the particular that one purports to single out. On one understanding of metaphysical primacy, we can associate things with natures and see if the nature of one thing makes reference to another. If so, the latter will be said to be relatively primary and the former secondary. We can then construct chains so that if the nature of A makes reference to B and the nature of B makes reference to C, then C will be primary, B secondary and A tertiary. According to capacitism, in hallucination and illusion the subject employs her perceptual capacities while failing to fulfill their function, and these capacities are by their nature defined in terms of success in the perceptual case. Thus, the perceptual case is relatively primary and the hallucination and illusion cases are secondary. For the reasons discussed under the fallibility condition, the asymmetry condition does not imply that we must have successfully used a perceptual capacity in the past to employ that capacity in hallucination. Another way of expressing the idea motivating the asymmetry condition is as follows: The fact that we can employ capacities while failing to single out particulars depends on the fact that we can employ such capacities to single out particulars. This idea is analogous to the idea that misrepresentation depends on representation. Indeed, if employing perceptual capacities yields representational content, the two ideas go hand in hand.18 The proposed asymmetric counterfactual analysis of perceptual capacities differs in significant ways from Fodor’s asymmetrical causal dependence account of mental representation (Fodor, 1987, 1990). According to Fodor, a mental state represents properties or objects only if it is reliably

Perceptual Capacities 153 tokened by the presence of the relevant properties or objects. A mental symbol represents, say, pigs only if it is reliably tokened by pigs. So reliability is a necessary condition for Fodor’s account: Symbols of cognitive systems represent because of regularities between those cognitive systems and environments. Such regularities also explain what it is for such symbols to represent in the first place. Like all tracking theories (Dretske, 1981; Millikan, 1984), Fodor’s account faces indeterminacy problems. It fails to ground determinate content, which is required not just for avoiding Quinean indeterminacy problems (e.g., undetached pig parts, pig timeslices), but also to allow for the possibility of misrepresentation (and thus for avoiding the “disjunction” problem) and for ruling out proximal contents (e.g., piggy retinal patterns). Fodor (1990) addresses these indeterminacy problems by adding several conditions to his original account. He stipulates (i) that the mental symbol must be actually caused (not just that it would be caused) by the object or property (i.e., by pigs) and (ii) that the mental symbol has actually been caused by the wrong kinds of objects or properties (i.e., non-pigs) and thus that misrepresentation is not simply possible but that it has actually occurred. Adding these extra conditions, however, undermines the power of the account to explain mental content. The key problem with accounts of mental content that depend on reliability conditions is the following: If a mental state M reliably represents P (e.g., pig), then M will also reliably represent the disjunction P v Q (e.g., pig or a bull terrier; pig or undetached pig part). After all, P and P v Q will be co-instantiated. The reliability relation does not cut finely enough to privilege P over the alternatives. In contrast to Fodor’s asymmetrical causal dependence account, capacitism does not face these problems since it does not depend on the reliability of perceptual capacities.

7.  The Repeatability Condition A perceptual capacity must be repeatable. More precisely: Repeatability Condition: A necessary condition for Cα to be a perceptual capacity is that Cα is repeatable. The repeatability condition implies that it must be possible to employ Cα in at least two distinct contexts for Cα to be a perceptual capacity. Now it might be that one possesses a perceptual capacity that one has—for whatever reason—employed only once, or indeed never. The requirement is not that one has in fact employed a perceptual capacity more than once, but that it is possible to employ that capacity in at least two distinct contexts. The contexts may differ in at least the following five ways. One way is with regard to the particulars singled out. In one context, the perceptual capacity Cα can be employed to discriminate and single out the particular α1; in another it can be employed to discriminate and single

154  Susanna Schellenberg out the particular α2, where α1 and α2 are numerically distinct particulars each of which Cα functions to discriminate and single out. Second, the contexts could differ with regard to whether the perceptual capacity is employed while fulfilling its function or employed while failing to fulfill its function. In one context, a perceptual capacity Cα can be employed while succeeding in singling out the particular α1; in another it can be employed while failing to single out any particular. Third, the contexts could differ with regard to the situational features that determine the conditions under which a particular is perceived—­ features such as lighting conditions, acoustic conditions or the angle and distance from which the particular is perceived. The perceptual capacity Cα can be employed to discriminate and single out the particular α1 under distinct situational features.19 Fourth, the contexts could differ temporally. The perceptual capacity Cα can be employed to discriminate and single out the particular α1 at time t1 and at time t2. Fifth, the contexts could differ spatially. The perceptual capacity Cα can be employed to discriminate and single out the particular α1 at location L1 and at location L2. In each of these five ways in which the contexts could differ, the same perceptual capacity Cα can be employed in two distinct contexts. As these examples of distinct contexts show, the bar for a perceptual capacity to be repeatable is low. Now it may be that at least some particulars are correlated with a unique perceptual capacity. This is plausible if one allows that perceptual capacities are quite high-level. Let’s assume that Robin possesses a perceptual capacity to discriminate and single out his mother. This perceptual capacity will be individuated by exactly one particular in the world. Nonetheless, the perceptual capacity is repeatable. After all, Robin can employ his capacity to single out his mother today and also tomorrow. Capacitism is neutral on whether perceptual capacities function to single out only low-level properties such as colors, shapes, sounds, smells and the like, or whether there are perceptual capacities that function to single out individual people, skyscrapers and pine trees as such. Nothing in the account presented in this chapter hinges on how the debate on whether perception represents only low-level properties or also high-level properties is resolved.

8. The Physical Base and Informational Base of Perceptual Capacities We can analyze perceptual states at three distinct levels: I. The mental state level II. The information-processing, computational level III. The physical, neural level

Perceptual Capacities 155 Employing perceptual capacities lies at the mental state level.20 Computational states, events and processes (as well as any other subpersonal functional states, events and processes) that support mental states lie at the information-processing level. Neural networks and neural activity (as well as other biological or mechanical structures, states, events and processes) in which the other two levels are realized lie at the physical level. What are the computational and neural underpinnings of employing perceptual capacities? What is the relation between mental states brought about by employing perceptual capacities and the non-mental states, events and processes in virtue of which they obtain? Any employment of a perceptual capacity has a physical base. Physical Base Condition: If a subject S is employing a perceptual capacity Cα, then there is a physical base of employing Cα that is constituted by physical states, events and processes (such as the neural activity) of S. The physical base condition allows for multiple realizability. So, the fact that employing perceptual capacities has a physical base does not imply an identity relation between employing a perceptual capacity and its physical base. Nor does it imply that there is an identity relation between the mental states constituted by employing perceptual capacities and the physical base of their employment.21 Any employment of a perceptual capacity has not only a physical base but also an informational base. Informational Base Condition: If a subject S is employing a perceptual capacity Cα, then there is an informational base of employing Cα that is constituted by the subpersonal psychological mechanism (information processing, computations and other subpersonal functional states, events and processes) of S. There are complex relations between the information-processing level and the physical level. After all, neural networks encode ­information. One central question is what the relation is (if any) between information-­processing modularity and neural modularity—assuming here standar­dly, though not uncontroversially, that there are information-­processing modules (Barrett and Kurzban, 2006; Evans and Frankish, 2009). ­Information-processing modules are informationally encapsulated, f­unctionally specialized computational mechanisms that are dedicated to perceptual or cognitive tasks: Specific perceptual discrimination, ­biological classification, face recognition, to give just a few examples (Fodor, 1983; Coltheart, 1999; Barrett and Kurzban, 2006; Carruthers, 2006). Neural modularity is a claim about the relation between information-processing modules and physical neural networks, namely, that there is a one-to-one mapping between information-processing modules and locations of neural activity. It has been argued that information-processing modules have localized neural bases and that evidence of neural modularity, and more specifically of neuroanatomical localization, is required to support claims

156  Susanna Schellenberg of information-processing modularity.22 There is, however, compelling evidence that information-processing modularity does not entail physical neural modularity.23 After all, information-processing modules are functionally characterized and could change over time—in response, for example, to damage (Segal, 1996). So, while at any given time there must be some neural structure (or analogous physical structure) that realizes each module’s processing mechanism and establishes its informational connections with other subsystems, these structures could change. Furthermore, distinct information-processing modules might be grounded in the same neural structures. As with any complex biological or informational systems, there may be considerable sharing of physical parts between information-processing modules (Carruthers, 2006). Moreover, given the flexibility of neural networks and physical structures more generally, any commitment to physical modularity should be rejected (Lloyd, 2011). Thus, there is good evidence that information-processing modules need not have localized neural bases and that neuroanatomical localization is not required to support claims of information-processing modularity (Frankish, 2009). And indeed, we can accept the physical base condition on the employment of perceptual capacities without endorsing any one-to-one mapping between information-processing modules and locations of neural activity. What about the relation between the mental state level and the information-processing level? The view that perceptual states are constituted by employing perceptual capacities fits neatly with computationalism, according to which personal-level mental states are grounded in computational states. Now some reductive versions of computationalism have it that mental states are fully analyzable in computational terms. According to such views, personal-level mental states can be deduced from computational states, events and processes: Mental states simply are computational states at a certain stage of information processing. We can accept that mental states are grounded in computational states, however, without endorsing such a reductive view. After all, states, events and processes on the mental level can be grounded in states, events and processes on the computational level even if no identity relations hold between the two levels. Accepting a grounding relation does not entail that personal-level mental states can be identified with or reduced to computational states, events and processes. Moreover, states, events and processes on the computational level can cause states, events and processes on the mental level even if no identity relations hold between the two levels. In short, mental states, events and processes can be grounded in, explained in terms of or obtain in virtue of computational states, events and processes without any identity relations holding between the two levels. This approach allows us to accept the existence of states, events and processes on both levels and to understand vision science and the cognitive sciences, more generally, as investigating the metaphysical and

Perceptual Capacities 157 explanatory dependencies between the two levels. On this approach, the focus is not on whether there are identity relations between states, events and processes on the two levels but on the causal and grounding relations between the two (Strevens, 2004; Craver, 2007; Godfrey-Smith, 2008; Silva and Bickle, 2009; Craver and Darden, 2013). This allows us to acknowledge that an account of information processing is a necessary element of any complete account of perception, while also acknowledging that central questions, such as the nature and source of perceptual consciousness and the epistemic force of perceptual states, cannot be adequately addressed solely at the computational level. As I have argued, employing perceptual capacities is grounded in subpersonal computational mechanisms and physical neural networks that encode information. Thus, capacitism entails that perceptual states can be scientifically explained in terms of informational and physical states, events and processes without thereby reducing perceptual states to those non-mental features. In this way, capacitism posits that perceptual states are genuinely mental yet can nonetheless be the object of scientific inquiry.24

9.  The Generality of Perceptual Capacities A perceptual capacity is general in that it can be employed to single out any particular of the type that the capacity functions to discriminate and single out. In the typical case, no specific particular needs to be singled out in any specific employment of a perceptual capacity.25 Any particular will do, as long as it falls under the type of particulars that the capacity functions to discriminate and single out. For example, the perceptual capacity Csquare can be employed to discriminate and single out any perceivable square object. In this sense, it is semantically general in much the way as the concept square is semantically general. Semantic generality should be distinguished from syntactic generality. While perceptual capacities are semantically general, they are syntactically singular: They function to single out particulars in the environment—not general kinds or universals. In this respect, they are akin to singular terms, such as demonstratives and indexicals. Not only are perceptual capacities syntactically singular but the perceptual states they yield are syntactically singular as well. 9.1  Perceptual Capacities and Modes of Presentation By employing a perceptual capacity in perception, we single out a particular in a certain way. Let’s say we are perceptually related to a triangle. We can single it out via its three-sidedness or via its three-corneredness. When we single it out via its three-sidedness we employ a different capacity than when we single it out via its three-corneredness. Similarly, when

158  Susanna Schellenberg we hear a cello in the midst of the cacophony of an orchestra, we can single it out in virtue of its rich timbre or its reverberating sound. When we see a ruby-red gemstone, we can single it out in virtue of its being red or in virtue of its being ruby-red. As I  have argued elsewhere, employing perceptual capacities constitutes perceptual content, and this content is structured by singular modes of presentation.26 Employing perceptual capacities parallels Fregean modes of presentation both with regard to being ways of singling out particulars and with regard to the fact that any particular can be singled out by employing a range of different perceptual capacities. The idea that content is constituted by employing perceptual capacities by means of which we (purport to) single out particulars is analogous to the Fregean idea that modes of presentation are a way of grasping or referring to particulars. A mode of presentation is the specific way in which a subject refers to a particular. While Frege introduces the distinction between sense and reference with a perceptual case, he does not develop the notion for perceptual content. His focus was never on lowly mental faculties like perception. Nonetheless, we can apply his view of modes of presentation to the case of perception. Applied to that case, the idea is that a mode of presentation is the specific way in which a subject singles out a perceived particular. Insofar as perceptual capacities are ways of singling out particulars, they can be understood as the mental counterpart of Fregean modes of presentation: As a mode of presentation is a way of referring to an object, employing a perceptual capacity is a way of singling out a particular. Moreover, just as there is a many–one relation between senses and references, there is a many–one relation between perceptual capacities and particulars. And while a mode of presentation is a component of a thought or a proposition, a perceptual capacity is a mental tool. According to Frege, concepts are mappings from objects onto truth-values (Frege, 1879). Similarly, perceptual capacities are mappings from particulars onto accuracy conditions. Not only do perceptual capacities parallel modes of presentation in their role of singling out particulars, insofar as employing perceptual capacities constitutes perceptual states, but they have a certain cognitive significance. So like Fregean modes of presentations, perceptual capacities play the dual role of having a cognitive significance and being a means of singling out particulars. One key motivation for introducing perceptual capacities and modes of presentation is to capture a fineness of grain in content that reference to mind-independent particulars alone could not achieve. Acknowledging that particulars are always singled out via employing perceptual capacities and grasped under modes of presentation makes room for the fact that any mind-independent particular can be represented in different ways. So any particular can be singled out via employing distinct perceptual capacities, and any particular can be grasped under different

Perceptual Capacities 159 modes of presentation. On a Russellian understanding, alternative possible modes of presentation can be expressed only insofar as one may have different cognitive attitudes to the same content. The way in which one perceives or thinks of the object is not expressed in the content proper. On the Fregean approach, every particular perceived will be represented under a mode of presentation. Due to this, the Fregean view avoids counterexamples to Russellian representational views.27 Consider a case in which you are looking at a page of graph paper, a page of symmetrically arranged tiles. You can see the tiles as being grouped. There are a number of ways the tiles can be grouped depending on which tiles are seen to be more prominent. Now let’s say that at time t1 you see one set of tiles as prominent and at time t2 you see another set of tiles as prominent ceteris paribus. In such a case, there is no difference in the environment: The tiles perceived are exactly the same at t1 and t2. The only difference is how the mind groups the tiles.28 Since there is no change in the environment to which you are perceptually related, it is not clear how a Russellian would account for the change in representational content. A Fregean has no problem dealing with such a case. A Fregean will say that you represent the tiles under different modes of presentation at t1 and t2. A more general advantage of the Fregean approach is that it accounts for the fact that thought is fundamentally perspectival. Applied to perception, it accounts for the fact that perception is fundamentally perspectival—­perspectival not only in that we perceive from a location and so in an egocentric frame of reference, but also in that we always perceive particulars under specific conditions (location, lighting conditions, acoustic conditions) with a specific set of perceptual capacities. There is always a way in which we discriminate and single out particulars in our environment. Consider Sasha, who hears jazz for the first time. When listening to John Surman’s recording of “Doxology” for the first time, she will not discern much. As she becomes an expert, she will discern significantly more when listening to the very same recording. One explanation is that she develops more fine-grained perceptual capacities that allow her, for example, (1) to discriminate between the sound of the trumpet and the sound of the piano even when they are playing at the same time, and (2) to hear differences between chords. More radically, we can say that we cannot perceive a particular in our environment without perceiving it from our location with our specific perceptual capacities.29 In this sense, we cannot perceive without being constrained by our perspective. The Fregean approach acknowledges this. There are two standard ways of thinking about Fregean modes of presentation. If one focuses on the role of modes of presentation as accounting for cognitive significance, then it is natural to think of them as de dicto. A de dicto mode of presentation is general in that it can be the very same regardless of what (if anything) the experiencing subject is

160  Susanna Schellenberg perceptually related to. If, by contrast, one focuses on the role of modes of presentation as a way of referring to a particular, then it is natural to think of them as de re. A de re mode of presentation is singular in that what particular (if any) the subject is perceptually related to has repercussions for the token content. A de dicto mode of presentation lays down a condition that something must satisfy to be the particular determined by the content. Chalmers, among others, understands Fregean senses in this way: “Fregean content is supposed to be a sort of phenomenal content, such that, necessarily, an experience with the same phenomenal character has the same Fregean content” (2006: 99, see also Thompson, 2009). A de dicto mode of presentation constitutes a way of representing mind-independent particulars irrespective of whether the relevant particulars are present. If the content of experiential states were constituted by de dicto modes of presentation, then the content of a perception, a hallucination or an illusion with the same phenomenal character would be (ep,h, i) where MOPdo is a de dicto mode of presentation of an object and MOPdF is a de dicto mode of presentation of a property. Such an account of perceptual content implies a two-stage view of determining reference: First, we represent a general content and in a second step, we refer to mindindependent particulars based on this content.30 Representing a de dicto mode of presentation is, on this view, independent of the second step, in which a particular may be determined. Such a two-stage view faces the problem of how the content grounds the ability to refer to external particulars. Insofar as a de dicto mode of presentation can be the very same regardless of what (if anything) the experiencing subject is perceptually related to, this way of thinking about content amounts to a version of austere representationalism and faces all the difficulties of that view. Any view on which perceptual content is constituted by de dicto modes of presentation fails to satisfy the particularity desideratum for the same reasons that austere representationalism does. This problem is avoided if perceptual content is analyzed as constituted by de re rather than de dicto modes of presentation. Understanding modes of presentation as de re is motivated by recognizing that modes of presentation play a dual role: They have a cognitive significance and they single out or refer to mind-independent particulars. Understanding perceptual content as constituted by de re modes of presentation recognizes that representing a particular is not independent of singling out the particular that is the referent of the sense. By contrast to de dicto modes of presentation, de re modes of presentation are singular in the good case. Now on one way of understanding de re modes of presentation, a subject can have a contentful experience only if she is (perceptually)

Perceptual Capacities 161 related to the very particular that she purports to single out. This view is a version of content disjunctivism.31 One problem is that the cognitive significance and the action-guiding role of experiential content is downplayed. When a subject hallucinates, the way things seem to her plays a certain cognitive role. If it seems to her that she is perceptually related to a white cup, she may, for example, reach out and try to pick it up. If one denies that hallucinations have representational content, this cannot be explained. It is not clear how the mere illusion of content could motivate the subject to act. Consider Harman’s example of Ponce de Leon who was searching Florida for the mythical Fountain of Youth (Harman, 1990). The Fountain of Youth does not exist, yet Ponce de Leon was looking for something particular. As Harman argues convincingly, he was not looking for a mental object. He was looking for a mind-independent object that, as it so happened, unbeknownst to him, did not exist. A  second problem—and the problem most salient for present purposes—is that, insofar as content disjunctivists hold that hallucinations do not represent, they leave unclear what explains the phenomenal character of hallucinations. So it is not clear how content disjunctivists satisfy the phenomenal sameness desideratum. While content disjunctivists acknowledge that a hallucination could seemingly have the same phenomenal character as a perception, they do little if anything to explain this phenomenon. The problems of disjunctivism are avoided if perceptual content is not understood as radically object-dependent. That would allow that hallucinations can have at least some kind of content. One way to develop such a view is to argue that the content of a hallucination involves a gap that in the case of a perception is filled by a particular. Recall that in Section  1, we distinguished between three levels at which to analyze mental states. The first level of analysis pertains to the function of the mental capacity. The second level of analysis pertains to the mental capacity employed, irrespective of the context in which it is employed. The third level of analysis pertains to the mental capacity employed, taking into account the context in which it is employed. Applied to the notion of perceptual content, we can say that the second level of analysis pertains to the content type of a mental state, while the third level of analysis pertains to the token content of a mental state. The content type and the token content are both constituted by capacities that have a certain function. So the first level of analysis explains how the content type and the token content are connected beyond the one being a token of the other. Harnessing this distinction, we can say that there is a content type that is constituted by the perceptual capacities employed. Consider the case in which a subject, let’s call her Miriam, employs two perceptual capacities, one that functions to single α particulars and one that functions to single out π particulars. Let’s say α particulars are cups α1 and the π particulars

162  Susanna Schellenberg are instances white. In such a case, the content type of the perceptual state brought about by employing these two capacities will be: (contentType) where MOPrα[___] is a content type that is constituted by employing the perceptual capacity Cα that functions to single out particulars α1, α2, α3,  .  .  .  αn; and MOPrπ[___] is a content type that is constituted by employing the perceptual capacity Cπ that functions to single out particulars π1, π2, π3,  .  .  . πn. If in employing the perceptual capacity Cα the particulars α1 is singled out, then the token content of the perceptual state will be MOPrα(α1). If in employing the perceptual capacity Cα the particulars α2 is singled out, then the token content of the perceptual state will be MOPrα(α2). If in employing the perceptual capacity Cα no particular is singled out, then the token content of the perceptual state will be MOPrα(__). So the content type MOPrπ[___] can be tokened by MOPrα(α1), MOPrα(α2) and MOPrα(__). Let’s say Miriam sees a white cup and so singles out α1 by employing Cα and π1 by employing Cπ. In this case, the token content of her perceptual state will be: (contente1) where MOPrrα(α1) is a singular mode of presentation of the cup α1 that is the product of employing a perceptual capacity that functions to single out the kind of object under which α1 falls. So “α1” is functioning as the name of an object. “MOPrrα” is a functional expression that expresses a function from objects to singular modes of presentation. MOPrπ(π1) is a singular mode of presentation of the property-instance π1 that is the product of employing a perceptual capacity that functions to single out instances of the property under which π1 falls. So while MOPrα(α1) is a de re mode of presentation of the object α1, MOPrπ(π1) is a de re mode of presentation of the property-instance π1. Now let’s say that Miriam hallucinates a white cup. In this case, the token content of her hallucinatory state will be: (contenth) where MOPrα(__) specifies the kind of object that would have to be present for the experience to be accurate, and MOPrπ(__) specifies the properties that this object would instantiate were the experience a perception rather than a hallucination. More specifically, MOPrα(__) is a gappy mode of presentation that is the product of employing a perceptual capacity that functions to single out objects of the kind that the hallucinating subject purports to single out while failing to single out any such

Perceptual Capacities 163 object. It accounts for the intentional directedness of the experience at a (seeming) particular object. MOPrπ(__) is a gappy mode of presentation that is the product of employing a perceptual capacity that functions to single out property instances of the kind that the hallucinating subject purports to single out while failing to single out any such propertyinstance. It accounts for the intentional directedness of the experience at a property-instance. In short, MOPrα(__) is a gappy, object-related mode of presentation and MOPrπ(__) is a gappy, property-related mode of presentation. So for a perceptual capacity to be employed baselessly amounts to the ensuing token content being gappy. There is nothing metaphysically spooky about gaps. The gap simply marks the failure to single out a particular. Both Miriam’s hallucinatory state and her perceptual state are characterized by: 32 (contentType) 9.2 Perceptual Capacities, Concepts and Nonconceptual Content A perceptual capacity can be understood either as a conceptual or a nonconceptual capacity. Which stance one takes will depend largely on how one understands the nature of concepts and their possession conditions. Depending on how concepts are understood, it is more or less plausible to think of perceptual content as conceptually structured. For this reason, the debate over whether perceptual content is conceptual or nonconceptual is almost entirely terminological. One of the advantages of analyzing perceptual states (and, as I will argue, perceptual content) as constituted by employing perceptual capacities is that it allows us to sidestep the issue of whether perceptual content is conceptual or nonconceptual.33 Concepts have been understood in terms of mental representations, stereotypes, functional roles and inferential roles, to name just a few standard views. Nonconceptual content has been understood in terms of image-like or map-like representations, as constituted by employing nonconceptual, perceptual capacities, or in terms of the idea that we represent, naked properties and objects. If concept possession requires the ability to draw inferences, then it is wildly implausible that the capacities employed in perception are conceptual capacities.34 After all, perception is a low-level mental faculty that we share with animals that have no inferential capacities. This implies that, if concept possession requires the ability to draw inferences, then it cannot be the case that all perceptual capacities are conceptual capacities. If, on the other hand, it is held that all perceivers possess concepts—even perceivers that have no inferential abilities or any other such high-level cognitive abilities—then it is more plausible that perceptual capacities are conceptual capacities. On such a view of concepts, the requirements for

164  Susanna Schellenberg concept possession are cognitively so minimal that it becomes unproblematic to say that a honeybee possesses concepts and hence unproblematic to say that perceptual capacities are conceptual. While the debate on whether perceptual content is conceptual or nonconceptual is almost entirely terminological, there are elements of the debate that are not terminological. Focusing on those elements, I argue that perceptual content is nonconceptual. The key motivations are to accommodate the fact that at least some aspects of perceptual content can be image-like or map-like and moreover to account for the richness and fineness of grain of perceptual experience.35 If perceptual content is constituted by employing such nonconceptual capacities, then perceptual content is nonconceptual. The thesis that perceptual content is constituted by employing perceptual nonconceptual capacities gives a substantive analysis of the nonconceptual content of perception. The thesis that perceptual content is nonconceptual is supported by the fact that on standard views of concepts, perceptual experience is richer and more fine-grained than our concepts. For example, the shades of color a perceiver is able to discriminate in perception are typically significantly more fine-grained than her color concepts. If that is right (and on most notions of concepts it is), then richness and fineness of grain of perceptual experience supports the thesis that perceptual content is nonconceptual. Additional evidence is provided by the fact that non-rational animals perceive. If non-rational animals do not possess concepts, then perceptual capacities cannot necessarily be conceptual. As mentioned, however, whether this additional evidence has any force depends on the notion of concept with which one is operating. The conceptualist might object that if singular thoughts or perceptual beliefs inherit their content from perception, then perceptual content must have the same structure as the content of belief. If that is right, then perceptual content must be conceptual rather than nonconceptual. In response, the nonconceptualist can say that such beliefs can be based on perception without their content being exactly like perceptual content. After all, the fact that perceptual beliefs are based on perception does not imply that perceptual content is conceptual. While it is plausible that at least some elements of perceptual content are similar to the content of a belief based on that perception, the similarity need not be a matter of both mental states having conceptual content. The conceptualist might object further that only something that is conceptually structured can justify beliefs; so, if perceptual experience justifies beliefs, then perceptual content must be conceptually structured. In response, the nonconceptualist can say that all we need for experience to play a justificatory role is that its content is propositionally structured. But content can be propositionally structured without being conceptually

Perceptual Capacities 165 structured. Moreover, there are reasons to question whether something must be propositionally structured in order to provide evidence. In sum, there is good reason to understand perceptual content and perceptual capacities as nonconceptual. The thesis that perceptual content is constituted by employing perceptual capacities allows for a substantive way of analyzing perceptual content as nonconceptual. However, the thesis is also compatible with understanding (at least some) perceptual capacities as conceptual capacities. Indeed, one of the benefits of analyzing perceptual content as constituted by perceptual capacities is that it allows one to sidestep the largely terminological debate over whether perceptual content is conceptual or nonconceptual.

10. Coda I have developed an asymmetric counterfactual analysis of perceptual capacities. The asymmetry stems from the primacy of the employment of perceptual capacities when the capacities fulfill their function over their employment when they fail to fulfill their function. The analysis is counterfactual, since (subtleties aside) one qualifies as possessing a perceptual capacity only if one would be in a position to discriminate and single out a particular of the type that the capacity functions to single out, were one perceptually related to such a particular. Moreover, the analysis is externalist insofar as capacities are individuated by the external, mindindependent particulars that they function to discriminate and single out.

Notes 1. See Cummins, 1985 for a good overview. 2. There are notable exceptions. See, for example, Cartwright, 1994 and Sosa, 2010. 3. For discussion, see Hawthorne and Manley, 2012. 4. For this reason, the account of capacities developed here does not face Dennett’s (1991) indeterminacy worries. 5. For a critical discussion of etiological accounts of function, see Nanay, 2010. As Nanay argues, such accounts are circular. 6. Schellenberg, 2013, 2017. 7. For discussions of the role of pre-attentive discrimination in perception, see Julesz, 1981, Watson and Robson, 1981, Sagi and Julesz, 1985, Malik and Perona, 1990, Krummenacher and Grubert, 2010 and To, Gilchrist et  al., 2011. 8. Accepting this is compatible with holding that perceptual knowledge results from the exercise of cognitive capacities operating on inputs received from perception. However, in Part III, I will develop a view of perceptual knowledge that does not put any such intellectualist conditions on perceptual knowledge. 9. I am here following Byrne and Hilbert (2003) in treating color properties, and similar such properties, as external, mind-independent intrinsic properties. My argument, however, easily generalizes to alternative views of color

166  Susanna Schellenberg as long as there are external, mind-independent properties, such as reflectance properties or wavelength emittence properties, that form the basis for perception of colors. 10. For a development of the notion of situation-dependent properties, see Schellenberg, 2008. For critical discussions, see Cohen, 2010 and Jagnow, 2012. 11. For a discussion of masking, see Johnston, 1992; for a discussion of finking, see Martin, 1996. 12. For the distinction between “might”-conditionals and “would”-condi tionals, see Lewis, 1973: 21–24. For a discussion of “could”-conditionals, including a discussion of whether they are in fact conditionals, see Austin, 1970: 211–213. See also DeRose and Grandy, 1999. 13. For discussions of this set of issues, see in particular Li et al., 2004, 2009, Scott et al., 2007. See also Luna et al., 2005, de Lafuente and Romo, 2005, Chowdhury and DeAngelis, 2008, Law and Gold, 2008, Kahnt et al., 2011. 14. See Caston, 2002 for a helpful discussion. Thanks to Victor Caston for helpful exchanges on Aristotle’s view of capacities and powers. 15. For distinctions analogous to the distinction between employing a capacity and possessing a capacity, see Schellenberg, 2007, Glick, 2012, Vihvelin, 2013 and Whittle, 2010. 16. Thanks to Matt McGrath for raising this question. 17. For a defense of this idea, see Schellenberg, 2018a. 18. For a helpful discussion of asymmetry arguments, see Marušić, 2016. 19. For a discussion of situational features, see Schellenberg, 2008. 20. I use “mental” to refer to personal-level states, events and processes and “information processing” to refer to states, events and processes that are at a subpersonal level. To avoid terminological confusions, it is important to note that some have used “mental” to refer to states, events and processes at the subpersonal, computational level (see e.g., Fodor, 1975). 21. The locution “of the subject S who is employing Cα” in the physical base condition need not be understood as implying that the physical base is a biological component of S. The physical base could be an implant. 22. For discussion, see Fodor, 1983 and Panksepp and Panksepp, 2001. Note that they use the terminology of mental modules rather than informationprocessing modules. For a helpful discussion of modularity, see Toribio, 2002. 23. For general discussion, see Koch et al., 2016. 24. For helpful discussions of computational accounts of perception, see Egan, 1992 and Cohen, 2010. 25. Exceptions are perceptual capacities that function to single out one unique particular, such as Robin’s perceptual capacity to single out his mother. 26. Schellenberg, 2011. 27. See, for example, Neander, 1998; Macpherson, 2006 and Nickel, 2007. 28. For a discussion of this case, see Nickel, 2007. 29. For an excellent discussion of how best to understand the mental capacities we bring to bear in perceptual experience, see Speaks, 2005. 30. For an argument against such a two-stage view of determining reference, see Johnston, 2004: 150f. Johnston does not distinguish between de dicto and de re modes of presentation, and as a consequence sees the problem articulated in the main text as a problem for any Fregean view tout court. As I will show, it is only a problem for a view on which Fregean senses are de dicto rather than de re. 31. For a defense of such a view, see Evans, 1982 and McDowell, 1984. 32. Schellenberg, 2018b.

Perceptual Capacities 167 33. For discussion of nonconceptual content, see Peacocke, 1998; Heck, 2000 and Speaks, 2005. For recent arguments for the idea that perceptual content is conceptually structured, see Glüer-Pagin, 2009 and Bengson et al., 2011. 34. For a view on which possessing concepts requires inferential capacities, see Brandom, 1994. 35. The key arguments in this book can, however, be accepted if perceptual capacities are understood as conceptual rather than as nonconceptual capacities.

References Aristotle (1984) The Complete Works of Aristotle (ed. J. Barnes), Princeton: Princeton University Press. Austin, J. L. (1946) “Other Minds” in Supplement to the Proceedings of the Aristotelian Society, 20, 148–187. Austin, J. L. (1970) “Ifs and Cans” in Philosophical Papers, 2nd ed., London: Oxford University Press. Ayala, F. (1970) “Teleological Explanations in Evolutionary Biology” in Philosophy of Science, 37, 1–15. Barrett, H. C. and Kurzban, R. (2006) “Modularity in Cognition: Framing the Debate” in Psychological Review, 113, 628–647. Block, N. (1978) “Troubles with Functionalism” in Minnesota Studies in the Philosophy of Science, 9, 261–325. Brandom, R. (1994) Making It Explicit, Cambridge, MA: Harvard University Press. Burge, T. (1979) “Individualism and the Mental” in Midwest Studies in Philosophy, 4, 73–121. Carruthers, P. (2006) The Architecture of the Mind, Oxford: Oxford University Press. Chalmers, D. (2006) “Perception and the Fall from Eden” in Perceptual Experience (ed. T. Gendler and J. Hawthorne), Oxford: Clarendon Press. Chomsky, N. (1995) “Language and Nature” in Mind, 104, 1–61. Coltheart, M. (1999) “Modularity and Cognition” in Trends in Cognitive Sciences, 3, 115–120. Craver, C. (2007) Explaining the Brain: Mechanisms and the Mosaic Unity of Neuroscience, Oxford: Oxford University Press. Craver, C. and Darden, L. (2013) In Search of Mechanisms: Discoveries across the Life Sciences, Chicago: University of Chicago Press. Cummins, R. (1985) The Nature of Psychological Explanation, Cambridge, MA: MIT Press. Cummins, R. (2011) “Ontological Minimalism about Phenomenology” in Philosophy and Phenomenological Research, 83, 1–40. Davidson, D. (1987) “Knowing One’s Own Mind” in Proceedings and Addresses of the American Philosophical Association, 60, 441–458. Dennett, D. C. (1991) Consciousness Explained, Boston: Little, Brown & Co. DeRose, K. (2009) The Case for Contextualism, Oxford: Oxford University Press. Devitt, M. (1981) Designation, New York: Columbia University Press.

168  Susanna Schellenberg Dretske, F. (1969) Seeing and Knowing, Chicago: The University of Chicago Press. Dretske, F. (1981) Knowledge and the Flow of Information, Cambridge, MA: MIT Press. Evans, G. (1982) The Varieties of Reference (ed. J. McDowell), Oxford: Oxford University Press. Evans, J. St. B. T. and Frankish, K. (2009) In Two Minds: Dual Processes and Beyond, Oxford: Oxford University Press. Fodor, J. (1975) The Language of Thought, Cambridge, MA: Harvard University Press. Fodor, J. (1983) The Modularity of Mind: An Essay on Faculty Psychology, Cambridge, MA: MIT Press. Fodor, J. (1987) Psychosemantics: The Problem of Meaning in the Philosophy of Mind, Cambridge, MA: MIT Press. Fodor, J. (1990) “A  Theory of Content, I  & II” in A Theory of Content and Other Essays (ed. J. Fodor), Cambridge, MA: MIT Press. Frankish, K. (2009) “Systems and Levels: Dual-System Theories and the Personal-Subpersonal Distinction” in Two Minds: Dual Processes and Beyond (ed. J. St. B. T. Evans and K. Frankish), Oxford: Oxford University Press. Godfrey-Smith, P. (2008) “Reduction in Real Life” in Being Reduced (ed. J. Hohwy and J. Kallestrup), Oxford: Oxford University Press. Goldman, A. (1976) “Discrimination and Perceptual Knowledge” in The Journal of Philosophy, 73, 771–791. Harman, G. (1990) “The Intrinsic Quality of Experience” in Philosophical Perspectives, 4 (ed. J. Tomberlin), Northridge: Ridgeview Publishing Company. Hawthorne, J. (2003) Knowledge and Lotteries, Oxford: Oxford University Press. Johnston, M. (1992) “How to Speak of the Colors” in Philosophical Studies, 68, 221–263. Kripke, S. (1972) Naming and Necessity, Cambridge, MA: Harvard University Press. Lewis, D. (1966) “An Argument for the Identity Theory” in Journal of Philosophy, 63, 17–25. Lewis, D. (1973) Counterfactuals, Oxford: Blackwell. Lewis, D. (1997) “Finkish Dispositions” in Philosophical Quarterly, 47, 143–158. Lloyd, D. (2011) “Through a Glass Darkly: Schizophrenia and Functional Brain Imaging” in Philosophy, Psychiatry and Psychology, 18, 257–274. Macpherson, F. (2006) “Ambiguous Figures and the Content of Experience” in Noûs, 40, 82–117. Martin, C. B. (1996) “Dispositions and Conditionals” in The Philosophical Quarterly, 44, 1–8. McDowell, J. (1984) “De Re Senses”  in The Philosophical Quarterly, 34, 283–294. Millar, A. (2008) “Perceptual-Recognitional Abilities and Perceptual Knowledge” in Disjunctivism: Perception, Action, Knowledge (ed. A. Haddock and F. Macpherson), Oxford: Oxford University Press. Millikan, R. G. (1984) Language, Thought and Other Biological Categories, Cambridge, MA: MIT Press.

Perceptual Capacities 169 Millikan, R. G. (1989) “In Defense of Proper Functions” in Philosophy of Science, 56, 288–302. Neander, K. (1991) “Functions as Selected Effects” in Philosophy of Science, 58, 168–184. Neander, K. (1998) “The Division of Phenomenal Labor: A Problem for Representational Theories of Consciousness” in Philosophical Perspective, 12 (ed. J. E. Tomberlin. Atascadero), CA: Ridgeview Publishing. Nickel, B. (2007) “Against Intentionalism” in Philosophical Studies, 136, 279–304. Panksepp, J. and Panksepp, J. B. (2001) “A Continuing Critique of Evolutionary Psychology: Seven Sins for Seven Sinners, Plus or Minus Two” in Evolution and Cognition, 7, 56–80. Peacocke, C. (1992) A Study of Concepts, Cambridge, MA: MIT Press. Peacocke, C. (1998) “Nonconceptual Content Defended” in Philosophy and Phenomenological Research, 63, 381–388. Pritchard, D. (2010) “Relevant Alternatives, Perceptual Knowledge, and Discrimination” in Noûs, 44, 245–268. Putnam, H. (1975) “The Meaning of Meaning” in his Philosophical Papers vol. 1: Mind, Language, and Reality, Cambridge: Cambridge University Press. Schaffer, J. (2005) “Contrastive Knowledge” in Oxford Studies in Epistemology (ed. T. S. Gendler and J. Hawthorne), Oxford: Oxford University Press. Schellenberg, S. (2008) “The Situation-Dependency of Perception” in The Journal of Philosophy, 105, 55–84. Schellenberg, S. (2011) “Ontological Minimalism about Phenomenology” in Philosophy and Phenomenological Research, 83, 1–40. Schellenberg, S. (2018a) “Perceptual Consciousness as a Mental Activity” in Noûs. Schellenberg, S. (2018b) The Unity of Perception: Content, Consciousness, Evidence, Oxford: Oxford University Press. Segal, G. (1996) “The Modularity of Theory of Mind” in Theories of Theories of Mind (ed. P. Carruthers and P. K. Smith), Cambridge: Cambridge University Press. Shoemaker, S. (1994) “Phenomenal Character” in Noûs, 28, 21–38. Silva, A. and Bickle, J. (2009) “Science of Research and the Search for Molecular Mechanisms of Cognition” in Oxford Handbook of Philosophy and Neuroscience (ed. J. Bickle), Oxford: Oxford University Press. Stanley, J. (2005) Knowledge and Practical Interests, Oxford: Oxford University Press. Strevens, M. (2004) “The Causal and Unification Accounts of Explanation Unified—Causally” in Noûs, 38, 154–179. Thompson, B. J. (2009) “Senses for Senses” in Australasian Journal of Philosophy, 87, 99–117. Wright, L. (1973) “Functions” in The Philosophical Review, 82, 139–168.

9 Thinking Differently About Thought Nicholas Georgalis

1. Introduction On birth, as a matter of fact, we enter a community in which a longestablished language is spoken that we neither speak nor understand, though we eventually succeed in doing both. I  have no idea just how these remarkable accomplishments occur, and they will not be discussed here. It is a fact, however, that we eventually succeed in expressing our thoughts, understanding one another’s utterances and ascribing thoughts to others. We do all this with the common language of our community. I take my starting point to be this language as embodied in knowledgeable and fluent speakers of English, together with our ordinary linguistic practices, including those governing thought attributions. We talk of the meanings of words within this framework. These meanings are just what any fluent and knowledgeable speaker of the language would understand them to have were she not engaging in philosophical discourse, “the face value of the words,” if you will (to borrow a phrase from Quine). When I speak of the meaning of a term, I just mean their customary linguistic meaning as it is determined through the established use of the language; this is henceforth referred to as the “standard meaning” which is more or less codified in dictionaries.1 Being publicly available, standard meaning is objective. Still, while the ordinary fluent speaker has adequate command and understanding of vast stores of such words, there are many others for which her command of their use or her understanding is much more limited though still adequate, e.g., “electron.” Nevertheless, such a speaker would have little difficulty uttering some meaningful sentences using that term that expresses her thought on that occasion, e.g., “Electrons have negative charge.” Frequently, the layperson’s significant use of such terms is parasitic on experts in the language community who do have an adequate understanding.2 The basic idea as it pertains to standard meaning is that within a language community there are several predicates associated with a term which amount to its standard meaning such that a speaker in the community may use the term to make meaningful utterances even

Thinking Differently About Thought 171 though she herself is quite ignorant of the range of applicable predicates. Knowledge of the predicates associated with a term is not the sole responsibility of a speaker, nor is it a condition for the speaker’s meaningful and correct use of the term; such knowledge is embedded in the community knowledge of the public language, and it is that community knowledge that secures standard meaning. Allow me, then, this talk of standard meaning, as it will suffice (I think) to enable me to make some important and sustainable points about the relations that obtain between subjectivity, thought and language, as well as to explain the crucially important differences between the contents of what one thinks, on the one hand, and the contents of a sentence appropriately used to express that, on the other.3

2.  A New Theory of Thought We use sentences to express our thoughts and to attribute thoughts to others. Except as noted in the comments above, components of sentences typically have rather determinate contents (standard meanings). The content of what one thinks sometimes lacks the relative determinacy of content of a sentence which appropriately expresses it.4 Parasitic uses of terms are typically like this. For example, suppose I use the sentence “Up quarks have color” to express a thought that I in fact have. The actual content of what I am thinking on this occasion is extremely sparse, as I know next to nothing about up quarks. Still, I would not and should not be faulted for using that sentence to express what I am thinking. In contrast, if an advanced elementary particle physicist were to express her thought on some occasion with the same sentence, we would expect a much greater coincidence between the content of her utterance and what she is actually thinking. Do we have the same thought? The answer seems to be yes and no. Insofar as the same sentence is appropriately used in accordance with our common linguistic practices to express what we each think, the answer is yes. If the question concerns the actual contents of our respective thinkings that each of us expresses with that sentence, the answer is no. Though what each of us is thinking is, according to our common linguistic practices, appropriately expressed by the same sentence, the thought expressed by that sentence certainly is differently manifested in her than it is in me. Our respective understandings of what we are each thinking is quite different. The grounds for the disparity in answers to whether two such agents have the same thought is the fact that the same sentence may be appropriately used to express what they are thinking even when what they are actually thinking is rather different. For this reason, it proves to be imperative that we distinguish thought-tokens from thoughts. What a thinker is actually thinking on a given occasion constitutes both her thought-token and its content. For purposes here, a thought-token is to

172  Nicholas Georgalis be understood as what is manifested to the thinker herself from her firstperson perspective, what she is actually thinking on a given occasion. I hasten to add that in saying this I am not claiming that what the thinker is actually thinking is completely and transparently evident to her from her first-person perspective, but only that something in her is manifested with sufficient clarity that it may be adequately expressed by the agent’s utterance of or assent to some particular sentence with standard meaning. Such manifestations are often imprecise, and they often lack the determinate content of the sentences that the agent may appropriately use to express them. For example, what exactly is the content of my thought-token expressed by the sentence “Up quarks have color”? Stating that the content of my actual thought-token is expressed by the content of the sentence “Up quarks have color” will not do, despite the fact that that sentence is appropriately used to express my thought-token. While that sentence may accurately express the content of an expert physicist’s thought-token on some occasion, it cannot do the same for me, as the actual contents of the physicist’s thinking and mine on a similar occasion (our respective thought-tokens) are significantly different in content. Still, I can appropriately use that sentence to express my thought-token’s content. I am assuming that whatever the content of my thought-token is, qua my thinking it, it is at least partially determined by my understanding. As my understanding of quarks increases, the content of my various thought-tokens regarding them will change and become richer and more accurately expressible by the standard meanings of the words “Up quarks have color”;5 thus, too, the actual content of what I  think, my thought-token’s content, aligns more closely with that of the physicist as my understanding of up quarks progresses. Still, as things currently stand, the rather determinate content of the sentence is not faithfully mirrored in my thought-token’s actual content; nevertheless, it is not incorrect or misleading of me, in accordance with our linguistic practices, to use that sentence to express my vague and inchoate thought-token. To see this more clearly, let us modify the quark case. It is easy to imagine a series of thinkers who have the thought expressible by “Up quarks have color” but each of whom has varying degrees of understanding of quarks, all short of that of the most advanced elementary particle physicist. What each of our agents is actually thinking on a given occasion constitutes their respective thought-tokens. This array of agents suggests the plausibility of holding that there is a sense in which the actual contents of their respective thought-tokens differ, even while it is in accord with our linguistic practices to express each of their thought-tokens with one and the same sentence. They each would assent to the sentence “Up quarks have color” and it would be in accord with our ordinary linguistic practices to use that sentence to characterize each of their thoughttokens or, as we would commonly say, their “thought.” The sentence’s

Thinking Differently About Thought 173 determinate content, as given by its standard meaning, does not exactly match the range of what would be the actual contents of the thoughttokens of the various thinkers, since these differ in their contents. The determinate content of the sentence would come closest to accurately expressing the content of the expert physicist’s thought-token.6 Given the variation in thought-token content described, it would be a mistake to suppose that all the various thought-tokens considered were instances of a single content type—notwithstanding the fact that all the different thought-tokens are appropriately expressed by one and the same sentence and the sentence itself signifies a single determinate content type. Thought-tokens are clearly episodic. Occurrent thinking involves something going on in us at a time. When at t1 I  think, for example, that quantum mechanics is complicated, what goes on in me is different from what goes on in me when at t2 I think that up quarks have color or when at t3 I  think that some apples are red. The expression “what goes on in me” is vague. Nevertheless, in cases such as those cited, there is no problem in identifying and distinguishing what we think on different occasions from a first-person perspective; e.g., at t1—t3 I express what I think in each of these cases by asserting the sentences, “Quantum mechanics is complicated,” “Up quarks have color” and “Some apples are red,” respectively. In general, we characterize a centrally relevant part of “what goes on in us” by the use of such sentences. Whatever that centrally relevant part of what goes on in us that occasions the use of a particular sentence to express what it is that we are thinking, it is this which I designate with the expression “thought-token.” So, thoughttokens may be partially identified by sentences that an agent would use (or uses) to appropriately express her thought-token on some occasion even though the exact contents of the two may differ. The variation in the contents of different thought-tokens that are expressed by the same sentence is widespread. It is not restricted to cases that involve sophisticated or arcane terminology such as “up quark.” The same variation in thought-tokens can be manifested even when an ordinary term, e.g., “elephant,” appears in a sentence expressing a thoughttoken. For, in the first place, for many of the words we commonly and correctly use, many of us would be hard put to provide an exact dictionary definition, the word’s standard meaning. This is sufficient reason to deny that the actual contents of our thought-tokens always match the standard meanings of the words used to express our thought-tokens. Second, even when one has mastery of the standard meanings of the words used to express one’s thought-token, this would not prevent some collateral information or misinformation regarding whatever is defined from intruding into the actual thinking of the thinker on any given occasion and, hence, partly constituting her thought-token’s content. Based on the above discussion, I  maintain that although different agents’ thought-tokens may all be expressed by one and the same

174  Nicholas Georgalis sentence, the differences in their understanding from one another are manifested in differences in the content of their respective thought-tokens and from the standard meaning of the sentence. Thus, we may conclude that whatever the components of the different thinkers’ thought-tokens are, they do not map one-to-one to the components of a sentence that appropriately expresses the various thought-tokens. There is a many–one relation between thought-tokens and the sentences that express them. It turns out that this typically overlooked many–one relation is of the utmost importance—it is a key to the solution of a number of problems in the philosophy of mind and language. (Later I will indicate the some of these problems and suggest how the solution would go in one case.) Thought-tokens are one thing; thoughts are another. While I  have stressed that the former is subjective and episodic, thoughts are neither. It is correct to say of someone that she has the thought, for example, that 2 + 2 = 4, that some apples are red or that quantum mechanics is complicated, whether or not she is entertaining the thought at the moment. Thoughts are attributable to agents whether they are asleep or awake, contemplating them or not. So, what are these thoughts and how are they related to thought-tokens? The reason I or others would attribute to me the thought that some apples are red is that I have the disposition to utter or assent to the sentence “Some apples are red” under appropriate conditions. It is not that there is stored in me some little thought which represents the content of the aforementioned sentence, waiting to be summoned.7 I can now state what on my view a thought is and how it is related to a thought-token: A thinker’s thought is to be identified with the standard meaning of the sentence that she would be disposed to utter or assent to in order to express a thought-token she has or would have under appropriate conditions. Thoughts are attributed to agents as a shorthand way of reflecting such dispositions. Thoughts are not things that agents have, though they do have thought-tokens. Thought-tokens require bearers; thoughts do not. When thought is understood in this way, the expression “thought” is a facon de parler. To be clear, I am not holding that we should understand thought as identical to standard meaning. What is proposed is that a thought is characterized by the standard meaning of some sentence under certain specific circumstances, those just indicated. Thought is not standard meaning simpliciter. Thought is identified, however, by the standard meaning of a sentence when and only when the latter could be correctly used in accordance with our common practices of thought attribution to express a thought-token that one has or would have under appropriate conditions. Thought is not something over and above that. We are conscious of episodic thought-tokens; they are mental, in the head and they map many–one to sentences that express them. In contrast, sentences expressing thought-tokens map one-to-one to the thoughts they

Thinking Differently About Thought 175 express, by definition. While thought-tokens are mental, neither thoughts nor standard meanings are mental entities; they are not in the head. Nor are they somewhere else; they are not things. Since the standard meanings of sentences are objective, so too are thoughts when understood as just proposed. Though agents do not literally have thoughts, we may figuratively speak as if they do and of two agents “having” the same thought, even while their thought-tokens may differ substantially, as the comparison between the physicist and me showed. I maintain that this theory based on the concepts of thought-token, many–one mapping and standard meaning is sufficient to account for thought and its attribution to oneself or others. The fruitfulness of the theory is further demonstrated by its successful application to a number of important problems in the philosophy of mind and language.8

3.  The Role of Understanding Differences in understanding on the part of different thinkers have played a significant role in support of my crucial claim that there is variation in their respective thought-tokens, even when these may be expressed by the same sentence. Thus, some further explanation of the role that understanding plays in all of this is in order. While thought-tokens are subjective, the assessment of whether an agent understands her thoughttoken or thought and the degree to which she does so is an objective matter. What, then, is it for the thinker to understand her thought-token or thought? This is a large and complicated question, and I do not presume to offer a full answer to it. I think I can say enough about it, however, to support my earlier claims about differences in understanding among various thinkers is sufficient reason to hold that there are differences in the content of their respective thought-tokens (even while these various tokens are appropriately expressed by the same sentence). Additionally, the identity I  maintain that holds between thought and standard meaning—­under certain conditions—forces a connection between understanding a thought and understanding a sentence. One important measure of understanding a thought germane to my claims is the ability to explicate it. Explicating a thought depends on knowing various relations between the sentences that appropriately express thought-tokens and other sentences of the language. How closely the actual content of a thinker’s thought-token aligns with the sentence content used to express her thought is directly proportional to her knowledge of these relations and, hence, to her understanding. The disparity may be great or little. When the disparity is great, the content of one’s thought-token, what one is actually thinking on a given occasion, only dimly reflects the determinate content of the sentence, its standard meaning, which is appropriately used to express her thought-token. Under these conditions, the thinker would have little understanding of

176  Nicholas Georgalis her thought. For example, if I were to be asked to explicate my thought that up quarks have color, I would have very little to say, whereas the expert physicist could undoubtedly say much more, more than many of us could comprehend. Among students of physics the ability to explicate this thought would vary according to their respective levels of education and, correspondingly, so too would their degrees of understanding. When I speak of understanding a thought or of differences in understanding it, this is the sense intended. Such differences in understanding of a thought by agents imply differences in the contents of their respective thought-tokens, even though these different thought-tokens are appropriately expressed by the same sentence.

4. Thought-Tokens, Sentences, Thought and Isomorphism The differences in the thought-tokens of various individuals reflect the subjective character of thinking, as opposed to the objective character of sentences that express thought-tokens. Since this particular subjective character of thought-tokens is typically overlooked, the many–one relations which obtain between thought-tokens and the sentences that express them are also not remarked on, let alone noticed. The neglect of the subjectivity of thought-tokens is abetted by the fact, illustrated above, that the same sentence with its determinate standard meaning may be (rightly) used to express what the various thinkers are thinking, despite the fact that there may be wide variation in the actual contents of what they are thinking. When these factors are overlooked, the common sentence used to express the various thought-tokens washes out the individual variation; consequently, attention is focused on the sentence ascribed to the thinker rather than what she is actually thinking. The content of the sentence used to express the thought-token serves as a proxy for what the thinker is actually thinking. As a result, we speak of the thinker’s “thought” as mapping nicely (one-to-one) to the sentence that expresses the thinker’s thought-token. This practice is harmless so long as thought is treated abstractly. When it is, my theory accommodates the point, for given the concept of thought presented above there is no problem with saying, for example, that the expert physicist and I have the same thought. Indeed, it is a consequence of my theory and amounts to no more than saying that on some occasions the same sentence may appropriately be used to express our respective thought-tokens, even though the latter differ in content. So, I agree that no problems need ensue by neglect of thought-tokens when one considers thought abstractly for, by definition, there is a one-to-one mapping between the components of a thought and the components of a sentence

Thinking Differently About Thought 177 that expresses the thought. We may express this feature as the Isomorphism Assumption: (IA) The component contents of a thought map one-to-one to the components of a sentence that expresses it. Among the things that I have been at pains to establish, however, is that there is no such isomorphism between thought-tokens and sentences that express them. Thought-tokens must not be neglected, for when we move away from an abstract consideration of thought and turn our attention to a number of important problems that arise in the philosophy of mind and the philosophy of language, it turns out that just what the thinker is actually thinking on a given occasion, her thought-token, is not only the crux of the problem, but its accommodation is the key to its solution. What a thinker is actually thinking—her thought-token—is at the core of problems such as Kripke’s puzzle about belief, substitution in attitude ascriptions and Donnellan’s so-called referential/attributive distinction, to name a few obvious instances.9 When problems turn on what an agent is actually thinking, neglecting the thinker’s thought-token is utterly inexplicable. How could we be discussing what a thinker is actually thinking if we ignore what actual content is manifested to her? We should not be misdirected to the thinker’s thought and ignore her thought-token. Two notable examples of philosophers who embrace Isomorphism Assumption (IA) but entirely neglect thought-tokens and the many–one relation to the sentences that express them are Frege and Burge. Burge illustrates this neglect in the following passages: “Just as whole thatclauses provide the content of a person’s attitude, semantically relevant components of that-clauses will be taken to indicate notions that enter into the attitude (or the attitude’s content)” (Burge, 1973: 75). And on the same page: “Semantically relevant components of embedded thatclauses in belief ascriptions indicate notions that are elements of mental contents.” He says nothing in this long paper by way of explanation or identification of an individual’s “notions that are elements of mental contents” other than what is in these passages. He simply assumes (1) that there is an isomorphism between thought and sentences used to express or ascribe the thought ([IA] applies to thought), but more finely, he also assumes (2) that semantically relevant parts of the sentence and the thinker’s individual notions (the “elements of [her] mental contents”) are similarly isomorphic. To briefly frame a difficulty for (2), I revert to the quark case: It may be correctly said that Georgalis believes that up quarks have color and Stephen Hawking believes that up quarks have color. On my view, we have the same thought and so there is no need to dispute the claim that semantically relevant components of each of the embedded that-clauses

178  Nicholas Georgalis indicate the same contents of our shared thought. But is it remotely plausible to hold, as Burge would, given the quotation above, that our respective mental contents are the same? One might argue that it is, since it is certainly correct to say that the subject of what we are each thinking is indicated with the expression “up quarks.” True enough but note that for Burge it is not the orthographic or phonetic component that is at issue but the “semantically relevant components of that-clauses” (ibid., italics added) that is relevant. The semantics of “up quarks” in my head, I can assure you, bears almost no resemblance to the semantics of that term as it would be manifested in any physicist’s head. My use of the term is strictly parasitic and has very little semantic content in my head.10 There is no reason to think that our respective individual notions have the same content and ample reason to dispute it. In short, though Burge discusses what he calls “individual notions,” he completely neglects the individual. He neglects thought-tokens and the variation they are subject to from one individual to another, as well as how their specific contents may differ from the contents of the sentences that express them. On Burge’s construal of “individual notion,” there is nothing specific to the individual, nothing individual about the individual notion, hence nothing that reflects the subjective character of an individual’s notions, but if individual notions are elements of the agent’s mental contents, any differences in these among agent’s must be taken into account. If I  am right, this is wrong, and wrong in several important ways. In my “Rethinking Burge’s Thought Experiment,” I argued that Burge’s “Individuation and the Mental” neither establishes anti-individualism nor refutes individualism. I  make no attempt here to rehearse my or his arguments.11 My main point here regarding Burge is to show that he accepts IA but neglects thought-tokens and the many–one mappings between them and the sentences expressing them. I take the quotations to have shown this, while the subsequent arguments are to suggest why such neglect is unsatisfactory.

5. A Critical Comparison of the Proposed Theory of Thought With Frege’s Theory Another influential source of this neglect may be traced back to Frege. To simplify here, Frege views thought as an abstract entity that may be expressed by a sentence; components of the sentence express Fregean senses, which, in turn, are components of the thought. Thought, for Frege, is clearly independent of anything going on in a thinker. He allows only that a thinker may “grasp” a thought and he denies that a thinker has thoughts. Thoughts are senses that have the additional feature of being either true or false. (Frege, 1997: 327–328)

Thinking Differently About Thought 179 Frege does distinguish what he calls “Ideas,” which are very much like my concept of thought-tokens, from thoughts, but he is adamant in ruling them out of his finished theory. There is no place for ideas or thought-tokens and many–one mappings between thought-tokens and the sentences that express them in Frege’s final theory. Frege’s exclusion of thought-tokens or ideas in his finished theory is of particular interest, since he was acutely aware of the subjective variation in the understanding of terms among agents, particularly with regard to names. Though Frege discusses this variation at great length and the problems it engenders, he does not in the end deal with it; instead, he proposes its elimination by an unsatisfactory ad hoc strategy. Let me explain these several claims. One place where Frege confronted the problem of subjective variation is in his discussion of the Dr. Lauben case. He worried whether two individuals could understand one another when the thought expressed by the speaker on uttering, say, “Dr. Lauben was wounded,” may be different than the thought grasped by the hearer. This would happen, Frege held, if the agents associated different senses with a name. Since these senses would be constituents of their respective thoughts, the thought expressed by the speaker would be different than the thought understood by the hearer. Indeed, Frege went so far as to suggest that in such a situation they might be speaking different languages, despite appearances to the contrary. He also worried that the one of their thoughts could be true and the other false. (Frege, 1997: 333) Unfortunately, after discussing problems that arise from the subjective variation in the understanding on the part of various agents, he ultimately abandons any attempt to face them; instead, he simply proposes that we should stipulate unique senses to proper names. Such a proposal is not a solution to the problems posed—it is an evasion. Moreover, as I have elsewhere argued in detail, this fails to even avoid the problems since the same sort of subjective variation that occasioned the problem in the first place arises once again regarding any supposed stipulated sense.12 It was of paramount importance to Frege to eliminate subjective variation in how agents understood terms, for he was convinced that this would lead to the collapse of the objectivity of science and mathematics, indeed, objectivity in general. I will not go into detail here beyond explaining how his worries are a consequence of his theory of thought and need not be a consequence of a theory of thought that accommodates subjective variation, as mine does.13 Toward that end I explain how this comes out in his efforts to distinguish the inner (subjective) world from the outer (objective) world and correspondingly in a distinction he drew between ideas and thoughts. The central features of ideas are that sentient beings have ideas, ideas need an owner and each idea has only one owner. It is of the utmost importance for Frege that thought have none of these properties, for he

180  Nicholas Georgalis was convinced that if it did, this would be the demise of objectivity. Here, as one example, is what he says in defense of thought being objective and why it must be distinct from subjective ideas: If other people can assent to the thought I express in the Pythagorean Theorem just as I do, then it does not belong to the content of my consciousness. I am not its owner; yet I can, nevertheless, acknowledge it as true. However, if what is taken to be the content of the Pythagorean Theorem by me and by somebody else is not the same thought at all [which would be true, he argues, if thought were an idea, since an idea can have only one owner], we should not really say “the Pythagorean Theorem,” but “my Pythagorean Theorem, “his Pythagorean Theorem’, and these would be different, for the sense necessarily goes with the sentence. (Frege, 1997: 336) He further concludes that one of our respective “theorems” could be true and one false. “Truth would then be confined to this content [of an individual consciousness and of hers alone] and it would remain doubtful whether anything at all similar occurred in the consciousness of others” (Frege, 1997: 336). (Frege makes a similar point in his discussion of the Lauben case, as noted earlier) (Frege, 1997: 333). It is easy to see why such consequences would be catastrophic for objective knowledge. It is also easy to see that what I have called “thoughttokens” are what Frege would call Ideas, for conscious agents have them, they require an owner and each thought-token has but one owner; these are the key features of Fregean ideas, cited above. But while my theory of thought is committed to thought-tokens having these properties, the disastrous consequences Frege envisions do not ensue, since thought-tokens are mapped many–one to the sentences that appropriately express them and thought-tokens are distinguished from thought. While the content of an agent’s thought-token does “belong to the content of [her] consciousness,” it does not follow on my theory that the content of the sentence she uses to express her thought-token, nor the thought (my sense thereof) that would be ascribed to her or self-ascribed, similarly resides in her consciousness: Objectivity of thought is maintained. It is Frege’s exclusion of thought-tokens (or Ideas) and, particularly, their many–one relations to sentences that express an agent’s thought that drives the disastrous consequences he identified. Let me indicate in a bit more detail how my theory avoids the loss of objectivity, despite the fact that thought-tokens are constituents of an agent’s consciousness. The subjective variation that Frege was so rightly worried about but in the end simply evaded is precisely what drives my introduction of the concept of thought-token and the many–one relations of them to sentences. The thought-tokens of the two individuals

Thinking Differently About Thought 181 in the Lauben case differ in content, but in the described circumstance it would be in accordance with our linguistic practices for each to appropriately express their respective thought-tokens with the same sentence, “Dr. Lauben was wounded.” That they think of Lauben differently accounts for the differences in their respective thought-tokens but not in their thoughts (my sense thereof). On my theory, while their thoughttokens differ in content, they appropriately map many–one to the same sentence which expresses their common thought. All this is in accordance with the facts of subjective variation, our common linguistic practices of thought ascription and the objectivity of thought. The situation in the Lauben case is analogous to the case discussed earlier where it is assumed that I  and the expert physicist think of up quarks quite differently yet we appropriately express our different thought-tokens with the same sentence. On my account, this allows us to conclude that we have the same thought. Yes, we understand the thought differently, but these differences in no way require us to assert “My up quarks have color” and, say, “Hawking’s up quarks have color,” while eliminating our ability to truly say “Up quarks have color,” as Frege similarly suggested in the case of the Pythagorean Theorem. The claim here that we have the same thought is in accord with the facts of the case and our common linguistic practices of thought ascription, as it is in the Lauben case. In fact, Frege’s conclusion that the speaker and hearer in the Lauben case have different thoughts is not in accord with our practices of thought ascription, though it is in accord with his theory. So much the worse for his theory—at least as applied to natural language and thought ascriptions.14

6.  Brief Comparison of Sense and Standard Meaning Where Frege relies heavily on sense, I rely on standard meaning. Is there a difference? Recall that for Frege thoughts are objective; they are neither in our consciousness nor in the physical world, nevertheless, we “grasp” them. How do we do this? Frege’s answer: The “thought . . . gets clothed in the perceptible garb of a sentence and thereby we are able to grasp it” (Frege, 1997: 328). It is important to note that Frege insists that when he talks of sentences, he is not thinking of them as formal items, but as meaningful ones; a sentence for Frege is not simply its orthographic or phonetic features.15 Moreover, thoughts, for Frege, are a proper subclass of senses; they are senses that are either true or false. The constituents of thought are also senses and, hence too, we grasp senses via their perceptible garb, namely, meaningful linguistic items. But what are senses and how does the “perceptible garb” enable us to “grasp” them? If we do grasp the alleged sense expressed by a linguistic item, we do so by knowing its standard meaning. The “perceptible garb”

182  Nicholas Georgalis is not the mere shapes of letters or sounds of utterances. For example, the standard meanings of the locution “the evening star” is clear to those in our language community and it is this standard meaning that allows us to “read off” the supposed sense, as noted earlier. Without our already understanding the standard meaning of some linguistic expression we would not have a clue as to which Fregean sense is supposedly expressed by it. Importantly, even with this, it remains unclear just what these alleged senses are beyond standard meanings. In any case, to the extent that Frege’s answer as to how we grasp senses works, it is in virtue of our prior understanding of the standard meaning of an expression. The “perceptible garb” of a linguistic item that allegedly enables us to grasp sense is its standard meaning. Of course, when it comes to proper names an appeal to standard meaning does not appear to be as apt as it is when descriptions are at issue, since our ordinary linguistic practices do not assign meanings to names, standard or otherwise. But since Frege has already stipulated that a unique sense is to be associated with a proper name to circumvent the subjective variation in the understanding of terms, he can deploy this feature to overcome the fact that our ordinary linguistic practices do not assign meanings to names. So, does Frege’s theory have an advantage over mine on this point? No, let us see why. Though a theory of proper names appears to be lurking just below the surface, we need not here attempt any adjudication between competing theories of proper names, descriptivists versus direct reference. The issue before us in this section is not which theory of proper names is correct but simply—given Frege’s theory—whether the concept of sense adds anything useful that is not already provided by standard meaning. The point I wish to make here—the superiority of talk of standard meaning to talk of senses—can be made independently of any theory of proper names, for when Frege speaks of senses associated with a name, stipulated or not, he is speaking of various properties that the referent of the name may have or be believed to have by an agent. These properties are expressed with predicates that do have standard meaning, and it is the latter that, again, would allow us to grasp the alleged sense of proper names. Therefore, the point made in an earlier paragraph regarding senses expressed by descriptions with standard meanings applies as well when the issue is alleged senses associated with proper names, at least from the vantage of Frege’s theory: Appeal to sense presupposes standard meaning both when the linguistic item is a predicate and when it is a proper name. In the end, then, it is standard meaning that does the work that Frege thinks sense does, since that is what provides the “perceptible garb” of linguistic items which, he tells us, enables us to grasp its alleged sense. One should be forgiven, then, for wondering what, if anything, sense adds to standard meaning. The conclusion is unavoidable: It is the standard

Thinking Differently About Thought 183 meaning of terms that is doing all the work for Frege; senses appear to be nothing but an idle ontological overlay. Thus, standard meaning has distinct advantages over Fregean senses, particularly since any grasp of the latter depends entirely on the former and it is the former that does the work. Moreover, any fluent speaker has a ready, if sometimes only vague or incomplete, understanding of standard meaning. It is this standard meaning that I  have relied upon, together with the fact that there is subjective variation among individuals regarding their degree of understanding of it. I think that Frege’s attempt to evade—rather than to accommodate—the subjective variation in an agent’s understanding of terms via his stipulation strategy is fundamentally the undoing of his theory of thought, sense and reference.16

7.  Concluding Remarks I am not alone in seeing Frege’s ultimate neglect of subjective variation in the understanding of terms among agents as a serious issue for his philosophical views. No less an authority on Frege than Michael Dummett sees this fact as highly problematic for Frege, as he came to think that an accurate evaluation of Frege’s philosophy waits on settling the issues that are thereby raised. I quote Dummett: Since writing . . . [Frege: Philosophy of Language], I have come to see that the relation of the meaning of a word in the language shared by a community of speakers and the understanding that an individual speaker has of it is far more problematic than I then supposed, as is also the question whether an understanding of a word or of a language is genuinely a case of knowledge. I further believe that these are prime examples of questions that need to be settled before we can accurately evaluate Frege’s philosophy of language. But they are very difficult questions, demanding extensive discussion; and they are questions about which nothing explicit is said by Frege. (Dummett, 1973: xiii) Frege thought his stipulation strategy could successfully evade the significant problems that he recognized arose from the subjective variation in the understanding of terms on the part of different agents. I have argued that this strategy is unsuccessful. But successful or not, in the end it says nothing of the relation between an individual’s subjective understanding of a term and its meaning in her language community. I concur with Dummett’s conclusion that this is a serious lacuna in Frege’s work, as it is for any theory of thought. The problem of explaining “the relation of the meaning of a word in the language shared by a community of speakers and the understanding

184  Nicholas Georgalis that an individual speaker has of it” is central to the development of my theory.17 Much of this chapter has been devoted to it. The idea of the “language shared by a community of speakers” is accommodated in my account of standard meaning. Standard meaning does all the legitimate work that Fregean sense was intended to do. My concept of thoughttoken is introduced precisely to accommodate the subjective variation in individuals’ understanding of the sentences they use to express their thought-tokens. Many–one mappings between thought-tokens and a sentence that expresses them is what is required to account for the relation between the subjective variation among individuals’ understanding of their community’s language and the latter’s standard meaning. The many-one mappings of thought-tokens to sentences is what accommodates the subjective nexus between thought and language, while leaving the objectivity of thought intact. The resulting theory is in accord with the facts of thought ascription and our common linguistic practices that govern it.18

Notes 1. This of course says virtually nothing about what meaning is or what is required for it. All I need now is that there is an established use and community understanding of our various linguistic items that allows us to speak of their standard meanings and references, and this does not typically lead to any irresolvable disputes. I hold the meaning skeptics at bay, at least for now. (See Georgalis, Mind, Language, and Subjectivity, chapters 11–12.) 2. I discuss the parasitic uses of terms in Mind, Language, and Subjectivity, chapter 9, particularly as this applies to proper names. John Searle utilized the idea of parasitic uses of proper names in his “Proper Names,” as did Saul Kripke in his Naming and Necessity. Searle’s Intentionality, 248–249, extends the idea of parasitic uses to general terms, as I did above with “electron.” He maintains that there can be parasitic uses of both natural kind and artifactual terms. The idea of reliance on experts is prominent in Putnam’s “The Meaning of ‘Meaning,’ ” where he introduces the idea of the division of linguistic labor. 3. Whenever I  say “A  sentence is appropriately used to express a thoughttoken,” this is to be understood as, appropriate in accordance with our common linguistic practices of thought ascription, whether it be to oneself or another. 4. When I  speak of sentences I  will use the terms “content” and “meaning” interchangeably. When I speak of thoughts I will avoid the term “meaning,” as “the meaning of a thought” can have a wider significance than lexical meaning or content. 5. While this may indicate a kind of “content holism” with regard to thoughttokens, this would not necessarily imply a kind of “meaning holism” with regard to language, though I am sympathetic to both. 6. I do not mean to suggest that the expert physicist represents some ideal endpoint for this kind of case, where the content of the thought-token and the content of the sentence expressing it are in perfect harmony. Even among experts there is variation in understanding, from minor to substantive disagreements about parts of theories, which may be reflected in the actual

Thinking Differently About Thought 185 content of what they think on a given occasion. Also, where there is no disagreement, there may be some collateral information contained in one of the theorist’s thought-tokens not contained in the thought-tokens of the other. 7. A simple reason to reject some “storage model” of our thoughts is that it would seem to require the most implausible thesis that we have infinite storage capacity, for we have the standing thoughts: 1 is a natural number, 2 is a natural number and so on, not to mention the myriad other more interesting thoughts that we may be said to have. 8. Both the theory itself and its application to a number of problem are extensively developed in my Mind, Language, and Subjectivity. These include Kripke’s puzzle about belief, substitution in propositional attitudes, Frege’s puzzle about identity, the referential/attributive distinction, rule-following and securing determinate meaning, among others. 9. These are among the problems to which my theory is applied in my Mind, Language, and Subjectivity. 10. Similarly, for Burge’s cases. While the isomorphic correspondence between semantically relevant components of that-clauses and elements of an agent’s mental contents that he appeals to may well hold for experts there is no such isomorphism that holds for novices or the ill-informed. A further word regarding the locution “the subject of what an agent is thinking” and its variants is in order. In my The Primacy of the Subjective and Mind, Language, and Subjectivity, I  argue that such expressions are inherently ambiguous between a subjective and objective reading. For the former reading I introduced the concept of minimal content. It is the subject or topic of an agent’s thought as she conceives it. This is distinguished from the objective content of an agent’s thought, where this concept is defined as the subject of an agent’s thought that an objective observer would attribute to the agent. In the aforementioned works, I argue that while the two contents often do not differ, sometimes they do, and when there is a difference, the problems that ensue require explicit accommodation of the difference to achieve adequate resolution. 11. In my “Rethinking Burge’s Thought Experiment,” I  made much of the divergence between an individual’s mental content and the sentence content expressing it, contrary to Burge’s implicit assumption that there is no difference (as the above quotations manifest). Moreover, this unstated and unsupported assumption of his is one of the key assumptions needed for his conclusion. My argument there was expanded and included in The Primacy of the Subjective, chapter 7, where I used the concept of minimal content to make my case, but I still lacked the ideas of the thought-token and many– one mappings. What I have said here on the relation between thought-tokens and sentences may shed additional light on my earlier arguments. (On the locution “minimal content,” see previous note.) 12. In Mind, Language, and Subjectivity, chapter 4, I develop this point in detail. This chapter is an extensive critical discussion of Frege’s theory of sense, reference and thought, with comparisons to mine. 13. For details, again, see Mind, Language, and Subjectivity, chapter 4. 14. One might object here that Frege was not so much concerned with natural language but with an “ideal language” for mathematics and science. True enough, but his theories are applied to natural language, for example, substitution in propositional attitudes. Frege himself applied his theory to the latter, and it is the applicability of his theory to natural language and thought that is my concern.

186  Nicholas Georgalis 15. Frege is explicit on this: “What is it that we call a sentence? A  series of sounds, but only when it has a sense” (italics added, Frege, 1997: 327). 16. Again, considerably more detailed argument both against Frege’s theory and in defense of mine is offered in my Mind, Language, and Subjectivity. 17. I have not and do not address the other problem Dummett identified in Frege’s philosophy: “[T]he question whether an understanding of a word or of a language is genuinely a case of knowledge,” as it is not central to my theses. 18. A version of his paper was presented at the 2nd International Conference on Philosophy of Mind (University of Minho, Braga, Portugal, 2016). I’d like to thank the organizers, Raquel Pereira, Steven S. Gouveia and Dr. Manuel Curado, for allowing an expanded time slot for my presentation and the participants for their comments, particularly Tim Crane and Torin Alter.

References Burge, Tyler (1973) “Individuation and the Mental” in Midwest Studies in Philosophy, 4, 73–123. Dummett, Michael (1973) Frege: Philosophy of Language, New York: Harper & Row. Dummett, Michael (1981) The Interpretation of Frege’s Philosophy, Cambridge: Harvard University Press. Frege, Gottlob (1997) “Thought” in The Frege Reader (ed. Michael Beaney), Oxford: Blackwell Publishers. Georgalis, Nicholas (1999) “Rethinking Burge’s Thought Experiment” in Synthese, 118, 145–164. Georgalis, Nicholas (2006) The Primacy of the Subjective: Foundations for a Unified Theory of Thought and Language, Cambridge: MIT Press. Georgalis, Nicholas (2017/2015) Mind, Language, and Subjectivity: Minimal Content and the Theory of Thought, New York: Routledge. Kripke, Saul (1972) Naming and Necessity, Cambridge: Harvard University Press. Putnam, Hilary (1975) “The Meaning of ‘Meaning’ ” in Mind, Language, and Reality, New York: Cambridge University Press, 215–272. Searle, John (1958) “Proper Names” in Mind, 67, 166–173. Searle, John (1983) Intentionality, New York: Cambridge University Press.

10 Immediate and Reflective Senses Angela Mendelovici

1. Introduction It is sometimes held that thoughts and other intentional mental states have two distinct semantic features—sense and reference—where reference is a matter of picking out or targeting specific items and sense is a matter of how subjects conceive of these items. While I  think this is more or less the right picture, I want to suggest that we need to make a further distinction between two different kinds of senses: immediate senses, which are what we are immediately aware of in having an intentional mental state, and reflective senses, which are what we understand of an intentional mental state’s (putative) referent upon reflection. This chapter has two aims: The first is to argue that thoughts and other mental states have immediate and reflective senses and that, in many cases, they are distinct. The second is to suggest that the phenomenal intentionality theory, a theory of intentionality based on phenomenal consciousness, provides the best account of immediate and reflective senses.1

2.  Two Kinds of Senses This chapter is concerned with the semantic features of mental states. Although my claims are meant to apply to various kinds of mental states, my main arguments focus on the case of thoughts, the mental states we are in when we do what we commonly call “thinking.” Thoughts include occurrent states such as judgments and occurrent desires (desires we are currently entertaining) but not perceptual experiences or standing states, such as beliefs we are not currently entertaining. Thoughts seem to “say” or represent something, target things in the world, and be assessable for truth and accuracy. For example, a thought might represent that Eleni is intelligent, target Eleni herself, and be true just in case Eleni is in fact intelligent. In discussing semantic features, it is useful to assume a distinction between semantic features and the items that have them, which are the vehicles of representation, or representations, for short. As I  am using

188  Angela Mendelovici the terms, mental states such as thoughts, beliefs, desires, and many perceptual states are examples of representations. It is also useful to assume that most thoughts are complex in that they are composed of other representations, which I will call concepts.2 For example, a thought that cats have tails might be partly composed of the concept cat, which might also be involved in other thoughts, such as the thought that Mitten is a cat. Roughly following Frege (1948), thoughts and other representations are sometimes taken to have two kinds of semantic features: sense and reference. On one intuitive way of characterizing this distinction, a representation’s referent is what it picks out, singles out, or refers to, while its sense is the way in which this (putative) referent is conceived. While Frege argued that senses are publicly available abstract entities that determine reference, on my use of the term “sense,” I  will not assume that senses have any of these features. A representation’s sense is merely how its subject conceives of its (putative) referent. One gloss on Frege’s arguments for the existence of senses is that we need them in order to account for semantic differences that are not differences in reference. For example, the thought that Hesperus is bright and the thought that Phosphorus is bright involve all the same referents but differ in meaning or “cognitive significance.” We can account for this difference by saying that the two thoughts contain two distinct concepts, Hesperus and Phosphorus, with different senses. The sense of Hesperus is , while the sense of Phosphorus is . The two concepts pick out the same thing but involve two different ways of conceiving of it. My contention is that, while this is more or less the right picture, we need to make a further distinction between two different kinds of sense corresponding to two different ways of understanding what it is to conceive of a (putative) referent in a certain way. One way of understanding the notion of conceiving of a (putative) referent is as capturing what is in some sense “before one’s mind” when we use a concept or have a thought. This is what is running through our minds, what is immediately available to us or what we are aware of in using a representation. Call this a representation’s immediate sense. To say that is the immediate sense of Hesperus is to say that when we use the concept Hesperus, the content is running through our minds. Another way of understanding the notion of conceiving of a referent is as capturing our reflective intuitive understanding of a referent, an understanding we might retrieve upon brief or prolonged reflection. Call this a representation’s reflective sense. To say that is the reflective sense of Hesperus is to say that when we use the concept Hesperus, the content is available upon sufficient reflection of the right sort. For instance, if we were to think , upon sufficient

Immediate and Reflective Senses 189 reflection we should be able to further elucidate our thought as , accepting this as what we meant, intended, or targeted by the first thought.

3.  A Distinction with a Difference So far I  have merely introduced the notions of immediate and reflective senses. In this section, I argue that there are immediate and reflective senses and that they are distinct. I do so by considering a variety of examples, focusing mainly on concepts and arguing that they involve two distinct things respectively answering to the notions of immediate and reflective senses.3 Number Concepts. My first example is adapted from David Bourget’s (2017) discussion of number concepts. When we think about numbers, especially large numbers, we are not immediately aware of the numbers themselves but rather of the numerals that stand for them, perhaps in addition to a rough ballpark sense of their magnitudes. Bourget offers a couple of reasons to think this: First, as far as what is in our immediate awareness is concerned, numerals like “8,346,182” could stand for entirely different numbers than they in fact stand for. Second, the same arithmetical equation can seem more or less obvious depending on the notation used to symbolize it. For example, 70 × 100 = 7000 is obvious when presented in decimal notation but less obvious when presented in binary notation (1000110 × 1100100 = 1101101011000). All this suggests that the immediate senses of most of our number concepts are not numbers but something more like numerals. However, our reflective understanding of numbers goes beyond what’s in our immediate awareness. There are different possible views of exactly what this understanding consists in. Perhaps we understand numbers as multiplicities, abstract entities with various features, whatever entities satisfy certain axioms, something to do with counting or the succession relation, or even sui generis entities. If all this is right, then our number concepts have distinct immediate and reflective senses. Their immediate senses are something like numerals, while their reflective senses encompass a richer understanding, such as that of a multiplicity or an abstract object. Stipulated Concepts. Suppose that, while developing a philosophical argument, you stipulate . You thereby acquire a new concept, Ent, which you can use in your further deliberations. When you have a thought involving Ent,

190  Angela Mendelovici is not running through your head in its entirety. Instead, what is running through your head—call it —might consist in purely verbal or perceptual imagery of the word “Ent,” a schematic, gist-like, or partial understanding of , or perhaps a new simple or sui generis content. Different people or the same people on different occasions might enjoy different immediate senses associated with the concept Ent. ’s lack of detail and complexity is what makes your stipulation useful to you, lessening your cognitive load by allowing you to use a simpler content to stand for one that is more complex. In the extreme case, these kinds of stipulations allow you to engage in purely formal reasoning about complex, substantive issues, just as we can reason about large numbers by manipulating numerals that stand for them. At the same time, your reflective understanding of the target of Ent is the more complex . This is what you stipulated is to stand for, and it can be easily retrieved by you at any point in your deliberations. If all this is right, then Ent’s immediate sense is , whatever that ends up being, while its reflective sense is . The same holds in other cases of stipulated concepts. Concepts with Definitions. Similar points hold for non-stipulated concepts that have definitions, such as the concepts grandmother, odd number, and supervenience. These concepts’ definitions arguably capture our reflective understanding of their targets, but they do not run through our heads every time we use the concepts. For example, when you think , you are not immediately aware of any definition of an odd number—for instance, you are not immediate aware of the property of being divisible or the number two. Likewise, when you think , you are not immediately aware of possible worlds or other elements in supervenience’s definition. Although a defined concept’s definition is not usually running through your mind when you use it, there is something running through your mind, so defined concepts also have immediate senses. In the case of supervenience, for example, this immediate sense might be or involve some verbal or perceptual imagery (e.g., the word “supervenience” or mental imagery of one item above another), a schematic, gist-like, or partial grasp of the definition of supervenience (e.g., an abstract idea of some kind of logical dependence), or simple or sui generis contents, not understandable in terms of other contents (e.g., as far as our immediate understanding of supervenience is concerned, supervenience might be a

Immediate and Reflective Senses 191 simple or sui generis relation, its own kind of thing). As in the case of concepts with stipulated definitions, different people or the same people on different occasions might enjoy different immediate senses associated with the concept supervenience. If all this is right, then concepts with definitions have immediate and reflective senses and they are usually distinct. Prototype, Exemplar, and Theory Concepts. Many everyday concepts like table and chair arguably do not have definitions. One reason to think this is that it is notoriously difficult to produce satisfactory definitions for such concepts. Another reason is that there is a host of empirical findings suggesting that most concepts do not have a definitional structure.4 Alternative views of such concepts take them to represent or involve the representation of prototypes, which are (perhaps weighted) lists of features that members of a category tend to have or that are predictive of belonging to a category (Rosch, 1975; Rosch and Mervis, 1975), sets of exemplars, which are individual items or subcategories that fall under a category (Medin and Schaffer, 1978); Brooks (1978), or theories about a category (Murphy and Medin, 1985; Carey, 1985; Gopnik and Wellman, 1994). If the prototype, exemplar or theory theories are right about certain concepts, then our reflective understanding of the categories they pick out likely consists in the information stored in our prototypes, exemplars, or theories of the category, respectively. For example, how we conceive of the target of our concept bird upon reflection might be a matter of our bird prototype, our stored exemplars of birds or our theoretical information about birds, depending on the theory. But when we use the concept bird, we are not immediately aware of a bird prototype, multiple exemplars of birds, or theories of birds. Still, there is something running through our minds when we think about birds. This might be a matter of verbal imagery corresponding to the word “bird”; mental imagery of a bird; a schematic, gist-like or partial understanding of our prototypes, exemplars, or the information in our theories; or a simple or sui generis content.5 If all this is right, then concepts with an exemplar, prototype, or theory structure have distinct immediate and reflective senses. What we are immediately aware of in using such concepts differs from our reflective understanding of their targets. Perceptual Experiences. In many perceptual experiences, immediate and reflective senses do not come apart. Many of the targets of perceptual experience are presented to us just as we reflectively understand them. Suppose you visually experience something as having a specific shape for which you have no corresponding concept. Your entire grasp of the

192  Angela Mendelovici nature of the shape is in some sense immediately before your mind’s eye. What you immediately understand of the shape is the same as what you reflectively understand of it. So, the experience’s immediate sense is the same as its reflective sense.6 But there are two kinds of cases in which the immediate and reflective senses of a perceptual experience can come apart. The first involves the representation of high-level properties, like the properties of being a cat or a computer (see Siegel, 2005). Suppose you see a computer as a computer. And suppose further that your reflective understanding of computers includes a definition, prototype, set of exemplars, or theory of computers. When you see the computer as a computer, you are not immediately aware of such definitions, prototypes, exemplars, or theories. But, at the same time, you are immediately aware of more than just the lowlevel features that your experience represents. In seeing the computer as a computer, you might be aware of verbal imagery corresponding to the word “computer,” a generic or schematic image of a computer, a gist-like or schematic understanding of your reflective conception of computers, or a simple or sui generis content. Something similar holds in other cases of high-level property representation in perception. If this is right, then the immediate and reflective senses of perceptual representations of highlevel properties differ. Immediate and reflective senses also come apart when we have a conceptual understanding of low-level features represented in perception that goes beyond our perceptual grasp of those features. For example, suppose you perceptually experience a square. Your reflective conceptual grasp of squareness might involve a definition of a square as a closed plane figure with four sides of equal length that meet at right angles. Your experience, however, does not involve an immediate awareness of this definition. What you are immediately aware of is a particular experienced shape, perhaps together with verbal or perceptual imagery, or some additional gist-like appreciation of squareness. In some cases, an experience’s immediate sense can be in tension with its reflective sense. Suppose that your color experiences represent primitive colors, while your color concepts represent dispositions to cause color experiences. Suppose now that you experience an apple as red. You are immediately perceptually aware of primitive redness. However, what you understand upon reflection is a disposition to cause experiences of redness. Your experience’s immediate sense is , but its reflective sense is .7 In summary, the immediate senses of many perceptual experiences are the same as their reflective senses. In such cases, we are immediately aware of exactly what constitutes our reflective understanding of a representation’s referent. In other cases, however, immediate senses and reflective senses come apart. I’ve considered various concepts and other representations

Immediate and Reflective Senses 193 and argued that they have immediate and reflective senses. We’ve seen that in the case of concepts and perceptual representations that involve something like a “conceptual” element, the two kinds of sense come apart. Since thoughts are composed of concepts and assuming that the senses of thoughts involve or are a function of the senses of their constituent concepts, it is reasonable to conclude that thoughts also have immediate and reflective senses and that these senses often come apart.8

4. Against Tracking and Functional Role Theories of Immediate and Reflective Senses Immediate and reflective senses are semantic features of mental states, so we might turn to contemporary theories of intentionality for an account of them. In this section, I will very briefly consider two prominent contemporary theories of intentionality—tracking and functional role theories—and suggest that their main ingredients cannot account for either kind of sense. In the next section, I will consider an alternative theory— the phenomenal intentionality theory—and argue that its ingredients can. According to tracking theories, what a representation represents is or is determined by what it tracks, where tracking is detecting, carrying information about, or otherwise appropriately corresponding to states of affairs or other items in the environment. On many tracking theories, the relevant tracking relation is a causal relation (Stampe, 1977; Dretske, 1995, Fodor, 1987; Neander, 2017), though on others it is not (Millikan, 1984). One might suggest that we can account for immediate and reflective senses in terms of tracking relations. The problem with such an account is that, in many cases, what we track comes apart from how we understand something both immediately and reflectively. For example, a child’s color concepts might track various surface reflectance properties or dispositions to affect us in certain ways, but when she experiences colors, she is only immediately aware of a primitive, sui generis property of objects. So, what her perceptual color representations track doesn’t match their immediate senses. Her reflective understanding of colors might also be as of primitive, sui generis properties, so what her color concepts track doesn’t match her reflective understanding of colors either. If this is right, the tracking theory cannot account for either kind of sense.9 Let us now consider functional role theories, which state that a mental state’s content is or is determined by its inferential or other kind of functional role (see Harman, 1987; Block, 1986). For example, the representation bachelor might represent in virtue of being involved in the following inferences: (B1)

From judging o is a bachelor, one is likely to judge o is a man.

194  Angela Mendelovici (B2) From judging o is a bachelor, one is likely to judge o is unmarried. (B3) From judging o is a man and o is unmarried, one is likely to judge o is a bachelor. Unfortunately, functional roles are ill-suited to accounting for immediate senses. For example, the concept Ent might be taken to be involved in the following key inferences: (E1) From judging Ent, one is likely to judge physicalism is committed to the a priori entailment of the phenomenal facts by the physical facts. (E2) From judging physicalism is committed to the a priori entailment of the phenomenal facts by the physical facts, one is likely to judge Ent. But this pattern of inferences most closely captures Ent’s reflective sense, not its immediate sense. The problem is that Ent behaves as if it stands for its stipulated definition—indeed, that is why it can be so useful.10 A functional role approach is more promising as an account of reflective senses, as illustrated by the case of Ent. However, such an account faces a challenge, which is a special case of what is sometimes called the “problem of collateral information” (Block, 1986): Not all of a representation’s inferential or other functional roles determine its reflective sense. For example, in addition to (B1)–(B3), bachelor might play the following role: (B4) From judging o is a bachelor, one is likely to judge o lives alone. But (B4) does not play a role in determining bachelor’s reflective sense, which, presumably, is something like , not . The functional role approach needs a way of distinguishing the functional roles that play a role in determining reflective senses from those that merely capture collateral information we have about a representation’s target. I will later suggest that an account of reflective senses that invokes both functional roles and phenomenal consciousness can meet this challenge. For now, however, I want to suggest that an account of reflective senses in terms of functional roles alone is incorrect because it neglects the relationship between immediate and reflective senses. Reflective senses are available upon reflection. If we were to reflect sufficiently, they would be immediately available to us, just like immediate senses. This suggests that a representation’s having a reflective sense is a matter of our being disposed to bring to mind a particular immediate sense. The immediate sense need not be the immediate sense of the

Immediate and Reflective Senses 195 representation in question but might instead be the immediate sense of another representation. For example, Ent’s having the reflective sense is a matter of our being disposed to retrieve a distinct complex representation with that content as its immediate sense. All this accords with the phenomenology of retrieval of reflective senses. Reflective senses, when they are retrieved, seem like the same kind of thing as immediate senses. Like immediate senses, retrieved reflective senses are immediately available to us. On this picture, one representation’s reflective sense is another representation’s immediate sense.11 If this picture of reflective senses is correct, then an account of reflective senses in terms of functional roles alone is incorrect because it does not make reference to immediate senses, aiming instead to account for reflective senses directly.

5. Accounting for Immediate and Reflective Senses: The Phenomenal Intentionality Theory This section outlines an alternative approach to intentionality, the phenomenal intentionality theory,12 and argues that it has the resources needed to neatly account for both immediate and reflective senses. According to the phenomenal intentionality theory (PIT), there is a kind of intentionality, phenomenal intentionality, that is nothing over and above phenomenal consciousness, the “what it’s like” (Nagel, 1974) aspect of mental life, and all other kinds of intentionality are derived from it. Phenomenal intentionality might be identical to, constituted by, grounded in, or in some other way nothing over and above phenomenal consciousness, and derived intentionality might derive from it in various ways.13 For example, a perceptual intentional state with the content might be identical to a reddish squarish phenomenal state, and a concept of redness might derive its content from its relation to such possible or actual phenomenal intentional states. We can call contents that are phenomenally represented phenomenal contents and contents that are derivatively represented derived contents. In the previous example, the perceptual state has a phenomenal content and the concept has a derived content. On the face of it, PIT appears to have the right structure to accommodate both kinds of senses. Let us say that the PIT approach to immediate and reflective senses is the claim that immediate senses are phenomenal contents and reflective senses are derived contents. It is plausible that immediate senses are phenomenally represented. Immediate senses are immediately available to us upon having a thought or other representational state in the same way that phenomenal characters are typically available to us upon having a phenomenal experience. This accords with the view that immediate senses are phenomenal

196  Angela Mendelovici contents. Relatedly, taking immediate senses to be phenomenal contents goes some way toward explaining why they are immediately available to us: They are phenomenally conscious. It is also plausible that reflective senses are derived from phenomenal contents. In the previous section, I argued that having a reflective sense is a matter of being disposed to entertain an immediate sense upon reflection. On this picture, immediate senses are more basic than reflective senses, and reflective senses are derived from them. If this is right and if immediate senses are phenomenal contents, then reflective senses are derived from phenomenal contents, as the PIT approach claims. Many details of the PIT approach remain to be worked out. First, since thoughts have immediate senses, the PIT approach requires that they also have phenomenal characters and that these phenomenal characters match their immediate senses. Second, the PIT approach must offer an account of precisely how reflective senses derive from immediate senses. Third, something must be said about nonconscious occurrent states, which might be taken to have immediate senses but no phenomenal characters. In the next three subsections, I will briefly consider these issues in turn. 5.1  The Phenomenology of Thought Since thoughts have immediate senses, the PIT approach requires them to have phenomenal characters that match their immediate senses. Whether thoughts have phenomenal characters and of what sort is a hotly debated issue.14 The debate usually focuses on what is sometimes called the cognitive phenomenology thesis, the thesis that thought has a proprietary and individuative phenomenology. Thought has a proprietary phenomenology just in case the phenomenology of thought is “cognitive,” or non-sensory, i.e., different in kind from the phenomenology of sensory mental states like perceptual experiences and emotions. Thought has an individuative phenomenology just in case thoughts with different contents have different phenomenal characters and thoughts with different phenomenal characters have different contents. For present purposes, we can largely avoid the debate over the cognitive phenomenology thesis. This is because the kind of phenomenology of thought required for PIT to account for immediate senses needn’t be proprietary and it needn’t be individuative in the sense that is least plausible. It need not be proprietary because as far as what we’ve said about immediate senses is concerned, immediate senses could be sensory in character. For example, they might consist in verbal or perceptual imagery, which, quite plausibly, would only require verbal or perceptual phenomenology. Whether this view is viable depends on whether verbal and perceptual contents can plausibly capture both the immediate senses of thoughts and the reflective senses that derive from them. In other words, it depends on

Immediate and Reflective Senses 197 whether we think a kind of empiricism is true of immediate and reflective senses.15 Whether the PIT approach requires the phenomenology of thought to be individuative depends on how we understand what it is for thought’s phenomenology to be individuative. We said that thought has an individuative phenomenology just in case thoughts with different contents have different phenomenal characters and thoughts with different phenomenal characters have different contents, but we didn’t specify what are the relevant thought contents. Two different understandings of being individuative are relevant to our purposes, one on which we understand the relevant kind of contents to be reflective senses and one on which we understand them to be immediate senses. The phenomenology of thought is individuative-with-respect-to-reflective-senses just in case thoughts with the same reflective senses have the same phenomenal characters and thoughts with the same phenomenal characters have the same reflective senses. The phenomenology of thought is individuative-with-respect-toimmediate-senses just in case thoughts with the same immediate senses have the same phenomenal characters and thoughts with the same phenomenal characters have the same immediate senses. Since the PIT approach takes immediate senses, but not reflective senses, to arise from phenomenal character alone, it requires the phenomenology of thought to be individuative-with-respect-to-immediate-senses but does not require it to be individuative-with-respect-to-reflective-senses. This is arguably a quite plausible view of the phenomenology of thought. To see this, consider a case that challenges the view that thought has an individuative phenomenology: Dimitri and Dimitra have different reflective understandings of what it is to be a father. For Dimitri, a father is a biological father, whereas for Dimitra, a father is a person with a male gender identity and a parental role. This difference in Dimitri and Dimitra’s concepts of a father is not always reflected in the phenomenal characters of their thoughts about fathers, especially when questions of biological fatherhood or parental roles are not particularly salient. When Dimitri and Dimitra think the thoughts they would express with “Marius is a father,” their thoughts sometimes have the same phenomenal characters. This is a counterexample to the phenomenology of thought being individuative-with-respectto-reflective-senses, since Dimitri and Dimitra’s thoughts have the same phenomenal characters but different reflective senses. But it need not be a counterexample to the phenomenology of thought being individuativewith-respect-to-immediate-senses. If Dimitri and Dimitra’s thoughts have the same phenomenal characters, then it is quite plausible that Dimitri and Dimitra are immediately aware of the same thing and so that their thoughts have the same immediate senses. This example shows that relatively deep theoretical differences of the sort exemplified by Dimitri

198  Angela Mendelovici and Dimitra make a difference for reflective senses, but they need not make a difference for phenomenal characters or immediate senses. This suggests that the phenomenology of thought is individuative-withrespect-to-immediate-senses, as required by the PIT approach, but not individuative-with-respect-to-reflective-senses. In summary, although the PIT approach to immediate and reflective senses requires there to be a phenomenology of thought, this requirement is not as controversial as it might seem, since the phenomenology of thought need not be proprietary or individuative-with-respect-to-reflective-senses. Even opponents of the cognitive phenomenology thesis might agree that thoughts have the kind of phenomenology required. The PIT approach requires not just that thoughts have phenomenal characters but also that their phenomenal characters can plausibly be identified with their immediate senses. In many cases, such an identification is not implausible. For example, Ent’s immediate sense might simply be a verbal representation of the word “Ent” together with a gist-like awareness of physicalism’s commitment to a priori entailment, and its phenomenal character might involve some verbal phenomenology of the word “Ent” and a gist-like cognitive phenomenal character. Similarly, a concept of a large number might have an immediate sense consisting of an auditory or visual depiction of the numeral that stands for it and a matching auditory or visual phenomenal character. Something similar holds in other cases. While it might be unclear what exactly are the immediate senses of representations like bird and supervenience, it is not implausible that they match their phenomenal characters in the way required. Immediate senses might consist in verbal or perceptual imagery, gists or aspects of reflective senses, or primitive or sui generis contents. This allows for a variety of views of the phenomenal character of any given representation on which it matches its immediate sense: First, the representation might have a verbal immediate sense and a matching verbal phenomenology (e.g., supervenience might have the word “supervenience” as its immediate sense and a matching quasiauditory verbal phenomenal character). Second, the representation might have a perceptual immediate sense and a matching perceptual phenomenology (e.g., supervenience might have an immediate sense consisting in mental imagery of one item above another and a matching visual phenomenal character). Third, the representation might have a simple or sui generis immediate sense and a matching simple or sui generis phenomenal character (e.g., supervenience might have a simple or sui generis immediate sense, not understandable in terms of any other contents, and a matching simple or sui generis supervenience-ish phenomenal character). Fourth, the representation might have an immediate sense that is a component or aspect of its reflective sense (which might be a matter of perceptual and verbal imagery, simple or sui generis components, or

Immediate and Reflective Senses 199 something else) and a matching phenomenal character that is a component of the phenomenal character corresponding to its reflective sense (e.g., supervenience might have the immediate sense , where this is a matter of verbal, or simple or sui generis contents and a phenomenal character consisting in matching perceptual, verbal, or simple or sui generis phenomenal characters). Finally, the representation might have an immediate sense consisting of a schematic or gist-like understanding of at least some aspects of its reflective sense and a similarly schematic or gist-like phenomenal character (e.g., supervenience might have an unarticulated or gist-like grasp of its definition as its immediate sense and a similarly gist-like phenomenal character). These options can be combined in various ways. First, a single representation might have an immediate sense that is a combination of the above options and a similarly combined phenomenal character. For example, supervenience’s immediate sense and phenomenal character might contain both verbal and sui generis components. Second, different representations might have different kinds of immediate senses and phenomenal characters. Perhaps bird’s immediate sense and phenomenal character consist in perceptual imagery and perceptual phenomenal character, respectively, but knowledge’s immediate sense and phenomenal character are sui generis. Third, different subjects or the same subject on different occasions might have representations with different immediate senses and phenomenal characters, even if they have the same reflective senses—just like Dimitri and Dimitra.16 One potentially attractive view is that the immediate senses and phenomenal characters of many representations are both simple, in that they do not involve component parts, and gist-like, in that they involve hints of at least some aspects of their reflective senses. This might be possible if immediate senses (phenomenal characters) are internally unstructured, in that they do not involve proper parts that are also immediate senses (phenomenal characters) and at the same time externally structured, in that they differ from various other actual or possible immediate senses (phenomenal characters) along various dimensions.17 We can think of such externally structured immediate senses (phenomenal characters) as being located in multidimensional spaces of immediate senses (phenomenal characters) organized by similarity. For example, the immediate senses (phenomenal characters) of perceptual color representations are arguably internally unstructured: They do not have or decompose into proper parts that are also immediate senses (phenomenal characters). For example, the phenomenal character of an experience of green arguably isn’t composed of phenomenal characters corresponding to particular hues, saturations, brightnesses or distinct full-fledged colors.18 But it is externally structured: It differs from the

200  Angela Mendelovici phenomenal characters of other color experiences with respect to hue, saturation and brightness. The same arguably holds for the experience’s immediate sense. More speculatively, perhaps the immediate senses (phenomenal characters) of some concepts can be located in the same multidimensional spaces as their reflective senses (their reflective sense’s phenomenal characters). For instance, perhaps the immediate sense of knowledge has some features in common with those of justification and belief and its similarity to those immediate senses accounts for the sense we have that the immediate sense of knowledge is a gist-like grasp of at least some components of its reflective sense. All this is compatible with the immediate sense of knowledge being internally simple—it might not have immediate senses such as and as parts, in the same way that the immediate senses of perceptual color representations do not have immediate senses corresponding to hues, saturations, and brightnesses as parts. The same arguably holds for the concept’s phenomenal character. My preferred view of the immediate senses and phenomenal characters of most concepts is that they are simple and gist-like in this way, perhaps also in some cases involving a verbal component and some mental imagery. This accords with the phenomenology of consciously entertaining immediate senses. When we think , we seem to have both a grasp of supervenience as its own unique thing as well an inarticulate glimpse of its reflective sense. In any case, my main claims are compatible with other views of the immediate senses and phenomenal characters of concepts and other representations. 5.2  How Reflective Senses Derive From Phenomenal Contents The PIT approach takes reflective senses to derive from phenomenal contents. But how exactly does this derivation work? In Section  4, I  suggested that a representation’s reflective sense is a matter of the immediate senses we are disposed to retrieve upon reflection. For example, upon reflection of , we are disposed to retrieve the immediate sense , which is Ent’s reflective sense. This suggests a way in which reflective senses derive from phenomenal contents: A representation’s reflective sense derives from the phenomenally represented immediate senses we are disposed to retrieve upon reflection of its immediate sense. The central challenge for such an account of reflective senses is to make sense of what it is for an immediate sense to be retrieved upon reflection. This cannot just be a matter of being disposed to entertain it after entertaining the representation in question’s immediate sense, since we might be similarly disposed to entertain immediate senses that are not

Immediate and Reflective Senses 201 relevant to determining reflective senses. For example, upon entertaining the immediate sense , we might be disposed to entertain immediate senses such as , , and , but only the first of these is bachelor’s reflective sense. The challenge is analogous to the functional role theory’s problem of collateral information discussed in Section  4. The situation, however, is importantly different for the PIT approach than for a functional role approach, since the PIT approach does not aim to understand reflective senses in entirely non-semantic terms but instead accepts phenomenal intentional states as part of its explanatory base. My proposal takes its cue from non-mental cases of derived representation. Words, signs and other non-mental items can come to have contents thanks to antecedently contentful mental states. For example, the word “dog” might mean and stop signs might mean because we take “dog” to mean and stop signs to mean . This taking is at least partly a matter of our representational states, such as our beliefs, intentions, and stipulations. I want to suggest that the relation between mental representations and their reflective senses is similar to the relation between non-mental derived representations and their contents. Just as we take words and signs to have certain contents, so too do we take mental representations to have certain contents beyond their immediate senses. On this proposal, bachelor has the reflective sense because we take it to represent . This taking is a matter of being disposed to accept as equivalent to, what is meant by, or a further elucidation of the immediate sense of bachelor, . This might involve being disposed to entertain the immediate sense , or . Such thoughts—call them cashing out thoughts—say that one content is an elucidation, unpacking, precisification, expansion or more generally a cashing out of another. In many cases, it takes multiple cashing out thoughts to retrieve a full elucidation of a representation’s immediate sense. For example, we might be disposed to have cashing out thoughts relating to , to , and to . Assuming (probably falsely) that we are not disposed to have cashing out thoughts relating and to further contents, is what we might call bachelor’s full cashing out. This content reflects our best understanding of bachelor’s target—it is bachelor’s reflective sense. My suggestion, then, is that a concept’s reflective sense is its full cashing out.19 This provides a way of solving our analogue of the problem of collateral information: What singles out and not as the reflective sense of bachelor is that we are disposed to have cashing out thoughts specifying that

202  Angela Mendelovici cashes out into but we are not disposed to have cashing out thoughts specifying that cashes out into . This picture of reflective senses is cursory and much more needs to be said before its viability can be assessed. However, I hope this section has at least shown in outline how reflective senses might be derived from phenomenal contents.20 5.3  Nonconscious Occurrent States The third issue for the PIT approach concerns nonconscious occurrent states that are arguably representational in some sense. Occurrent states are states that are active, used or undergone, such as perceptual states or thoughts. Nonconscious states are mental states that lack phenomenal character. Some nonconscious occurrent states are arguably representational in some sense, e.g., some states involved in language processing or early perceptual processing. Since nonconscious occurrent representational states are nonconscious, they presumably lack phenomenal characters, which commits the PIT approach to saying that they also lack immediate senses. But, since nonconscious occurrent representational states are occurrent and representational, it seems they should have immediate senses—when we have such states, their contents should be immediately available to us or “running through our minds,” even if we are not consciously aware of them. Nonconscious occurrent states pose a challenge for PIT in general, not just for the PIT approach to immediate and reflective senses. Accordingly, the positions the PIT approach might take toward them correspond to the positions available to PIT, which are the following: (1) They phenomenally represent their contents (e.g., Pitt, forthcoming; Bourget, 2010a; Mendelovici, 2018: ch. 8); (2) they derivatively but not phenomenally represent their contents (e.g., Bourget, 2010a; Kriegel, 2011) and (3) they do not phenomenally or derivatively represent their contents, though they might “represent” them in some other sense (e.g., Searle, 1990; Mendelovici, 2018: ch. 8). For the PIT approach, option (1) allows that nonconscious occurrent states have immediate senses and, if the right dispositions are in place, reflective senses. One might worry that this option requires that we can have phenomenal states that we are not introspectively aware of, but, given that there exist many phenomenal states that any given individual is not consciously aware of (e.g., the phenomenal states of other subjects), it is not wholly implausible that our own heads harbor phenomenal states that are likewise isolated from our introspective abilities. Option (2), on which nonconscious occurrent states derivatively represent their contents, results in the view that nonconscious occurrent states lack immediate senses but nonetheless have reflective senses. On this

Immediate and Reflective Senses 203 view, when a representation gets used nonconsciously, we are not immediately aware of anything. However, despite lacking immediate senses, nonconscious occurrent states have reflective senses, which derive from immediate senses. Since nonconscious occurrent states lack immediate senses, their reflective senses cannot be derived from their own immediate senses but must be derived from the immediate senses of other states, either other states of the same subject or other states of other subjects (as on Kriegel’s (2011) view). Option (2), however, is not compatible with my view of reflective senses described above, since my view requires that a representation with a reflective sense have an immediate sense that can partake in experiences of cashing out. For the PIT approach, option (3), on which nonconscious occurrent states neither phenomenally nor derivatively represent their alleged contents, results in the view that nonconscious occurrent states lack both immediate and reflective senses. Of course, they might still have various neural and functional properties and they might still track various worldly items. In virtue of these features, we might even say that they represent, perhaps in a loose or extended sense of “represent,” one that does not amount to their having immediate or reflective senses. Option (3) is arguably the least extreme of the three options. It is very much in line with what we might consider to be the standard view of nonconscious occurrent states: It accepts that such states lack phenomenal characters, that they have neural and functional properties, that they track various items and even that they “represent,” insofar as representation is merely a matter of their neural, functional, or tracking features. The main potential disagreement between option (3) and the standard view is presumably over whether the nonconscious “representation” it allows for amounts to the having of immediate or reflective senses or whether phenomenal properties are required. In other words, the main disagreement is over whether the PIT approach itself is true of phenomenally conscious intentional states—which is something I’ve already argued for. So, rather than being the most extreme of the three options, the third option arguably represents the smallest reasonable departure from what we might take to be the standard view. Which of the three options the PIT approach should adopt depends in part on whether there is independent reason to ascribe immediate or reflective senses to nonconscious occurrent states. My preferred view is a combination of (1) and (3) on which some nonconscious occurrent states have phenomenal contents and hence immediate and perhaps reflective senses, while others have neither phenomenal nor derived contents and hence neither immediate nor reflective senses. For example, states involved in nonconscious problem solving might have immediate senses we are not consciously aware of, while states in early visual processing might have neither immediate nor reflective senses.21

204  Angela Mendelovici

6. Conclusion I have argued that there are immediate and reflective senses and that the two senses come apart in almost all cases of thought and in many cases of perceptual representation. I  explored various ways we might try to account for immediate and reflective senses and suggested that the most promising way is the PIT approach. One way to put the view of immediate and reflective senses that I’ve proposed is this: Representations have immediate senses, which are phenomenal contents. Some representations, in addition to having immediate senses, are treated as if they stood for the immediate senses of other representations, which is, roughly, a matter of having dispositions to accept the immediate senses of those other representations as equivalent to or further elucidations of their immediate senses. When we treat a representation as if it stood for the immediate senses of other representations, those other representations’ immediate senses are its reflective sense. The end result is a picture on which thoughts and other representational states provide us two perspectives on the world, an immediate one, one that is running through our minds, and a derived one, one that is available upon reflection.22

Notes 1. Many of the ideas in this chapter are discussed in greater detail in chapter 7 of Mendelovici 2018. 2. This is to assume something like a language of thought (see Fodor 1975). 3. We might also argue for the distinctness of immediate and reflective senses in a way that parallels our gloss on Frege’s argument for the existence of senses: We need immediate senses that are distinct from reflective senses in order to account for differences that are not differences in either reflective senses or referents. For example, suppose that we reflectively understand a bachelor as an unmarried man. The thought that John is a bachelor and the thought that John is an unmarried man involve the same reflective senses and referents but are different thoughts. We can account for this difference by saying that the two thoughts involve two distinct concepts, bachelor and unmarried man, with distinct immediate senses. The difference between the two thoughts lies in what we are immediately aware of in having them. One might object that the difference between the two thoughts can be accounted for by the different vehicles they involve. While the two thoughts do involve different vehicles, this does not fully account for the difference between them because the difference is a semantic difference, a difference in what we are thinking. In order to account for this semantic difference, we need to posit a difference in the semantic features involved in these cases. (See also Mendelovici 2010: §13.4 for a related discussion of how distinguishing between immediate and reflective senses best resolves the paradox of analysis.) 4. For an overview, see Murphy 2004. 5. There is an important caveat: Some aspects or parts of the prototypes, sets of exemplars, or theories that are associated with a given concept might not be available upon reflection but might nonetheless do other work that would

Immediate and Reflective Senses 205 justify their inclusion in a concept’s prototype, set of exemplars, or theory. For instance, they might only be involved in categorization—for example, we might categorize objects as dogs based partly on their having a certain dog smell, but it might not be part of our reflective understanding of dogs that they have such a smell. Indeed, it is even possible for no parts of a concept’s prototype, set of exemplars, or theory to constitute our reflective understanding of its target. This might be the case with concepts with definitions that nonetheless have associated prototypes, exemplars, or theories, such as the concept grandmother. 6. See also Bourget’s (2017) example of perfectly grasping the shape of a line drawing. 7. Suppose that objects have dispositions to cause color experiences but do not have primitive colors. Then your experience is both accurate with respect to its reflective senses and inaccurate with respect to its immediate sense. For further discussion, see also Mendelovici 2018: Appendix D. 8. It is also possible to argue for the claim in the case of thoughts in much the same way as I’ve argued for the claim in the case of concepts: through the consideration of various cases. For example, we might directly argue that the reflective sense of the thought that the mental supervenes on the physical differs from its immediate sense. 9. For objections to tracking theories along these lines, see Mendelovici 2013, 2016, 2018: ch. 3 and Mendelovici and Bourget forthcoming. 10. I have only considered “short-arm” (Harman 1987) functional role theories, which take functional roles to be internal roles. “Long-arm” (Harman 1987) functional role theories appeal to causal or other relations to the environment in addition to internal roles. Including such relations does not help us accommodate immediate senses, since there are many cases in which immediate senses do not match anything we are interestingly related to in the environment (e.g., the case of color representations mentioned earlier). See also Mendelovici 2018: §4.3. 11. I have suggested a picture on which reflective senses are a matter of dispositions to entertain immediate senses. An alternative picture would take reflective senses to be somehow inherent in representations themselves. One reason to prefer the dispositional picture is that it makes better sense of why we need to retrieve our representations’ reflective senses in order to complete certain tasks. For example, in order to judge whether a necessary God supervenes on everything, we might need to retrieve the reflective sense of supervenience. Another reason to prefer the dispositional picture is that it makes better sense of how we can sometimes fail to retrieve our reflective senses. For example, Gabriel might suddenly find that while he can entertain the immediate sense of supervenience just as he could before, and, e.g., think thoughts like , he is no longer able to unpack the concept. This would be odd if the reflective sense of supervenience were somehow inherent in the concept itself. (See also Mendelovici 2010: ch. 10, especially n. 23.) 12. See Loar 2003, Strawson 1994, Siewert 1998, Horgan and Tienson 2002, Horgan and Graham 2009, Pitt 2004, Farkas 2008, Kriegel 2011, Bourget 2010a, 2010b, Mendelovici 2010, 2018, Bourget and Mendelovici 2016, and Mendelovici and Bourget 2014, forthcoming. 13. Phenomenal intentionalists have proposed various derivation mechanisms. For example, Horgan and Tienson (2002) suggest that wide intentional states are derived from phenomenal intentional states together with grounding presuppositions and facts about the world, Bourget (2010a) takes

206  Angela Mendelovici non-phenomenal intentional states to derive from phenomenal intentional states via descriptive reference and other derivation mechanisms, and Kriegel (2011) maintains that non-conscious intentional states are derived from the phenomenal intentional states of an ideal rational agent. 14. See Bayne and Montague 2011 for a representative collection of essays. 15. See Prinz 2002 for a recent defense of empiricism. Although Prinz combines his empiricism with a tracking view of intentionality, it can also be quite plausibly combined with the PIT approach. 16. For further discussion, see Mendelovici 2018: §7.2.4. 17. This terminology comes from Chalmers 2016. 18. See Roelof 2014 for arguments to the contrary and Mendelovici 2019 for discussion. 19. A representation’s full cashing out is an immediate sense or at least a content specified by an immediate sense (see Mendelovici 2018: §7.3.1, p. 142). There is good reason to think that, at least for concepts whose target we claim to somewhat understand, we can reach such an understanding in terms of immediate senses. If we couldn’t, then we would have no claim to understanding them. It is an open question, however, what kinds of immediate senses we will find in a representation’s full cashing out. For instance, we might find perceptual contents, but we might also find unanalyzable nonperceptual contents, perhaps like or . 20. This kind of view is developed in more detail in Mendelovici 2018: ch. 7. Note that there I employ a definition of “intentionality” on which derived contents, and hence reflective senses, do not qualify as genuinely intentional contents, though they qualify as representational in a looser, everyday sense. 21. For more details, see Mendelovici 2018: §8.4. Note that there “nonconscious states” is used to pick out mental states that are not phenomenally conscious, while here it is used to pick out mental states that we are not consciously aware of. 22. Many thanks to David Bourget for helpful comments on earlier drafts and to audiences at Oakland University’s philosophy department, Victoria University at Wellington philosophy department, and PhilMiLCog 2014 at the University of Western Ontario for helpful discussion.

References Bayne, T. and Montague, M. (2011) Cognitive Phenomenology, New York: Oxford University Press. Block, N. (1986) “Advertisement for a Semantics for Psychology” in Midwest Studies in Philosophy, 10 (1), 615–678. Bourget, D. (2010a) “Consciousness is Underived Intentionality” in Noûs, 44, 32–58. Bourget, D. (2010b) The Representational Theory of Consciousness, PhD diss., Australian National University. Bourget, D. (2017) “The Role of Consciousness in Grasping and Understanding” in Philosophy and Phenomenological Research, 95 (2), 285–318. Bourget, D. and Mendelovici, A. (2016) “Phenomenal Intentionality” in The Stanford Encyclopedia of Philosophy. Brooks, L. R. (1978) “Nonanalytic Concept Formation and Memory for Instances” in Cognition and Concepts (ed. E. Rosch and B. B. Lloyd), Hillsdale, NJ: Erlbaum, 169–211.

Immediate and Reflective Senses 207 Carey, S. E. (1985) Conceptual Change in Childhood, Cambridge, MA: MIT Press. Chalmers, D. J. (2016) “The Combination Problem for Panpsychism” in Panpsychism (ed. L. Jaskolla and G. Bruntrup), Oxford: Oxford University Press, 179–214. Dretske, F. (1995) Naturalizing the Mind, Cambridge, MA: MIT Press. Farkas, K. (2008) “Phenomenal Intentionality Without Compromise” in The Monist, 91 (2). 273–293. Fodor, J. A. (1975) The Language of Thought, Cambridge, MA: Harvard University Press. Fodor, J. A. (1987) Psychosemantics, Cambridge, MA: MIT Press. Frege, G. (1948) “Sense and Reference” in The Philosophical Review, 57 (3), 209–230. Gopnik, A. and Wellman, H. M. (1994) “The Theory” in Mapping the Mind: Domain Specificity in Cognition and Culture (ed. L. A. Hirschfeld, and S. A. Gelman), Cambridge: Cambridge University Press, 257–293. Harman, G. (1987) “(Non-Solipsistic) Conceptual Role Semantics” in New Directions in Semantics (ed. E. LePore), London: Academic Press, 55–81. Horgan, T. and Graham, G. (2009) “Phenomenal Intentionality and Content Determinacy” in Prospects for Meaning (ed. R. Schantz), Amsterdam: De Gruyter, 321–344. Horgan, T. and Tienson, J. (2002) “The Intentionality of Phenomenology and the Phenomenology of Intentionality” in Philosophy of Mind: Classical and Contemporary Readings (ed. D. J. Chalmers), Oxford: Oxford University Press, 520–533. Kriegel, U. (2011) The Sources of Intentionality, Oxford: Oxford University Press. Loar, B. (2003) “Phenomenal Intentionality as the Basis of Mental Content” in Reflections and Replies: Essays on the Philosophy of Tyler Burge (ed. M. Hahn and B. Ramberg), Cambridge, MA: MIT Press, 229–258. Medin, D. L. and Schaffer, M. M. (1978) “Context Theory of Classification Learning” in Psychological Review, 85, 207–238. Mendelovici, A. (2010) Mental Representation and Closely Conflated Topics. PhD diss., Princeton University. Mendelovici, A. (2013) “Reliable Misrepresentation and Tracking Theories of Mental Representation” in Philosophical Studies, 165 (2), 421–443. Mendelovici, A. (2016) “Why Tracking Theories Should Allow for Clean Cases of Reliable Misrepresentation” in Disputation, 8 (42), 57–92. Mendelovici, A. (2018) The Phenomenal Basis of Intentionality, New York: Oxford University Press. Mendelovici, A. (forthcoming) “Panpsychism’s Combination Problem is a Problem for Everyone” in Handbook of Panpsychism (ed. W. Seager), New York: Routledge. Mendelovici, A. and Bourget, D. (2014) “Naturalizing Intentionality: Tracking Theories Versus Phenomenal Intentionality Theories” in Philosophy Compass, 9 (5), 325–337. Mendelovici, A. and Bourget, D. (forthcoming) “Consciousness and Intentionality” in Oxford Handbook of the Philosophy of Consciousness (ed. U. Kriegel), Oxford: Oxford University Press.

208  Angela Mendelovici Millikan, R. G. (1984) Language, Thought and Other Biological Categories, Cambridge, MA: MIT Press. Murphy, G. L. (2004) The Big Book of Concepts, Cambridge: MIT Bradford. Murphy, G. L. and Medin, D. L. (1985) “The Role of Theories in Conceptual Coherence” in Psychological Review, 92: 289–316. Nagel, T. (1974) “What Is It Like to Be a Bat?” in The Philosophical Review, 83 (4), 435–450. Neander, K. (2017) A Mark of the Mental: A Defence of Informational Teleosemantics, Cambridge: MIT Press. Pitt, D. (2004) “The Phenomenology of Cognition or What Is It Like to Think that P?” in Philosophy and Phenomenological Research, 69 (1), 1–36. Pitt, D. (forthcoming) “Unconscious Thought” in The Quality of Thought. Prinz, J. (2002) Furnishing the Mind: Concepts and their Perceptual Basis, Cambridge: MIT Bradford. Roelofs, L. (2014) “Phenomenal Blending and the Palette Problem” in Thought: A Journal of Philosophy, 3 (1), 59–70. Rosch, E. (1975) “Cognitive Representations of Semantic Categories” in Journal of Experimental Psychology: General, 104 (3), 192–233. Rosch, E. and Mervis, C. (1975) “Family Resemblances: Studies in the Internal Structure of Categories” in Cognitive Psychology, 7 (4), 573–605. Searle, J. (1990) “Consciousness, Explanatory Inversion and Cognitive Science” in Behavioral and Brain Sciences, 13, 585–642. Siegel, S. (2005) “Which Properties Are Represented in Perception?” in Perceptual Experience (ed. T. Szabo Gendler and J. Hawthorne), Oxford: Oxford University Press, 481–503. Siewert, C. (1998) The Significance of Consciousness, Princeton: Princeton University Press. Stampe, D. W. (1977) “Towards a Causal Theory of Linguistic Representation” in Midwest Studies in Philosophy, 2 (1), 42–63. Strawson, G. (1994) Mental Reality, Cambridge, MA: MIT Press.

11 The Unity of Unconsciousness1 Tim Crane

1. Conscious and Unconscious Intentionality It is widely accepted in philosophy, psychology and cognitive neuroscience that some mental states or episodes or processes are conscious, and many are unconscious. This raises two immediate questions about the relationship between the conscious and the unconscious. First, in what ways do conscious mental states resemble unconscious mental states, and in what ways do they differ? And second, how do consciousness and the unconscious interact? My aim in this chapter is to address these two questions. My interest here is specifically in intentional or representational mental states, states that represent the world. So, my question is: How is the way unconscious mental states represent the world related to the way conscious mental states represent it? If we follow the common, if not universal, habit of calling the way a state represents the world its content, then we can rephrase this question thus: What is the relation between the content of unconscious mental states and the content of conscious mental states? The usual answer to this question these days is that these contents are the same kind of thing. The content of an unconscious belief (for example) is a proposition, a bearer of truth-value, and the content of a conscious belief is the same. We can see that this view might not be totally straightforward, though, when we consider its relationship to the representationalist view that the phenomenal character—that is, the way having an experience seems to a subject in a conscious experience—is identical with its representational content. For, as David Chalmers (2004: Section 2) has pointed out, how can this be if the content of a conscious belief (say) is the very same thing as the content of an unconscious belief? Some representationalists will reject the assumption that the content of a conscious intentional state is the same kind of thing as the content of an unconscious state. Tye (1995), for example, argues that the contents of conscious intentional states are abstract, nonconceptual and poised for action, and this is what differentiates them from the contents

210  Tim Crane of unconscious intentional states. One difficulty with this proposal is to explain why a propositional content having these features cannot also be the content of an unconscious state; another difficulty is to explain how states with conceptual content can be phenomenally conscious. I agree that the contents of conscious and nonconscious states must be different; but I go further than Tye and others, who still hold on to the usual “propositional attitude” picture for both kinds of state. I will argue here that the way in which unconscious states represent the world is significantly different from the way conscious states represent the world. Concentrating for the purposes of this chapter on the state of belief, my proposal is that ascriptions of unconscious beliefs are partial characterizations of what I call a subject’s “world view”—their entire attitude toward reality. A world view can be less specific, coherent and determinate than ordinary psychological ascriptions often suggest. These ascriptions serve a modeling and therefore simplifying function and should not therefore be thought of as corresponding directly to individual belief states with fixed or determinate content. Where conscious intentionality is concerned, things are very different. Conscious intentional states can also be modeled by propositions, but they have in addition a distinctive way of representing the world which is very different from the way the unconscious represents the world: Conscious intentional content is generally more specific, more determinate and more individuated than the content of the unconscious world view. But my interest here in this chapter is more on the unconscious than the conscious: My central claim is that unconscious states of mind do not have content in the way that conscious states do. This is the first theme of this chapter. My second theme is the nature of the unconscious itself. Many approaches to the mind appeal to the idea of the unconscious—philosophy, cognitive science and psychoanalysis—but there is no clear agreement about what they mean by this. My proposal about the relationship between the conscious and the unconscious provides a story not only about how mental states are brought to consciousness but also about why all these different approaches to the unconscious are genuinely about the same thing—hence my title.

2. Belief: The Usual View I start with the widespread view that beliefs are mental states that represent the world and that the way they represent the world is by having representational or intentional content; their intentional content is a proposition, something which is true or false. This view is sometimes summed up by saying that belief states are relations to propositions and implies that for each belief you have there is a distinct proposition to which you are related.

The Unity of Unconsciousness 211 This picture of beliefs often—though not invariably—goes along with the view that the underlying mechanism for these states is a computational language, a “language of thought” in Jerry Fodor’s famous image. Fodor (1975) put forward his language of thought hypothesis as an empirical claim about the structure of the mind/brain. It is fair to say that in the last 40 years psychology and cognitive neuroscience have not discovered anything like a language of thought, though these disciplines use the terminology of mental representation and computation without compunction. Since there are no compelling empirical reasons to believe in a language of thought, I would like to pass over this specific issue and concentrate more generally on the idea of individual belief states. Beliefs, it is often said, are “individuated by their contents.” I take this to mean that individual belief states are distinguished from one another by the propositions which give the ways they represent the world. States are standardly thought of as instances of properties or relations, with numerically distinct states being numerically distinct instances. So your and my beliefs that Princip assassinated the Archduke are usually treated as instances of the same property, or the same relation (the relation expressed by the word “believes”) to the proposition Princip assassinated the Archduke. These are two distinct states, just as this belief of mine is a distinct state from all my other beliefs. The distinct states are then characterized in various further ways—as dispositions to behave, or to utter sentences, as aiming at truth, as being fine-grained or coarse-grained in their contents, as being externalistically individuated, etc. This, in any case, is the textbook picture of belief. If distinct beliefs are numerically distinct states, then a natural (if naïve) question arises: How many beliefs do we have? Given how natural this question is, it is striking how rarely it is asked, and even more striking how few serious answers have been given. Plainly an answer like “1,758,876” seems absurd. When Bob Dylan was asked by a journalist how many protest singers there were, he replied “either 136 or 142.” His answer was, of course, intended to poke fun at the question. So someone might similarly be inclined to poke fun at the naïvety of this question. Yet I  still want to insist that the question should be pursued. For if beliefs really are numerically distinct states, individuated by their contents, why shouldn’t there be an answer in principle to the question of how many beliefs you have? What is supposed to be wrong with this question in principle, given the standard view of beliefs? After all, for objects to be individuated is for them to be distinguishable from one another and therefore (in principle) countable. Why shouldn’t it be the same with states? And yet there does seem to be something fishy about the idea that you might have a specific, definite number of beliefs, even if it is difficult to figure out what it is. It doesn’t seem to be like the question of the number of hairs I have on my head. So rather than dismissing the question and retaining the usual picture of beliefs, perhaps we should

212  Tim Crane consider the obscurity of the question to be a reason for dispensing with the usual picture. The standard answer to the naïve question—insofar as there is one—is that we need to distinguish “core” beliefs from “derived” beliefs. Sometimes this is expressed by saying that we have some beliefs which are “explicit” and some which are “implicit” or tacit; or some which are “actual” and some which are “potential.” These distinctions are not all getting at the same thing, but I will ignore the differences here. For none of these approaches have come up with anything like a criterion, however vague or imprecise, to distinguish the implicit (etc.) from the explicit beliefs. Which are our basic or explicit beliefs? How do we go about settling this? It could be said that this is an empirical question, but even if empirical psychologists were interested in specifying which beliefs are explicit and which are implicit—which to the best of my knowledge they are not—they would still need to start with a clearly implementable idea of what an explicit belief is. And with the exception of the unworkable idea of a “sentence in a language of thought,” this is what we don’t have. (see Dennett, 1975) We have our practice of belief ascription, of course. Some belief ascriptions are true, others are not. But despite its centrality in our psychological self-conception, our commonsense practice of belief ascription does not give us any firmer grip on the idea of an explicit belief. An explicit belief is not just one that is consciously available, or one that is easy to ascribe. I  can now pluck a belief out of the air, so to speak, and truly ascribe it to myself or to others. I know that I believe that some houses are bigger than others—I only have to consider the question to know that I believe it, and that you believe it too. Does this mean it is a core or explicit or basic belief? Surely not—it’s not a particularly important fact for me, it’s just clearly a consequence of other things I believe. It’s a derived belief, if anything is, and yet there it is, immediately accessible for ascription by you or conscious self-ascription by me. Small children and animals present particular problems in this context, since they cannot put their beliefs into words. As Daniel Dennett says, “a child can demonstrate his understanding of addition by reeling off sums without being able to formulate or understand propositions about the commutativity of addition. His performance indicates that he has caught on to commutativity, but should we say that among his beliefs is the belief that addition is commutative?” (Dennett, 1975: 46) Maybe we should say not, because he does not understand the word “commutative.” Certainly, he does not, but should we then assume that someone has a concept of Fs only when they understand a word for Fs? That isn’t part of the standard view. The view encounters similar difficulties in accounting for the beliefs of animals. We see the dog chasing the cat up the tree, but does the dog’s belief involve a proposition about trees in the way that ours does? We

The Unity of Unconsciousness 213 think trees are living things, as opposed to houses (say) which are not. So our beliefs about trees are embedded in a network of beliefs that distinguish the living from the non-living and distinguish them because of (e.g.,) the ability of living things to reproduce, our ability to make them grow and so on. Surely the dog has none of this. In what sense, then, does the dog have any beliefs about trees? But even ignoring children and animals, the standard picture of belief faces questions when it is applied to our effortless (and surely true) everyday ascriptions of beliefs to people. Often, we ascribe the same belief to a number of people, using sentences with a common linguistic meaning, even when those to whom the beliefs are ascribed differ significantly in how they take the world to be. Consider three people, Alice, Bert and Carol, who are looking at the Houses of Parliament in London from Westminster Bridge. They all believe that this is the British parliament, but they believe very different things about it. Alice is an expert on the British constitution and has a full knowledge of how the parliamentary system works in the UK. Bert mistakenly thinks that the British Parliament is the “English parliament,” perhaps because he has seen the Scottish Parliament building and because he was brought up in the United States, he regularly uses the word “British” to mean “English.” Carol thinks that the British government is based in the Houses of Parliament, since she does not distinguish in her mind between government and Parliament. These views are so different, the inferences they will draw from their various true and false beliefs are so varied—and yet they all believe that this is the British Parliament, as they stand on Westminster Bridge. How can there be a single belief state that they have in common if the concepts they have—of Parliament, or Britain, or government—and the consequent inferential properties of their beliefs are so different? Of course, there are familiar answers to these questions. We might distinguish, for example, between the core essential (analytic or constitutive) content of the belief and the “collateral information” associated with its objects; or we might say that these phenomena give us reason to take an externalist or directly referential view of belief content; or we might just give up on the idea that the ascription of the same belief to all these people is literally true. But none of these answers is without its problems. Some philosophers (most effectively, Stich, 1983) have taken these phenomena—animals, children and the holism of belief, etc.—to give us reason to dispense with the commonsense notion of belief altogether. But another response is that the problem is not with the notion of belief itself but with the philosophical conception of it which philosophers have developed from their reflections on the practice of belief ascription. Might there be another way of thinking about beliefs, a way of thinking that better accommodates these facts about belief? To put it another way, if we think that belief ascriptions in these cases are literally true, what is it that makes them true, if not a collection of individual, countable belief

214  Tim Crane states? What is the relationship between the truth of the belief ascription and the psychological reality it describes?

3. Unconscious Belief and the Subject’s World View According to Richard Wollheim, “[T]he role of belief is to provide the creature with a picture of the world it inhabits. Not, of course, any picture of the world, but . . . a picture that depicts the world more or less as it is.” (Wollheim, 1999: 13) The “picture” of the world which Wollheim talks about is obviously not a conscious picture; we do not have conscious awareness of everything we believe. Philosophers sometimes say that some beliefs are conscious and some are not. In other work I  have rejected this idea, and I  have argued that all beliefs are unconscious. (Crane, 2013) There is, of course, the phenomenon of consciously thinking something, but conscious thinking does not amount to believing, even if what one is thinking is what one believes. In the Philosophical Investigations, Wittgenstein observed that “believing is not thinking” and called this “a grammatical remark,” which I take to mean that it is a necessary or conceptual truth. (Wittgenstein, 1953: Section 574) My own argument for this conceptual truth was based on the idea that belief is not an occurrence, a point anticipated by Kent Bach 35 years ago: Philosophers sometimes distinguish between occurrent and dispositional senses of “believe,” but I will use the term “believe” only for the dispositional sense and reserve the word “think” for the wouldbe occurrent sense. I say “would-be” because I deny that occurrent believing is believing at all, or in my terminology, that thinking that p is either necessary or sufficient for believing that p.  .  .  . Unlike thoughts, beliefs are states, not occurrences. (Bach, 1981: 35–34) What Bach does not say explicitly here is that occurrences of thinking are conscious, and dispositions are not. Clearly, a disposition cannot itself be conscious if it is something that can persist through (e.g.,) sleep or other forms of unconsciousness. What can be conscious, though, is the exercise of a disposition—and conscious thinking can be the result of such an exercise. Here I  will assume this distinction between belief as unconscious and thinking as conscious. The idea that beliefs are dispositions (to behave or to speak) is, of course, an idea familiar from behaviorist and functionalist theories of mind. Roughly speaking, what someone does is determined by what they want, together with the way they believe the world to be. Someone who crosses the room enthusiastically to greet you is doing so because they recognize (and therefore believe) that you are their long-lost friend and want to spend time in your company, or because they recognize you as a

The Unity of Unconsciousness 215 long-lost enemy yet want you to think otherwise or because of some other combination of beliefs and wants or desires. Any one type of action, as Geach (1957) famously showed, could be the product of different beliefs, given different desires. This gives us the beginnings of an answer to our question about how to reconcile ordinary belief ascriptions with the complexity of individual psychology. I noted above that we are perfectly happy to attribute the belief that this is the British Parliament to Alice, Bert and Carol in the situation I described, even though what each of them is disposed to infer from this belief is quite different. As the functionalist story says, the very same belief can give rise to many different other states or actions, depending on the other mental states one has: the belief is a “multi-track” disposition. In this way, the functionalist story makes sense of the variety of possible outcomes stemming from the same belief state. But how do propositions—conceived of as abstract objects specifying the contents of beliefs—fit into the functionalist story? My proposal, which I have defended elsewhere (Crane, 2015), is that we should think of propositional attitude attributions as models, in the sense described by the philosophy of science (and not in the sense of Tarskian semantics, pace Dennett, 1982: 154). A model is a (concrete or abstract) object used to draw attention to some features of a system under investigation and to make the study of the system more tractable. When Rutherford compared the atom to the solar system, this was intended to draw attention to the structural relationship between the atom’s nucleus and its electrons. Similarly, a propositional model draws attention to features of your belief, and different kinds of proposition can be used to draw attention to different features. Models idealize and simplify in order to highlight some structural feature of the thing being modeled. Propositions too can simplify—as applied to our example above, we can attribute to different people the belief that this is the British Parliament, when they are standing on the banks of the Thames, even if they have very different (and even incompatible) ideas about what the British Parliament is, or what the Thames is. From this point of view, it doesn’t matter that Alice and Bert have such different ideas about Britain. Using the proposition to attribute this belief to them makes it salient that it is that building which is the object of their belief, however they happen to conceptualize it. The case we are considering here is the belief system—the subject’s unconscious representation of the world, a representation that aims to “depict the world more or less as it is,” in Wollheim’s words. Let’s call this the subject’s world view. Rather than thinking of each attribution of belief as corresponding to a single belief state, then, let’s think instead of the belief system as embodying the subject’s whole belief state, their entire doxastic orientation toward the world. The world view is an unconscious structure that embodies all the dispositions associated with what you believe. So instead of thinking of a single disposition being associated

216  Tim Crane with a distinct propositional content, the propositions which we employ in belief ascriptions are used to model aspects of this total structure. Why say that this total structure is something psychological? Some writers, notably John Searle (1992), have argued that there is no unconscious intentionality: There is only conscious intentionality and its neural basis. I reject Searle’s way of thinking here on the grounds that psychology and cognitive science appeal unproblematically to unconscious mental structures and processes. (for some examples, see Kihlstrom, 1987; Wilson, 2002; Hassin et al., 2005) Treating a creature’s world view as belief amounts to this: The creature has a faculty or capacity whose purpose is to represent the world as it is. The world view has a representational character—it is directed on the world, and its “direction of fit” is mind-to-world, the terminology Searle (1983) derives from Anscombe. This is a psychological or intentional characterization of this faculty or capacity, not a neural one, although it is compatible with a further neural characterization—I leave that question open. This idea of the world view is really just another way of expressing the functionalist insight that mental states form a complex network of interrelated dispositional and causal connections—except that I refrain from labeling particular nodes in the network as individual belief states. There is a network of causal and dispositional relations, and we use propositions to model them. This modeling is the essence of individual belief ascriptions, but the truth-maker for these ascriptions need not be individual belief states, as the standard view understands them. I’d like to describe briefly now how this approach gives better accounts of (i) the holistic nature of belief, (ii) the beliefs of animals and children, and (iii) contradictory and indeterminate beliefs. First, the holism of the intentional. This is the fact that a thinker’s intentional states must always be part of a wider connected network of intentional states which influence one another and contribute to their intentional features. We need to hold two facts stable: First, that different people can be ascribed the same beliefs—that is, beliefs with the same propositional contents—and second, that their related states and associated ideas and beliefs may be very different from one another. The worry for the standard view of propositional attitudes is that if propositions genuinely individuate states, then they must provide identity conditions for the states. This means that any states with the same propositional contents must be identical. But how then can Alice, Bert and Carol all count as believing that the British Parliament is on the Thames if these beliefs have relational properties—their relations to all their other beliefs —- that are so different? The picture of propositional attitude attributions as models explains how. The model is only supposed to isolate certain aspects of the subjects’ world views: that they are representing the UK Houses of Parliament, that they are representing it in a certain way—say, as being on the banks

The Unity of Unconsciousness 217 of the River Thames. This is good enough to make the proposition that is the British Parliament a model of each of their beliefs, despite the other relational ways in which these beliefs differ. Thinking of propositional attitudes as models explains, then, the phenomenon of unity over difference in our propositional attitude attributions. Second, the picture gives a better account of what is going on when we attribute propositional attitudes to human infants and animals. Most of us will agree with Normal Malcolm’s claim that a dog can believe that a cat went up a certain tree (Malcolm, 1973). This can be literally true of the dog. But we should not have to say, because of this, that the dog believes that what the cat went up is a tree in the sense that we believe things are trees. If we treat the proposition the cat went up the tree as a model of the dog’s belief, then we can explain its behavior—hanging around under the tree, barking, etc.—by appeal to this proposition. Whatever is the correct account of canine psychology, it is enormously plausible that the dog will have a world view—its attempt to register the facts about its environment. In anything that counts as a psychological organization, there must be some component which is responsive to information about the world and whose function is to represent things correctly. But you may want to insist—”the cat went up the tree,” is that really what the dog believes? My answer, assuming my conception of what it is to believe something, is: Yes, of course. But this doesn’t mean that the content of this part of the dog’s world view cannot be captured in some other way, by using some more theoretical psychological vocabulary. These different ways of describing the same belief should all be compatible, but it is also important that different theoretical purposes will require different levels of determinacy or precision. In their classic discussion of the alarm calls of vervet monkeys, Cheney and Seyfarth claim that “the meaning of the leopard alarm is, from the monkey’s point of view, only as precise as it needs to be” (Cheney and Seyfarth, 1990: 169). This might suggest that it is the monkey’s decision, as it were, to make its call more or less precise. But of course, it is the theorist’s decision. A theorist may attribute a “disjunctive” content to a monkey’s alarm call, but all this means is that this kind of call is provoked by predator X or predator Y; not that the monkey itself has (in some obscure sense) settled on a less committed content. The third advantage of this approach is that it gives a more realistic account of the obscurity contained in what we actually believe. A lot of what we believe is incomplete, partial, confused and even contradictory. The single proposition-plus-individual belief state picture makes it hard to see how this can be the case, tending to attribute these features to our knowledge of our belief states rather than to the states themselves. But consider Dennett’s poignant example of Sam the art critic, who “extols, buys, and promotes mediocre paintings by his son” (Dennett, 1975: 39).

218  Tim Crane Does Sam really believe that the paintings are good because he has been blinded by his love for his son? Or does he not believe they are good but pretends they are, out of loyalty and love? Dennett comments that even if a “neurocryptographer” were to discover an explicit symbolic representation in his brain, this would not necessarily help answer these questions: [I]f our neurocryptographer were able to determine that Sam’s last judgement on his deathbed was “My consolation is that I fathered a great artist,” we could still hold that the issue between the warring hypotheses was undecided, for this judgement may have been a self-deception. (Dennett, 1975: 49) Let’s extend the story a little. Sam is asked to nominate, anonymously, the ten best artists of the up-and-coming generation for a major prize. Let’s suppose that the nomination process is so secure and secret that no one will ever find out whom Sam nominates, and the prize is a significant honor which would advance his son’s career significantly. But let’s suppose that Sam declines to nominate his son. Is this because he really does not believe that the son is a great artist? Or is it because he does believe it, but he is a man of such integrity that he does not want it on his conscience that he has promoted his son anonymously? Does Sam, or does he not, believe that his son is a great artist? In a complex case like this, there may be no straightforward yes/no answer to this question. The right description, it seems to me, is that Sam’s world view involves a large collection of dispositions in relation to his son, not all of which are consistent, and the spelling out of which will involve so much complexity that it simply may not be possible to settle on one sentence that describes the belief. Stuart Hampshire has described this kind of situation well: A man may think that he believes p, whilst his behaviour can only be explained by the hypothesis that he believes not-p. . . . Perhaps the confusion in his mind cannot be conveyed by any simple account of what he believes; perhaps only a reproduction of the complexity and confusion will be accurate. (Hampshire, 1975: 123) So we need to be able to say that it may simply be indeterminate whether Sam believes that his son is a great artist. But this is not because there are no psychological facts about what he believes—it’s rather because there are too many. Complexity and confusion can go right to the bottom of our world view.

The Unity of Unconsciousness 219 Not all cases are like this, of course. Many aspects of our world view are straightforward, and the dispositions involved can be modeled effectively with a single proposition. But even in these straightforward cases, the holism of the intentional guarantees that there need not be one single dispositional profile associated with each proposition believed. Modeling by a third party is one way that a determinate content can be imposed upon the world view. The other is, of course, self-interpretation: Introspection or the self-attribution of beliefs. In the recent philosophical literature, this subject is often discussed under the heading of “self-knowledge.” As Quassim Cassam points out (2015: vii), this is not an entirely happy label, since the knowledge discussed is not normally knowledge of one’s self but knowledge of one’s beliefs and other attitudes. I agree, but here I want to focus on another aspect of self-knowledge in this sense. Often when we are asked what we believe, we can answer the question straight off—”do you believe that it is raining now?” and so on. But there are some cases where we have to work out what we believe. Often when we reflect on our beliefs about some complex matter—a political or theoretical question, for example, or a situation involving our relationships with those close to us—we find ourselves with confused and often contradictory or unresolved attitudes. (The situation is the same for what we want or desire, but I am focusing only on belief in this chapter.) The picture of “self-knowledge” suggests that there are determinate facts about our beliefs and we just need to employ some method to find out what they are. But often when we introspect (whatever that exactly means) we find only confused and unclear beliefs. In these kinds of cases, bringing our beliefs to consciousness may require us to resolve this confusion by deciding to express ourselves in one way or the other. In this way, the conscious acknowledgment or expression of a belief has a different kind of content—more fixed, more determinate—than the content of much of the world view. Again, it’s easier to see what I mean here by contrasting it with the picture of fully determinate belief states (whether understood as “sentences in the head” or as fully determinate dispositions). That picture implies that all the facts about our beliefs are there waiting to be discovered and that the difficulty we have in articulating them is purely epistemological. On this view, discovering what you think is one thing; making up your mind is entirely another. The alternative I am proposing is that there is no sharp distinction between these two activities. Discovering what you believe can resolve indeterminacy and unclarity in your world view and producing a conscious judgment that settles things as far as you are concerned. This view treats the determinate content of a belief as “coming into being as we probe,” as Michael Dummett once said of a different kind of case (mathematical reality: Dummett, 1959).

220  Tim Crane The clearest way in which the content of a world view can become the content of a conscious judgment is when one says something, either aloud or to oneself. This is the truth behind the often-quoted aperçu, “How can I tell what I think till I see what I say?,” attributed variously to E.M. Forster (1927: 99) and Graham Wallas (1926). When we use words to express our beliefs, often by uttering a sentence, we impose an order on what we think which can then make our subsequent dispositions more determinate. We need to recognize “the hard edge of determinacy our verbal output substitutes for the fuzziness of our convictions” as Dennett (1975: 48) nicely put it. But we can also do this without putting it into words—we can make a conscious judgment without explicitly asserting a sentence. The central phenomenon here is interpretation. To model an aspect of a world view with a proposition is to interpret it. To bring a belief to consciousness is to interpret something about your world view, where interpretation is a matter of taking a system—you—and making sense of how you represent the world in general. From the third-person point of view, you take the subject’s belief system or world view and isolate some feature of it by relating it to a proposition. From the first-person point of view, recovering what your beliefs are is a matter of interpreting yourself—finding out what is the best way to express or describe your beliefs. This can involve making up your mind or figuring out what is straightforwardly contained within it, anyway. Interpretation can bring determinate content to consciousness or it can render determinate what is not yet determinate. Saying this is not the same as endorsing an “interpretationist” conception of intentionality in general—as if all facts about intentionality can be explained by appealing to the idea of interpretation. (William Child [1996] attributes this interpretationist view to Davidson and Dennett.) For one thing, interpretation itself is a mental or intentional activity, so we cannot expect all intentionality to be explained in terms of interpretation. The interpretation of one’s own belief is a state or episode of consciousness, and this has its intentionality independent of any further act of interpretation. Conscious acts of interpretation do not require interpretation to give them content.

4. The Unconscious in General How do these ideas about belief relate to the unconscious in general? The notion of unconscious mentality figures largely in at least three distinct theoretical areas: Cognitive science, the philosophical interpretation of commonsense psychology and psychoanalysis. It is worth asking whether the notion is used in the same way in all these areas. Is there a common, unified notion of the unconscious?

The Unity of Unconsciousness 221 With a few notable exceptions, psychoanalysis has not received the kind of discussion in analytic philosophy of mind as cognitive science and commonsense psychology have. So it is an interesting fact that in their introductory expositions of the unconscious, both philosophers and cognitive psychologists frequently credit Freud with the insight that there are unconscious intentional states. Here are a few examples. Jerry Fodor says: It used to be universally taken for granted that the problem about consciousness and the problem about intentionality are intrinsically linked: that thought is ipso facto conscious, and that consciousness is ipso facto consciousness of some or other intentional object. . . . Freud changed all that. He made it seem plausible that explaining behaviour might require the postulation of intentional but unconscious states. Over the last century, and most especially in Chomskian linguistics and in cognitive psychology, Freud’s idea appears to have been amply vindicated. (Fodor, 1991: 12) Much more recently Peter Carruthers has observed that “almost everyone now accepts. . . (post-Freud) that beliefs and desires can be activated unconsciously” (2016: Section 1). In psychology too, Freud often gets the credit for the notion of the unconscious: “[T]he idea that a large portion of the human mind is unconscious,” writes Timothy Wilson, “is not new and was Freud’s greatest insight” (Wilson, 2002: 3). And yet Freud’s specific ideas themselves are largely ignored in both analytic philosophy of mind and in psychology. After acknowledging Freud’s insight, Wilson goes on to say that “the modern adaptive unconscious is not the same as a the psychoanalytic one” (Wilson, 2002: 5). Psychoanalysis as a theoretical enterprise operates largely outside the mainstream of psychology and philosophy—as I  see it, there is more communication between cognitive science and mainstream philosophy of mind than there is between either endeavor and psychoanalysis. But this raises a puzzle—if psychologists and philosophers of mind ignore Freud’s ideas, why do they pay tribute to him for his great discovery? What are they actually crediting Freud with? Or to put it another way: What did Freud actually discover? Not the unconscious, as is now widely acknowledged. If anyone thinker should be credited with making explicit the idea of unconscious intentionality, it is Leibniz, who in his New Essays (1704) talked about the petites perceptions which are “either too minute and too numerous, or else too unvarying” to attract our awareness (Leibniz, 1996: 53). Leibniz drew attention to these phenomena in part to undermine the Cartesian view that the soul is always conscious. Also influential was Hermann von

222  Tim Crane Helmholtz, who argued that perception must be explained in terms of unconscious inference—disciplined transitions among intentional states (Helmholtz, 1853). Other nineteenth century thinkers like Gustav Fechner (1801–1887) and Friedrich Albert Lange (1828–1875) made use of ideas of unconscious mentality in their theorizing. Freud was not the first. Paul Katsafanas has argued recently that a distinction can be made between those theories of the unconscious which treat it as essentially the same kind of thing as the conscious, and those which see it as something of a very different kind—even as something opposed to, or set against, consciousness (Katsafanas, 2016: 17–21). As Katsafanas says, some theories of the first kind treat consciousness itself as something which is the upshot of the combination of unconscious states. Leibniz, for example, argued that when some petites perceptions are “combined with others they do nonetheless have their effect and make themselves felt, at least confusedly, within the whole” (Leibniz, 1996: 53). But those that treat the conscious as essentially the same kind of thing as the unconscious need not think of the former as constituted out of the latter; there are other ways to think of the unconscious on this model. Freud’s view of the unconscious is of the second kind distinguished by Katsafanas: The unconscious is “a system with its own rules and principles” (Wollheim, 1971: 13). One central Freudian theme, for example, is that of repression. Ideas which cannot be admitted into consciousness are repressed; these ideas are resisted by the conscious mind, with resistance being lowered only in the context of dreams and in therapy. There is a struggle between the unconscious drives and the conscious resistance of their products, and what is moving things in the unconscious is very different from what moves us consciously. This is a very different conception of the unconscious from that which treats unconscious intentional states as essentially the same kind of thing as conscious states, just not as powerful (as Leibniz did). Looked at in this way, Freud’s original achievement was not simply to draw attention to the distinction between intentionality and consciousness—as Fodor, Carruthers and many others claim—but rather to defend a conception of the unconscious as something of a wholly different nature from consciousness. The contemporary conception of the relationship between conscious and unconscious intentionality is, in this respect, more similar to Leibniz’s than to Freud’s. Unconscious propositional attitudes represent the world in the same way that conscious ones do—by having propositional content. What makes them conscious is something else—qualia, higher-order representation, etc.—which is not anything to do with their intentional content (how they represent the world). In rejecting here this conception of the relationship between conscious and unconscious intentionality, I am therefore thinking of the unconscious in broadly Freudian rather than Leibnizian terms.

The Unity of Unconsciousness 223 My aim here, however, is not to defend Freud’s specific view of the mechanisms of the mind, with its postulated structures of id, ego and superego. I  do think that many of the phenomena psychoanalysis describes are real—at least in the sense that among the deepest human motivations are certain drives that individuals do not fully understand or recognize from simple or superficial introspection, which often result from past experience now forgotten. The aim of psychoanalytic therapy is to uncover these motivations, and whether this kind of therapy is generally effective is a question I cannot consider here. What I want to draw attention to instead is how the psychoanalytic conception of the relationship between the conscious and the unconscious—whatever its merits as a theory or as therapy—conforms to the picture I sketched in the previous section. The key point in common, again, is interpretation. The psychoanalytic process (or indeed, any therapy aimed at uncovering drives or desires which are not immediately manifest) involves the patient reflecting on things that have happened to them, or things that they have done, in order to expose the repressed desires or drives and make their behavior more intelligible to them. This essentially involves creating a narrative of their lives and revealing those ideas or drives which are repressed and play a role in the explanation of their behavior. I agree with James Hopkins, then, when he argues that psychoanalytic interpretation is, in a certain way, a process which is continuous with that of commonsense psychology (Hopkins, 1982). But this is not because psychoanalytic interpretation simply takes on board the standard philosophical conception of the propositional attitudes and their contents and applies it to the psychoanalytic unconscious. It’s rather because commonsense psychological interpretation essentially involves the kind of mixture of discovery and imposition of structure which is also there in the psychoanalytic session. Just as “self-knowledge” can involve making up your mind about what you believe, so the recovery of unconscious drives can be part of a story that you tell to make you intelligible to yourself. Your unconscious drives are part of your unconscious motivational system. Your attempts to introspect your world view can involve creation as well as discovery, and the same applies to your attempts to recover your motivations by interpretation. (Remember that I am restricting the term “world view” to your belief system; the motivational system is something distinct, although many of the same things are true of it.) My hypothesis is that cognitive science, commonsense psychology and psychoanalysis all operate with a conception of the unconscious as something that has a causal structure which is modeled by the attributions of propositional content by the theorist or the therapist. In order for these attributions to provide a successful interpretation of the unconscious mind there is no requirement that they correspond in any straightforward

224  Tim Crane sense to individual inner states with determinate propositional contents. The same is true of the subject’s own first-person reflections on their world view, and on their unconscious motivations.

5. Conclusion The standard philosophical picture of the propositional attitudes is a mixture of important truths—the functionalist truisms about dispositional profiles, the ideas of direction of fit and representational content—and some unrealistic and misguided dogmas, such as real relations to abstract propositions, the pursuit of a compositional semantics for intentional states or for sentences in a language of thought. The latter ideas have recently come under attack because of their inability to handle consciousness—for surely any plausible story about consciousness must incorporate a story about the subject’s intentional representation of the world. The way to approach this question, I  have argued here, is to abandon certain aspects of the standard picture of the propositional attitudes and to work out an alternative picture of unconscious intentional content. It should be a constraint on any such picture that it make sense of the way the unconscious is conceived across many different kinds of theory of the mind.

Note 1. This chapter derives from my 2016 Presidential Address to the Aristotelian Society and was first published in the Proceedings of the Aristotelian Society 2017. It is reprinted here with the permission of the Aristotelian Society and Oxford University Press. Three footnotes have been cut for this version. Talks based on this material have been given at the 2014 Phenomenal Intentionality conference in Dublin organized by Marta Jorba and Dermot Moran, the University of Wisconsin Madison, Harvard University, the University of California Berkeley (the 2015 George Myro Lecture) and at the University of Minho, Braga. I am grateful to participants on these occasions for their reactions, and especially to Farid Masrour, Antonia Peacocke, Susanna Siegel, Peter Simons and Elliott Sober, as well as to my PhD students Ali Boyle and Dan Brigham. I am especially grateful to Kati Farkas, Guy Longworth and Alex Moran for their very helpful written comments. Work on this chapter was supported by a grant from the John Templeton Foundation, New Directions in the Study of the Mind. Finally, thanks to Rory Madden for suggesting the title.

References Bach, Kent (1981) “An Analysis of Self-Deception” in Philosophy and Phenomenological Research, 41, 351–370. Carruthers, Peter (2016) “Higher-Order Theories of Consciousness” in The Stanford Encyclopedia of Philosophy (Fall 2016 Edition), (ed. Edward N. Zalta), http://plato.stanford.edu/archives/fall2016/entries/consciousness-higher/

The Unity of Unconsciousness 225 Cassam, Quassim (2015) Self-Knowledge for Humans, Oxford: Oxford University Press. Chalmers, David (2004) “The Representational Character of Experience” in The Future for Philosophy (ed. Brian Leiter), Oxford: Oxford University Press). Cheney, D. L. and Seyfarth, R. M. (1990) How Monkeys See the World: Inside the Mind of Another Species, Chicago: University of Chicago Press. Child, William (1996) Causality, Interpretation and the Mind, Oxford: Oxford University Press. Crane, Tim (2013) “Unconscious Belief and Conscious Thought” in Phenomenal Intentionality: New Essays (ed. Uriah Kriegel), Oxford: Oxford University Press, 156–173. Crane, Tim (2015) “The Mental States of Persons and their Brains” in Mind, Self and Person, Royal Institute of Philosophy Supplement, 76 (ed. Anthony O’Hear), Cambridge: Cambridge University Press, 253–270. Dennett, Daniel C. (1975) “Brain Writing and Mind Reading” in Minnesota Studies in the Philosophy of Science, 7, 403–415, reprinted in Dennett (1981) Brainstorms, Hassocks: Harvester. Dennett, Daniel C. (1982) “Beyond Belief” in Thought and Object (ed. Andrew Woodfield), Oxford: Oxford University Press. Dummett, Michael (1959) “Truth” in Proceedings of the Aristotelian Society, 59, 141–162. Farkas, Katalin (forthcoming) “The Boundaries of the Mind” in Routledge History of the Philosophy of Mind: The 20th Century (ed. Amy Kind), London: Routledge. Fodor, Jerry A. (1975) The Language of Thought, Hassocks: Harvester. Fodor, Jerry A. (1991) “Too Hard for our Kind of Mind” in London Review of Books, 13, June 27. Forster, E. M. (1927) Aspects of the Novel, reprint (1976) Harmondsworth: Penguin. Gardner, Sebastian (1993) Irrationality and the Philosophy of Psychoanalysis, Cambridge: Cambridge University Press. Geach, P. T. (1957) Mental Acts, New York: Humanities Press. Hampshire, Stuart (1975) Freedom of the Individual, Princeton, NJ: Princeton University Press. Hassin, Ran, Uleman, James S. and Bargh, John A. (eds.) (2005) The New Unconscious, Oxford: Oxford University Press. Hopkins, James (1982) “Introduction” in Philosophical Essays on Freud (ed. James Hopkins and Richard Wollheim), Cambridge: Cambridge University Press. Katsafanas, Paul (2016) The Nietzschean Self, Oxford: Oxford University Press. Kihlstrom, John F. (1987) “The Cognitive Unconscious” in Science, 237, 1445–1452. Leibniz, G. W. (1996) New Essays on Human Understanding, Oxford: Oxford University Press. Malcolm, Norman (1973) “Thoughtless Brutes” in Proceedings and Addresses of the American Philosophical Association, 46, 5–20. Searle, John (1983) Intentionality, Cambridge: Cambridge University Press. Searle, John (1992) The Rediscovery of the Mind, Cambridge MA: MIT Press.

226  Tim Crane Speaks, Jeffrey (2015) The Phenomenal and the Intentional, Oxford: Oxford University Press. Stich, Stephen P. (1983) From Folk Psychology to Cognitive Science, Cambridge, MA: MIT Press. Tye, Michael (1995) Ten Problems of Consciousness, Cambridge, MA: MIT Press. Wallas Graham (1926) The Art of Thought, London: Jonathan Cape. Wilson, Timothy (2002) Strangers to Ourselves, Cambridge, MA: Harvard University Press. Wittgenstein, Ludwig (1953) Philosophical Investigations, Oxford: Blackwell. Wollheim, Richard (1971) Freud, London: Fontana. Wolheim, Richard (1999) On the Emotions, New Haven: Yale University Press.

12 Phenomenal Experience and the Thesis of Revelation Michelle Liu

1. Introduction The term “revelation,” introduced by Mark Johnston (1992), is used in the debate about the metaphysics of color properties to refer to the view that the whole nature of color properties is given in color experience. This chapter is not concerned with the thesis of revelation in the philosophy of color but with a parallel idea that also goes by the name “revelation” in the philosophy of mind. This is the claim that the whole nature of qualia (phenomenal properties) is given in phenomenal experience. This latter thesis of revelation is by no means unfamiliar to philosophers of mind, although the name “revelation” has not always been used to identify it. Discussions on the idea of revelation can be traced to Descartes (1985: 215–217) in his writings on the notion of a clear and distinct idea and Hume (2000: 136) in the Treatise.1 One can also find contemporary discussions on revelation in relation to the topic of phenomenal experience and the physicalist-dualist debate about the nature of consciousness (Strawson, 1989; Lewis, 1995; Nida-Rümelin, 2007; Stoljar, 2006, 2009; Damnjanovic, 2012; Goff, 2015, 2017; Majeed, 2017; Trogdon, 2017). Two things are worth noting regarding the contemporary literature on revelation. First, revelation is often thought of as being in tension with physicalism. For instance, Lewis (1995) takes revelation to be incompatible with physicalism. More recently, Philip Goff (2015, 2017) has appealed to the thesis of revelation to argue against physicalism. Of course, whether there is an incompatibility between revelation and physicalism, and how this incompatibility is to be understood provided that there is one, depends on how we cash out revelation and physicalism (see Damnjanovic, 2012; Trogdon, 2017; Stoljar, 2018). The second thing worth noting about the contemporary literature on revelation is that revelation is often thought of as an intuitive claim. Among those who think this way about revelation are physicalists who also acknowledge the incompatibility between revelation and physicalism. (Lewis, 1995: 142; McLaughlin, 2003: 378; Braddon-Mitchell,

228  Michelle Liu 2007: 287; Hill, 2014: 199–200) For instance, Lewis (1995) and Braddon-Mitchell (2007) take revelation to be intuitive because they think that it is part of the ordinary conception of experience. Despite thinking that it has an intuitive appeal, these physicalists nevertheless deny the thesis of revelation in order to safeguard physicalism. Of course, the aforementioned claims about revelation, namely, that it serves as a crucial premise in an argument against physicalism and that it is intuitive, are points of contention in the literature. Regarding the former, one might dispute the force of the argument from revelation against physicalism. For instance, one might query whether the argument proves that all versions of physicalism are false; one might also argue that revelation is in fact implausible (see Damnjanovic, 2012; Trogdon, 2017; Stoljar, 2018). With respect to the intuitiveness of revelation, one might question whether revelation is in fact intuitive or part of the ordinary conception of experience as is often claimed (see Stoljar, 2009). In order to gain a better understanding of these debates about revelation we need, first and foremost, to clarify the idea of revelation itself. While there are frequent mentions of revelation in the philosophy of mind, as we have just seen, there is room for further discussion of what the thesis of revelation precisely amounts to. The starting point of the thesis is the claim that by having a phenomenal experience, one is in a position to introspect that experience and to form certain thoughts—make certain judgments, achieve certain knowledge—which are “revelatory” about the experience. The sense in which such thoughts are “revelatory” requires careful articulation. According to Lewis (1995), the kind of thought about phenomenal experience under consideration amounts to knowing the essence of the qualia of a given experience. A main goal of this chapter is to look into Lewis’s remarks in detail and clarify what these revelatory introspective thoughts about phenomenal experience are supposed to be. According to the formulation presented in this chapter, revelation is a thesis about qualia, which are the phenomenal properties of our experience. The thesis claims that by having an experience-token with a particular quale Q, one is in a position to form a certain thought or know a certain truth, namely, “Q is X,” where the predicate “X” captures the essence of Q, although it may be hard to put into words. Having clearly articulated the thesis of revelation, I  turn to explain how the thesis of revelation thus understood is incompatible with versions of physicalism. Overall, this chapter aims to make important clarifications that shed light on existing debates regarding revelation. The structure of the chapter is as follows. In Section  2, I  examine the thesis of revelation as presented in Lewis’s 1995 paper, “Should a Materialist Believe in Qualia?” Lewis’s paper is particularly important to the literature on revelation because ideas in this paper have given

Phenomenal Experience, Thesis of Revelation 229 rise to much discussion on the topic (see Stoljar, 2006, 2009; Majeed, 2017). Section 3 goes beyond Lewis’s remarks on revelation and puts forward a more precise formulation of the thesis of revelation. Sections  4 and 5 make two further clarifications regarding the thesis. In Section 4, I make clear that the thesis of revelation involves a knowledge-attribution de dicto regarding knowledge de dicto that a particular quale Q is X, where the predicate “X” captures the essence of Q. In Section 5, I clarify how the thesis of revelation might plausibly entail the claim that qualia are simple. Section 6 explains how revelation thus understood is incompatible with certain versions of physicalism. Section 7 concludes the chapter.

2.  Lewis on Revelation In his article “Should a Materialist Believe in Qualia?” Lewis discusses the thesis of revelation, which he calls “the Identification Thesis.” Lewis’s account of revelation is framed in terms of the notion of qualia. The term “qualia” in this context refers to the phenomenal character or the whatit-is-likeness of our experience. Qualia are phenomenal properties of particular events of experiencing. Different types of phenomenal experience are individuated by their qualia. Token experiences with the same qualia belong to the same type of experience.2 Although “quale” and its plural, “qualia,” are technical terms, the concept, as Lewis sees it, is nevertheless part of our folk psychology. As a functionalist, he (1995: 140) regards the term “qualia” as “a name for the occupants of a certain functional role that is spelled out in our tacitly known folk psychology.” According to Lewis, the folk-psychological role of qualia includes at least the following four theses: i. Qualia cause our “abilities to recognize and to imagine experiences of the same type” (Lewis, ibid.: 141); ii. Qualia are “responsible for responses of pleasure, disgust, etc.” (op. cit.); iii. Qualia are “responsible for judgements of similarity-distance, e.g., the judgement that these two colour-samples nearly match whereas those two contrast strikingly” (op. cit.); and, iv. “We identify the qualia of our experiences.” (op. cit.) While the first three components of the qualia-role are clear, the fourth one requires elaboration, especially with respect to the notion of identification. Lewis elaborates on this component of the qualia-role: Folk psychology says, I think, that we identify the qualia of our experiences. We know exactly what they are—and that in an uncommonly

230  Michelle Liu demanding and literal sense of ‘knowing what’. If I have an experience with quale Q, I know that I am having an experience with quale Q, and I will afterwards remember (unless I happen to forget) that on that occasion I had an experience with quale Q. It is by producing this identifying knowledge that a novel experience confers abilities to recognize and imagine. (op. cit.) Consider the experience of tasting the Australian spread Vegemite. In having such an experience, one might have various thoughts about the experience and come to know numerous things about it, e.g., “This experience is interesting,” “I have had this experience before,” “It is similar to the experience of tasting the British spread Marmite,” etc. Among these thoughts is, Lewis would say, the following one: “I (now) know exactly what the phenomenal character of the experience of tasting Vegemite is.” This latter knowledge is an instance of what Lewis (op. cit.) calls “the Identification Thesis”—“We know exactly what [qualia] are.” According to Lewis’s understanding of folk psychology, knowing exactly what the quale of the experience is confers one’s abilities to remember, recognize and imagine experiences of the same type. But what does Lewis mean when he says that “we know exactly what [qualia] are”? Lewis clarifies: I spoke of ‘an uncommonly demanding and literal sense of “knowing what” ’. Let me elaborate. I say that according to the Identification Thesis, the knowledge I gain by having an experience with quale Q enables me to know what Q is—identifies Q—in this sense: any possibility not ruled out by the content of my knowledge is one in which it is Q, and not any other property instead, that is the quale of my experience. Equivalently, when I have an experience with quale Q, the knowledge I thereby gain reveals the essence of Q: a property of Q such that, necessarily, Q has it and nothing else does. (Lewis, ibid.: 142) According to Lewis, by having an experience with a particular quale, we are in a position to know exactly what that quale is in the sense that we are in a position to know the essence of the quale. This is what Lewis calls “an uncommonly demanding and literal sense of ‘knowing what.’ ” This demanding sense of “knowing what a quale is” contrasts with a less demanding sense of “knowing what.” (Lewis, ibid.: 143) In everyday contexts, we can say that one knows what something is even if one does not know the essence of that thing. One knows what water is in an everyday non-scientific context insofar as one knows a cluster of descriptions about water, and knows how to identify ordinary samples of water, etc.

Phenomenal Experience, Thesis of Revelation 231 To know what water is in this less demanding everyday sense of “knowing what” does not require one to know the essence of water, i.e., know that water is H2O.

3.  The Formulation What Lewis calls “the Identification Thesis” is just the thesis of revelation (see Stoljar, 2006, 2009). As we have just seen, Lewis formulates the thesis of revelation as the following: [W]hen I  have an experience-token with quale Q, the knowledge I thereby gain reveals the essence of Q: a property of Q such that, necessarily, Q has it and nothing else does. (Lewis, 1995: 142) There are two things to notice in Lewis’s formulation of the thesis of revelation. First, Lewis is adopting a modal account of essential properties and of essence, according to which an essential property of a thing is just a necessary property of that thing, and the essence of a thing is a property that is both necessary and sufficient for being that thing. One might not want to commit to this simple modal account of essence given its well-known problems, discussed by Kit Fine (1994). Fine (ibid.: 4–5) provides several counterexamples to the modal account. For instance, Socrates has the necessary property of belonging to the singleton set that has Socrates as its only member, i.e., {Socrates}. But intuitively, it is not part of the essence of Socrates, that is, it is not an essential property of Socrates, that he belongs to the singleton set {Socrates}. Fine, for his part, opts for the real definitional account of essence, which can be traced back to Aristotle and characterizes the notion of essence on the model of definition. According to this account, “the notion of definition has application to both words and objects— . . . just as we may define a word, or say what it means, so we may define an object, or say what it is” (Fine, ibid.: 2). Here, in cashing out the thesis of revelation, we can simply take the notion of essence in the intuitive Aristotelian/Finean sense, conceived on the model of definition.3 The second thing to notice in Lewis’s formulation of the thesis of revelation is that, on the face of it, the essence of Q, according to Lewis, is a second-order property. The notion of qualia here is equivalent to the notion of phenomenal qualities or phenomenal properties. The essence of Q is then a property of a property. To know exactly what Q is, for Lewis, is to know that phenomenal property Q has the (second-order) property X where X is the essence of Q (that is, Q has X and nothing else does). Without resorting to talk of second-order properties, I  propose to understand “knowing exactly what a quale is,” i.e., knowing the essence

232  Michelle Liu of a quale, as knowing some truth.4 (This makes the task of formulating the thesis of revelation simpler and is in fact what Lewis goes on to say.)5 We can say that one knows exactly what P is if and only if one knows that “P is thus and so” where “thus and so” defines P, that is, captures the essence of P. For instance, one knows exactly what the property being triangular is if and only if one knows a truth that captures what the property (by definition and essentially) is, e.g., “Being triangular is having a three-sided closed shape.” The whole proposition is a definition of the property being triangular, stating what it is to be triangular. The predicate “having a three-sided closed shape” captures the essence of triangularity. In the same way, knowing exactly what quale Q is amounts to knowing the truth “Q is X,” where “X” is a predicate that captures Q’s essence. The thesis of revelation says that one is in a position to know this truth about Q merely on the basis of having an experience with quale Q. However, in the case of a quale, if the thesis of revelation is true, one might think that the truth “Q is X” is hard to put into words. It seems that with many experiences we find ourselves devoid of words to describe the qualia or the phenomenal characters of these experiences. Let us call the characteristic quale (phenomenal property) of experiences of red things “phenomenal redness.” If, as I  have or recall, an experience with that quale, I attempt to define what the quale phenomenal redness is, it might seem that all I am able to say is that “It is that,” using a demonstrative to refer to the phenomenal character of the phenomenal redness experience. But intuitively, the demonstrative “that” is merely a placeholder for the rich understanding of phenomenal redness I have which I am unable to put into words. The fact that I cannot put my understanding into words does not mean that I don’t know exactly what phenomenal redness is. Given our discussion, we can sharpen Lewis’s idea of revelation and formulate the thesis in the following way: (Revelation) By having an experience-token with phenomenal property Q, one is in a position to know a truth, namely, “Q is X,” where the predicate “X” captures the essence of Q, although it may be hard to put into words.6 Several things are worth pointing out with respect to the above formulation of revelation. First, the knowledge we have with respect to quale Q is a kind of knowledge-that or knowledge de dicto. The dictum is “Q is X.” More needs to be said about this knowledge de dicto. I will discuss this in the next section. Second, the knowledge de dicto “Q is X” is not just any truth about Q; it is a statement about what makes Q the property it is—the predicate “X” captures the essence of Q. More needs to be said about the notion of “capture” at issue. I  will address this in Section  5 when discussing Lewis’s argument from the thesis of revelation to the

Phenomenal Experience, Thesis of Revelation 233 simplicity of qualia. Third, the relevant knowledge about the essence of Q is obtained under a certain circumstance, namely, by having the experience. The knowledge at issue is made available in virtue of undergoing the experience alone. Fourth, the thesis of revelation does not imply that whenever one undergoes an experience with quale Q one automatically knows the essence of Q; it only implies that one is in a position to know the essence given that one is not distracted, is able to attend to the experience, etc.

4. Knowledge De Dicto Versus Knowledge De Re In the last section, we have seen that the thesis of revelation is about a kind of knowledge de dicto. This point is worth dwelling on. In this section, I shall clarify that the thesis of revelation is best understood as involving a knowledge-attribution de dicto regarding knowledge de dicto that a particular quale Q is X, where the predicate “X” captures the essence of Q. To aid our discussion, I appeal to the following two distinctions (see Broackes, 1986): 1. knowledge de dicto vs. knowledge de re; 2. knowledge-attribution de dicto vs. knowledge-attribution de re.7 (1) is a distinction between different kinds of knowledge. Knowledge de dicto consists in a relation between the knower and a certain proposition, whereas knowledge de re consists in a relation between the knower and a thing (see Sosa, 1970: 883).8 Distinction (2) is one between different types of reports of knowledge. Knowledge-attributions de re, but not knowledge-attributions de dicto, permit substitution of co-designating terms salva veritate (i.e., preserving the truth-value of the proposition). Given the two distinctions, one must not assume that knowledge de dicto is always reported with knowledge-attributions de dicto, or knowledge de re reported with knowledge-attributions de re. Which kind of knowledge-attribution to use often depends on the context in which the knowledge report is made, to whom the report is made, the purpose of the report, etc. If a thinker thinks about something, she thinks about it under some mode of presentation, e.g., thinks of Venus as Hesperus or as Phosphorus. But a report of the thinker’s thinking might not specify that mode of presentation. In general, a knowledge-attribution de dicto specifies the mode of presentation of the subject’s knowledge, whereas a knowledge-attribution de re does not. As such, if we are interested in the mode of presentation, we should report the knowledge at issue with an attribution de dicto. For instance, suppose Tom thinks of Hesperus as Hesperus and has knowledge de dicto that Hesperus appears in the evening sky. Suppose also that Tom does not think of Hesperus as Venus. In such a case, it

234  Michelle Liu would still be correct to report Tom’s knowledge in the following way: “Tom knows, of Venus, that it appears in the evening sky.”9 But this attribution de re does not specify the way Tom thinks about Venus and consequently is silent about whether or not Tom knows de dicto that Hesperus appears in the evening sky. It seems that Tom’s knowledge is best reported in the de dicto way—“best reported” in the sense that the report is as accurate and complete a guide as possible with respect to the knowledge at issue, with respect to the way in which Tom thinks about Venus (Broackes, 1986: 375). With the two distinctions above, let us return to the topic of revelation. The thesis of revelation says that by having an experience with quale Q, one is in a position to know a truth or a dictum, namely, “Q is X,” where the predicate “X” captures the essence of Q. It seems that this knowledge de dicto is accurately reported by a knowledge-attribution de dicto as below: (D) By having an experience with quale Q, S is in a position to know that Q is X, where the predicate “X” captures the essence of Q. To report it with a knowledge-attribution de re fails to completely capture the kind of knowledge at issue. A knowledge-attribution de re in this case looks something like this: (R) By having an experience with quale Q, S is in a position to know of X, where the predicate “X” captures the essence of Q, that Q is it. Compare (R) with the de re report, “Tom knows of Venus that Hesperus is it.” Suppose Tom thinks of Venus as Phosphorus and knows de dicto that Hesperus is Phosphorus. In such a case, the de re report is true. But this does not imply that Tom knows de dicto that Hesperus is Venus. The de re report is compatible with Tom having no idea what Venus is. Equally, (R), while stating that S is in a position to know de re of X, that Q is it (where the predicate “X” captures the essence of quale Q), does not imply that S is in a position to know de dicto that Q is X (where the predicate “X” captures the essence of quale Q). In general, one can have knowledge de re of the essence, E, of something, Φ, without knowing de dicto that Φ is E. Let’s agree that water is H2O, where the expression “H2O” captures the essence of water. Someone who has no knowledge of chemistry may still know de re of H2O—thought of as water—that water is it, although she does not know de dicto that water is H2O. Similarly, there could be cases of knowledge de re of X that Q is it without knowledge de dicto that Q is X. In such cases of knowledge de re, X is thought of under some mode of presentation but not as X. So, (R) is compatible, in a perfectly good sense, with S having no idea of what X is, and with S

Phenomenal Experience, Thesis of Revelation 235 having no knowledge de dicto that Q is X. In Section 6, where we cash out the incompatibility between revelation and physicalism, we will see that for a physicalist, “X,” which captures the essence of Q, is a physical/ functional predicate, and while a physicalist can endorse (R), she cannot endorse (D). To sum up, revelation is not the claim that experience merely puts one in a position to know de re of X, under some mode of presentation— not necessarily as X—that Q is it (where the predicate “X” captures the essence of quale Q) (see Stoljar, 2009). Rather, revelation is the claim that experience puts one in a position to know de dicto that Q is X (where the predicate “X” captures the essence of Q). This knowledge de dicto is best reported with a knowledge-attribution de dicto in the manner of (D). The knowledge de dicto that “Q is X” may not be such that it can readily be expressed in words, but it does involve a rich understanding of X, which is what Q essentially is.

5.  Revelation and the Simplicity of Qualia Another notion that is worth pausing on is the notion of capture in the formulation of the thesis of revelation which I have presented at the end of Section 3. Recall the formulation: (Revelation) By having an experience-token with phenomenal property Q, one is in a position to know a truth, namely, “Q is X,” where the predicate “X” captures the essence of Q, although it may be hard to put into words. The word “capture” has a number of meanings. “Capture” here should not be taken just to mean “refer to, latch onto.” It is meant to convey the idea of “representing accurately in words or pictures” as in “A vivid photograph captures the scene.”10 For instance, the truth “Being triangular is having a three-sided closed shape” captures or describes accurately what the property of being triangular is. Note that this truth describes the internal structure of the property of triangularity—having three sides and being closed. If a property P has an internal structure, assuming that a property’s structure is essential to that property, then a truth that captures the essence of property P must also describe the internal structure of P. If a truth that captures the essence of P does not describe P as having an internal structure, then P does not have an internal structure, i.e., it is simple. Recall that in our formulation of the thesis of revelation, the truth “Q is X” is supposed to describe Q’s essence. In cashing out revelation, we agreed that “X” might be difficult (even impossible) to put into words.

236  Michelle Liu In cases where Q seems indescribable, we might just say “this” (pointing inwardly to an experience of Q we have in mind). The intuition that at least in some cases we cannot say more about Q’s essence than “Q is this” arises from the fact that in these cases, in having an experience with quale Q, we do not seem to notice Q to have any structure. Given revelation and assuming again that a property’s structure is essential to that property, if quale Q had internal structure, then by having an experience with Q, one would be in a position to know that Q had that structure. The description of that structure would be given in stating that Q is X, where the predicate “X” captures the essence of quale Q.11 If there is no structure to notice with respect to Q, then Q has no structure, i.e., Q is simple. Indeed, in the following passage, Lewis remarks on the simplicity of qualia, which he takes to follow from the thesis of revelation: If we know exactly what the qualia of our experiences are, they can have no essential hidden structure—no ‘grain’—of which we remain ignorant. (If we didn’t know whether their hidden 'grain' ran this way or that, we wouldn’t know exactly what they were. Whatever we might know about them, we would not fully know their essence.) But if nothing essential about the qualia is hidden, then if they seem simple, they are simple. We may assume that if a property is structural, then it is so essentially. Then it is a consequence of the Identification Thesis that if we fail to notice structure, there is no structure there to notice. But we do fail to notice structure. So the simplicity of the qualia is a consequence of the Identification Thesis (inter alia), and so a derivative part of the folk-psychological concept of qualia. (Lewis, 1995: 142) The above passage can be formulated into the following argument: (a) Structure is part of the essence of a property. (b) By having an experience-token with phenomenal property Q, one is in a position to know a truth, namely, “Q is X,” where the predicate “X” captures the essence of Q. (c) If Q has internal structure, then by having an experience with Q, one would be in a position to know that Q has internal structure. (d) There is at least a range of cases such that in having an experience with Q, one is not in a position to know that Q has internal structure. (e) Some qualia have no internal structure (i.e., they are simple). Premise (a) seems intuitive. If a property has a certain structure, it has it essentially—a property lacking that structure would not be the same property. Premise (b) is the thesis of revelation. Premise (c) follows from Premise (a) and Premise (b). Premise (d) seems intuitively true—we just do not notice any structure in having experiences such as phenomenal

Phenomenal Experience, Thesis of Revelation 237 redness, hearing middle C, etc. (Some other qualia do seem to have structural features, e.g., the quale of a certain complex visual experience.) Conclusion (e) then follows from Premise (c) and Premise (d). So for Lewis, the thesis of revelation also entails that (some) qualia are simple (see also Adams, 1987).

6.  Revelation and Physicalism Having clarified the thesis of revelation, in this section I turn to address the primary significance of revelation in the literature, namely, its incompatibility with physicalism. As already mentioned in Section 1, how the incompatibility between revelation and physicalism is precisely understood depends on how the thesis of revelation and the doctrine of physicalism are respectively cashed out. We have formulated the thesis of revelation, but we have not clarified what physicalism precisely entails. While there is no uncontentious definition of physicalism (see Stoljar, 2010), most physicalists would agree that the phenomenal properties of conscious experiences are purely physical or functional—a physical/functional description exhausts the nature of qualia. So according to physicalism standardly conceived, the only truth that captures the essence of Q is some physical or functional truth, i.e., “Q is X,” where “X” is a physical or functional predicate, as in, for instance, “Phenomenal redness is R-fiber firing in the brain,” “Itchiness is functional role F (realized by physical properties P1, P2, P3, . . .),” etc. In cashing out the incompatibility between revelation and physicalism, I will focus on standard versions of physicalism, namely, versions according to which qualia have physical or functional essences. The structure of this section is as follows. First, I  address Lewis’s remarks on the matter. Then I turn to consider role functionalist versions of physicalism. I shall argue that role functionalism is incompatible with revelation. Finally, I lay out a formulation of the incompatibility between the thesis of revelation and standard versions of physicalism. 6.1.  Lewis, Revelation and Physicalism Lewis (1995: 142) says that the thesis of revelation is part of the folkpsychological concept of qualia and “seems obvious” (that is, intuitive). However, he rejects revelation because it is incompatible with physicalism, which he commits to. After spelling out the thesis of revelation, Lewis moves on to discuss why physicalists cannot accept revelation. He explains: If, for instance, Q is essentially the physical property of being an event of C-firing, and if I  identify the qualia of my experience in the appropriate ‘demanding and literal’ sense, I come to know that

238  Michelle Liu what is going on in me is an event of C-firing. Contrapositively: If I identify the quale of my experience in the appropriate sense, and yet know nothing of the firing of my neurons, then the quale of my experience cannot have been essentially the property of being an event of C-firing. A materialist cannot accept the Identification Thesis. If qualia are physical properties of experiences, and experiences in turn are physical events, then it is certain that we seldom, if ever, identify the qualia of our experiences. Making discoveries in neurophysiology is not so easy! (op. cit.) The idea here should be straightforward. Recall that for Lewis (1995: 142), the term “qualia” is “a name for the occupants of a certain functional role that is spelled out in our tacitly known folk psychology.” We have seen that Lewis allows that the thesis of revelation (his “identification thesis”—component (iv) in Section 2) describes one component of the qualia-role. Let Q be the phenomenal property of a pain experience (its painfulness), and let X be Q’s essence and the predicate “X” capture Q’s essence. Physicalists like Lewis take X to be a physical property that plays the functional role of quale Q. So “X” is a physical predicate, e.g., “being an event of C-fibers firing.” But if the functional role of qualia includes the thesis of revelation—that is, includes our being in a position to know the essences of qualia simply by being the subject of experiences that have the qualia among their properties—then we will not be able to find any physical properties that fulfill the role. In having an experience with quale Q, there is no physical property X such that just by having the experience alone, one can know the truth “Q is X,” where “X” is a physical predicate that captures the essence of Q. In having a pain experience, I am certainly not in a position to know de dicto that the painfulness of my experience of pain is its being an event of C-fibers firing. What this means is that no physical property can play the full qualia-role, because no physical property can play the revelation component of the qualia-role. As a result, Lewis drops the thesis of revelation from the full-blown functional definition of qualia which we saw in the beginning of Section 2. What remains, then, are components (i), (ii) and (iii). i. Qualia cause our “abilities to recognize and to imagine experiences of the same type”; ii. Qualia are “responsible for responses of pleasure, disgust, etc.”; iii. Qualia are “responsible for judgements of similarity-distance, e.g., the judgement that these two colour-samples nearly match whereas those two contrast strikingly.”

Phenomenal Experience, Thesis of Revelation 239 These remaining components constitute the nearest functional role which some physical properties can indeed satisfy. So to the question “Should a materialist believe in qualia?” Lewis responds: Yes: he should believe in imperfect but good-enough deservers of the name, occupants of the part of the folk-psychological role we get by leaving out the Identification Thesis [i.e., revelation]. And no: he should not believe in perfect deservers of the name, occupants of the entire role. (Lewis, ibid.: 142–143) As a physicalist, Lewis denies that there is anything that perfectly satisfies our folk-psychological conception of qualia. But he proposes that physical properties of physical events (such as the property of being an event of C-fibers firing) can deserve—imperfectly, but well enough—the name “qualia.” 6.2.  Revelation and Functionalism As we saw, Lewis is a functionalist who takes qualia to be occupants of functional roles spelled out in folk psychology. As we also saw, Lewis subscribes to what is known as “the identity theory,” according to which qualia are just physical properties such as being an event of C-fibers firing. Lewis is thus a realizer functionalist, who identifies qualia with physical properties that realize the corresponding functional roles (see also Lewis, 1966).12 Realizer functionalism is distinguished from role functionalism, which identifies qualia with functional role properties that may be multiply realized by different physical properties playing the same functional role (e.g., in differently constituted creatures). In our discussion on Lewis, we saw that revelation is incompatible with realizer functionalist physicalism and the claim that qualia have physical essences. More needs to be said about whether revelation is incompatible with role functionalist physicalism and the claim that qualia have functional essences.13 In the remainder of this section, by “functionalism,” I  simply mean role functionalism. Note that different versions of functionalism cash out the notion of a functional role differently. According to scientific functionalism or psychofunctionalism, in the terminology of Block (1978), functional roles are cashed out in the vocabulary of cognitive science or neuroscience. According to what is known as “commonsense functionalism,” functional roles are cashed out in the non-technical, everyday vocabulary of commonsense psychology. It is straightforward that revelation is incompatible with psychofunctionalism. In having an experience with a particular quale Q, there is no functional property X cashed out in technical,

240  Michelle Liu scientific vocabulary such that just by having the experience alone, one can know that Q is X. It is not so straightforward that the thesis of revelation is incompatible with commonsense functionalism. All else equal, being in a position to know a piece of everyday commonsense seems more likely than being in a position to know something technical and scientific. A commonsense functionalist would say that “X” is a functional predicate couched in the non-technical vocabulary of commonsense psychology. She might also insist that one is indeed in a position to know such a functional truth “Q is X” by having an experience with quale Q.14 However, I shall argue that revelation is incompatible with commonsense functionalism.15 In cashing out the thesis of revelation, we mentioned the idea that the essence-capturing truth “Q is X” may be hard to put into words. This is supposed to capture the intuitive idea that phenomenal experiences have indescribable phenomenology.16 Now commonsense functionalists claim that qualia can be fully defined by their functional roles. One might think that intuitively, such functional roles do not seem to capture the “je ne sais quoi (I know not what)” character of qualia.17 But if the essence of a quale is functional (or physical), then it can be put into words. Commonsense functionalists would, of course, deny that any qualia are indescribable and insist that if one states the full functional role of a quale in the everyday vocabulary of commonsense psychology, then one has successfully given a definition of what that quale is essentially. In spelling out the incompatibility between revelation and commonsense functionalism, I shall grant that the thesis of revelation does not entail, but merely allows, that the essence-capturing truth “Q is X” is hard to put into words.18 I shall also set aside the general concern that, intuitively, a state’s having a certain functional role does not guarantee that the state have any phenomenology at all. In order to satisfy the thesis of revelation, the commonsense functionalist would have to say that by having a pain experience with phenomenal property Q (where Q is the painfulness of pain), a subject is in a position to know the commonsense functional role of Q. That is, the subject is in a position to achieve de dicto knowledge of the form “Q is X,” where “X” specifies the commonsense functional role of painfulness, which captures the essence of painfulness. I shall argue that in having a pain experience with phenomenal property Q, one is not in a position to know the commonsense functional role of Q; one is not in a position to know de dicto such a truth “Q is X,” where the predicate “X” captures the essence of Q in commonsense psychological terms. Commonsense functionalists define qualia in terms of their functional roles by drawing on what is common knowledge about mental states with these qualia (Braddon-Mitchell and Jackson, 2007: 52). According

Phenomenal Experience, Thesis of Revelation 241 to these functionalists, this common knowledge has three parts: “There are input clauses—clauses that say what sorts of events cause mental states in people; output clauses—clauses that say what sort of behaviours are caused by mental states; and internal role clauses—clauses that describe the internal interactions of mental states” (Braddon-Mitchell and Jackson, ibid.: 48). The input clauses for quale pain would include the clause “bodily damage causes experiences with quale pain”; output clauses would include the clause “experiences with quale pain cause bodily movements that relieve pain and minimize damage”; internal clauses would plausibly include the clause “experiences with quale pain cause desires to relieve pain, which typically lead to desires to take painkillers (if one also believes that taking painkillers would be likely to relieve the pain)” (Braddon-Mitchell and Jackson, ibid., 52–53). Giving a definition of what quale pain is in commonsense functional terms requires specifying all three kinds of clauses. As we saw in Section  3, according to the thesis of revelation, the knowledge de dicto that Q is X, where the predicate “X” captures the essence of Q, is gained in virtue of having the experience alone. Let us agree that one is in a position to know the tripartite commonsense functional role of quale pain. Still, it seems implausible that it is in virtue of having a pain experience alone that one is in a position to have such knowledge, which, according to commonsense functionalists, captures the essence of quale pain. Consider, first, the internal clauses which say that the painfulness of pain causes a desire to relieve pain and motivation toward actions that the subject believes would make the painfulness of pain more likely to stop. The details of such actions and motivations (to take painkillers, for example), are unlikely to be known just in virtue of having a painful experience. If “causing the desire to take painkillers” is built into the internal clauses that specify part of the functional role of quale pain, then the functional role will not be revelation compatible, as we shall say. It is just not true that in virtue of having a pain experience alone, one is in a position to know that quale pain is X where the predicate “X” includes “painkillers.” Of course, the commonsense functionalist has a ready response to this point. They should say that what is core and primary in the internal clauses of the functional role of quale pain is nothing specifically about painkillers but just that the painfulness of pain causes the desire that it stop. Consider, second, the input clauses of the functional role of quale pain which say that bodily damage causes experiences with quale pain. It does not seem to be the case that in virtue of having a pain experience, that is, in virtue of feeling pain, I am in a position to know that it is bodily damage that causes my experience to have a particular phenomenal character. Nothing about bodily damage is revealed to one in having pain experiences like headaches, stomach aches, cramps, etc. Even in cases

242  Michelle Liu where one feels pain in one’s finger as a result of its being cut, having the sensation of pain in one’s finger does not on its own reveal that one has a cut on one’s finger which gives rise to the sensation. According to proponents of commonsense functionalism, the input clauses that form part of the definition of quale pain explicitly mention “bodily damage” (Braddon-Mitchell and Jackson, ibid.: 52). But a functional role with such input clauses will not be revelation compatible. In having the phenomenal experience of pain alone one is not typically in a position to know that quale pain is X where the predicate “X” includes “bodily damage.” Now we can grant that there are commonsense clauses about quale pain that one is in a position to know in virtue of having a painful experience, including the output clause “the experience with quale pain causes one to wince, to have reflexive movements of avoidance or withdrawal,” the internal clause “the experience with quale pain causes one to want the experience to stop” and perhaps others. But it is far from obvious that the clauses of this reduced, revelation-compatible functional role specification are adequate to capture the essence of quale pain. Specifically, these functional role clauses might not distinguish the painfulness of pain from other qualia. For example, the phenomenal properties of experiences such as biting into a slice of lemon, seeing something extremely bright, hearing a loud fire alarm going off, etc., all typically cause one to wince and to have reflexive movements of avoidance or withdrawal. In all these cases, the subject also typically desires the relevant experience to stop. But while these experiences are examples of distress, they are typically not considered as experiences of pain—one would not usually say that one is in pain if one has one of these experiences. The claim that commonsense role functionalism is compatible with the thesis of revelation clearly faces a dilemma. If a functional role specification is to capture the essence of quale pain, then it must individuate quale pain, in the sense that no other quale has that functional role. A functional role specification that includes enough of what is commonly known about the role of pain to meet this individuation requirement is plausibly not revelation compatible. And a functional role specification that is sufficiently reduced to be revelation compatible is plausibly no longer individuative. This pattern certainly generalizes beyond the example of the painfulness of pain. Consider how a commonsense functionalist would spell out the functional roles of the phenomenal properties of various color experiences. In his paper “Naming the Colours,” Lewis (1997) mentions the following two clauses as being central to the commonsense functional roles of worldly colors and corresponding properties of color experiences: D1 Red is the surface property of things which typically causes experience of red in people who have such things before their eyes.

Phenomenal Experience, Thesis of Revelation 243 D2 Experience of red is the inner state of people which is the typical effect of having red things before the eyes. (Lewis, ibid.: 327) D1 and D2 inter-define red and experience of red. But exactly similar clauses inter-define green and experience of green. So the putative functional role of the pair does not individuate that pair. The pair is very far from being the only pair such that X is the surface property of things which typically causes Y in people who have such things before their eyes and Y is the inner state of people which is the typical effect of having X things before the eyes. In order to break out of the circle of inter-definition and distinguish experience of red from other color experiences, D2, which is the input clause for experience of red, needs to be augmented with something about common examples of worldly surfaces that are red, and so (according to D1) cause experience of red. But it seems that having an experience of red does not, by itself, put the subject in a position to know about such examples. For instance, one might define experience of red as being “typically caused by the color of a British pillar box” (Lewis, ibid.: 335). But surely, in having an experience of red alone, one is not in a position to know anything about British pillar boxes nor in a position to know that phenomenal redness is typically caused by seeing a British pillar box.19 This second example helps to sharpen the dilemma facing commonsense functionalists who claim that their position is compatible with the thesis of revelation. If commonsense functionalists define qualia in functional terms by appealing to everyday concepts of worldly objects and states like “painkillers” or “bodily damage” in the case of quale pain and “British pillar box” in the case of phenomenal redness, then commonsense functionalism is incompatible with revelation. But if commonsense functionalists avoid making references to these things in spelling out the commonsense functional roles of qualia, then it seems no longer clear that, with only the vocabulary that would not create a problem for maintaining revelation, they can give adequate functional definitions that individuate qualia. It does not seem that the reduced functional definition of a given quale Q, which the subject is in a position to know by having an experience with quale Q, is fine-grained enough to distinguish Q from other qualia.20 Given this dilemma, our conclusion must be that commonsense role functionalism—like realizer functionalism (and the identity theory) and scientific role functionalism (psychofunctionalism)—is incompatible with the thesis of revelation. 6.3.  The Argument From Revelation Against Physicalism At this point, it would be useful to formally lay out the incompatibility between the thesis of revelation and physicalism. Here I will present the

244  Michelle Liu incompatibility between the two in the format of an argument, namely, an argument from the thesis of revelation to the falsity of physicalism. (Here physicalism is understood as entailing the claim that qualia have physical or functional essences. If there are positions that do not entail this claim, but are nonetheless worthy of the name “physicalism,” then whether those versions of physicalism are incompatible with revelation is a separate question.)21 1. If S has an experience-token with quale Q, then S is in a position to know de dicto that “Q is X,” where the predicate “X” captures the essence of Q. 2. If physicalism is true, then qualia have physical/functional essences. 3. If qualia have physical/functional essences, then by having an experience-­token with quale Q, S should be in a position to know de dicto that “Q is X,” where “X” is a physical/functional predicate which captures the essence of Q. 4. It is not true that by having an experience-token with quale Q, S is in a position to know de dicto that “Q is X,” where “X” is a physical/ functional predicate which captures the essence of Q. 5. Physicalism is false. Premise (1) is entailed by the thesis of revelation. Premise (2) states what the essences of qualia would be if physicalism were true. Premise (3) follows from Premise (1) and Premise (2). Premise (4) is supported by the considerations discussed in Sections 6.1 and 6.2. Premise (5) then concludes from Premise (3) and Premise (4) that physicalism is false. Given our clarification of the distinction between knowledge de dicto and knowledge de re of qualia in Section 4, it is worth noting that physicalists, in denying revelation, can still say that by having an experience with quale Q, one is in a position to know de re of X, where the predicate “X” captures the essence of Q, that Q is it (see Lewis, 1995: 143). A  physicalist might say that there are two distinct modes of presentation of the essence of the phenomenal property Q—a physical/functional mode of presentation and a phenomenal mode of presentation, and that the mode of presentation that reveals the essence of Q is the physical/ functional one (see Loar, 1997: 603). So “X,” which captures the essence of Q, is a physical/functional predicate. Such a physicalist can say that by having an experience with quale Q, one is in a position to know de re of physical/functional property X (but under its phenomenal mode of presentation) that Q is it, e.g., know de re of the property of being an event of C-fibers firing that the painfulness of a pain experience is that property.22 This is analogous to the case of Tom’s knowledge of Venus. Suppose again that Tom thinks of Venus as Phosphorus and not as Venus and that Tom knows de dicto that Hesperus is Phosphorus. In such a

Phenomenal Experience, Thesis of Revelation 245 case, we can still say “Tom knows of Venus that Hesperus is it.” Tom has knowledge de re of Venus (even without any idea what Venus is) insofar as he thinks of Venus as Phosphorus. One can also see that the doctrine of the simplicity of qualia, which we discussed in Section  5, is in tension with physicalism. If qualia are physical or functional properties, which are presumably not simple, then qualia cannot be simple. Hence, physicalists also ought to reject the idea that qualia are simple (see Lewis, 1995: 142). In Section 5, we saw that there is an argument from revelation to the conclusion that some qualia are simple. The argument crucially relies on the premise that there is at least a range of cases such that in having an experience with Q one is not in a position to know that Q has internal structure. A commonsense functionalist who claims their position to be compatible with revelation is likely to reject this premise and claim that in all cases, by having an experience with Q, one is indeed in a position to know the complex structure of Q as a result of being in a position to know the complex commonsense functional role of Q.23 This allows such a commonsense functionalist to reject the entailment from the thesis of revelation to the claim that some qualia are simple. But as we saw, it is unclear that commonsense functionalism is compatible with revelation in the first place.

7. Conclusion In this chapter we have offered a detailed discussion on the thesis of revelation and its incompatibility with physicalism. Much of our discussion focused on Lewis’s remarks on revelation. Despite Lewis’s illuminating discussion in his 1995 paper, it is only more recently that the thesis of revelation has been brought up explicitly in relation to the discussion on whether physicalism is true (see Stoljar, 2006, 2009; Majeed, 2017; Trogdon, 2017; Goff, 2017).24 But in the literature, as already mentioned, revelation and physicalism are not always thought to be incompatible (see Damnjanovic, 2012; Stoljar, 2018). Of course, the issue here depends on how one cashes out the thesis of revelation and how one formulates physicalism. We have just seen that the thesis of revelation, as we have formulated it here, is indeed incompatible with standard versions of physicalism, which take qualia to have physical/functional essences. Whether the argument from revelation against physicalism is compelling depends, of course, on whether the thesis of revelation itself is compelling, an issue that we have not touched on in this chapter. At the beginning of the chapter I mentioned that revelation is often thought of as intuitive, even by avowed physicalists. But the claim that revelation is intuitive is not without contention (see Stoljar, 2009). Whether revelation is intuitive, and in what sense it is intuitive if it is, are issues that await further discussion. I am inclined to think that revelation is intuitive in the

246  Michelle Liu sense that it is part of our ordinary conception of experience and that this is evident from the way we talk about experiences.25 Even if revelation turns out to be an intuitive claim, whether it is in fact true is a separate question. There are few direct arguments against revelation in the literature.26 Physicalists like Lewis, who acknowledge the intuitiveness of revelation, end up denying it simply because of its incompatibility with physicalism. Further discussion on the plausibility of revelation is thus called for.27

Notes 1. Both can be read as endorsing versions of revelation. Descartes (1985: 215, 217) thought that upon a careful reflection, we are capable of having clear and distinct ideas of experiences such as sensations. To have a clear and distinct idea of something, according to the standard interpretation of Descartes, is to know, at least to some extent, what belongs and does not belong to the essence of that thing (see Gewirth, 1943; Patterson, 2008). One might also read Hume as advocating a thesis of revelation. Hume (2000: 9, 136) thought that impressions, which are roughly feelings, including “sensations, passions and emotions,” “appear, all of them, in their true colours” and “must necessarily appear in every particular what they are, and be what they appear.” 2. Of course, any experience-token belongs to many experience-types, of varying degrees of specificity. When we talk about experience-tokens of the same experience-type, we rely on context to indicate the relevant degree of specificity. 3. Revelation is about knowing the essence of phenomenal properties. Although in his original paper Fine was mainly concerned with individual-essence, one can extend the Finean conception of essence to apply to type-essence. 4. Here knowledge-what is understood as a kind of knowledge-that. This is the orthodox approach to knowledge-what in the literature (see Stanley and Williamson, 2001). Knowing what x is at least involves knowing a proposition which is the answer to the question “What is x?” 5. Immediately after the passage quoted above, Lewis writes: If, for instance, Q is essentially the physical property of being an event of C-firing, and if I identify the qualia of my experience in the appropriate ‘demanding and literal’ sense, I come to know that what is going on in me is an event of C-firing. Contrapositively: If I identify the quale of my experience in the appropriate sense, and yet know nothing of the firing of my neurons, then the quale of my experience cannot have been essentially the property of being an event of C-firing. (Lewis, 1995: 142) Note that being an event of C-firing is, strictly speaking, not a property of the quale Q which is the qualitative character of pain experience, i.e., quale pain. The property being an event of C-firing is more appropriately understood as a property of pain-experience-token itself. 6. On some accounts, our rich understanding of phenomenal properties is embodied in a demonstrative-like, or recognition-based, phenomenal concept (e.g., Chalmers, 2010). The thesis of revelation, in recent literature, has been formulated via the notion of phenomenal concepts (see Goff, 2015, 2017; Trogdon, 2017). The present formulation does not depend on the

Phenomenal Experience, Thesis of Revelation 247 notion of phenomenal concepts and captures the intuitive idea behind the thesis. 7. The same distinctions apply to other propositional attitudes such as belief, desire, etc. 8. Knowledge de dicto and knowledge de re are not mutually exclusive. For instance, I know that the sun will rise tomorrow. I not only have knowledge with respect to a dictum, namely, “The sun will rise tomorrow” but I also have knowledge with respect to a res, namely, the sun. 9. In having knowledge de dicto that “Hesperus appears in the evening sky,” Tom also has knowledge de re of Hesperus/Venus. Such knowledge de re can also be reported with a knowledge-attribution de dicto—in this case, namely, “Tom knows that Hesperus appears in the evening sky.” 10. According to Brian Loar (1997), the word “capture” is ambiguous. Loar (ibid.: 603) points out two different senses in saying that “a concept captures the essence of the referent.” He writes: These are equivocating uses of “capture the essence of.” On one use, it expresses a referential notion that comes to no more than “directly rigidly designate.” On the other, it means something like “be conceptually interderivable with some theoretical predicate that reveals the internal structure of” the designated property. But the first does not imply the second (op. cit.). For Loar, who is an a posteriori physicalist, a phenomenal concept can be said to capture the essence of the phenomenal property in the sense that it directly refers to the referent, but it does not tell us anything about the nature of the referent. This is not how I have intended to use the term “capture.” On the other hand, a physicalist like Loar would say that a physical concept captures the essence of the phenomenal property, in the sense that it tells us what the property really is in a way that reveals the nature of the property. This is the sense of “capture” at issue here. The thesis of revelation says that the truth “Q is X,” which one knows through having an experience of the type Q and which is hard to put into words, is a truth that captures, in this second sense, the essence of Q. 11. It seems that whenever we notice our phenomenology to have structures, we can describe the phenomenology, including its internal structure, to some extent which goes beyond using a mere demonstrative “this.” We might say things like: “There is a bit of phenomenal redness on the left of my visual field. Next to it, there is some phenomenal blueness,” etc. From this, it does not follow that a complex, structured experience can be fully captured in a verbal description. The simple constituent phenomenal properties will likely not be captured in words without resorting to demonstratives and/or metaphors. 12. Lewis is also an a priori physicalist, who holds the view that the conditional “If P then Q,” where P stands for the totality of physical truths and Q stands for a given phenomenal truth, is true a priori. One can also be an a posteriori physicalist and hold the view that “If P then Q” is only true a posteriori. Many proponents of a posteriori physicalism explicitly reject revelation (see Loar, 1997; McLaughlin, 2001, 2003; Papineau, 2007; Balog, 2012). Like Lewis, they reject it simply because it renders physicalism false. 13. Both identity theorists and realizer functionalists would agree that phenomenal properties have physical essences. In contrast, role functionalists take phenomenal properties to have functional essences. 14. I am indebted to Frank Jackson for pressing me on this issue. 15. There are problems with commonsense functionalism that are independent of the thesis of revelation (see Block, 1978).

248  Michelle Liu 16. The idea that qualia are indescribable can be traced to Descartes, who thinks that sensations have an “I-know-not-what” (“je ne sais quelle” in French) aspect (see Cottingham, 2012). For a recent discussion on the indescribable character of experience, see Camp (2006). 17. This, I  think, points to the common objection against functionalism, that knowing what the functional role of pain is does not seem to amount to knowing what pain feels like, and the essence of the painfulness of pain does not seem to be exhausted by its functional role. 18. The idea that mere words are unable to communicate the essence of phenomenal property Q is not officially part of the thesis of revelation. In cashing out revelation, I merely left open the possibility that the essence-capturing truth “Q is X” may be hard to put into words. As we have seen, the reason that the identity theory or realizer functionalist physicalism is incompatible with the thesis of revelation is not that revelation requires the essences of qualia to be indescribable—it does not explicitly require that. 19. It is interesting to note that in his paper “Naming the Colours,” Lewis (1997: 338–389) discusses the idea that adding revelation to the functional definition of color experience would help individuate experience of red and distinguish it from other color experiences. However, he rejects this approach because he takes revelation to be false by materialistic lights. 20. Block (1978: 300) raises a similar worry that common knowledge can fail to distinguish two sensations that feel different. 21. For instance, one might argue that physicalism defined as a grounding thesis does not entail that qualia have physical/functional essences (see Stoljar, 2018). Whether revelation is incompatible with grounding physicalism, i.e., physicalism cashed out as a grounding thesis, requires separate treatment. 22. A physicalist can say that the essence of Q, which is a physical/functional property, can be thought about in a physical/functional way, i.e., as X, or in a phenomenal way, as, say, Y. Such a physicalist can say that by having an experience with quale Q, one is in a position to know de dicto that Q is Y. Note that “Y,” which is a phenomenal predicate, refers to but does not capture the essence of Q. 23. Note that such a commonsense functionalist operates on a broad conception of qualia where the functional roles of qualia form at least part of their essences. One might say that on a narrow conception of qualia which treats functional roles as being separate from the essences of qualia, we do not notice some qualia to have structures in having the relevant experiences. 24. One does find something similar to the thesis of revelation being used as an objection against the phenomenal concept strategy (see Levine, 2001: 84; Nida-Rümelin, 2007; Goff, 2011, 2015). This is the worry that the phenomenal concept strategy cannot account for the idea that phenomenal concepts afford us substantive knowledge about the nature of phenomenal properties. For replies to this challenge from a posteriori physicalists, see Schroer (2010), Díaz-León (2014), Elpidorou (2016), and Taylor (2018). 25. In an unpublished paper, I  put forward an argument for the intuitiveness of revelation which takes the form of an inference to the best explanation. I contend that sentences such as “I know what pain feels like, but I don’t know what the feeling of pain really is” are odd, and such oddity is best explained by the hypothesis that the thesis of revelation is part of our ordinary conception of experience. Given this inference to the best explanation, revelation is part of our ordinary conception of experience and is thus intuitive.

Phenomenal Experience, Thesis of Revelation 249 26. Stoljar (2018), in his recent review of Goff’s book Consciousness and Fundamental Reality, has raised an argument against the thesis of revelation. 27. I am very grateful to Gonzalo Rodriguez-Pereyra, Bill Child and especially Martin Davies for their helpful comments on drafts of this chapter.

References Adams, Robert Merrihew (1987) “Flavors, Colors, and God” in The Virtue of Faith and Other Essays in Philosophical Theology, Oxford: Oxford University Press, 243–262. Balog, Katalin (2012) “Acquaintance and the Mind-Body Problem” in New Perspectives on Type Identity: The Mental and the Physical (ed. S. Gozzano and C. S. Hill), Cambridge: Cambridge University Press, 16–42. Block, Ned (1978) “Troubles with Functionalism” in Minnesota Studies in the Philosophy of Science, 9, 261–325. Braddon-Mitchell, David (2007) “Against Ontologically Emergent Consciousness” in Contemporary Debates in Philosophy of Mind (ed. Brian McLaughlin and Jonathan Cohen), Oxford: Blackwell, 287–299. Braddon-Mitchell, David, and Frank Jackson (2007) The Philosophy of Mind and Cognition, Oxford: Blackwell. Broackes, Justin (1986) “Belief De Re and De Dicto” in The Philosophical Quarterly, 36 (144), 374–383. Camp, Elisabeth (2006) “Metaphor and That Certain ‘Je Ne Sais Quoi’ ” in Philosophical Studies, 129 (1), 1–25. Chalmers, David, J. (2010) “The Content of Phenomenal Concepts” in The Character of Consciousness, Oxford: Oxford University Press, 251–276. Cottingham, John (2012) “Intentionality or Phenomenology? Descartes and the Objects of Thought” in History of the Mind-Body Problem (ed. Tim Crane and Sarah Patterson), London: Routledge, 139–155. Damnjanovic, Nic (2012) “Revelation and Physicalism” in Dialéctica, 66 (1), 69–91. Descartes, René (1985) The Philosophical Writings of Descartes, Volume 1. (transl. John Cottingham, Robert Stoothoff and Dugald Murdoch), Cambridge: Cambridge University Press. Diaz-Leon, Esa (2014) “Do A Posteriori Physicalists Get Our Phenomenal Concepts Wrong?” in Ratio, 27 (1), 547–562. Elpidorou, Andreas (2016) “A  Posteriori Physicalism and Introspection” in Pacific Philosophical Quarterly, 97 (4), 474–500. Fine, Kit (1994) “Essence and Modality” in Philosophical Perspectives, 8, 1–16. Gewirth, Alan (1943) “Clearness and Distinctness in Descartes” in Philosophy, 18 (69), 17–36. Goff, Philip (2011) “A  Posteriori Physicalists Get Our Phenomenal Concepts Wrong” in Australasian Journal of Philosophy, 89 (2), 191–209. Goff, Philip (2015) “Real Acquaintance and Physicalism” in Phenomenal Qualities: Sense, Perception, and Consciousness (ed. P. Coates and S. Coleman), Oxford: Oxford University Press, 121–143. Goff, Philip (2017) Consciousness and Fundamental Reality, Oxford University Press.

250  Michelle Liu Hill, Christopher S (2014) Meaning, Mind, and Knowledge, Oxford: Oxford University Press. Hume, David (2000) A Treatise of Human Nature, (ed. David Fate Norton and Mary J. Norton), Oxford and New York: Oxford University Press. Johnston, Mark (1992) “How to Speak of the Colors” in Philosophical Studies, 68 (3), 221–263. Levine, Joseph (2001) Purple Haze: The Puzzle of Consciousness, Oxford University Press. Lewis, David (1966) “An Argument for the Identity Theory” in The Journal of Philosophy, 63 (1), 17–25. Lewis, David (1995) “Should a Materialist Believe in Qualia?” in Australasian Journal of Philosophy, 73 (1), 140–144. Lewis, David (1997) “Naming the Colours” in Australian Journal of Philosophy, 75 (3), 325–342. Loar, Brian (1990/97) “Phenomenal States” in Philosophical Perspectives, 4, 81–108; with revisions in The Nature of Consciousness (ed. N. Block, O. Flanagan and G. Güzeldere), Massachusetts: MIT Press, 597–616. Majeed, Raamy (2017) “Ramseyan Humility: The Response from Revelation and Panpsychism” in Canadian Journal of Philosophy, 47 (1), 75–96. McLaughlin, Brian (2001) “In Defense of New Wave Materialism” in Physicalism and its Discontents (ed. C. Gillett and B. Loewer), Cambridge: Cambridge University Press, 319–330. McLaughlin, Brian (2003) “Color, Consciousness, and Color Consciousness” in Consciousness: New Philosophical Perspectives (ed. Q. Smith and A. Jokic), Oxford: Oxford University Press, 97–154. Nida-Rümelin, Martine (2007) “Grasping Phenomenal Properties” in Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (ed. Torin Alter and Sven Walter), Oxford: Oxford University Press, 307–336. Papineau, David (2007) “Phenomenal and Perceptual Concepts” in Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism (ed. T. Alter and S. Walter), Oxford: Oxford University Press, 111–144. Patterson, Sarah (2008) “Clear and Distinct Perception” in A Companion to Descartes (ed. J. Broughton and J. Carriero), Oxford: Blackwell, 216–234. Schroer, Robert (2010) “Where’s the Beef? Phenomenal Concepts as Both Demonstrative and Substantial” in Australasian Journal of Philosophy, 88 (3), 505–522. Sosa, Ernst (1970) “Propositional Attitudes de dicto and de re” in Journal of Philosophy, 67, 883–896. Stanley, Jason and Williamson, Timothy (2001) “Knowing How” in Journal of Philosophy, 98 (8), 411–444. Stoljar, Daniel (2006) Ignorance and Imagination: The Epistemic Origin of the Problem of Consciousness, Oxford: Oxford University Press. Stoljar, Daniel (2009) “The Argument from Revelation” in Conceptual Analysis and Philosophical Naturalism (ed. D. Braddon-Mitchell and R. Nola), Cambridge: MIT Press, 113–138. Stoljar, Daniel (2010) Physicalism, London: Routledge. Stoljar, Daniel (2018) Review of Consciousness and Fundamental Reality, by Philip Goff, Notre Dame Philosophical Reviews, 9 February, 2018.

Phenomenal Experience, Thesis of Revelation 251 Strawson, Galen (1989) “Red and ‘Red’ ” in Synthese, 78 (2), 193–232. Taylor, Henry (2018) “Powerful Qualities, Phenomenal Concepts, and the New Challenge to Physicalism” in Australasian Journal of Philosophy, 96 (1), 53–66. Trogdon, Kelly (2017) “Revelation and Physicalism” in Synthese, 194 (7), 2345–2366.

Section III

Perception and Cognition in Aesthetics

13 Would You Buy Absence Art? Anya Farennikova

1.  Going, Going, Gone At a recent auction, Banksy’s painting sold for 1.4 million dollars.1 The painting depicted a girl with a balloon. As soon as the painting was sold, it self-destructed with a built-in shredder. The transformed piece consisted of half of the canvas hanging in shreds and an exposed chunk of cardboard. The news agencies report that the bidder decided to go through with the purchase. She accidentally bought a piece of absence art, but she hasn’t changed her mind now that she is aware of the nature of her purchase. The twenty-first century is not the first time that artists took a stab at absence art. Absence art began to boom in the 1950s and 1960s, with Rauschenberg, Eva Hesse and Warhol in the visual arts, Wright and Perec in literature, John Cage in music and Abramovic in performance art. Absences were painted and sculpted, installed and performed, splashed and cut. Their main purpose was straightforward: To put a dent in a theory of art by challenging the convention of what’s essential to art. Since then, absence art has grown in scope and courage. Its original message still stands strong: Art does not need material objects or any kind of objects at all. But during its growth, absence art also made an important discovery. While removing matter from art, absence artists came to see that absences aren’t just a tool for rebellion. They can be aesthetic objects and can have aesthetic value in their own right. Absence art transformed into a major international movement, with exhibitions around the world dedicated to the themes of disappearance, freedom and loss. A theme that emerges in many of these exhibitions is that absences can be genuine objects of wonder; they are not just as a method for art sabotage. With its inexhaustible capacity to subvert and surprise, absence art seems like a genuine and perhaps ultimate form of art. But it’s time for us to be completely honest. Absence art also triggers a strange feeling. It sometimes feels like a fraud.

256  Anya Farennikova Perhaps a part of this feeling has to do with how easy it is to create absence art. One can follow these instructions. Step one: Take any existing artwork or an ordinary object. Step two: Remove something from them and display it as art. That’s all it takes to create absence art. Absence art can be uninspiring and cheap, but the main problem is not that it’s so easily bad. The problem is that it may not be real. Absence art might be a fake. This suspicion gets manifest by a subtle sense we sometimes get when we view absence art. We’ve all been there. You go to a museum or a show and come across some work of absence art. You read the description on its wall label, study the catalogue and you learn what the absence symbolizes, interrogates and questions. You also learn how the absence has been crafted to accomplish these acts. It’s time to cut through this art-theoretic fog. An absence art, at bottom, is an art of nothing. And it’s hard to find reality in that. So, despite what our best art theory or wall labels declare, a nothing cannot be art. This chapter is an attempt to have a dialogue about the courage and the fraud of absence art. I’ll set up this dialogue by asking the following question: What is it that we buy, when we buy absence art? Understandably, this question may trigger surprise. What does monetary value have to do with the nature of absence art? Perhaps it’s better just to set up standard arguments in philosophy of art. That would be good, except we are dealing with a particularly serious challenge here. Art-theoretic forces are especially strong in the case of absence art. The rhetoric they create clouds absence art. Putting money on the line might help cut through the fog. This method is not new. There is a parallel test for the certainty of one’s belief in decision theory.2 There is a view that the best way to know what somebody believes is not by asking the person to recite her views; it’s by asking how much she is willing to bet on her beliefs. Betting is a litmus test of the strength of one’s beliefs and buying absence art is a litmus test of one’s real views about the nature of absence art. If you proclaim belief in astrology but are not willing to bet on your next horoscope, then you don’t really believe in astrology. Similarly, if something is holding you back from investing in absence art, then you don’t believe that absence art is real, regardless of how much art theory you are willing to cite. One of the goals of this chapter is to reveal a major problem with absence art. The second goal is to show that this problem can be removed by using the resources from perception theory. I’ll demonstrate how one can appeal to contemporary accounts of perception that can support the claim that absences can be aesthetic objects which can trigger aesthetic experience. In doing so, I hope to accomplish another goal. By showing that the problem with absence art can be solved using the resources in perception theory, I  want to advertise the idea that philosophy of perception and

Would You Buy Absence Art? 257 philosophy of art should have a dialogue more frequently (and as equals). Just as art theory can learn from theories of perception,3 philosophy of perception can learn from art. Dewey was right—we can understand experience better by understanding the dynamics in our appreciation of art. It’s time that we apply this principle to perception. I’ll show how we can understand perceptual experience better if we understand how it engages with absence art. These are the goals and here is the plan. In Section 2, I discuss central features of absence art. I explain how absence art differs from ephemeral art and reduce all the varieties of absence art to three main kinds. In Section 3, I show that absence art faces a major philosophical problem and outline two arguments against the reality of absence art. In Section  4, I  show that we can use the resources in perception theory to block a philosophical threat to absence art. In Section 5, I explain that theories of perception too can learn some important lessons from absence art. I conclude in Section 6 by discussing a hidden function of absence art. This function gives a surprising and final answer to the question, what it is that we buy, when we buy absence art?

2.  What’s Up for Purchase? 2.1  The Right Kind of Ephemera Raqib Shaw has a monumental artwork titled “Absence of God III.”4 The multimedia piece is a crimson circle executed in “acrylic, glitter, enamel and rhinestones.” The circle depicts warriors, flowers, birds and beasts. The mass of objects splashed on crimson is overwhelming. Here is a curious thing. It’s hard to see the absence of God in the acrylic, glitter and all the works. Perhaps this is the point. The artwork shows that God’s absence cannot be experienced directly: We can only see the consequences of God’s absence on the world. There is another possibility. The absence isn’t visible because this isn’t a work of absence art. Consider what the artwork depicts. It strikes us with a presence of objects. The crimson circle is about the aftermath of an absence. It’s about what’s left. If it’s the overwhelming presence that’s at the center of the artistic focus, then “Absence of God III” may not be a piece of absence art. Controversy is the bread and butter of absence art and there is also controversy in what should count as absence art. John Cage’s piece is famously cited as a piece of absence art, but a closer read of his explanation suggests that it’s more about what fills the gap, than about what’s gone. The label “absence art” gets thrown around generously. Absence art is often linked to the ephemeral: Inconstant and fragile objects, such as mirrors, light or a fog. It’s also associated with art that has an abnormal or

258  Anya Farennikova deviant subject. For example, conceptual art introduces mutations into ordinary objects. Due to its deviant subject matter, it’s often regarded as absence art. It seems that anything goes as absence art. That could be part of its nature. Absences call for radical freedom: They push boundaries and can let anything in. This seems to be their social function. However, this function can come apart from their role as objects of art. Ephemeral and deviant art may be associated with the same function: Of pushing the boundaries of art. However, they can do so by different means. They can put an unusual presence, not an absence, at the center of a display. If their focus is on what’s present and not on what’s missing, then it’s fair to question whether they constitute absence art. This highlights the need to have more clarity about what counts as absence art. I’ll start by outlining key differences between absence art and forms of ephemeral and deviant art. After that, I’ll look at the variation within absence art and discuss its two principal forms. Let’s begin with ephemeral art. Ephemeral art involves inconstant or fragile objects. They can be sheer, shifty and quick to dissolve. Here are some examples: A see-through crystal cube, a pillar of fog recreated in a room, a hallway of mirrors, or light tracing something on the wall. Ephemeral art uses unusual materials in order to challenge the conception of a suitable art medium or to highlight fragility of art. However, it can do so in a way that attracts attention to the positive qualities of an object and not to what it lacks. For example, a fog sculpture could represent impermanence and therefore count as absence art. But it can also just emphasize the unusual nature of the medium by showcasing its positive qualities: e.g., beautiful swirls of smoke. The same distinction applies to deviant art. An object can look unusual or strange because it’s missing some part and this absence can be in the center of focus. However, it can also look deviant because of what’s been added to it. The general principle is this: If nothing seems missing in an unusual object (deviant art) or in a normal object executed in unusual media (ephemeral art), then it’s probably not a work of absence art. The third contrast is more nuanced and it’s with memory art. Memory art is a popular subject of exhibitions and explores themes of grief, invisibility or loss. It can be a wall with lists of people who disappeared during a war or videos of buildings being destroyed. It can also be nonpolitical: It can be a sentimental photograph of something that’s lost or made obsolete. Intuitively, memory art seems like a paradigmatic case of absence art: If any art is about an absence, this one certainly is. However, as with ephemeral or deviant art, we need to consider what’s actually in focus. Memory art is often about an absence of an object, but it can also be about remembering an object that’s gone. It can use memory to honor a certain object or to bring it back to life.

Would You Buy Absence Art? 259 This doesn’t mean that memory art can’t be simultaneously about a presence and an absence. And this illustrates a more general point. Absence art is a complex phenomenon and its distinction from other forms of art won’t be clear-cut. Ephemeral art can use elements of absence art: Imagine a room that gets flooded by a fog, which dissipates, leaving the room vacant for some time. Absence art, in turn, can contain elements of ephemeral art: It can use unusual media or memory artifacts as means to “point” to what’s gone. And all these forms of art can play with each other in one complex experience. Let’s pause and summarize key points. A lot of contemporary art deals with non-traditional media, contexts and themes, and I cautioned against automatic subsumption of these forms of art to absence art. This leads us to a natural question. What counts as absence art? The following two criteria are relevant. First, the artwork features an absence: The absence is part of content of that work of art. Second, the absence is displayed in a prominent way: It’s not an accidental or peripheral feature and is central to that work of art. Artworks that meet these two criteria count as absence art. In order to deepen our understanding of absence art, let’s review some of its key forms. 2.2  Categories of Absence Art When considering absence art, it’s easy to get dazzled by its trickery and various forms of sabotage. Still, I think that we can reduce all the variation in absence art into two main forms. I’ll review them in turn. A lot of absence art works by giving a comment on traditional art. It plays with the concept of art by removing things. For example, it can remove presentational media from the art piece, it can place it in a new context, strip away the art piece from its presentational media, or do some damage to the artwork. And, as we’ve seen, absence art can go to an extreme and present us with nothing at all: An empty gallery5 or a canceled performance / show. I’ll call this extreme form of art meta-absence art. It comes in two basic types. The first type is radical absence art. This form of art offers a complete vacuum in place of art. When you enter an empty gallery, there is nothing to be seen—not even empty frames. The second type is less radical: It involves partial presence. Display materials are there, but they are missing the actual art. This can be a frame without a canvas, a pedestal, a blank video tape, an empty computer file, or a non-performing actor, standing idly instead of doing the show. I’ll refer to this form as partial absence art. What’s in common between radical and partial forms of absence art is their function: They make a comment about the nature of art. There are

260  Anya Farennikova two more extreme forms of absence art that can be ascribed to “metaabsence-art” (though less straightforwardly so). The first form is “stripped” art. The artwork is present but missing presentational materials: For example, a frame, a showcase or a wall label. The second is related. Artworks in this form of art have all classical elements of art: Display materials, media, content. However, the artworks have been intentionally damaged. We can call this form “disrupted” absence art. Meta-absence-art issues a powerful statement. It rebels against convention by using a method that couldn’t be more effective. But, as we’ve noted before, rebellion is not all there is to absence art. Absence art isn’t just in the business of criticizing other kinds of art. It can see an absence as an aesthetic object in its own right. Like other forms of art, it attempts to capture reality—what’s there, in the world. And some of the most important objects that we experience in the world are absences—losses, gaps and voids. This brings us to the second large category of absence art, which I call representational absence art. Representational absence art features an artwork, traditionally understood (a painting, a sculpture or an installation piece), which features6 an absence as part of a depicted object or a scene. For example, it can be a drawing of a mother grieving loss of a child, a sculpture of a table missing one of its legs. An example from the Renaissance would be a painting of a scene where the disciples discover that Jesus is not in the tomb. We’ve just reviewed various kinds of absence art. This by no means captures its full range. However, the usefulness of the exercise was that we were able to identify two main forms of absence art. The first is metaabsence art, where the artwork, by including an absence, makes a comment about traditional art (with absences being outside the content of that traditional artwork). The second is representational absence art, where absences are part of the content of an artwork, by being ascribed to objects or scenes that these artworks depict. Meta-absence art, in turn, has at least these three main forms: Radical absence art (nothing is present); vacated art (a blank drawing board) and disrupted absence art (damaged artworks and “stripped” art with missing presentational materials). These two forms of absence art can be distinguished by their function. It’s quite easy to see why the second form—representational art— counts as art. In this form of art, absences seem to be an obvious part of the aesthetic content of an artwork. But what makes meta-absence art something we can call art? In this form, absences are outside of classically understood aesthetic content. They are what’s done to art and don’t themselves seem to be part of art. Here is one obvious yet unpopular answer. The value of meta-absence art lies in its function: To raise questions about the nature of art. By

Would You Buy Absence Art? 261 removing objects from traditional art, radical, partial and disrupted absence art asks about how little is enough. In making these comments, they themselves become art. Here, then, is the general schema for this type of defense of meta-absence-art: If Artistic Object X Makes a Comment About the Nature of Art, Then X Is Art Part can be understood as a literal material part (as in sculpture or installation art), or a (re)presentation of an item that’s part of content of the artwork. The challenge for applying this scheme to absence art is to explain how absences can play a role of such a part, given that they are a lack of a part, or a nothing.7 There is also a less obvious way to argue that meta-absence art is art. And that’s by arguing that the absence in invisible artworks can function as a traditional aesthetic object. It can produce feelings of pleasure and it can look graceful or grand. In the case of Lana Newstrom’s empty gallery, the absences look attractive because they were beautifully lit up. These are standard aesthetic properties and it seems that we can ascribe them to absence art. If we can do so for meta-absence art, then this should be enough to vindicate its status. As we are about to see, this move also faces a problem and not just as part of a defense of meta-absence art. Just as it’s difficult to show how an absence can be part of some artistic meta-content, it’s difficult to show how an absence can be part of ordinary aesthetic content in traditional art. I would now like to show why the seemingly least problematic kind of absence art is in just as much trouble as its other forms.

3.  Nothing to Value 3.1.  On Not Learning Your Lessons You are invited to an exclusive gallery opening. A famous artist will be showing new work to a small group of people. The show starts at midnight and goes on until morning with an opportunity to purchase exhibited art. You arrive on time and pick up a brochure detailing the art in two exhibition rooms. When you enter the first room, you feel confused. There are empty frames stacked in the corners and wall tags describing art. But the actual artworks are absent. You check if the show got delayed but you’re assured that it’s on. You proceed to the next room and feel more confused. The room is supposed to showcase the centerpiece—a glowing, giant multimedia cube. But the room is totally empty. There is nothing there, not even the wall tags.

262  Anya Farennikova The bell is rung and everyone is ushered for a talk. The artist enters. She explains that she’s been exploring the theme of absence and that she interrogates this theme in this show with different kinds of media. She also notes that the large multimedia piece is absent in the second room but importantly, so is she. These two absences are her centerpiece. The talk triggers a lot of discussion and many preorder her work. You get intrigued by the idea of owning a piece of absence art and buy one of the empty frames. It’s time to raise some questions. Let’s start with this simple point. Usually, when we buy a piece of art, we buy something that we value aesthetically. But what is this something in the case of absence art? An absence in the center of a frame? What is the object that you value aesthetically? Here is a worry. Some artist pronounced a certain object “a work of absence.” That artwork is, presumably, what you’ve bought. But let’s face it. No one can really buy (or own) an absence. All you’ve got is a simple frame. One might wonder if only radical forms of absence art feel absurd. Perhaps other forms of absence art will do better. So, let’s see if this is the case. The saga continues. Thrilled by your recent purchase, you decide to expand your collection of absence art. You find just the right piece and place an order. Your partner didn’t love your previous purchase, but you think that this one might be a success. Unlike the empty frame, the artwork here is present and looks more like real art. Here is how it’s described: The artwork consists of two spheres positioned on a display. The first sphere is 30 cm, made of obsidian stone and rotates clockwise. The second sphere is 90 cm, gold and rotates counterclockwise. The only catch is that the second sphere is absent. The display houses only the first one. There is enough room for that second sphere, but you can’t tell if it was ever there. There is just some empty space. Your purchase finally arrives. You unwrap the box, call up your friends and set up the artwork. A couple of your friends declare it to be a beautiful piece of absence art. Most argue that it’s just a display with a sphere and it’s ridiculous to think that it contains anything else. Whatever it is, they insist, you didn’t buy “an absence” of a sphere. There is a bit of space on the display, but that’s just it, a space. Your confidence begins to wane. The more you look at the piece you’ve purchased, the less it looks like absence art. It’s just a display with a sphere. The piece still has some aesthetic value. But with the absence now missing, it begins to look a bit sad. 3.2.  How to Devalue Absence Art The purpose of the art collector story was to do two things. First, it was to underscore the sense in which absence art can appear unreal or absurd.

Would You Buy Absence Art? 263 Second, it was to show that this feeling of absurdity isn’t just an artifact of radical absence art. Once we inspect representational absence art more closely, it will also look absurd. My next goal is to try to articulate the underlying problem that these feelings highlight. There are two ways to back up the intuition that absence art isn’t real or is a sham. I will go through each argument and then review some responses that attempt to justify the idea of absence art. I’ll start with an argument that directly tackles the unreality of absences. The Metaphysical Argument Against Absence Art 1. Buying an artwork involves buying something that has aesthetic value; 2. Something can contribute to aesthetic value only if it exists; 3. Absences do not exist—they are a lack of something that exists; 4. Therefore, an absence cannot contribute to aesthetic value of an artwork; 5. Therefore, aesthetic value of absence art does not lie in an absence but in something else. This argument points to the fact that aesthetic appreciation, including other relations, such as buying or owning, are relations to existing things. “Thing” here is to be understood very broadly: A physical substance, a fictional entity or something abstract. Absences are lacks of all those things. They are not even fictions. They aren’t there in any sense. Absences cannot be owned and they cannot be valued, for the simple reason that they don’t exist. There’s much to criticize in this argument. But let’s not lose sight of its central point. Relations (buying, valuing, owning) require relata and it’s missing in our case. So when we are relating to an “absence” in absence art, we are in fact relating to something else. I’ll let this point sit. In the meantime, let’s consider the problem from a less abstract angle. Let’s take a look at the mechanics of art appreciation: What processing it requires and the role of the senses in that process. When we engage with art, it offers us something that we can relate to with the senses. When we hear a poem or view a piece of installation art, the senses detect a material object. An intrinsic part of that process is that they pick up on its form. The information about the form triggers a valenced assessment: We feel good about symmetric or circular objects and dislike edgier ones. With further processing, positively or negatively valenced perceptions of objects transform into aesthetic judgments. With more processing, these judgments become conscious and then turn into a complex, evolving experience of a work of art. The key point about this account is that higher-level experiences of aesthetic properties are grounded in lower-level sensory states. This grounding relation is missing in the case of absence art.8 The process we’ve just

264  Anya Farennikova described can’t happen with absences. To start, there is nothing for the senses to “grab” when they encounter an absence. For this reason, the senses cannot detect its form. This, in turn, means that the senses cannot use affect to judge the object to be positive or negative. With these core pieces of information missing, no aesthetic properties can be computed or inferred. This analysis highlights that absences cannot initiate an important stage in aesthetic process: Sensory experience. This allows us to formulate the second argument against absence art. Perceptual Argument Against Absence Art 1. Buying an artwork involves buying something that has aesthetic value; 2. Something can have aesthetic value only if it can elicit aesthetic experience; 3. Something can elicit aesthetic experience only if it can be perceived; 4. Absences cannot be perceived; 5. Therefore, absences cannot elicit aesthetic experiences; 6. Therefore, absences cannot contribute to the aesthetic value of an artwork; 7. Therefore, aesthetic value of absence art does not lie in the absence but in something else. As with the first argument, there’s a lot to criticize here. But let me flag what matters most. The argument is too demanding. Premise (2) places a strong requirement on what it means to be art; Premise (3) is too demanding about aesthetic experience.9 These requirements are often dropped in contemporary theories of art. I propose we keep both premises. Substituting perception with experience construed more broadly in Premise (3) won’t help our case. The problem of the senses having nothing to latch on to happens for all experience—­whatever mechanisms generate experience, they need something to latch on to. Premise (2) might be false as a general theory of art. However, there’s benefit in trying to show that absence art can count as art even on the strictest notion of the aesthetic. Loosening the criteria will make us miss something important about absence art: Its capacity to have classic aesthetic properties. We’ve just considered two arguments against absence art. The first argument highlights a metaphysical requirement on the relation of aesthetic appreciation and finds trouble with its relata. The second focuses on psychological implementation of aesthetic appreciation and finds trouble with its core stage. Let’s return to your apartment and the missing sphere. You want to bring back the magic, so you recite the mantra: “The artwork features

Would You Buy Absence Art? 265 the absence of a sphere and questions the nature of installation art.” But after we’ve done the anatomy of absence art, the mantra feels like a smokescreen. There is nothing there to ground your aesthetic experience. There is no absence art. What, then, do you buy when you buy absence art? You can’t buy an absence. So you are buying something else. 3.3.  Something Else Engaging with absence art is a bit like watching a magic show. We feel mesmerized by the craft that went into creating the magic. At the same time, we’re aware that the effect is not real. We’re being tricked. The arguments above revealed that we’re being tricked by absence art. Still, absence art can be a worthy investment. We’re happy to pay for a magic show even though magic isn’t real. Given that, it would be good to know where the money goes in absence art. If absences aren’t real, then what is the object of our investment? Let’s look at some ideas. The first option is that you invest into a material object. You buy a display with a sphere or a physical carrier, such as an empty frame. These items make your purchase real, because they involve something concrete. However, absence art seems to involve more than a chunk of wood. Compare the “missing sphere” artwork with an identical work that’s just about a single sphere. There seems to be an extra item in the first. What is it? This brings us to the second option. When you buy an empty frame, you invest into an idea. You invest into thoughts and feelings that the physical object stirs in you. On this view, absence art is fundamentally a mental object: Thoughts and feelings are psychological states. This, obviously, can’t be the whole explanation. Absence art isn’t just a set of mental states. When we buy absence art, we want to invest into what those mental states are about. In the case of the empty frame or the absent sphere, it’s something in the world. Here, then, is an improved interpretation: You invest in meaning. Absence art delivers a message, which can express various things. Meanings are neither straightforwardly physical, nor fully in the mind; they are somewhere between the mind and the world, so perhaps these transcendent entities can capture what’s going on in absence art. The problem is that it’s not clear which of whose meanings are bought. First, any artwork can mean an absence: An artist can take any object and just stipulate that. More importantly, we lack an account of what meaning is about, in the case of absence art. If absences don’t exist, then they can’t confer meaning on absence art. Let’s try a more encompassing option. When you invest into absence art, you purchase an experience. Buying absence art is like buying an Ayahuasca trip. What you pay for, in the end, is not some material object,

266  Anya Farennikova it’s what the object does to you—the whole thing. Buying an experience means getting something more global than individual feelings and thoughts. The holistic aspect is nice, but there is still a problem. Absence art can have a life-altering effect. But which object has this effect, if not an absence? What triggers that global experience? Let’s switch gears. One could object that this whole way of thinking about absence art is wrong. Absence art is not a mental or physical object or a fuse of the two. Its value lies in what has been done to an object: An action or an event. On this view, investing into absence art is investing into the past: An action toward an object (removing a sphere). Extending the view, it can also be about a future action: Degradation of an object, as in Going, going, gone. However, this explanation won’t fit paradigmatic cases of absence art. Certain artworks of absence exist because something hasn’t been done: Not painted, sculpted or performed. We won’t find the relevant absence events in their future or past. This brings us to the final option: Craft. Absence art, like most art, is a product of skill. Works of absence art have been designed and manufactured and there is craft in doing that. This craft is responsible for the capacity of absence art to cause ideas, thoughts and experiences. When you invest in absence art, you recognize that craft. Based on the previous explanation, craft includes making something but also not making. There is craft in leaving out. Still, this doesn’t explain absence art. What, exactly, is the craft if there is no absence? What is being manufactured in absence art that can explain its effect? Let’s review. We just went through seven interpretations of the value of absence art. The benefit is that there is nothing absurd about ideas on that list. People have already bought these kinds of items: A shirt owned by Justin Bieber, an online experience, an idea produced by somebody else. So it won’t be strange to use these ideas to explain why we go for absence art. The drawback is that every idea failed to account for the value of absence art by severing its connection to an absence. The core value of absence art must be an absence. It cannot not reside in something else.

4.  Absences as Objects of Perception 4.1  The Way Forward Let’s review the situation. Absence art elicits a lot of admiration and is generally regarded as a legitimate form of art. In fact, it grew so successful that people are now willing to invest in absence art.10 At the same time, we recognize that there is something unreal about absence art. There are two arguments to back up this feeling. The metaphysical argument states that absences can’t stand as objects in a relation,

Would You Buy Absence Art? 267 including the relation of aesthetic appreciation. The perceptual argument states that absences can’t provide suitable grounding for aesthetic judgments required for the appreciation of art. There is, however, one powerful piece of evidence against the impossibility of absence art: Experience. We know from experience that absences can be real. We know what it’s like to come to an empty home, witness devastation or to lose somebody we love. We know how these absences look, sound and feel. Absences can be real and irrevocably so. If absences are real, then they can confer reality on absence art. How, then, do we defend this inference against the two charges? What is this object we call “an absence” and how do we establish that it can be art? Two strategies naturally suggest themselves. One would be to seek further theoretical backing. There is likely to be a view in the philosophy of art which can render absences valid aesthetic objects and legitimate instances of art. Another strategy is to press for a pluralistic explanation that accords more with the practice of absence art. We should give up the search for a single grand theory. The interpretations of absence art attempted earlier (feelings, experiences or craft) may have failed individually, but they could succeed as a group. There is promise in both strategies. However, they ignore something important. They do not address fundamental tension we’ve identified in absence art: That absence art is a powerful tool for presenting an absence and that there is nothing in an absence for one’s experience to latch on to. The pluralistic approach sidesteps the first. Art theorizing can’t help with the second. So, while they may give some validation to those who want to invest in absence art, they are unsatisfactory as a theory of absence art. I propose to explore a different strategy. Instead of going deep into a theory of art, I suggest that we instead go deeper into a theory of perception. The perceptual argument identified a problem with absence art: Absences cannot be perceived. I propose to reevaluate this premise within the framework of contemporary theories of perception.11 My prediction is that once we see how these theories conceive of the function of perception, we will have good reasons to reject the pursuit of a theory of art. Let’s start with a bit of background first. There used to be a fairly popular view in philosophy of perception, on which perception had a very limited job. Its job was to detect basic sensory features, such as colors and edges and to deliver this information to higher-level cognitive systems. The job of cognition was to transform this information into a meaningful representation of the world by applying conceptual categories to the outputs of the perceptual stream. Where do absences fall on this view? They belong to cognition. Absences are not basic sensory features. Like meanings, they must be computed at a later stage. Their computation requires drawing inferences

268  Anya Farennikova from the perceptual output. Just as cognition infers that these colors and shapes mean “water,” it also infers that these colors and shapes mean that the water is gone. This approach to perception, obviously, is of no help to absence art. It implies that we cannot perceive absences, not in the natural environment, nor in art. Fortunately for absence art, perception research has moved forward and this approach to perception is no longer a default view. The emerging picture is that we should think differently about the function of perception. We find more and more evidence that the role of perception goes well beyond detecting rudimentary sensory features. It’s an intelligent system which uses stored knowledge to make sense of the world.12 This new view of perception holds promise both for the perception of absence and for absence art. I will now walk us through the basic steps. The first point involves thinking more carefully about the function of perception. Perceptual systems evolved in conditions of high uncertainty. They had to deal with the ambiguity and complexity of the world.13 Adapting to such an environment involved learning how to get by with low-quality sensory data. The solution was to make use of stored knowledge to produce reliable representations of the world. Given this pressure to be intelligent, sense-making wasn’t something that perception could afford to outsource to a slower process like thought. In sum, on contemporary approaches to perception, perception’s job is not to report colors and shapes, it’s to generate meaningful representations of the world. Perception is a meaning-making machine which maximizes its resources to make sense of what it observes. Here is the next step. We extend its sense-making function to absences. There is evolutionary value in perceiving absences in the world. Absences should be something that we are not able just to think about, but to observe. This brings us back to absence art. On earlier approaches to perception, the absence of a sphere cannot be seen. However, a different picture emerges from contemporary approaches to perception. Absences are the kinds of things that we should perceive. This gives us the motivation for perception of absence. We now need the know-how. How does perception register absences? I think that the basic mechanism in this process is a mismatch.14 I will briefly describe this mechanism and then pause to evaluate what it implies for absence art. When we engage with the world, we systematically form expectations. Many expectations are formed unconsciously because we automatically track patterns in the world. Others are initiated by our interests or needs: The desire to find shelter or food. Expectations, in turn, trigger searches: We start to look for an object we’ve come to want or expect. We search for an object by using object “templates”: Representations that enable an efficient search. Absence representations are generated when the perceptual system registers a mismatch between the template and what is being observed.

Would You Buy Absence Art? 269 On this picture, mismatches are the main vehicle that deliver absence content. There are two additional tools. Selective attention “objectifies” a region of space viewed through a mismatch, by transforming it into a cohesive unit: An absence object. Expectations make that object meaningful. Because you expect to see a sphere, you’ll see the absence of a sphere and not a cube. This, in brief, is the process of how the mind discovers absences in the world. Plausibly, the same process occurs when we view works of art. Anticipation of an artwork or viewing it directly will trigger certain expectations. Some will arise out of one’s knowledge of the genre or prior exposure to similar pieces of art. Others will be triggered automatically by an exposure to a pattern. For example, you might notice that one light is missing when you walk through an exhibition of neon lights. These expectations will activate a template of an object you’re looking for (consciously or not) and match it against the relevant bits of an artwork. Once a mismatch is registered, then perception, together with attention and expectations, will generate an experience of absence. This absencedetection mechanism happens when we view art in other media: Sculpture, music or performance art. It’s time to summarize. I’ve just presented an argument based on contemporary accounts of perception, that absences can be perceived. If absences can be perceived, then experiences of absence can provide suitable perceptual grounding for aesthetic experiences of absences presented in an artwork. Absences can be art. However, the defense of absence art is not yet complete. The account I’ve offered makes it possible for absences to generate aesthetic experiences and become aesthetic objects, but it does not explain how they do so. We have an explanation for how we can perceive an absence in a sculpture or a painting, but no explanation of how we can perceive it aesthetically. There is a gap between perception of an absence and viewing it as art. 4.2  The Aesthetic Bridge There is a missing step in the argument I’ve offered. The perceptual account of absences eliminates an important obstacle from absence art, but it does not explain how absences become art: What processes transform them into aesthetic objects. What is this extra step? Absences can look graceful or sublime. How do absences acquire these properties? How do we move from perceiving an absence to viewing it as art? My hypothesis is that there is a process working alongside mismatches that fulfills this role. This process is “inheritance”: Absences acquire aesthetic properties by inheriting those properties from other objects. I’ll now explain how this process works and return to evaluate two arguments against absence art.

270  Anya Farennikova To move absences from perceptual content into aesthetic content, we need to explain how absences acquire aesthetic properties. According to the inheritance proposal, this process unfolds in two steps. Step 1: Absences borrow perceptual features from other objects. Step 2: Absences inherit aesthetic properties associated with the borrowed perceptual features. Let’s unpack. Perceptual features are borrowed from two sources. One source is the world: Absences can take on properties of the space where they’re perceived to be situated. For example, an absence located in a red-lit room can itself look red. The second source is internal: Absences can take on properties of missing objects represented in object templates. For example, if a missing object was large, its absence itself can look large. This covers the perceptual stage of the inheritance process. The next step is aesthetic processing of borrowed perceptual features. My hypothesis is this. Certain perceptual properties (and their combinations) are automatically associated with certain aesthetic attributes: For example, a large size might be associated with looking sublime.15 So, by borrowing a certain perceptual feature, an absence will automatically inherit its associated aesthetic attribute: For example, vast absences in nature or art are likely to look sublime.16 To summarize, the inheritance hypothesis makes two claims: (a) absences inherit perceptual features from their environment and missing objects, and (b) this borrowing automatically triggers an attribution of the associated aesthetic features to absences. A quick note about the scope of this explanation before we move on. First, the inheritance hypothesis is not designed to explain all aesthetic of absence art. While I think that inheritance is one of the core aesthetic mechanisms in absence art, there are other interesting and important mechanisms in absence art worth studying.17 Second, inheritance can turn an absence into an aesthetic object, but that may not be enough to transform it into art. Radical absence art is a case in point. One way to defend this form of art is by filing it under conceptual art. It’s art because of the message it carries. But we now have an entirely new kind of defense. An absence featured in radical absence art is an aesthetic object. It’s art because of how it looks. Radical absence art can take on classical aesthetic properties: For example, the absence in Newstrom’s beautifully lit vacant gallery looks festive and delicate. For those who can’t stomach conceptual art explanations, this allows for a traditional defense of the most radical form of art. Still, one might resist this judgment about Newstrom’s gallery. Having aesthetic looks is not enough: An empty gallery may look aesthetically pleasing but it isn’t art. With these qualifications in places, here is the full defense of absence art. Absence art can generate perceptual experiences of absences by engaging a mismatch mechanism. These experiences can lead to aesthetic

Would You Buy Absence Art? 271 responses by engaging an inheritance mechanism. By using mismatches, we can perceive absences depicted in paintings, drawings, in music and installation art. By using the inheritance mechanism, we can appreciate their aesthetic qualities. Thanks to this dual process, absences can take on traditional aesthetic features and appear beautiful or sublime. Welcome formal absence art. We are now ready to tackle earlier arguments against absence art. The perceptual argument is no longer a threat. Against its two core premises, I’ve argued that absences can be perceived and I’ve also outlined the pathway for how they take on aesthetic properties. What about the claim in the ontological argument that absences don’t exist? The perceptual account I’ve offered helps block this argument as well. At minimum, my defense implies that absences can exist as perceptual objects. This leaves it open whether absences are constructed by the mind or discovered in the world. Regardless of the choice, if absences are perceptual objects, then they can function as objects in relations that we care about: Aesthetic experience and aesthetic appreciation. With the two arguments out of the way, we can finally answer the central question of the chapter: What is it that we buy, when we buy absence art? My account implies that we buy something real. However, what that object is depends on the form of absence art. In many cases of absence art, we buy a perceptual object: An absence located in a physical carrier. In case of purely conceptual absence art, we invest into an abstract object: An absence suggested or symbolized by some physical object or a description. This completes my defense of absence art. I’ve demonstrated that we can draw on perception theory to help verify its status as genuine art. I would now like to summarize broader lessons for the philosophy of art from perception theory and then explain what perception theory itself can learn from absence art.

5.  Dewey’s Expansion 5.1  Perception Illuminates Art The account I’ve defended offers a psychological explanation of absence art. It doesn’t push for the broadening of the notion of the aesthetic, nor does it ask for a philosophical revision of what constitutes an experience of art. It respects the traditional notions and offers an explication of absence art by the appeal to two psychological mechanisms: Mismatches which deliver perceptibility for absences and inheritance which confers on absences aesthetic looks. There are two direct benefits from applying perception theory to the analysis of absence art. First, perceptual theory helps validate the status

272  Anya Farennikova of absence art as genuine art. If absences can be perceived, then they can become art by generating aesthetic experiences. This confirms the reality of absence art. Second, perceptual theory brings to the fore an important form of absence art. If absences can be experienced perceptually, then their status as art doesn’t have to depend on the message they carry. It can be a function of how absences look. This implies that absence art isn’t all conceptual. There can be perceptual absence art. These points concern how we theorize about absence art. I would now like to outline some lessons for the philosophy of art more generally. One of the biggest lessons that the philosophy of art can draw from perception theory is that perception is intelligent. Its intelligence makes it possible for us to experience the world and it’s also deployed in our appreciation of art. Art might trigger raw emotions, but these emotions rest on an experience that’s fundamentally smart: A product of a system that’s trying to make sense of the world. The second point is related. Any art that seems to require a degree of knowledge for its appreciation shouldn’t be analyzed as necessarily involving high-level thought. Much of this knowledge is picked up at the level of perception. Art can be smart and not by means of being conceptual art. The final lesson concerns a formalist movement in the philosophy of art. Our investigation of the mechanisms of absence art implied that formal art, contrary to its defenders, is not fundamentally an art of presence.18 There can be form to an absence. There can be formal absence art.19 Significant form can inhere in ephemeral objects. Thanks to the inheritance mechanism, it isn’t just a product of the concrete. This gives a glimpse of how much philosophy of art can profit from engaging with the more current, scientifically grounded approaches to perception in psychology and cognitive science. It’s time to turn the tables. Perception theory itself has a great deal to learn from art. I’ll conclude the chapter by outlining some lessons from art for the accounts of perceptual experience. I’ll recruit Dewey to help me with that. 5.2  Art Illuminates Perception Dewey is famous for his works on pragmatism, but he has also done important work on the notion of experience as a psychological state. The core of Dewey’s approach is to isolate a general mental state he calls “an experience” and then show how it can be illuminated by being modeled after experiences we have when we engage with nature or art. In “Art as Experience,” Dewey argues that we can understand our general experiences better by seeing them through the lens of how certain aesthetic experiences unfold.20 When we view works of art, our

Would You Buy Absence Art? 273 experience starts at the inception phase, then it grows, culminates and ends once it reaches its full potential. This dynamic, according to Dewey, isn’t proprietary to the appreciation of art. It applies to certain kinds of ordinary experiences. When we interact with the world, the experiences we token aren’t just a series of static images: Representations of objects or scenes. These experiences are not static. They are unfolding “living” things. I should clarify an important point about this approach. Dewey’s approach to experience isn’t a proposal to increase what we take to be the minimal length of an experience. The call isn’t to adjust the timeframe of an experience; it’s about recognizing the unity of experience through its evolution. To adopt an aesthetic model is to recognize that experiences are complex because they unfold through stages. It’s also to recognize their depth. As the object is slowly revealed and changes with every shift of attention, its meaning also undergoes various shifts. Dewey proposes to use this aesthetic model as a lens through which to see other significant experiences in our lives. The implications are substantive. Once we learn to identify significant experiences, then we won’t be thrown into them mindlessly. We can start developing the craft of having an experience and turn having experiences into art. I’ll refer to Dewey’s aesthetic model of understanding experience as the “aesthetic lens.” I’d now like to show how the application of Dewey’s aesthetic model can help illuminate experiences that occur within perception. Absence art yields an interesting aesthetic lens and it can reveal important lessons about ordinary perception. One of the core lessons from absence art is that absences don’t have to be analyzed solely as abstract ideas. Absence art reveals absences as objects of experience: As units that we can explore with our senses. This goes against a tendency in philosophy of mind to view absences solely in terms of a logical operation of negation: As abstract facts in logical space.21 Absence art counters this approach by highlighting the role of absences as objects of the senses. This carries an important lesson for the philosophy of perception: Materiality does not matter for objecthood. We’ve already learned from the gestalt theory of perception about the dominance of form over matter in our perception of objects. Absences push this idea further. They are instances of form without the matter. Absences lack matter, yet they can be “objectified” and perceived as units and even have significant form. This suggests a new role for aesthetic attention.22 A piece of absence art can look like a painting of presence if one fails to direct attention to the painted surfaces in the right way. Aesthetic attention is critical for the discovery of absences in art. The next point also relates to attention. Absence art reveals the fragility of absences as objects of perception. In ordinary life, absences have a kind of slip. As soon as we notice

274  Anya Farennikova an absence, that experience often quickly dissolves. The same is true of art. Art manages to capture an absence, but the absence is still a volatile thing. It requires sustained aesthetic attention for its looks not to break. Absence art can help us see more clearly a general feature of perception: That fragility is a general feature of perceptual presence.23 Perceived physical objects constantly “break” with shifts of attention. Their looks get destabilized, their meaning changes and they disappear into the background. Alva Noë called this dynamic the “dance” of perception.24 This dance is hidden for ordinary physical objects, but it’s made palpable in absence art. By appreciating the fragility of absences in the context of art, we can learn to see how much ordinary physical objects “dance” in perception. The glue that brings it all together is expectations. Expectations play a critical role in our appreciation of art.25 They also have an important role to play in perception.26 Consider again the point of fragility that absence art highlights. But what explains fragility? We can view fragility in terms of surprise: Of setting and adjusting one’s expectations. This implies that there is hidden sensory phenomenology to all perception: It’s the phenomenology of having and renewing a sensory expectation. Going back to Dewey, having “an experience” is a matter of learning how to be surprised.

6.  An Empty Wallet Theories of perception are great, but in the end, it’s just you and your wallet. And a wallet can be a good measure of what you believe. You might say that you care about veganism and health, but if you line up for burgers at the airport, then it’s not really health or veganism that you seek. Similarly, with absence art. If you sing praises to the ingenuity of an empty frame, but then go for another impressionist print, then in your heart, you don’t believe that absence art is real. You probably don’t believe in absence art. I say, invest in absence art. Absences are extraordinary objects. In daily life, they are a hard catch: They slip away as soon as we see them. Art is magical because it makes absences stay. It doesn’t just know how to catch an absence, it also knows how to keep it safe. Part of absence craft is understanding the laws of expectation. An artwork features an absence because the artist knows what we expect. This knowledge determines how the artist shapes a physical object that will hold an absence. if properly shaped, the object will have two roles: It will function as an alert signal (“expect this!”) but also as a kind of a living frame: It holds and molds the absence. Absence art is the craft of expectations. Expectations set up the mold within which an absence is viewed or formed. It is also a craft of values. Our values ultimately determine why we see absences in art. I’d like to

Would You Buy Absence Art? 275 close the chapter by talking about values and what (perhaps) is the most important lesson that absence art can teach. Absences can confer reality on absence art. The most basic fact about aesthetic experience is that it’s evaluative. We experience beauty or grace because we are attuned to values in the world. This sets an interesting parallel between art and absence perception. There are reasons to think that absence perception is also fundamentally evaluative.27 While expectations play a big role in setting up our experiences of absence, it’s our personal values that fundamentally explain why experiences of absence occur. You can form an expectation to see an object and set a probability to seeing it. But ultimately, you experience its absence not because of probability estimation, but because that object was of value to you. I think that experiences of absence are evaluative and they can give a deep insight into our values if we take time to look. Experiences of absence don’t just record what’s missing in the world. These experiences are reflective: They reveal our values back to us. This revelation of value also happens in our experience of absence art. When you look at absences in art, you get to see, in a free, creative context, not just what’s missing, but what matters to you most. Given this view of experiences of absence, absence art faces a special challenge. It must construct a special space for an absence between freedom and expectations. On the one hand, it must capture the absence. The expectations engaged by an artwork must be robust enough to hold the absence still and render it visible to all. But it must also allow for aesthetic freedom: We should be free to explore and experience the absence subjectively. An absence can mean anything at all. Absences in art can be superficially beautiful and graceful. They can also be meaningful and deep. But one of their hidden values is what they mirror back to their viewers: what they show us about ourselves. So, an artist can stipulate their depth, or in the opposite, can let them roam free. But invariably, experiences of absence also capture what we value and expect. Their freedom is your freedom and their depth is your depth. It’s how much of yourself you’re willing to face when you look at an absence. So, what is it that you buy when you buy absence art? You buy a glimpse of yourself.

Notes 1. Reyburn, Scott. “Banksy Painting Self-Destructs After Fetching $1.4 Million at Sotheby’s.”  The New York Times, The New York Times, 6 Oct. 2018, www.nytimes.com/2018/10/06/arts/design/uk-banksy-painting-sothebys. html. 2. For example, de Finnetti (1974) advocated an operational approach to belief.

276  Anya Farennikova 3. See Nanay’s Aesthetics as Philosophy of Perception (2016) for a discussion of how aesthetics can benefit from a more empirically grounded approach. 4. Shaw, R. (2008). Absence of God III  .  .  . And His Tears of Blood Will Drown the Cities of Men II. [Acrylic, glitter, enamel and rhinestones on fine portrait linen]. 5. In 2014, there was an art exhibition, ascribed to artist Lana Newstrom, where she displayed invisible works. The gallery was empty. The show was later declared to be a hoax. 6. There are some complexities involved in defining what it means for an artwork to feature an absence as part of its content. See Farennikova, “Art of Absence,” ms., for a more detailed discussion. 7. Note that representational absence art can also use absences as comments on the nature of art. Consider a portrait of a headless monarch. Portraits, by definition, must depict persons but here, no live person is depicted. The missing head might just be a case of realism, but it can also function as a way to push the boundaries of the portraiture genre. 8. See Shelley (2003) for an excellent discussion of the problem of the grounding relation for non-perceptual art. His solution, however, will be different from the one I offer in this chapter. I will reject the claim that absence art (in many of its forms) is a case of non-perceptual art. According to the argument I’ll present, absence art will not count as an instance of premise X in the argument Shelley criticizes in his paper. 9. See Carroll (2012) for a review of approaches to aesthetic experience. 10. Banksy’s “Girl with a balloon” sold for $1.4 million. 11. See, for example, R. Gregory’s (1974, 1998) papers outlining constructivist approach to perception. Gregory launched the movement viewing perception as an intelligent system. See also Kersten et  al. (2004) and Feldman (2015) for a review of recent Bayesian approaches to perception, which view perceptual intelligence as an instantiation of Bayesian rationality. 12. For example, Gregory (1966) argued that “perception is not determined simply by stimulus patterns; rather it is a dynamic searching for the best interpretation of the available data.” 13. Farennikova, A. “Seeing Absence,”  Philosophical studies,  166(3) (2013: 429–454). 14. For a detailed argument, see Farennikova (2013). 15. The association of a large scale with sublime was posited in Burke (1757). Ishizu and Zeki (2014) used “large scale” as a category of the sublime in their study. 16. I describe this mechanism in more detail in Farennikova “Art of Absence,” ms. 17. See A. Farennikova (ms.) for a discussion of other possible mechanisms. 18. This, for example, is presupposed in Carroll’s (2011) review of formalism and Zangwill’ s (2000) defense of his brand of formalism. 19. See Farennikova, “Art of Absence” (ms.) for the defense of this claim. 20. Dewey (1934). 21. See Millkian (2004: 223–224) 22. See Nanay (2015). 23. Alva Noë talked about fragility as a core feature of perception in “Varieties of Presence” (Harvard University Press, 2012). I think that he’s right about this underappreciated feature, and I think that we can appreciate this feature of perception more by paying attention to how we experience absences in ordinary life or art. 24. “Experience is not something that happens in us. It is something we do. Experience itself as a kind of dance—a dynamic of involvement and engagement with the world around us.” Ibid., p. 130.

Would You Buy Absence Art? 277 25. Surprise also matters generally for our experience of art: There is a growing movement in neuroaesthetics that highlights the role of wonder in how we view art. See, for example, Kesner (2014), Muth and Carbon (2016). 26. See, for example, Seriès and Seitz (2013). 27. Farennikova (2014).

References Burke, E. (1757) A Philosophical Enquiry into the Origin of our Ideas of the Sublime and Beautiful, London: R. and J. Dodsley. Carroll, N. (2011) “Formalism” in The Routledge Companion to Aesthetics. Routledge. Carroll, N. (2012) “Recent Approaches to Aesthetic Experience” in The Journal of Aesthetics and Art Criticism, 70 (2), 165–177. De Finetti, Bruno (1990) Theory of Probability, Vol. I. New York: Wiley. Dewey (1934) Art as Experience. Reprinted in 1989,  John Dewey: The Later Works, 1925–1953. vol. 10 (ed. Boydston, J.), Carbondale: Southern Illinois University Press. Farennikova, A. (2013) “Seeing Absence” in  Philosophical Studies,  166 (3), 429–454. Farennikova, A. (2014) Art of Absence. Manuscript. Farennikova, A. (2018) “Perception of Absence as Value-Driven Perception” in Evaluative Perception (ed. A. Bergqvist and R. Cowan), Oxford: Oxford University Press. Feldman, J. (2015) “Bayesian Models of Perception: A  Tutorial Introduction” in Handbook of Perceptual Organization (ed. J. Wagemans), Oxford: Oxford University Press. Gregory, R. L. (1966) The Psychology of Seeing, London: Weidenfeld and Nicolson. Gregory, R. L. (1980) “Perceptions as Hypotheses” in Philosophy of Psychology, London: Macmillan. Gregory, R. L. (1997) “Knowledge in Perception and Illusion” in Philosophical Transactions of the Royal Society of London B: Biological Sciences, 352 (1358), 1121–1127. Ishizu, T. and Zeki, S. (2014) “A  Neurobiological Enquiry into the Origins of our Experience of the Sublime and Beautiful” in Frontiers in Human Neuroscience, 8, 891. Kersten, D., Mamassian, P. and Yuille, A. (2004) “Object Perception as Bayesian Inference” in Annual Review Psychology, 55, 271–304. Kesner, L. (2014) “The Predictive Mind and the Experience of Visual Art Work” in Frontiers in Psychology, 5. Millkian, R. (2004) Varieties of Meaning, Cambridge, MA: MIT Press. Muth, C. and Carbon, C. C. (2016) “SeIns: Semantic Instability in Art” in Art & Perception, 4 (1–2), 145–184. Nanay, B. (2016) Aesthetics as Philosophy of Perception, Oxford: Oxford University Press. Nanay, B. (2015) “Aesthetic Attention” in Journal of Consciousness Studies, 22 (5–6), 96–118. Noë, A. (2012) Varieties of Presence, Cambridge, MA: Harvard University Press.

278  Anya Farennikova Reyburn, S. (2018) “Banksy Painting Self-Destructs After Fetching $1.4  Million at Sotheby’s” in The New York Times, 6 October  2018, www.nytimes. com/2018/10/06/arts/design/uk-banksy-painting-sothebys.html. Shaw, R. (2008) Absence of God III . . . And His Tears of Blood Will Drown the Cities of Men II. [Acrylic, glitter, enamel and rhinestones on fine portrait linen]. Shelley, J. (2003) “The Problem of Non-Perceptual Art” in The British Journal of Aesthetics, (43) 3, 363–378. Seriès, P. and Seitz, A. (2013) “Learning What to Expect (in Visual Perception)” in Frontiers in Human Neuroscience, 7, 668. Zangwill, N. (2000) “In Defence of Moderate Aesthetic Formalism” in The Philosophical Quarterly, 50 (201), 476–493.

14 Penetrating Beauty Knowledge, Culture and Context in Aesthetic Perception Jesse Prinz

1. Introduction Perception is, uncontroversially, in the service of cognition: It provides us with information about the external world, which allows us to make decisions. Much more controversial is the claim that cognition can operate in the service of perception. This is nowhere more apparent than in the domain of art. Knowledge can help us know what to look for in artworks and it can guide visual experience. Visual experience of art can also be impacted by other senses and by emotions. Indeed, much of our vocabulary for making assessments straddles the boundary between seeing and feeling. We describe artworks as stunning, sumptuous, attractive and hypnotic. Even the word “beautiful,” I will argue, can be regarded as both perceptual and emotional. In these ways, visual encounters with art are impure—they are not just visual, but also cognitive and affective. Once we recognize this fact, questions arise about the modularity of aesthetic experience. Are the emotions and thoughts that arise in the context of art mere consequences of what we see, or is the connection somehow more intimate? I will argue that thoughts and feelings can impact visual experience: They penetrate the perception of art. This is not the first effort to argue for aesthetic penetration—i.e., the possibility that the perception of aesthetic qualities in visual artworks may be influenced by something other than the visual inputs (see, e.g., McMahon, 2012; Nanay, 2015; Stokes, 2014). Indeed, as we will see, there is a long-standing view among critics and art historians that learning impacts the perception of art. I stand on the shoulders of these predecessors, using different lines of evidence and examples. Like Nanay and Stokes, my case for penetration in the aesthetic domain will be both empirical and conjectural. There is a large body of research indicating that vision can be impacted by non-visual processes (other modalities, emotion and cognition), but little of this work directly investigates art. That said, much of art history and criticism aims to enrich experience and therefore there are countless suggestive examples. I will explore some

280  Jesse Prinz of these with the aim of illustrating possible examples of penetration in the aesthetic domain. The examples I  will use are speculative, and my claim that they are cases of penetration will not convince diehard modularists, who say that perception can never be impacted by other faculties. I submit the examples as ripe for investigation, not as a demonstrative proof. But I  will sketch a general case against modularity and show these cases to be especially plausible against the background of those considerations. Too, for readers already opposed to modularity, the examples will serve to illustrate how penetration may contribute to the experience of art. Those who recognize that aesthetic experience involves an interplay of form, feeling and meaning still tend to implicitly endorse a form-first thesis, that seeing spurs us to feel and think. It would be a substantive discovery to show that there is also much traffic the other way, from feeling and meaning to form. That multi-directionality will be my focus here. I will begin with a brief look at the modularity debate; then I will discuss the nature of beauty and related aesthetic concepts and finally I will present a range of examples in which factors other than form impact the appearance of artworks. My examples will be drawn from the history of Western and non-­ Western art. The use of non-Western examples is not just intended as a corrective to a Western bias in analytic aesthetics but also to demonstrate a subsidiary thesis: Seeing, I  want to say, is both culturally and historically inflected. When we look at art, culture functions as a kind of lens, and this can lead us to misperceive works form other places—to neglect features that were important to their intended audiences. The non-­Western cases also underscore a way in which penetration can be valuable. We have much to gain by increasing our knowledge of unfamiliar art. Such knowledge can help us see works in new and better ways. In viewing art, inputs from outside the visual system do not necessarily distort or conceal the objects in our gaze—they can also reveal important aesthetic features.

2.  The Modularity Wars 2.1  Evidence for the Penetrability of Perception Modularity has long been a polarizing topic in the philosophy of mind and cognitive science. Here I will present an opinionated overview of the debate. Though modularity is often defined as having many dimensions (Fodor, 1983), it has, in recent discussions, been most associated with one principle: The hypothesis that perceptual systems are informationally encapsulated. For the modularist, perceptual systems receive inputs from the external world and then process those inputs using rules and representations that are built into them without being impacted by other

Penetrating Beauty 281 systems. Some modularists limit encapsulation to early stages of perceptual processing (Fodor, 1983), while others suggest that every stage of perception may be insulated from external effects (Firestone and Scholl, 2016). Modularity may also be characterized in terms of the direction of information flow. If modularity is true, then information flows up from the sensory transducers, through various stages of perceptual processing and then generates outputs that can be use used for cognition, to evoke emotions and to drive action. Information, on this view, does not flow down or even sideways. Opponents of modularity reject this. They posit both “top-down” effects from cognition on perception, which I will call “vertical penetration,” as well as what I  will call “lateral penetration.” Lateral penetration includes cross-modal effects in which one sense modality impacts another; bodily effects, in which action or behavioral plans impact perceptions and emotional effects, in which emotions and other affective states impact perception. Empirical work on the modularity debate usually proceeds as follows. Some researchers conduct experiments that purport to show vertical or lateral effects on perception and then modularists devise and test alternative explanations of these effects. I cannot summarize this vast literature here, but I will mention a few representative findings. Evidence for cross-modal effects includes a study by Shams et al. (2000) in which a briefly flashed disk seems to double into two flashes when accompanied by two beeping sounds. Research on action effects has obtained numerous findings linking bodily abilities to perceived physical magnitudes. For example, visual estimates of distance can be affected by posture, what tools are being used, handedness and jumping ability (Proffitt and Linkenauger, 2013). Work on emotional effects indicates that different emotions impact perception in different ways: Fear can make balconies seem higher, sadness can make hills seem steeper and disgust can make objects look closer (Stefanucci et al., 2011). Positive moods, in comparison, can evoke global or holistic visual processes, make positive stimuli more salient and impact the speed and accuracy of peripheral vision (Vanlessen et al. 2016). Vertical effects are a bit trickier to study, since there is little agreement on what counts as cognition. I like to define cognition functionally as those mental processes or representations that can come under executive control. These include language, beliefs, expectations and elaborative classification (when we go beyond immediate recognition and reflect on what attributes something has). So defined, examples of vertical effects on perception may include the following: Linguistic labels can effect categorical perception of color and emotional expressions (Roberson and Davidoff, 2000); the belief that athletes see better improves visual acuity after exercising (Langer et al., 2010); expectations increase speed, confidence and false-positives and classification of distorted images (Król and El-Deredy, 2011) and status cues influence race

282  Jesse Prinz categorization (Freeman et al., 2011). The details in each of these cases would require considerable interrogation to establish that they involve cognitive effects on perception, but, on the face of it, they seem to. So along with other results I have mentioned, there seems to be much prima facie evidence for penetration. 2.2  Deflecting Modularist Rebuttals Proponents of modularity have not been convinced by the mass of evidence that aims to establish vertical and lateral effects on perception. The objections usually revolve around alleged methodological flaws in the studies, along with alternative explanations. For example, it is suggested that the results are post-perceptual or perhaps even the result of demand characteristics: Participants giving the answers they think the experimenters want to hear (Firestone and Scholl, 2016). Modularists try to refute alleged cases of penetration by designing studies that first replicate the original results and then make the results disappear (i.e., obtaining null results) by altering demand characteristics. I cannot review these objections here, but I want to indicate why I find them wholly unconvincing. There are, to be sure, many flawed studies, but the critics often systematically ignore many studies that carefully anticipate and address precisely the objections that they raise (Witt, 2016). The method used to refute evidence for penetration is also dubious; a replication and a null result cannot serve as a refutation since a replication provides further support and null results are scientifically uninterruptable and simply confirm that performance on perceptual tasks is malleable. As for the claim that the effects are post-perceptual, critics tend to ignore the vast literature showing top-down effects in the brain (Gilbert and Sigman, 2007) or results that use psychophysical measures such as afterimages (Lupyan, 2015). There are also various well-established mechanisms by which penetration can take place. First, we can intentionally generate sensory images, and most parties agree that sensory imagery reactivates cells in perceptual pathways. Second, we can alter perceptual thresholds by using language or goals to co-classify sensory inputs. Co-classification can lead to the storage of perceptual prototypes, which can in turn impact category boundaries (inputs near the boundary of a prototype are shifted in the direction of its center). Third and most importantly, information outside perceptual systems can shift attention. Modularists sometimes unacceptably disqualify such effects, saying that attention is a matter of shifting where one looks rather than changing perceptual processing (Fodor, 1983). This is a hopelessly misguided view of attention. Attention can be covert, without any change in gaze, and there are decades of incontrovertible evidence showing that attention alters activity in perceptual pathways at the cellular level (including tuning curves, receptive field sizes, response thresholds and center-surround structure). Behaviorally,

Penetrating Beauty 283 it is also well established that attention can enhance the representation of specific sensory features and specific objects. This shows that attention recruits perceptual representations (cf., Stokes, 2014; Mole, 2015). Once one recognizes that attention works in this may, the modularity hypothesis begins to look hopeless. Another reason to doubt the modularity hypothesis stems from the fact that many important aspects of perceptual processing seem to benefit from penetration. These include the following: Predictive coding (the use of prediction to enhance an input); visual search (trying to find specific objects or features in the visual field); goal-directed vision (identifying features that serve current action plans); visualization (re-activating visual states in the service of planning or problem solving); cross-model enhancement (combining sensory information to resolve degraded inputs); skilled classification (using knowledge, goals and corrective feedback to develop a rapid perceptual classification abilities) and hazard detection (quick responses to objects that may be threatening). These are all cases where, as a matter of principle, perceptual systems would benefit from information flowing into perceptual systems. It is difficult to imagine how they (or the empirical findings that have emerged from their investigation) could arise without penetration. They are so commonplace that it seems reasonable to assume that penetration is not just possible but pervasive. Against this, modularists like to say that there are also principled reasons for denying that penetration is possible. In particular, it would undermine the evidential value of perception (Fodor, 1983). This concern is misplaced in several ways. First, penetration often improves perception (as in the case of cross-modal enhancement or hazard detection); second, perceptual errors do in fact arise (as when wardrobe misleads about race); third, post-perceptual interference can also undermine perceptual evidence and, finally, no proponent of penetrations suggests that beliefs entirely override sensory inputs. This last point often undermines the most famous argument for modularity. Fodor (1983) argues that informational encapsulation best explains the fact that beliefs cannot override visual illusions, such as the Müller-Lyer figure, in which equal lines appear to differ in length. An alternative explanation is that such figures are presented clearly and rely on overlearned perceptual associations, so they simply trump potentially corrective feedback. Beliefs can impact perception when there is degradation or ambiguity. This seems to occur in the case of bistable images and in all the studies surveyed above. The foregoing considerations would not persuade a committed modularist, but for the uncommitted, I hope they establish that modularity is a radical view. Though sometimes presented as a sensible and orthodox position, modularity actually flies in the face of countless empirical findings and is difficult to reconcile with some of the most basic operations of ordinary perception, such as visual search. In what follows, I will assume

284  Jesse Prinz penetration is possible and explore how it may contribute to the experience of art.

3.  Aesthetic Perception 3.1  Beauty and Beyond Having made a quick case for the penetrability of perception outside the aesthetic domain, I now want to turn to the topic of aesthetic perception. The term “aesthetic perception” can be used in two ways: The perception of objects created for aesthetic contemplation (i.e., works of art, design and décor, for example) and the perception of aesthetic features. To focus this inquiry, I will consider the intersection of these meanings: The perception of aesthetic features in works of art. The term “aesthetic feature” is vexed, since opinions differ on what it means to call a feature aesthetic and whether there are any distinctively aesthetic features. To bypass those thorny debates, I will simply define the concept by example. The preeminent aesthetic feature is beauty, though I will mention others as well. Beauty was once the main topic in discussions of the evaluation of art, but it has drifted out of the spotlight with the rise of modern and contemporary art, which is sometimes deemed anti-aesthetic or non-aesthetic. I think beauty remains a common aspiration in art-making, as it has been both historically and globally, though there are also other meritorious features of artworks that I will touch on in this discussion. Beauty will provide a core case. Theories of beauty abound. Classically, it was associated with harmony. With the rise of modern Western philosophy, it was conceived as the balance of uniformity and diversity and, with Kant, we arrived at the idea that beauty involves the free play of the intellect and the imagination (a very non-modular idea!). I will not dwell on these proposals from philosophy’s past and will instead begin with a simpler psychological suggestion. Beauty, on the view I will recommend, is positive form. Positive form refers to formal qualities of something that are perceived in a positively valenced way (I leave aside the complication of beauty outside the sensory domain). So characterized, beauty is best conceived as something subjective. Positivity is an evaluative assessment. Form can be positive to one person and not to another. I will leave aside the question of whether there can be standards of taste. I would venture further that positivity is not so much judged as felt. For if we define positive form as form that we judge to be good, then there would be a question of what such judgments consist in; we would need a theory of what goodness is and this is something for which few could provide an answer, as divergence in taste suggests that aesthetic goodness cannot be reduced to objective features that are

Penetrating Beauty 285 recognized by everyone who makes beauty judgments. Therefore, I join the venerable sentimentalist tradition that sees beauty as a feeling rather than a judgment. Good form is form that is experienced in an emotionally positive way. One can immediately notice that the good form theory implies a kind of pluralism about beauty. Not only will there by differences in taste (which I would call relativism rather than pluralism), there are also differences in the forms and feelings that constitute good form. I think it is useful to regard “beauty” as a superordinate term. We can distinguish species of beauty. These species vary along two dimensions: The goodness (construed here as an emotional component) and the form. We have a rich aesthetic vocabulary that tracks these variations. Let me illustrate with some possible examples. First, we use words that distinguish kinds of beauty based on the kinds of formal features we are assessing. “Grace” implies forms that flow; “graphic” implies forms that are bold and legible; “balance” and “harmony” refers more to the distribution of formal elements; “splendor” requires a kind of formal richness; “cute” forms are petite or child-like; “refined” implies forms that are delicate or mannered; “vivid” draws attention to spectral aspects of form or clarity of light. There are also terms that track stereotypically gendered features, such as “handsome” and “pretty,” which we use in reference to people. All of these may be regarded as species, grounds, or dimensions of beauty. We can defend a beauty verdict by appeal to such terms and each plausibly involves an emotional response to a formal feature. What strikes one viewer as graceful may not impress another. Second, there are species of beauty that vary on the emotional dimension. “Attractive” forms are those that instill sensory pleasure; “sexy” and “sensuous” forms arouse more lascivious emotions; “charm” is associated with a more playful delight; “breathtaking” forms cause awe or wonder. Beauty can also be moving, somber, or profound and these adjectives imply differences in emotional tone. In addition to these terms, which I regard as species of beauty, there are terms of positive appraisal which tend not to be so-classified. Forms can be described as “dynamic,” “energetic,” “serene,” “playful,” “cool,” “chic” and “sublime.” These terms vary in their emphasis on form (e.g., “dynamic”) or feeling (e.g., “playful”), or both (e.g., “serene”), but each seems to apply in contexts where formal qualities have a positive emotional impact. I think it is convenient to classify these as species of beauty for that reason, but this is mere stipulation. Perhaps there is a way to divide formal goodness into beauty proper and other varieties—a project I won’t undertake here. Species variation increases further if we move beyond the critical vocabulary of English and consider other languages. Consider, for example, the range aesthetic terms in Japanese (cf. Richie, 2007: 53). These include karumi (the beauty of simplicity), kurai (a cold dignified beauty),

286  Jesse Prinz jimi (tastefully understated), iki (an urbane chicness), yūgen (profound grace), sabi (the beauty of ephemerality), wabi (satisfaction with poverty or imperfection) and the compound wabi sabi (rustic beauty). Japanese also includes a number of compounds modifying kawaii, the word for cuteness. My favorite is guro kawaii, which combined cuteness with gore to yield aesthetically pleasing imagines that are at once adorable and violent. For each term one can identify formal constraints and concomitant features. Cultural norms contribute to determine what forms are good. For example, in the tradition of kintsugi, broken ceramics are repaired in a very visible way, which is prized because it conveys impermanence, manifesting sabi—a feature we do not normally aestheticize in Western aesthetics, though such norms can be transmitted across cultures. The main point here is that beauty has many variants, and there is much plasticity in what forms we experience positively and the kind of positive feelings evoked. 3.2  Evaluative Seeing I have sketched an approach to beauty that is both pluralist and sentimentalist. Before turning back to the main thread of aesthetic penetration, I want to make one more observation about the nature of beauty, which bears on the broader theme of the modularity debate, which can be described as the question of perceptual purity. Modularists say that each sense modality generates representations that are pure in the sense that they have not been impacted by inputs that originate outside of that particular modality. This is the position I rejected above and will return to below. There is also another question involving purity: We can ask whether perceptual states can constitutively include non-perceptual representations. One might describe the first form of impact as corruption and the latter as contamination—though of course I mean to use these terms in a neutral, or at least tongue-in-cheek way. Having endorsed the view that vision can be corrupted, I now want to blaspheme further by saying that it can be contaminated. Indeed, I want to say that beauty is a kind of contamination. By this I mean to say that the emotional concomitants of beauty are not mere consequences of the forms we perceive but instead get bound up with those forms in a way that allows us to say they comprise cohesive mental states (Prinz, 2012). When we see something as beautiful, we don’t first see it and then assess it as beautiful. Rather, we see beauty. This is first and foremost a dictate of phenomenology. We experience beauty as a feature of beautiful objects, not as a reaction in ourselves. Beauty is projected onto the world (Santayana, 1896). If the phenomenological thesis is right, it demands an explanation. What accounts for that fact that we experience gracefulness, sexiness, cuteness and so on, as features of the world? I think the best answer must implicate psychological

Penetrating Beauty 287 mechanisms. In particular, the mental states that underlie experiences of beauty must be hybrid states—states that have both visual and affective components. There are different theories in cognitive science about how mental states can be bound together to form hybrid states that are experienced as cohesive as opposed to, say, sequential. One class of theories posits neural dynamics implementing the components in such a way that they get “coupled” together. Dynamic coupling refers to a causal co-dependency relation in which two states unfold in time such that each is dependent on the other. If we think of emotions and visual states as implemented by neural firing patterns, coupling between the two would arise if these patterns enter into a kind of causal resonance. For example, the neurons might oscillate in ways that are “phase-locked”—with the oscillations of one coming into temporal register with the other. Other mechanisms of binding are possible as well, but they need not detain us. The key point here is that we experience beauty as inherent in the objects we perceive, and that can be explained by saying our visual states get bound up—however binding works—with evaluative states, such as emotions. If this is right, then beauty can be regarded as contaminated perception. Experiences of beauty are underwritten by mental states that are not purely visual but instead include both visual and emotional components. That, in summary, is the psychological approach to beauty that I want to endorse as we return to the main thread of aesthetic penetration. Against this background, I will now argue that beauty is penetrated. If the contamination story is right, there is already a sense in which aesthetic experience departs from the paradigm of purity that animates modularist theories. I now want to show that seeing art is not just contaminated but also corrupted.

4.  Aesthetic Penetrations 4.1  Kinds of Aesthetic Penetration In Section 2, I offered some general reasons for thinking that perception can be penetrated, and in Section 3, I presented a sentimentalist approach to beauty. I  now want to bring these pieces together and suggest that our perceptions of art, including especially perceptions of beauty, can be influenced by processes outside of vision. The idea that the visual experience of art can be corrupted by external sources of influence is not new. For well over a century, Western critics have used the phrase “ways of seeing” in conjunction with art. The earliest instance I have found is in Philip Gilbert Hamerton’s 1878 Encyclopedia Britannica entry on “Drawing.” He suggests that artists in “primitive” societies have different ways of seeing. A decade later, John

288  Jesse Prinz Evan Hodgson published an article in which he suggests that “[a] picture is an artist’s way of seeing and feeling” (Hodgson, 1888: 362). Another influential suggestion comes from Wölfflin (1950: 11), who teaches that “vision itself has a history.” Gombrich (1960: 239) goes even further, railing against Ruskin’s conviction that there can be an “innocent eye.” Baxandall (1972: chap. 2) replaces Ruskin’s phrase with “the period eye,” suggesting historical differences in how people see art. Wartofsky (1981) coins the phrase “visual intentionalities” to capture the role of intentions in seeing. More recently, Nanay (2015), citing some of these predecessors, argues that attention has a history, which influences the experience of art. There isn’t space to examine these proposals or the examples that have been hitherto discussed. I want to build on this impressive tradition by describing ways in which experiences of beauty can be externally influenced, with special attention to the impact of culture. If beauty is a hybrid state, then aesthetic penetration can arise through external influence on both the perceptual and emotional components, as well as via the ways in which the emotions that comprise experiences of beauty may influence visual components. Of these three, the second might seem least surprising. After all, it is uncontroversial that emotions can be influenced by non-emotional states, including cognition. Thoughts can give rise to emotions. Still, even here, penetration is potentially interesting. After all, we tend to think of beauty as visually driven. On the analysis give above, it is forms that arouse positive emotions. If there are cases where what we perceive as good form can be influenced in either the visual or emotional aspect by something non-visual, that would force a revision of the intuitive, bottom-up view of beauty. In proposing examples of aesthetic penetration, it will be useful to use a distinction introduced earlier between lateral and vertical penetration. Lateral cases involve impact on beauty that is not cognitive, but rather from extraneous emotions, non-visual perceptual states or motor responses. These lateral effects will qualify as cases of penetrated beauty whenever the visual-emotional beauty response is influenced by something other than visual inputs and the emotions that normally arise solely in response to those inputs. Vertical cases will involve cognitive impact on experiences of beauty—including impact on both the visual and emotional components. I will consider both kinds of cases in turn. Cultural influences can arise for both, but they will be especially prominent when we turn to vertical penetration. As noted at the outset, the examples that follow are speculative. I will present these as possible cases of penetration, meaning that further evidence would be required to confirm that perception is contaminated. Most examples are drawn from art history, where no such testing takes place. At the outset, I will also make reference to some empirical results, but even here direct evidence for penetration has not been sought. Against

Penetrating Beauty 289 the background of evidence for penetration, I think it is plausible to suppose that these are genuine cases, especially given the claim that attention can directly alter perception. Still, I am content to see this an exercise in motivating a research program. Penetration offers a promising way to cash out the idea that perception is culturally and contextually inflected. 4.2  Lateral Aesthetic Penetration To begin, let us consider some possible cases of lateral penetration. Here I will be brief, since vertical penetration is both more controversial and, I would submit, more crucial to understanding aesthetic experience. My discussion of lateral aesthetic penetration will draw more heavily on empirical results and less on art history, since most apparent cases of arthistorical interest are vertical: They involve an imposition of knowledge. As a first example of lateral penetration, let me mention some empirical literature indicating that aesthetic judgments can be impacted by extraneously induced emotions. On the theory of beauty sketched above, this might be explained by supposing that extraneous emotions alter the emotions that are component parts of experiences of beauty. Consider a study by Isen and Shalker (1982) in which participants are asked to evaluate photographs after they “find” ten cents that has been planted by experimenters. Aesthetic preferences increase for those who find the change. This suggests that the happiness from that small bit of good fortune leaks into the pleasure experienced in looking at the photographs. Pleasure is not the only emotion that can also have an aesthetic impact. In a study I conducted with Eskine and Kacinik, we found that fear induction using a startling zombie film clip increases sublimeness ratings for abstract paintings by El Lissitzky (Eskine et al., 2012). It may be that the chills of fear combine with the emotions aroused by the works, making them feel stirring, rather than merely pleasant. There is also some empirical work investigating the impact of motor responses on aesthetic judgments. This may play a role in a study with Seidel in which we asked participants to rate artworks mounted at different wall positions: At, above or below eye level. (Seidel and Prinz, 2018a) When hung above eye level, ratings went up and when hung below eye level, ratings went down. The embodied state of gazing upward may increase esteem, while gazing down has the opposite effect. If so, motor responses are impacting emotions, which may, in turn, contribute to experiences of beauty. We did not investigate the underlying mechanisms, but this remains a possibility for future research. More interesting for participants in the modularity debate, perhaps, are cases where visual perception is impacted. As noted above, there is research indicating that emotions can impact vision, including the finding that happy moods induce more holistic visual processing (Gasper and Clore, 2002). This may mean pleasurable beauty experiences (as when

290  Jesse Prinz we see artworks as pretty) may result in seeing the global structure of a work more readily than we would in the case of works that are regarded neutrally. There is also evidence that fear can narrow the focus of attention, not just reducing its spatial expanse but also impacting which visual features get noticed (Wegbreit et  al., 2014). Forms of beauty that are unsettling, including perhaps some instances of the sublime, may tend to have this effect. Fear, as noted previously, can also increase visual estimates of height and other magnitudes, which may entail that items depicted in sublime works look larger than they otherwise would. Finally, there are interesting cases of what may be intermodal penetration in art. My collaborator and I conducted a study in which we found that scary music makes dangerous objects more salient in bistable figures (Prinz and Seidel, 2012). Film directors intuitively exploit such effects when they use music to help create a suspenseful atmosphere. Sound can also increase visual intensity (Stein et al., 1996). Similar principles may pay a role in some video and performance art. For example, in Light/ Dark (1977), Marina Abramovic and Ulay take turns slapping each other in the face; the slapping sound may make the impact appear more violent than it would otherwise. In the case of static visual objects in a gallery setting, there is less opportunity for trans-modal effects, but this may still occur imaginatively. For example, when one learns the title of Duchamp’s With Hidden Noise (1916), which consists of a ball of string clamped between two sheets of metal, one might imagine shaking it and listening to it. These effects of visual imagination change the affordances of the work (shaking does not come to mind without the title) and that may impact how we attend to it. Consider, too, African masks. Many (though by no means all) are used in masquerade performances. For those familiar with the intended use of these objects, seeing them might awaken multisensory associations; bringing the traditional music to mind may help one visualize the way the mask would appear in a masquerade—fringed by raffia, perched atop a head and moving in a characteristic way. To see an object with known multisensory associations may awaken different imaginative visualizations, which in turn impact the aesthetic quality of what’s perceived. A mask associated with playful or entertaining masquerades (as performed, for example, by the Guro of Côte d’Ivoire) may have a different kind of beauty than one associated with masquerades that are somber, spiritual or even menacing (as with Kuba funeral rites in Zaire). This last example will serve as a segue into my discussion of vertical penetration. For those acquainted with a masquerade tradition, seeing a mask on display will awaken sensory (as well as motoric and emotional) associations that could potentially impact the aesthetic experience (both visually and emotionally). This would qualify as lateral penetration. For those of us who do not have such acquaintance, there is still some room for learning about masquerade traditions, and this learning

Penetrating Beauty 291 might provide us with some capacity to generate associations in imagination that impact our encounters with masks in museum settings. Western museums tend to present African masks and statuary in ways that are uniform, decontextualized and exotic (dramatic lighting is used to instill a sense of mystery). Learning that a mask actually is used in bright daylight, is colorfully painted and is used in a performance with certain sounds and movements can help us rescue these items from their misleading display conditions and imagine them in full splendor. Such knowledge might reveal expressive qualities (is the mask smiling or scowling?) and alter perceived brightness (perhaps we focus on light rather than shadow when viewing a mask with the knowledge that it is performed in the daylight). Such perceptual alterations along with the affective tone of other sensory associations would potentially impact whether a mask is seen as playful or solemnly sublime. If mediated by top-down knowledge rather than a lifetime of bottom-up experience, these would qualify as vertical effect. I turn to more examples now. 4.3  Vertical Aesthetic Penetration Vertical penetration occurs when cognition alters the quality of a perceptual experience. As the mask example illustrates, this seems to be a common occurrence when we learn about art. Few people these days are content to see an exhibition without reading something about it. Reviews, placards and brochures help viewers understand what they are seeing. In cases of older art, there is often extensive scholarship that can be consulted. Art enthusiasts are conditioned to believe that such information can deepen appreciation. Some of that depth has little to do with what we experience in the gallery. For example, humdrum biographical information about an artist may have little bearing on how the work appears. But other information does seem to have the potential for impact. When we see the work of an “outsider artist” as the outpourings of someone who battled with mental illness, patterns that looked merely decorative may start to look compulsive, hypnotic or strange, and scribblings that would be dismissed as amateurish might take on aesthetic allure. Though such effects have not been systematically investigated by psychologists, they do seem to be a regular aspect of our encounters with art. Here I will offer an informal taxonomy of different kinds of cases. The outsider art case belongs with a class of cases that involve revelations about an artist’s motivations or mental states. Seeing Frida Kahlo’s self-portraits with the knowledge of her chronic pain adds a poignancy that might otherwise be lost. Such poignancy can also be conferred by learning about the message that an artist intends to convey. Ai Weiwei’s Remembrance (2009) looks like a cheerfully colorful wall of children’s backpacks, but it takes on a grim cast when we learn that it is a memorial

292  Jesse Prinz to students who died in a Sichuan earthquake and a political protest against corruption in the building of schools. In some cases, the relevant knowledge is difficult to come by. Was the Venus of Willendorf a fertility amulet or a piece of Paleolithic pornography? And how are we to regard the sexual statuary that adorns the façade of the Khajuraho Temples in Chhatarpur (begun in the tenth century CE)? To look at an image for arousal is different than looking at it for worship, and knowledge of the intended context can alter the experience. These effects may be both emotional (titillation vs. veneration) and visual, determining which features are most vivid. In the case of the Venus of Willendorf, any fluctuation in feature salience is likely to involve endogenous shifts of feature-based attention rather than overt gaze, since she is quite tiny. A second kind of vertical impact arises when we engage in visual search. For example, when I first saw Hokusai’s “great wave,” I didn’t notice the small boats beneath its crest, nor the vista of Mt. Fuji on the horizon—the latter is revealed by the print’s official title, Mt. Fuji off the Coast of Kanagawa (ca. 1829–1833). With a bit of information, the boats become highly salient. Likewise, one’s visual experience of the penance panel at Māmallapuram (ca 650 CE) alters when one is told that there is a cat imitating a monk somewhere in the complex relief. Here overt shifts in gaze will be facilitated by a visual template that makes both cats’ and monks’ postures more salient, allowing one to find the devilish feline in the visual disarray. In contrast with these cases, where we actively search for things, there are also cases where learning can sensitize us to features that might otherwise go unnoticed. Consider the Japanese aesthetic concepts introduced in Section 3.2. Those who know the concept of wabi may start to notice signs of imperfection, such as a flaw in the glazing of a pot or a dry flower in an otherwise opulent ikebana arrangement and view these positively. It is fascinating to learn that such aesthetic concepts are also deeply associated with the Japanese tea ceremony, which may appear to outsiders as impeccably ordered and even decadent. Knowledge is needed to recover the humility of the ceremony and its setting and this, in turn, can impact what is observed and felt. Learning the aesthetic norms of a culture can help us make sense of what we see, and this can impact the seeing itself. Let’s shift from Japan back to India for further examples. Indian figurative sculptures in the Hindu, Buddhist and Jainist traditions look different from European sculptures in ways that may be difficult at first to articulate. Some insight comes when we learn that, in these traditions, smooth bodily surfaces were preferred to surfaces that show veins or muscles. In this respect, an Indian statue is almost the opposite of, say, a statue by Michelangelo (Nardi, 2006: 60). Armed with this knowledge, smoothness of the body becomes both more salient and perhaps more admired. Knowledge can also inform our experience of Indian art from the Moghul period. Moghul

Penetrating Beauty 293 miniatures easily charm unfamiliar viewers, but they also have features that are best appreciated with an understanding of Moghul aesthetic conventions. As Muslims, the Moghuls eschewed figurative art, except in the case of secular paintings, and they excelled in creating elaborate geometric patterns. When this aspect of their artistic output is brought to mind, it is easier to notice that the painting, which superficially appears quite naturalistic, can be viewed in a way that accords with this taste for geometric abstraction. Notably, horizon lines typically rise to the top of the picture plane, allowing a distribution of objects that begins to look more decorative than depictive. The result is a kind of flattening that might be missed if one focused on the pictorial elements (Minissale, 2009: xxv, 10, 23, 35, 70, 80). Consider one more case. The Asante of Ghana (1983) have an aesthetic norm that values shininess. This is not a trait that we, as Western viewers, tend to notice when looking at statuary, but learning this fact about the Asante can make luster more noticeable and may go some way toward helping us discern which of two Asante figures is a more impressive work, from the perspective of its creators. The shinier one may begin to look more obviously different and superior. In these examples, knowledge plays a role in deciding what makes an artwork good. In other cases, knowledge plays a more basic role; it tells us what we are seeing. This if often necessary in recent Western art, as a result of both abstraction and conceptualism. First consider a case of abstraction: If we see a transitional painting by Mondrian and learn that it represents a tree, the latticework of lines may look more organic and anchored to the base of the picture frame than it would look if seen as a pure abstraction. Conceptual art is often even harder to comprehend by sight alone. Rauschenberg’s Erased de Kooning (1953) looks like a blank page with some indiscernible traces of erased marks. These marks might go unnoticed without the title and they take on the appearance of a de Kooning sketch for those familiar with the artist’s work. Or consider Danh Vo’s We The People (2010–2014) which seems to consist of a large number of unidentifiable copper castings, which have been sold to different collections around the world These look different when we learn it is a full-size replica of the Statue of Liberty, divided into segments. When we see a piece we try to figure out its place in the whole, assigning intrinsic axes based on its intended orientation. A harder case is Sherrie Levine’s masterpiece of appropriation, After Walker Evans (1981) faithful photographic reproductions of the famed photographer’s Depression Era portraits. The original photos are deeply moving, but these copies, which are indistinguishable evoke different emotions: Intellectual pleasure or even annoyance. The transparency of photography is disrupted and we no longer focus on the depicted individuals but rather on various surface features that might disclose that these are mere copies—portraits of photos, rather than of persons.

294  Jesse Prinz Knowledge of what we are looking at can also be impacted by critical vocabulary. For example, when critics classify a work as belonging to Picasso’s blue period, we may see the work being more blue than it actually is: Detection thresholds for color may change. Or consider the terms used to describe paintings of Jackson Pollock. At the height of abstract expressionism, these were called “action paintings” by some critics and “all over paintings” by others. The first term draws attention to their method of creation and may make the splatters of color look more explosive, untidy or dynamic. Compare “action” to the less violent phrase “drip paintings.” The term “all over paintings” draws attention to the fact that Pollock’s compositions lack focal objects and extend to all borders of the canvas. This is easy enough to see by any viewer, but the phrase may invite a diffuse spread of attention rather than focus on details of the splattered paint. Gaze may be affected but need not be; such differences in attention can occur even when gaze is fixed. Knowledge is also necessary to know what we are seeing when we view artworks from other cultures. To take a single case, consider Mende initiation masks, which depict female faces atop what look like thick rings. The experience and emotional impact changes when we learn that the small lips and downcast eyes are intended to encourage demure behavior in young women, and the rings are rolls of fat, which represent aspirations to abundance. What once seemed to be mere stylistic choices are now recognized as depictions of corpulent and silent women—a cultural ideal that may be difficult for Western viewers to fully comprehend and embrace. For the Mende, these reflect beauty norms, but some Western viewers with different beauty norms may need to view the masks more abstractly to preserve full aesthetic pleasure. In these examples, knowledge relates specific facts and this, in turn, helps us classify individual features. There are also cases where experience of a work is informed by rich bodies of information—the kind of information one might acquire by reading an article or a book rather than a word or a sentence. Consider Tung Ch’i-ch’ang (1555–1636), a Chinese painter, calligrapher and aesthetic theorist who has been the subject of countless detailed studies. At first encounter, his landscape paintings can appear awkward and even inept, but, for those who seek to understand this celebrated an influential master, there is much more than meets the untrained eye. Tung Ch’i-ch’ang belongs to a long tradition of “literati” painters, non-professional artists who trained for their own edification. Painters in this tradition came to increasingly value individualism in style, and they did not seek to paint nature as it appears but rather to capture inner spirit. In this respect, they departed from Confucian traditions, which were more naturalistic. Tung, who wrote copiously about art, was influenced by Zen Buddhism (called Ch’an in China) and he distinguished two schools of painting based on two versions of Zen, which emphasize gradual and sudden enlightenment, respectively. He argues

Penetrating Beauty 295 that the latter produced better art, because it is more spontaneous and less rigid (Cahill, 1988). Tung played a significant role in a seismic shift in Chinese landscape painting in which qi (roughly originality) replaces zheng (norms or standards). (Burnett, 2013) At the same time, Tung lobbied for a “Great Synthesis” in which techniques of previous generations would be incorporated into contemporary art (ibid., 326). Careful scrutiny of his brushstrokes reveals the influence of artists from the seventh through the fourteenth centuries. He also argues that painting should be continuous with calligraphy and, thus, the quality of the strokes becomes more important than the resemblance to nature. Such knowledge allows us to view Tung landscapes differently than we otherwise would. We attend to the brushwork and get less hung up about the way he distorts nature. Tung’s paintings are not depictions of nature so much as depictions of paintings—full of archaic references to past masters and urgently advancing his aesthetic philosophy. In recognizing this, Tung’s awkward mountains get seen anew as intriguing assemblies of texture that reference a millennium of Chinese art. In a similar spirit, Tung’s contemporary Shitao painted a work called 10,000 Ugly Inkblots (1685), which visually anticipates Pollock by over two centuries. Though representational, the ironic title invites viewers to attend to the paint rather than what is painted and to reconsider norms of beauty. Such works may be offputting when seen without the backstory, but they warrant and reward careful attention. Both visual and affective responses seem to change as one gains understanding of the artistic endeavor. Knowledge not only deepens our understanding of individual artists and movements but also aids us in making comparisons. For example, students of the Italian Renaissance learn that Venetian painters became quite accomplished with color (Humfrey, 1995). They had roots in the opulent aesthetics of Byzantium, and they had access to many pigments as a port city. They were also early adopters of oil paint, since the humid climate wasn’t ideal for frescos. This led to the development of both a rich palette and sophisticated techniques for layering and blending colors. Some artists worked directly on canvas without the aid of preparatory drawings. Florence, in contrast, was more focused on principles of disegno or design, which included clarity of drawing, perspective and other techniques that emphasize line more than the modeling of surfaces. Knowing this, I  submit, can impact aesthetic experience insofar as it makes these features more salient. To a casual observer, Filippo Lippi and Giovanni Bellini may look very similar, but a viewer equipped with this knowledge may immediately notice the crisp contours in Lippi and the vibrant, smoothly blended pigments in Bellini. These features may even appear more pronounced than they actually are. Distinctions of this kind are important throughout art history and contribute to understanding and appreciating art. Consider two textbook examples from China. It is sometimes suggested that artists from the Yuan

296  Jesse Prinz Dynasty, such as Huang Gongwang and Wu Zhen, tend to produce landscapes that are simpler than landscapes by artists such as Ma Yuan and Guo Xi from the Song Dynasty which preceded it (Zehou, 1988: 227). Such a pronouncement may surprise a causal viewer, since their works look equally complex to the untrained eye, but the Yuan may have a few commonplace elements and little motion in comparison to some Song artists who liked twisting lines and imposing forms. Learning about these changes in style can make paintings that appear relatively similar begin to look more different. Another distinction concerns northern and southern styles of landscape painting (Sullivan, 1999: 173). In the northern styles, lines tend to be angular and jagged. Practitioners include Jing Hao, Guan Tong, Lin Cheng and Fan Kuan. In the south, where mountains have softer contours, a more curvilinear style emerged. This is exemplified by Dong Yuan and his pupil, Juran. When a novice encounters monumental Chinese landscape paintings, she may seem them as interchangeable: A  central mountain shrouded in mist and dotted by trees. But experts see small differences: The angularity of lines, the brushstrokes uses, the height of the horizon and many other small details. These will impact perceptions of beauty. A novice may have no preference in ordering, but experts may see details that move and impress. One of the great authorities on Chinese art in English, James Cahill, claimed that Xia Gui was a traditional landscape painter, but I would venture that the beauty of his work would be lost on someone new to the genre. Cahill can see things in the work that blow him away, which others entirely miss as they observe with comparative ignorance. For some final examples of art historically informed comparisons, let’s turn from Imperial China to early modern Japan. Japan, too, has a long history of painting and, in the fifteenth century it began to develop pictorial styles that aimed to be distinctively Japanese (called yamato-e). These styles developed in various directions. Early on, Japanese painting tended to make ample use of empty space, leaving sizeable blank regions (yohaku). This was especially evident in work inspired by Zen Buddhism. But, in the Edo period, beginning at the tail end of the seventeenth century, a new form of mass art emerged that would depart dramatically from the Zen aesthetic: The woodblock print (later known as ukiyo-e). Early ukiyo-e tend to be relatively sparse in composition, but over time, their complexity increased. Ultimately, they repudiated emptiness (kū) and space (mu) and pursued compositions that are so crowded with colors, patterns and forms that they can be difficult to parse (Matsunosuke, 1997: 71). This difference, between the breathable and crowded imagery, is easy enough for a novice to see, but knowledge and a historical chronology may aid with the appreciation of both. Harder for novice eyes are the many distinctions that exist between ukiyo-e. Consider, for example, the changes that marked the transition from the Edo period into the Meiji period when the Emperor’s power was restored and

Penetrating Beauty 297 international contact increased (1868–1912). There are various stylistic changes, such as bolder compositions, harsher pigments and various signs of Western influence, but these are easy to miss for a novice. Even subtler are the differences between prints made in Edo (soon to become Tokyo) and Osaka. These are often similar in composition, but experts often find Osaka prints easy to identify at a glance. There are subtle difference in how faces are portrayed and a tendency toward more rounded postures and forms. Once a print is identified as coming from Osaka, such features become all the more apparent and they enter into appraisal; the perceived beauty of a work can depend on whether it is truly representative of the norms operative in its region of origin. Experts often operate according to a kind of dog-show principle that privileges good instances of a given kind. Seeing a print with expert eyes will impact both what gets seen and the quality of the forms. As a final possible example of vertical penetration, let me turn from information about a work to information about its maker. There are many cases where learning something about an artist can impact our appraisal. We already saw some example, when we talked about artist’s psychology and intentions. In those cases, the knowledge tells us something about the work. There are other cases, however, where information about the artist is less directly linked to the work but has some impact nonetheless. Here I will return to empirical evidence and mention two studies. In one, Seidel and I found that assessments of the very same image changed when we told people it was a reproduction of a Van Gogh painting as compared to when we told people it was by a Van Gogh forger (Seidel and Prinz, 2018a). We didn’t ask about beauty, but we asked about affective features such as amazement and interest. All of these got lower ratings in the forger condition. We also asked participants to estimate the size of the reproduction while looking at it. Participants in the forgery condition gave lower size estimates, implying that the attribution might have a perceptual as well as emotional impact. In another study, we presented participants with paintings by Hitler and other Nazi artists (Seidel and Prinz, 2018b). We either revealed the identity of the painters or we gave misinformation, attributing them to morally neutral individuals or to Holocaust rescuers. Here emotional appraisal went down for the Nazi artists and up for the rescuers, as compared to the neutral condition. Moral attitudes seem to leak into the appraisals of the work even though they are, arguably, unrelated. Again, we neglected to test beauty assessments and we didn’t take any visual measures, but I would predict some impact on both. Findings like this are open to different interpretations. Perhaps participants love Hitler’s paintings but don’t want to admit it. Without further work, we can say only that vertical effects are a possibility. Future work can settle the issue. Against the background of the empirical research summarized earlier, the sentimentalist theory of beauty and the many examples from art history,

298  Jesse Prinz I speculate that both fame (the master vs. forger study) and infamy (the Hitler study) can have an impact of how artworks look both visually and aesthetically.

5. Conclusions In this discussion, I tried to make a case for the lateral and vertical penetration of aesthetic perception. I  sketched a sentimentalist theory of beauty according to which beautiful things have formal properties that arouse positive emotional responses. The range of formal qualities and emotions is large and culturally conditioned; beauty has culturally specific subtypes. Perceptual and affective dimensions of beauty are bound together, giving the impression that aesthetic qualities are out there to be perceived. Both dimensions, I  argued, are influenced by extraneous factors, including moods, motor actions, perceptions in other modalities and, most strikingly, knowledge. In defense of the thesis that knowledge impacts aesthetic perception, I mostly draw on examples from art history and criticism, though I ended with some empirical results. I discussed a range of possible cases. Knowledge can reveal motivations and intentions, drive visual search, cultivate sensibilities to features, teach us about local norms, orient us to critic’s classifications, clarify what it is we are seeing, deepen our appreciation of art-historical significance, educate us about important distinctions and alert us to concerns about artists’ character. In each case, I propose that knowledge may impact both what we perceive and what we feel, and thus knowledge has the potential to make works seem more or less beautiful. Critics will complain that my examples are speculative. I agree. Testing would be welcome. I offered a general defense of cognitive penetration, which, if successful, would show that penetration is possible in principle. The examples show that penetration may take many forms and is likely to be quite pervasive. Critics may also complain that most (though not all) of my examples of contaminated perception involve shift of ­attention: Knowledge can impact what features we notice. I  argued earlier that attention characteristically involves changes in perceptual processing, so such effects would qualify as cases of penetration (see also Stokes, 2014). I would also like to point out that there is a way in which the debate about penetration is moot. Proponents of modularity want to uphold Ruskin’s innocent eye thesis: The idea that we can see things in a pure way, as they really are. In a sense, this thesis is untenable even if we grant that attention does not directly impact perceptual processing. If attention simply directs the direction of the gaze, that is enough to ensure that knowledge can dramatically impact the experience of art. Vision depends on the eyes, which are instruments of human agency that can be directed by knowledge and intentions. Two individuals with the same visual acuity and processing mechanisms may use their eyes very differently.

Penetrating Beauty 299 Outside of the aesthetic domain, modularists worry that such freedom of movement will undermine the epistemic integrity of vision. The aesthetic case is a clear counterexample. We often need knowledge to know how to look properly at art. Without it, we systematically misperceive. Eyes that are more innocent, in the sense of being more ignorant, are also more inept. With knowledge we stand a better chance of seeing the aesthetic qualities that the creators and original consumers of artworks wanted us to see. The penetration of beauty is not a hazard but a hope—something we should cultivate if we aim to view works in a way that approximates the experiential ends for which they were created.

References Baxandall, Michael (1972) Painting and Experience in Fifteenth Century Italy: A Primer in the Social History of Pictorial Style, Oxford: Oxford University Press. Cahill, James (1988) “Tung Ch’i-ch’ang’s ‘Southern and Northern Schools’ in the History and Theory of Painting: A  Reconsideration” in Sudden and Gradual: Approaches to Enlightenment in Chinese Thought (ed. Peter N. Gregory), Hawaii: University of Hawaii Press, Honolulu (Studies in East Asian Buddism, 5), 429–446. Duchamp, Marcel (1916) With Hidden Noise, The Louise and Walter Arensberg Collection, Philadelphia Museum of Art, Philadelphia, PA. Eskine, Kendall J., Kacinik, Natalie A. and Prinz, Jesse J. (2012) “Stirring Images: Fear, Not Happiness or Arousal, Makes Art More Sublime” in Emotion, 12 (5), 1071–1074, doi:10.1037/a0027200. Firestone, Chaz, and Scholl, Brian J. (2016) “Cognition Does Not Affect Perception: Evaluating the Evidence for ‘Top-Down’ Effects” in Behavioral and Brain Sciences, 39, 1–77, doi:10.1017/S0140525X15000965. Fodor, Jerry A. (1983) The Modularity of Mind, Cambridge, MA: MIT Press. Freeman, Jonathan B., Penner, Andrew M., Saperstein, Aliya, Scheutz, Matthias and Ambady, Nalini (2011) “Looking the Part: Social Status Cues Shape Race Perception” in PLoS ONE6, 9, doi:10.1371/journal.pone.0025107. Gasper, Karen, and Clore, Gerald L. (2002) “Attending to the Big Picture: Mood and Global Versus Local Processing of Visual Information” in Psychological Science, 13 (1), 34–40, doi:10.1111/1467–9280.00406. Gilbert, Charles D. and Sigman, Mariano (2007) “Brain States: Top-Down Influences in Sensory Processing” in Neuron, 54 (5), 677–696, doi:10.1016/j. neuron.2007.05.019. Gombrich, Ernst (1960) Art and Illusion: A Study in the Psychology of Pictorial Representation, Princeton, NJ: Princeton University Press. Hodgson, John E. (1888) “Old Arts and Modern Thoughts: Poetical Treatment” in The Magazine of Art, 11, 361–365. Humfrey, Peter (1995) Painting in Renaissance Venice, New Haven, CT: Yale University Press. Isen, Alice M. and Shalker, Thomas E. (1982) “The Effect of Feeling State on Evaluation of Positive, Neutral, and Negative Stimuli: When You ‘Accentuate

300  Jesse Prinz the Positive,’ Do You ‘Eliminate the Negative’?” in Social Psychology Quarterly, 45 (1), 58–63, doi:10.2307/3033676. Król, Magdalena Ewa, and El-Deredy, Wael (2011) “When Believing Is Seeing: The Role of Predictions in Shaping Visual Perception” in Quarterly Journal of Experimental Psychology, 64 (9), 1743–1771, doi:10.1080/17470218.2011. 559587. Langer, Ellen, Djikic, Maja, Pirson, Michael, Madenci, Arin and Donohue, Rebecca (2010) “Believing Is Seeing” in  Psychological Science, 21 (5), 661– 666, doi:10.1177/0956797610366543. Li, Zehou (1988) The Path of Beauty: A Study of Chinese Aesthetics, Oxford, UK: Oxford University Press. Lupyan, Gary (2015) “Object Knowledge Changes Visual Appearance: Semantic Effects on Color Afterimages” in Acta Psychologica, 161, 117–130, doi:10.1016/j.actpsy.2015.08.006. McMahon, Jennifer A. (2012) “The Aesthetics of Perception: Form as a Sign of Intention” in Essays in Philosophy, 13 (2), 404–422, doi:10.7710/1526–0569.1428. Minissale, Gregor (2009) Images of Thought: Visuality in Islamic India 1550– 1750, Cambridge: Cambridge Scholars. Mole, Christopher (2015) “Attention and Cognitive Penetration” in The Cognitive Penetrability of Perception: New Philosophical Perspectives, 218–237, doi:10.1093/acprof:oso/9780198738916.003.0010. Nanay, Bence (2015) “Cognitive Penetration and the Gallery of Indiscernibles” in Frontiers in Psychology, 5, doi:10.3389/fpsyg.2014.01527. Nardi, Isabella (2006) Theory of Citrasutras in Indian Painting: A Critical Reevaluation of Their Uses and Interpretations, Abingdon, UK: Routledge. Nishiyama, Matsunosuke and Groemer, Gerald (1997) Edo Culture: Daily Life and Diversions in Urban Japan, 1600–1868, Honolulu, HI: University of Hawaiʻi Press. Prinz, Jesse and Seidel, Angelika (2012) “Alligator or Squirrel: Musically Induced Fear Reveals Threat in Ambiguous Figures” in Perception, 41 (12), 1535–1539, doi:10.1068/p7290. Proffitt, Dennis R. and Linkenauger, Sally A. (2013) “Perception Viewed as a Phenotypic Expression” in Action Science, 171–197. doi:10.7551/mitpress/ 9780262018555.003.0007. Richie, Donald (2007) A Tractate on Japanese Aesthetics, Berkeley: Stone Bridge Press. Roberson, Debi and Davidoff, Jules (2000) “The Categorical Perception of Colors and Facial Expressions: The Effect of Verbal Interference” in Memory & Cognition, 28 (6), 977–986, doi:10.3758/bf03209345. Santayana, George (1896) The Sense of Beauty, New York, NY: Charles Scribner’s Sons. Seidel, Angelika and Prinz, Jesse (2018a) “Great Works: A Reciprocal Relationship between Spatial Magnitudes and Aesthetic Judgment” in Psychology of Aesthetics, Creativity, and the Arts, 12 (1), 2–10, doi:10.1037/aca0000100. Seidel, Angelika, and Prinz, Jesse (2018b) “Impure Aesthetics” in Advances in Experimental Philosophy of Aesthetics (ed. Florian Cova and  Sébastien Réhault), London: Bloomsbury. Shams, Ladan, Kamitani, Yukiyasu and Shimojo, Shinsuke (2000) “What You See Is What You Hear” in Nature, 408 (6814), 788, doi:10.1038/35048669.

Penetrating Beauty 301 Stefanucci, Jeanine K., Gagnon, Kyle T. and Lessard, David A. (2011) “Follow Your Heart: Emotion Adaptively Influences Perception” in Social and Personality Psychology Compass, 5 (6), 296–308, doi:10.1111/j.1751–9004.2011.00352.x. Stein, Barry E., London, Nancy, Wilkinson, Lee K. and Price, Donald D. (1996) “Enhancement of Perceived Visual Intensity by Auditory Stimuli: A  Psychophysical Analysis” in  Journal of Cognitive Neuroscience, 8 (6), 497–506, doi:10.1162/jocn.1996.8.6.497. Stokes, Dustin (2014) “Cognitive Penetration and the Perception of Art” in Dialectica, 68 (1), 1–34, doi:10.1111/1746–8361.12049. Sullivan, Michael (1999) The Arts of China, Oakland, CA: University of California Press. Vanlessen, Naomi, De Raedt, Rudi, Koster, Ernst H. W. and Pourtois, Gilles (2016) “Happy Heart, Smiling Eyes: A Systematic Review of Positive Mood Effects on Broadening of Visuospatial Attention” in Neuroscience  & Biobehavioral Reviews, 68, 816–837, doi:10.1016/j.neubiorev.2016.07.001. Wartofsky, Marx (1981) “Sight, Symbol and Society: Toward a History of Visual Perception” in Philosophic Exchange, 12 (1), 23–38. Wegbreit, Ezra, Franconeri, Steven and Beema, Mark (2014) “Anxious Mood Narrows Attention in Feature Space” in Cognition and Emotion, 29 (4), 668– 677, doi:10.1080/02699931.2014.922933. Witt, Jessica K. (2016) “Action Potential Influences Spatial Perception: Evidence for Genuine Top-down Effects on Perception” in Psychonomic Bulletin  & Review, 24 (4), 999–1021, doi:10.3758/s13423-016-1184-5. Wölfflin, Heinrich (1950) Principles of Art History: The Problem of the Development of Style in Early Modern Art, Mineola, NY: Dover Publications.

15 Gist Experience Dena Shottenkirk

I walk into The Metropolitan Museum of Art in New York City, make my way up the wide, stone front staircase and go into gallery 635. It is a large-seeming room, with lighting that is neither dim nor bright and about two dozen paintings hanging on the walls. And when I look to my left, I see a row of Velázquez paintings hanging on the wall. I walk over to the middle one. What do I  see? And how do I  see it? What is the process of that experience? This is, of course, a subset of a fundamental and reoccurring question in philosophy: How do we experience the world? How, in other words, does the external world come to reside in us? The traditional view is that there is the act of perception and then there is the process of cognition and that the two procedures are mutually exclusive of one another. In this view, perception provides the basic fodder for thought, but thought shouldn’t assist in the construction of perception (e.g., to think you see something you don’t in fact see is a hallucination, called such because it is an instance of failed perception), and perception is simply unable to assist in thought—sensory experience is restricted to the embodied realm. Objects are to be seen in perception, and then thought is to use that data in abstraction and concept formation. There have been some recent moves against this traditional view. I  agree with this opposition and enter the dialogue by examining gist in perceptual experience, using the aesthetic experience as illustrative of perceptual experiences in general.

1.  Gist: The Science Gist is the first few hundred milliseconds of a perceptual experience, generally defined as perceptual data gained in the first 300 msec. There is wide agreement that what is gained at this moment is surprisingly full. It includes both low-level features and higher-level information. The former includes spatial relations and color, and the latter includes knowledge of the semantic category, e.g., whether it is a street scene or indoor scene and

Gist Experience 303 whether the scene is orderly and harmonious. Gist is, in essence, wholescene recognition, and our brains seem interestingly designed to see this before we see discrete, individual objects. Gist is that part of the perceptual experience that is given immediately and often nonconsciously. If a scene is presented for a second, it is easy to remember, but if it is shown for only 300 msec, about half of the scene is not remembered (Potter and Levy, 1969: 12). This notion of “remember” was tested by the subjects’ ability to recall the image. The question that arose at this point (e.g., 1970s) was whether those unremembered gist experiences at 300 msec were experienced at all. Using rapid serial visual presentation (RSVP), seminal research tested whether subjects seeing images at