Thinking and Perceiving: On the Malleability of the Mind 2020053855, 2020053856, 9781138729384, 9781138729391, 9781315189895
302 57 3MB
English Pages [331] Year 2021
Polecaj historie
![Perceiving Truth and Value: Interdisciplinary Discussions on Perception as the Foundation of Ethics [1 ed.]
9783666573200, 9783525573204](https://dokumen.pub/img/200x200/perceiving-truth-and-value-interdisciplinary-discussions-on-perception-as-the-foundation-of-ethics-1nbsped-9783666573200-9783525573204.jpg)
Citation preview
THINKING AND PERCEIVING
On the Malleability of the Mind
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Human beings are in contact with the world through their minds. One can make sensory perceptual contact with the world: One sees the tree and hears its leaves flutter. And one makes cognitive contact with the world: One forms beliefs about the tree, memories of how it was in the past, and expectations of how it will be in the future. Can the first, perception, be influenced in important ways by the second, cognition? Do cognitive states such as memories, beliefs, and expectations affect what one perceives through the senses? And what is the importance of these possible relations to how we theorize and understand the human mind? Possible cognitive influence on perception (sometimes called “cognitive pene tration of perception”) has been long debated in philosophy of mind and cognitive science: Some argue that such influence occurs, while others argue that it does not or cannot. In this excellent introduction and overview of the problem, Dustin Stokes examines the following: •
The philosophical and scientific background to cognition and perception
•
Contemporary ways of distinguishing cognition and perception
•
Questions about the representational content of perception versus cognition
•
Distinct theories of mental architecture: modularity versus malleability
•
Consequences for epistemology, philosophy of science, and aesthetics
•
Philosophical and scientific research on perceptual attention
•
Perceptual skill, learning, and expertise
•
Perceptual content, objectivity, and cultural bias.
Additional features, such as chapter summaries, suggestions for further reading, and a glossary, make Thinking and Perceiving an ideal resource for students of philoso phy of mind and psychology, cognitive psychology, and cognitive science. Dustin Stokes is Associate Professor of Philosophy at the University of Utah, Salt Lake City, USA. He works primarily on the philosophy and science of perception, imagination and imagery, and creativity. With Stephen Biggs and Mohan Matthen, he is editor of Perception and Its Modalities (2014).
NEW PROBLEMS OF PHILOSOPHY
Series Editor: José Luis Bermúdez
“New Problems of Philosophy is developing a most impressive line-up of topical volumes aimed at graduate and upper-level undergraduate students in philosophy and at others with interests in cutting edge philosophical work. Series authors are players in their respective fields and notably adept at synthesizing and explaining intricate topics fairly and comprehensively.” — John Heil, Monash University, Australia, and Washington University in St. Louis, USA “This is an outstanding collection of volumes. The topics are well chosen and the authors are outstanding. They will be fine texts in a wide range of courses.” — Stephen Stich, Rutgers University, USA
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The New Problems of Philosophy series provides accessible and engaging surveys of the most impor tant problems in contemporary philosophy. Each book examines either a topic or theme that has emerged on the philosophical landscape in recent years, or a longstanding problem refreshed in light of recent work in philosophy and related disciplines. Clearly explaining the nature of the problem at hand and assessing attempts to answer it, books in the series are excellent starting points for undergraduate and graduate students wishing to study a single topic in depth. They will also be essential reading for professional philosophers. Additional features include chapter summaries, further reading and a glossary of technical terms. Cognitive Phenomenology
Self-Deception
Elijah Chudnoff
Eric Funkhouser
Disjunctivism
Relativism
Matthew Soteriou
Maria Baghramian and Annalisa Coliva
The Metaphysics of Identity
Empathy
André Gallois
Heidi Maibom
Consciousness
Moral Disagreement
Rocco J. Gennaro
Richard Rowland
Abstract Entities
Perception
Sam Cowling
Adam Pautz
Embodied Cognition, Second Edition
Thinking and Perceiving
Lawrence Shapiro
On the Malleability of the Mind Dustin Stokes
For more information about this series, please visit: www.routledge.com/New-Problems-of Philosophy/book-series/NPOP
THINKING AND PERCEIVING
On the Malleability of the Mind
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Dustin Stokes
First published 2021 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 605 Third Avenue, New York, NY 10158 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 Dustin Stokes The right of Dustin Stokes to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Library of Congress Cataloging-in-Publication Data Names: Stokes, Dustin, author.
Title: Thinking and perceiving / Dustin Stokes.
Description: Abingdon, Oxon ; New York, NY : Routledge, 2021. | Series: New
problems of philosophy | Includes bibliographical references and index. Identifiers: LCCN 2020053855 (print) | LCCN 2020053856 (ebook) | Subjects: LCSH: Philosophy of mind. | Perception. | Cognition. | Thought and thinking. Classification: LCC BD418.3 .S77 2021 (print) | LCC BD418.3 (ebook) | DDC 128/.2—dc23 LC record available at https://lccn.loc.gov/2020053855 LC ebook record available at https://lccn.loc.gov/2020053856 ISBN: 978-1-138-72938-4 (hbk) ISBN: 978-1-138-72939-1 (pbk) ISBN: 978-1-315-18989-5 (ebk) Typeset in Joanna and Scala Sans by Apex CoVantage, LLC
Copyright © 2021. Taylor & Francis Group. All rights reserved.
To my mother, who was the first to instill in me a love for the spoken word, and who has never failed to support me, with love and earnest encouragement, through my life’s pursuits.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
9
Taylor & Francis
Taylor & Francis Group http://taylorandfra ncis.com
CONTENTS
Acknowledgements List of figures
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Introduction
xi
xv
1
1 Thought and perception: distinctions, similarities, and relations 1.1 “Defining” by ostension 1.2 The importance of cognitive influence on perception:
an initial gloss 1.3 Reductive and revisionary theories 1.4 Perception and thought: grounds for distinction Subject activity Sensory organ activity Presence of appropriate stimuli Conceptual requirements Phenomenology Summary 1.5 Mental content and its importance Determining the content of mental states The importance of mental content and its variety Concepts and perceptual content The admissible contents of perceptual experience 1.6 Summary Further reading
10
11
16
18
23
23
25
25
26
28
29
30
30
32
34
36
39
40
2 Mental architecture: the case for modularity 2.1 The modularity of mind: the essentials Domain specificity of modules Mandatoriness and speed of modules Informational encapsulation: the essence of modularity
44
49
51
51
52
VIII
CONTeNTS
2.2 The modularity of mind: arguments 2.2.1 Arguments from the stability of perception 2.2.2 Arguments from the reliability of perception 2.3 Summary Further reading
54
55
61
70
70
3 Top-down effects on perception: the initial case for malleability 3.1 Value-influenced perception 3.2 Doxastic influences on perception 3.3 Social categories, stereotypes, and perception 3.4 Intention, action, and perception 3.5 Language and top-down effects on vision 3.6 Summary: a stalemate of sorts Further reading
74
75
78
79
80
81
83
84
4
86
88
89
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5
Towards a consequentialist understanding of cognitive penetration 4.1 Definitions and the problem of cross-talk 4.2 Two definitions and their application Divergent verdicts: the definitions applied to cases
of value-influenced perception Divergent verdicts: the definitions applied to cases of diachronic
changes in perception 4.3 Consequences of cognitive penetration 4.4 Consequentialism 4.5 Conclusion and summary Further reading Attention and cognitive influence on perception 5.1 Rethinking the role of attention in cognitive effects on perception 5.2 Four questions about attention-involving, cognitive-
perceptual phenomena 5.3 Two arguments for attention-involving, cognitive penetration
of perception 5.3.1 An argument from directness 5.3.2 An argument from consequences 5.4 Summary: moving beyond cognitive penetration Further reading
6 Perceptual expertise I: mental architecture 6.1 From worse to better perceivers 6.2 Perceptual experts are perceptual experts 6.3 Perceptual expertise as a genuinely perceptual phenomenon 6.4 Perceptual expertise as genuine cognitively sensitive expertise 6.5 Situating perceptual expertise in theories of the architecture
of mind 6.6 Summary: towards the malleability of mind Further reading
91
94
99
105
111
112
115
117
119
130
131
135
138
139
141
144
148
151
162
166
169
169
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONTeNTS
7 Perceptual expertise II: epistemology 7.1 An epistemology of perceptual expertise: initial sketches 7.2 expertise and epistemic virtue 7.3 Perceptual improvement and biological function 7.4 Objections and qualifications 7.5 Summary: thinking improves perceiving Further reading
173
175
179
182
194
202
203
8
Consequences of perceptual malleability 8.1 The generality of perceptual expertise 8.2 Theory-laden perception and scientific investigation 8.3 Theory-laden perception, implicit bias, and epistemology 8.4 Objectivity, accuracy, and perceptual content 8.5 expertise and the admissible contents of experience 8.6 Summary Further reading
206
207
210
216
223
231
234
235
Conclusion epilogue: the arguments from understanding Glossary Bibliography Index
238
245
257
270
300
IX
Copyright © 2021. Taylor & Francis Group. All rights reserved.
9
Taylor & Francis
Taylor & Francis Group http://taylorandfra ncis.com
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ACKNOWLEDGEMENTS
This book is the culmination of more than a decade of my research. Over that time, I have benefitted enormously from countless conversations with friends and colleagues. I’ve always considered philosophy and science to be a collective and cooperative effort. In fact, this sounds almost trivially true, resisted perhaps only by the most stubborn (and thankfully antiquated) arm chair philosopher. In any case, or I should say in my case, this book would not have been possible without the collective wisdom and constructive feed back of a great many people. Of special mention, for their continued support, professional acumen, and simply exceptional philosophy, are two of my teachers: Dominic McIver Lopes and Mohan Matthen. In their own ways, they have offered invaluable lessons and guidance over nearly twenty years. I thank Dom especially for inspiring and then helping to cultivate my interest in issues at the intersec tion of aesthetics and cognitive science. And I thank Mohan especially for inspiring me to pursue the philosophy of perception, and to do so in ways richly informed by empirical study. Two earlier graduate teachers of mine, to whom I am indebted for their encouragement and patience, are Tamar Szabo Gendler and the late Phil Peterson. I also want to thank some under graduate teachers of mine, in particular, Bill Brown, the late Bert Helm, and Jack Knight. In 1998, Jack boldly advised me to abandon my plans to go to law school and consider pursuing a career in academic philosophy. Here I am, Jack.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
XII
ACkNOwLeDGeMeNTS
There is a small handful of professional colleagues that have discussed, at length, many of the issues and analyses that populate this book. I am grate ful for their friendship and for their minds: Vince Bergeron, Steve Downes, Fiona Macpherson, Chris Mole, Bence Nanay, Jeanine Stefanucci, Jonna Vance, and Wayne Wu. A special note to Vince, who is my closest friend and an exceptional collaborator. Thank you for your scientific understanding, your attention to detail, and your joie de vivre. I am also especially grateful to my colleagues here at Utah, for their con sistent support, their collegiality, and their open-mindedness. It’s one thing to have a large number of exceptionally talented philosophers and scientists to share the office with. All the better when they are simply just good people. I am proud to call you my friends and colleagues. Thanks also to the many wonderful students I have had here at Utah, and before that at the University of Toronto. The list of individuals who have shaped my thinking on thought and perception through conversation, criticism, and correspondence is longer than I capture here. But here is some of that list (and my genuine apologies in advance if I miss somebody): Greyson Abid, Murat Aydede, David Bain, Emily Balcetis, Katerina Batinaki, Edison Barrios, Tim Bayne, Jake Beck, Erin Beeghly, Anna Bergqvist, Stephen Biggs, Anna Borghi, Derek Brown, Dan Burnston, Marisa Carrasco, Roberto Casati, Ariel Cecchi, Dave Chalm ers, Phillipe Chuard, Kevin Connolly, Becko Copenhaver, Rob Cowan, Tim Crane, Sarah Creem-Regehr, David Davies, Felipe De Brigard, Ophelia Deroy, Jérôme Dokic, Trafton Drew, Santiago Echeverri, Anya Farennikova, Peter Fazekas, Joerg Fingerhut, Chaz Firestone, Craig French, Matt Fulkerson, Rob Goldstone, Bruce Glymour, Steve Gross, Matt Haber, Jim Hamilton, Jakob Howhy, Spencer Ivy, Aaron Kenna, John Kulvicki, Heather Logue, Gary Lupyan, Jack Lyons, Edouard Machery, Raamy Majeed, Francesco Marchi, Lije Millgram, Albert Newen, Lex Newman, Shaun Nichols, Matt Nudds, Casey O’Callaghan, Nico Orlandi, Elliot Paul, Keith Payne, Denny Proffitt, Jesse Prinz, Diana Raffman, Athanasios Raftopoulos, Madeleine Ransom, Gurpreet Rattan, Indrek Reiland, Taren Rohovit, Kateryna Samoilova, Jonah Schupbach, Sonia Sedivy, Nick Shea, Susanna Siegel, Charles Siewert, Nico Silins, Barry Smith, A. David Smith, Charles Spence, Cindy Stark, Kathleen Stock, Dave Strayer, Jim Tabery, Evan Thompson, Anna Vaughn, Sebastian Watzl, Jonathan Weinberg, Keith Wilson, Hong Yu Wong, Ben Young, and John Zeimbekis. I am grateful to have had the opportunity to present some of this material at a number of excellent universities and institutions, including the Antwerp
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ACkNOwLeDGeMeNTS
Centre for Philosophical Psychology, Berlin School of Mind and Brain, Uni versity of Bochum, Bristol University, Inter-University Centre-Dubrovnik, Carleton University-Ottawa, University of Crete, Institut Jean-Nicod-Paris, Institute of Philosophy-London, Kansas State University, Glasgow Univer sity, University of Nevada-Reno, and University of Utah. And at so many stimulating workshops and conferences, including the Annual Philosophy of Science Conferences-Dubrovnik, the Annual Meeting of the American Society for Aesthetics, the Pacific Division Meeting for the American Society for Aesthetics, the Pacific Division Meeting for the American Philosophical Association, the Canadian Philosophical Association Annual Meeting, the Cognitive Science Society Annual Meeting, the Bled Philosophical Confer ence on The Intersection of Epistemology and Philosophy of Mind-Slove nia, the Cognitive and Cross-Modal Effects on Vision Workshop-Glasgow, the Annual Meeting of the European Society for Philosophy and Psychol ogy, the Annual Meeting of the Society for Philosophy and Psychology, the Evaluative Perception Conference-Glasgow, the Conference on Impure Perception-Berlin, the International Summer School in Cognitive Science and Semantics-Perception-Riga, the Top-Down Influences in Perception-Re thinking the Senses Workshop-Glasgow, and the Workshop on Cognitive Penetrability and Predictive Coding-Bochum. Thank you to the series editor, José Bermudéz, and the editors at Routledge – Tony Bruce and Adam Johnson – for their encouragement, feedback, and patience. I’m excited for this book to be part of the excellent New Problems of Philosophy series. Thanks also to three anonymous peer reviewers, the con sequence of which is a much-improved book. Finally, thanks to my family: to Mom, my brother Ryan, and Jenn. To my son Xae. I hope that some day you will find value in some of these pages. To Brady, a lifelong friend. And to my Steph, who has stood by me in finishing a book through personal tragedy, a country rife with social and political turmoil, and, now, a global pandemic. I am grateful for your love, smile, and unfailing optimism.
XIII
Copyright © 2021. Taylor & Francis Group. All rights reserved.
9
Taylor & Francis Taylor & Francis Group http://taylorandfra ncis.com
FIGURES
2.1 2.2 2.3 3.1 5.1 5.2
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5.3 6.1 6.2 6.3 6.4 6.5 7.1
The Müller-Lyer illusion “Mooney” image 1 “Mooney” image 2 Example of binocular rivalry stimulus The duck–rabbit image Examples of stimuli for visual search task in studies
on competition in visual attention The Rubin goblet Greebles Examples of composite task structure for faces, Greebles,
and musical notes A part–whole task Spatial change vs. feature change task Results of visual short-term memory studies on car
experts vs. car novices Example of a spatial occlusion task: whole-body vs.
hip-only conditions
56
68
69
78
118
122
129
147
153
155
156
158
191
Copyright © 2021. Taylor & Francis Group. All rights reserved.
9
Taylor & Francis
Taylor & Francis Group http://taylorandfra ncis.com
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INTRODUCTION
To be in the world is to be in contact with it. This is not a distinctively human phenomenon, or even one exclusive to living organisms. Tables and chairs, sticks and stones, soundwaves and bits of matter are all objects that are in contact with other worldly objects. An intuitive notion of contact can help us to understand similarly intuitive features – which isn’t to say unimportant or unreal features – of the animate and inanimate world and, importantly, about the human mind. Try to imagine an object in the world that is not in contact with anything else. You may find this hard to do. You can, of course, imagine an object, perhaps any object, all alone in some science fictional void. Philosophers allege to be very good at this kind of thought experiment. But this is not to imagine a lone object in the world. So then one might think of abstract objects. For example, some believe that there exist objects beyond those in the phys ical world, such as numbers. But you don’t step on a number on your way to the café or bump into one as you get a drink of water in the middle of the night. Numbers are not, in an important sense, in the world. One way to make sense of this non-worldly existence of a number is to say that it is not in contact with worldly objects, whether it is you or me, or a stick or a stone. Worldly objects, then, seem to be objects that can be in contact with others. This may just be what it means to say that something is in the world. This raises a new question: What are the different ways that objects can be in contact with one another?
Copyright © 2021. Taylor & Francis Group. All rights reserved.
2
INTRODuCTION
From an intuitive folk perspective, objects are in contact when they are “in touch” with one another. My coffee cup sits on my desk. My feet are touching the floor. And so on. And, intuitively, objects that are in contact can often move one another. The coffee cup can be moved across the desk by a third object, say my hand. It is now in contact both with the desk and with my hand. Not all objects can move themselves, but most can be moved by some other object(s). Physics largely vindicates these and other intuitions about contact, identifying types of spatial relations between objects and how various forces can effectively exploit those spatial relations for causal ones. The coffee cup moved by the hand is described by the physicist as a contact force where, strictly speaking, the objects do not touch but electrons near their surfaces do. When this happens, the rest/motion state of one object is changed by another. Although it may use terms other than “contact”, modern physics further identifies a number of ways that physical entities interact – electrical interactions, magnetic forces, gravitational pulls – which further clarify how those entities are in contact with one another. The phys ical world of objects is a world of entities in contact. Physics illuminates what folk intuition starts with (that, and more, of course). That said, the notion of “contact” used here is not, or at least is not restricted to, a technical one like one(s) identified by the physicist, but is instead an intuitive notion useful for identifying differences and similarities between the ways that we situate ourselves, as creatures with very sophisti cated minds, within the world. It includes causal relations but is not exclu sive to them. The pebble on the ground is in contact with the ground, but perhaps neither is causally affecting the other. Maybe we want to describe this as a potential causal relation: For instance, if the ground shakes, the peb ble will be moved. Contact can also occur at great distance. The ultraviolet light from the sun doesn’t move the (macroscopic) object, but waves of light do get absorbed by most objects and, consequently, they heat. We naturally say of such a situation that the sun is “hitting” or “touching” it. This, too, is an intuitive form of contact between entities. Now consider objects with minds: a bird, a dog, a dolphin, or us. As phys ical objects, such creatures can obviously be in straightforward physical con tact with others. They can also be moved causally by other objects and can themselves animate other objects. But they can importantly do more than this and, like the sun, from considerable distance. The bird that hears a song from its mate across the field is in touch with its mate. The dog that sees the frisbee as it flies through the air is in contact with the frisbee. The dolphin maintains contact with its peers by transmitting and receiving meaningful
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INTRODuCTION
noises underwater. These are familiar ways that an animal can be in contact, through its senses, with the world. Contact with the world is not distinctive of humans. But what may be dis tinctively human is the richness and variety of the contact we can make with the world, with our bodies and minds. Most obviously, humans use their bodies to carry out actions: to walk across the grove, reach for and grasp an apple, take a bite of the apple. Perhaps less actively, one is also in contact with the world through one’s sensory perceptual capacities, just like other animals. With some sensory modalities, this contact may seem less direct than bodily action. But here, too, there is exchange of energy. With audition, for example, sound waves in the immediate environment cause vibrations in the eardrum; this pressure is amplified to affect fluids in the inner ear – the cochlea; activity in these fluids causes vibrations in tiny fibres in the coch lea; and finally, certain frequencies of those vibrations activate tiny hair cells which, through the cochlear nerve, send electrical energy to the cerebral cortex. So although through a long causal chain, there is a clear sense in which the auditory experience that one enjoys is a way of being in contact with the world outside of oneself. This sensory transduction likewise occurs for experiences in other sensory modalities. Perhaps the least directly, humans can be in contact with the world by using various modes of cognition: We form beliefs about the objects in our environment, memories of how it was in the past, and expectations of how it will be in the future. The reader might pause here to consider the importance of how we make contact with our world and with one another. Imagine how dramatically different, per haps inhuman, our existence would be if we did not enjoy the richness of contact that our minds afford. One way to understand this book is as an attempt to make sense of perceptual contact and cognitive contact and their relations. And this is no mere academic or scientific exercise. What perceptual contact gives one, when one’s sensory systems are work ing properly, is information about one’s environment. The visual experi ence of the apple on the tree carries information, for the perceiver, about important features of one’s world. So, in addition to, and plausibly because of, the exchange of energy between world and perceiver, there is an exchange of information between world and perceiver. This is what cognitive contact provides as well, even if it is contact that does not (always) involve exchange of energy. When I am recalling my experience of the apple grove yesterday, I maintain contact with that part of the world by carrying information about it. This is true of any kind of accurate thought that one has about the world: Those thoughts stand in informational relations with the objects, features,
3
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4
INTRODuCTION
and events that they are about. Put in another idiom, perceptual and cogni tive contact are (re)presentational connections with the world. This suggests a way of marking a distinction that some find intuitive and that philosophy and psychology have tended to mark in theory: the distinc tion between perception and cognition, between, say, seeing an apple versus thinking about an apple. The first seems to require exchange of energy, while the second does not. Except in rare cases of hallucination, any par ticular perceptual experience is one that requires some exchange of energy between world and perceiver. Some feature of the world has to cause some kind of response in one’s sensory organs. This is not true for any particular cognitive state that one might have. One can have beliefs, memories, expec tations, and so on about things with which one is not presently in that kind of contact (that is, where no relevant exchange of energy is occurring). But note, at the same time, how both involve maintaining contact through information: My visual experience and my memory of the apple both carry information about that worldly object. Which matters more: the difference with respect to energy or the similarity with respect to information? And how will either contribute to vindicating (or undermining) the distinction between thinking and perceiving? If we consider contact in the basic sense of two objects causally relating and in the sense of sensory contact, we see that there is not a clean or easy distinction to be made between the first and the second, at least not with respect to the exchange of energy. But in philosophy and cognitive science, it is assumed by many that there is some clean distinction between the kind of informational contact with the world that, say, vision affords and the kind of informational contact that, say, belief affords. Again, cognition is supposed to be distinct from sensory perception, and moreover the two are supposed to function in importantly independent ways. This is often described with computational language. The computations performed by vision, once it has received information (via energy) from the external worldly stimulus, run independently of any computations being run by cognitive systems. The central question to be discussed over the chapters that follow is whether and how this is true. Are cognition and perception so isolated? Or given that they are both rich and important means by which humans achieve and maintain informational contact with the world, are they connected and intertwined in ways that depart from the theoretical orthodoxy? Put most simply, how much, and in what ways, does thought influence perception? In Chapter 1, thought and perception will be distinguished with more pre cision, with a number of possible criteria for distinction discussed. For now,
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INTRODuCTION
just grant the intuitive distinction between cognitive states like belief or judg ment, on the one hand, versus sensory perceptual experiences like seeing or hearing, on the other. The question is how the first, cognition, might influ ence the second, perception. Consider some example cases. Some of these are phenomena about which no party would disagree. That is, they actually happen, and we can be confident about this from both an intuitive and scien tific perspective. Others are more controversial. It’s the controversial cases that will populate this book, but we do well to identify a range of cases, moving from the least to the most controversial so that, ultimately, we can identify structural similarities and differences between them and, with any luck, better understand how the mind works and how it puts us in contact with the world. 1 The first time you saw the Müller-Lyer illusion you, like most subjects, perceived one line as longer than the other. But then someone informed you that this was an illusion and that, in fact, the lines were of the same length. You believed your informant (perhaps they provided you with demonstra tive proof), and as a result you come to believe that the lines are of the same length. But because it is a persistent visual illusion, the change in belief does not change your visual perception: You still visually perceive the lines such that one appears longer than the other. Here cognition is affecting cognition, but without affecting visual experience. 2 You’ve hired a fancy interior designer. He has brought you an item for your living room: It’s a bright pink, faux animal rug. “Faux animal hair is back. And this is ‘millennial pink’. It’s fab.”, he says. You take one look at the thing and you feel queasy. “Get that thing out of my sight!” you implore. Your hired design hand complies, rolling the rug and removing it from view. Here a cognitive state, say a judgment or evaluation, causes you to issue a command, and as a result the available visual stimuli change and, con sequently, you have a distinct visual experience: no more millennial pink experience for you! 3 You’re at a modern art gallery. You’ve found yourself in a room full of Pollocks. You don’t like Pollock. You know the drill: You are told that it’s an “action painting” and that the marks on the canvas are a record of Pollock’s movement. But it still looks like a chaotic mess to you. You decide to skip this room and walk to the next. Ahhh . . . Rothko. Peaceful. Orderly. Simple. You have a seat and take in the calm room. Here cognitive states (or a decision partly constituted by such states) causes an action, and as a result of that action you have distinct perceptual experiences.
5
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6
INTRODuCTION
4 I’m writing this chapter using word processing software on my desktop computer. I try to stay focused on the current task, but it’s also World Cup season. I have the current match streaming on a small window in the upperleft portion of my screen. I abandon writing for a few minutes and shift my focus to the match. Here what I believe (and desire) about available visual information causes me to shift my attention from one portion of my visual field to another and, consequently, to have a different visual experience. 5 You go to pick your friend up at the airport on an extremely busy travel day. You meet her in the baggage claim terminal. There is a sea of baggage, most of it darkly coloured, black or navy, and so you begin aimlessly, visually scanning. “My bag is orange”, your friend says to you. Upon receipt of this information, the few brightly coloured pieces of luggage pop out, becoming salient in your experience. Here, what you believe – that the item of interest is orange – causes atten tion to select items with that feature. This aids visual search and, on one interpretation*, plausibly generates a different visual experience than the one you had prior to learning the colour of your friend’s luggage and as a result of your forming a belief about the orange luggage. 6 You hold an item in each hand. In the right hand is an American quar ter; in the left hand is a metal disc of the exact same size and shape. You hold both before me in plain sight and ask which is bigger. I report that the quarter appears bigger. We repeat this procedure multiple times, with coins of different values, and I consistently report that the coin is larger than what (in actual fact) is an identically sized metal disc. Here what I believe – that one object is a bit of valuable currency and another is not – is influencing the reports that I make about the size of objects. On one interpretation*, this belief (or beliefs) is affecting the visual appearance of things, and that is why I report as I do. It is affecting low level-vision, namely, the representation of the size of objects, such that I see the coin as larger than the analogous metal disc. 7 Consider the expert ornithologist. Before taking up birdwatching, to this individual, birds looked like, well, birds: objects with roughly the same shape, texture, and size (at least relative to other objects). But after the ornithologist has achieved a high level of expertise, she is able to categorize birds with a high level of specificity – this one is an eastern kingbird, that one an eastern phoebe – and to distinguish individuals one from another. She can do this rapidly and on the basis of her visual experience of birds.
INTRODuCTION
Here, what the expert has learned about categories of bird clearly influ ences what and how she is able to make judgments about birds. On one interpretation*, this learning is or has also influenced how she sees birds. For the expert, but not for the novice, visual perception picks up – repre sents – high-level categorical information about birds. The expert doesn’t just judge that a particular bird is an eastern kingbird, she visually perceives this. Cases 1–4 are not controversial. Each of these phenomena are actual, and they possess the mental features attributed to them, as described. And there are countless other cases that take the structure of such cases. Cases like 5–7 do occur under some description. But there is substantial controversy regard ing how such cases should be explained in terms of their mental features. A relevant controversial interpretation has been offered for each of these cases, marked with an *. The theoretical aim of this book is to defend cases like 5–7 as occurring and as being best explained in something like the ways given by the con troversial interpretations (those marked with an *). Contrary to orthodoxy in philosophy and cognitive science, the claim is that thinking influences perceiving not just by causing intermediate action or overt attentional shifts but in more nuanced and direct ways. These influences on perception are deeply important for philosophical and scientific theorizing both, as we will see. They are also important for how we conceive of ourselves as being in the world. To make this case, though, a broad survey of some of the existing, relevant conceptual and experimental work in perception studies is needed.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Chapter preview The goal of Chapter 1 is to provide the reader with some intuitive, and then more sophisticated, understanding of how philosophers distinguish thought or cognition, on the one hand, from perception or sensory experience, on the other; identify important similarities and differences between the two broad categories of mental process; and, finally, identify possible interactions between these processes and their importance. Chapter 2 critically analyzes the modularity theory of perceptual systems. This theory denies that one’s thoughts can influence one’s visual experiences in any interesting or direct way. Vision, like other perceptual systems, is argued to be functionally independent of cognitive states like belief. The chapter identifies the central commitments of that theory and its clearest and most formidable arguments. It criticizes those arguments and concludes that they are insufficient to support the theory and the orthodox place it holds
7
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8
INTRODuCTION
in contemporary philosophy of perception and cognitive science. Chapter 3 couples with Chapter 2, offering a brief survey of types of empirical cases that have been employed to oppose modularity. Chapters 4 and 5 attempt to shift the discussion of possible cognitive effects on perception. Chapter 4 focuses on definitions of the “cognitive penetration” of perception, suggesting that the variety of extant definitions, given their differences, may ultimately miss what is most important to all theorists involved in the debate(s). The suggestion, indeed, the methodo logical prescription, is to look not for phenomena that satisfy this or that definition, but instead for cognitive-perceptual relations that bear some of the consequences of interest to all parties in the relevant debates. Chapter 5 focuses on what kinds of mental process or mechanism are relevant when asking if thinking affects perceiving. The mechanism of central interest is covert, selective attention. The chapter argues that when attention takes this form (by contrast to overt, spatial attention) and plays a role in cognitive effects on perception, we plausibly have instances of important and/or direct cognitive influence on perception. Chapters 6 through 8 attempt a more dramatic shift to the discussion on cognitive effects on perception. The phenomenon of central emphasis is per ceptual expertise. The attempted shift is multi-faceted. First, it encourages an emphasis on possible and, it will be argued, actual cases of perceptual improvement. This contrasts with the standard and dominant emphasis (in this particular literature, and in philosophy of perception more generally) on negative or neutral cases, cases where we, at worst, suffer illusion or hallucination and, at best, just perceive “normally”. This involves both an architectural component and an epistemological one. By appeal to a large set of empirical studies and a wide range of experimental methods, Chapter 6 argues that perceptual experience can change in important ways as a conse quence of concept-rich, cognitive learning. Chapter 7 then argues that such cases are not only epistemically good, but they are epistemically virtuous, involving acquisition of cognitively enhanced perceptual skills, and ones for which the expert is to be credited. Chapter 8 then teases out some of the most interesting consequences from those architectural and epistemological analyses, concerning scientific investigation and theory-ladenness, implicit bias and stereotype, and objectivity and perceptual content. The second attempted shift, important enough to foreshadow here, is a dialectical one. Since relevant concepts have not yet been provided for the reader, the foreshadowing will be light. In the literature on cognitive effects on perception (sometimes called the “cognitive penetration” literature),
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INTRODuCTION
a standard assumption is made by theorists on both sides of the debate. That assumption is that there is a default theory of perception, at least with respect to how thought may affect perception. That theory is modularity. Any theorist who wants to claim that cognition importantly influences perception thus tends to defend cases as providing counterexamples to the default theory. Advocates of the default theory then argue that those coun terexamples fail. (The reader will find examples of this in early chapters of the book.) For this theoretical situation to be well grounded, for a theory to be the default, it must enjoy powerful arguments or powerful explana tory and predictive purchase, ideally both. The critical mode of this book is that the putative default theory enjoys neither. The positive mode of this book – and the chapters on perceptual expertise are central here – is to pivot to cases of cognitive improvement of perception and a mental archi tecture that characterizes perception as malleable. And the cases are positive both in the sense that they are epistemically virtuous and in the sense that they are explained and argued on their own merits. The second shift, thus, is away from a proof-by-disproof strategy to a proof-by-proof strategy. Put another way, the theoretical approach advocated here, even if incomplete, is an alternative approach, and its value is not beholden to providing or being a counterexample to the default. How and whether thinking affects perceiving is a deeply important ques tion. Of course it is of scientific interest: To understand the human mind is to understand how we best distinguish its processes, how those processes interact, and what this implies for how and what we may know about the world. And so in the philosopher’s terms, this book is one on both men tal architecture and the epistemology of perception. But there is a more human interest, and one that will increasingly surface over the course of the chapters that follow. How we make contact with the world, and with one another, is of the most basic of importance. We can make both sensory con tact and cognitive contact with the world. The first is traditionally supposed to be determined by the biological nature of our sensory systems, while the second is at least partly determined by us, what we have learned, our expe riences, and so on. The most basic claim of the book is that this is mistaken and importantly so. Our sensory contact with the world can also change and in a way that is importantly affected by the cognitive contact that we have, or have had, with the world. Thinking does not just affect perceiving; think ing improves perceiving. If this is true, it changes not only how we should theorize the mind; it changes how we should understand, as individuals, our place in and contact with the world.
9
1
THOUGHT AND PERCEPTION :
DISTINCTIONS, SIMILARITIES,
AND RELATIONS
Copyright © 2021. Taylor & Francis Group. All rights reserved.
1.1 “Defining” by ostension 1.2 The importance of cognitive influence on perception: an initial gloss 1.3 Reductive and revisionary theories 1.4 Perception and thought: grounds for distinction 1.5 Mental content and its importance 1.6 Summary There are different ways of making contact with the world. Perhaps most obviously, one can perform actions on the objects and features of one’s sur roundings: One can kick the football, grab the coffee cup, pat the dog on the head, and so on. One can also relate to, be in touch with, make contact with the world via one’s mind. When one sees or hears or smells parts of the world, this is sensory contact with the world. And one can maintain contact with the world cognitively by making judgments about it, or forming beliefs about it, or imagining a pleasant event of the past. Both sensory contact and cognitive contact involve mental states that are about the world. They both involve being in contact through, or by carrying, information. When one
THOuGHT AND PeRCePTION
sees the apple tree or when one believes that there is an apple tree on the other side of the house, one carries information about the world. This chapter proceeds as follows. First by appeal to ordinary and theo retical practices, Section 1.1 provides a preliminary list of mental phenom ena that are typically categorized as perception and those that are typically categorized as cognition. Not all of the complications that come with this task will be decisively addressed here, and so the chapter then moves on to analyze ways that paradigmatic cognitive states and paradigmatic perceptual states might relate as a conceptual or empirical matter. Section 1.2 offers some initial characterization of the importance of possible relations between cognition and perception and of the putative distinction between the phe nomena. Section 1.3 considers the philosophical view that perception just is, in some sense, a kind of belief and the psychological view that there is no clear distinction between perception and cognition. Section 1.4 consid ers five possible, non-exclusive modes for distinguishing thought from per ception. The reader should hereby acquire a handful of useful concepts for thinking about the remainder of the big questions of the book, centrally, how thought may importantly affect perception, how cognitive contact may influ ence sensory contact. Finally, Section 1.5 focuses on an important related set of issues: how both cognitive states and perceptual states are representational states with content and how the natures of such content may be the same or different. Here, too, the philosophical problems and technical concepts employed to discuss them will prove useful for navigating the problems and analyses, and their importance, that follow in the rest of the book.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
1.1 “Def ining” by ostension The strategy of this section is to see how much we can learn about percep tion and thought, and how they are distinguished, by considering our more ordinary concepts, as well as some more sophisticated practices of theoriz ing about them. Begin by considering how we ordinarily think and talk about the rele vant mental concepts, beginning with perception. A common way to think about perception is as a person’s understanding of or stance on things. We say things like “That speech changed my whole perception of things”; “The East’s perception of the West has shifted”; “They perceived their imminent doom.” Likewise for terms specific to particular sensory modalities, we say things like, “I see what you mean” or “I hear what you’re saying.” We might call this notion perception-as-outlook. But if we derived our notion of perception
11
Copyright © 2021. Taylor & Francis Group. All rights reserved.
12
THOuGHT AND PeRCePTION
from these linguistic practices, it either won’t deliver an unequivocal verdict (which should not be expected from an ordinary language analysis) or will not deliver a meaning that accords with more theoretical uses of the term(s). But there is another set of uses that points to the standard understanding of perception in philosophy and psychology. We also use perceptual terms in ways reserved for sensory experiences and at the exclusion of non-sensory mental states. For example, suppose you are disputing with a friend whether another friend was at the party. You might say something like “I think she was there.” And your friend, still unconvinced, might reply “Ok, but did you see her there.” Here, the use of the perceptual term “see” is used to exclusively denote visual experience. The only kind of evidence that will do is visual evidence. What you judge, or believe, or think simply isn’t relevant to this use of the perceptual term. Sim ilarly for other sensory modality terms. Attempting to persuade your partner that you are getting a raise, he might reply “Did you actually hear your boss say ‘raise’?” And if that auditory experience wasn’t had, then you simply lack the requested perceptual evidence. The linguistic practice, then, is one where if we want to denote percep tion in a sense more restricted than perception-as-outlook, we often employ terms specific to distinctive sensory modalities, to olfaction, audition, vision, and so on. Practice in psychological science follows this pattern in impor tant ways. In descriptions of research projects, research labs, and relevant research literature, psychologists often describe the target phenomenon in terms more specific than “perception”: they speak of “vision”, or “audition”, or “touch”. This is partly for a simple reason: Those phenomena are their research specializations! But whether intentional or not, the practice also serves to disambiguate the language in a way that mirrors ordinary practices. Psychologists, and to a lesser degree ordinary language practices, similarly use general sensory terms when they mean the more restricted sense of per ception, they refer to “sensory perception” or “sense perception”. Following these practices, perception is sensory perception. Although even here there are complications, perceiving includes, in the paradigmatic instances, experiences had via the Aristotelian senses: seeing, hearing, touch ing, smelling, and tasting. It plausibly also includes proprioceptive experi ences, and perhaps further some type of “inner” awareness of one’s bodily activity.1 To start, we can take “perception” to denote any of the mental types on this list. It is a capacity for sensory representation by, as it is sometimes put, an individual. It “is a referential and attributive ability to represent basic mind-independent features of the environment. It is a capacity for objective
THOuGHT AND PeRCePTION
representation” (Copenhaver 2013: 1065). As Tyler Burge has argued, this representational role – which is objective in the sense that it involves attrib uting features to particular objects in the perceiver’s environment – is the function of perceptual systems (2010).2 This leaves space for the possibility of unconscious perception and sub-personal perceptual processes. That said, the form of perception of central interest in this book is conscious per ception. Perceptual experience is conscious sensory representation.3 Such mental states are conscious, at least, in the sense that they have a phenomenology; there is something that it’s like, for the perceiver, to be in them. The notion of objective sensory representation and sensory phenomenology come together in an important way. As Fred Dretske once put it:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
the way things phenomenally seem to be (when, for instance, one sees or hallucinates an orange pumpkin), are – all of them – properties the expe rience represents things as having. Since the qualities objects are repre sented as having are qualities they sometimes – in fact (given a modicum of realism) qualities they usually – possess, the features that define what it is like to have an experience are properties the objects we experience (not our experience of them) have. (Dretske 2003: 67)
Perceptual experience is thus transparent, where the sensory appearances of objects and events involve attribution of features to those objects and events (Harman 1990). You perceive the pumpkin, not your perception of it, as orange. Your conscious awareness is of the mind-independent object – the pumpkin – not of your experience. Phenomenality is thus subjective in the sense that there is something that it is like for the perceiver; but it is objec tive in the sense that “it” (the object perceived, not the experience itself) is perceived to be that way. Now consider cognition or thought. First, a note on why it will be assumed that both of these terms refer, more or less, to the same broad category of mental phenomena. This assumption bumps up against our ordinary uses of the terms, since they do not obviously align. This is partly for the trivial rea son that “thought” is a familiar term to just about anyone, while “cognition” is a more theoretical term, even if it is entering more into the fold of popular usage. What kinds of mental activities do we refer to when using “thought” and its cognates? Sometimes we clearly use it to describe cases where we are mentally jogging through various ideas or possibilities, when we are reason ing, and when we are pondering over a decision. We say things like “Let’s
13
Copyright © 2021. Taylor & Francis Group. All rights reserved.
14
THOuGHT AND PeRCePTION
think about where we should take a summer holiday”; “I thought the correct answer was 1626”; or most simply, “I’m thinking about it” to denote that a decision may be forthcoming but one hasn’t arrived there yet. We also some times use the terms in a more restrictive, or at least somewhat different, way, to denote that we have some commitment to the truth of some state of affairs but perhaps without complete certainty. For example, I might say “I think the game is on Sunday morning, not Saturday” or, with less commitment and probably indicating an admission of error, “I really thought the game was on Sunday morning.” Here again ordinary language analysis is inconclusive. But we can note a few things. First, at least these uses all seem distinct from the restrictive use of perceptual terms and what they denote: None of the con cepts of thought at work here appear to be sensory in any way. Second, the kinds of activities described here match up fairly well with those that phi losophers and cognitive scientists denote with “cognition” and its cognates. Cognitive science today is broad in its scope of inquiry. But more tradi tionally (at least going back to the middle and later 20th century, when the discipline was just being formed), the emphasis was certainly on thought, as described earlier. The central point of debate in the discipline then, and in some quarters now, is whether genuine artificial intelligence was achievable. Can a machine think? Can it enjoy intelligence of the kind that humans clearly do and that arguably no other non-human animal does? Accordingly, empir ical studies of cognition in the discipline centred around attempts to build artificial systems capable of reason, of language use, of problem-solving, of decision-making. Note that this comports well with how thought was described earlier. The debates about the possibility of genuinely intelligent computers or machines do not concern sensory capacities – engineers and computer scientists have been quite successful in developing systems that respond to the environment through sensory proxies. The debates concern the capacity for thought or cognition. Following this theoretical practice, and the fact that what the cognitive scientist calls “cognition” and what the folk call “thought” seem largely to be the same thing, we will use these terms interchangeably. So what is thought or cognition? A relatively restrictive way to char acterize thought is just as those mental events or processes that relate to or somehow contribute to knowledge. Belief is the paradigmatic example. Although there are distinct theories of belief, just about any theory will understand belief as a state or attitude towards some proposition, P, where one possesses a conviction that P is true. If I believe that Sweden won the match, I am likely to report this proposition when relevant, form further
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
beliefs about further stages of the tournament, and so on. Furthermore, most take beliefs to be dispositional states, in the sense that a belief will man ifest in behaviour (for example, asserting that P is true) only in relevant circumstances. This allows for a subject to have many beliefs, but only a few are operative or occurrent at any one time. Beliefs are thus the central example of what philosophers call doxastic states. Other fairly standard can didates for doxastic states, insofar as they also involve some commitment to the truth of the proposition mentally represented, are judgments, opinions, expectations, and memories. A less restrictive way to characterize thought is to tie it to reasoning and action. Thus, doxastic states like those noted earlier would be included, since they contribute to our decision-making and reasoning and, often enough, to how we act. But doxastic states do not exhaust those that contribute to decision-making and action planning. We decide and act on the basis of non-doxastic states like desires, goals, wishes, and intentions. A belief that the restaurant is on 4th Avenue does not, by itself, cause me to walk to 4th Avenue. To perform this action, I need, at least, some motivational state – say a desire to have dinner at the restaurant. This is why, in traditional action theory, my behaviour would be explained both by the belief and the desire about the restaurant. And perhaps less obviously, we use imagination to con sider possibilities and often factor those possibilities into a decision. In ren ovating my home, I imagine various construction and decoration projects and, only after contemplating those imagined circumstances, I begin some of those projects. Additional distinctions might be made here to further cat egorize these non-doxastic states. Suffice it to say that for our purposes, on a more liberal construal, these non-doxastic states are also properly under stood as thought or cognition. The reader will note that these mental processes and types do not exhaust the full variety of human mental activity. For example, we can form mental imagery, and this seems to have features of both perception (for example, visual imagery is subjectively similar to visual perception) and of cogni tion (for example, we use imagery to reason about, and act on, the world). Emotional states, too, are difficult to categorize. They have traditionally been treated as non-cognitive (from Socrates to, at least, Descartes), while they enjoy rich phenomenology without always being bound to sensory perception. Those outstanding questions (and others) to one side, this initial analy sis provides some sense of which mental states and processes are typically categorized as perceptual and which cognitive. The first list will include
15
16
THOuGHT AND PeRCePTION
at least vision, audition, touch, taste, olfaction, and proprioception. The second list will include belief, memory, judgment, desire, reasoning, and decision-making. We want to consider why the lists look as they do, and how or whether cognition and perception are distinguished as such, and how they are the same and how they are different.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
1.2 The importance of cognitive inf luence on perception: an initial gloss Determining those similarities and differences is foundational to under standing the importance of possible cognitive effects on perception. That is the topic of the next two sections. Before coming to that, though, we might ask why it matters. What is the importance of the cognition/perception dis tinction, and how might the first affect the second in interesting ways? What counts as “interesting” will be discussed in richer detail later (especially in Chapter 4). For now, here is an intuitive sketch of some ways that cognitive effects on perception might be of significant importance. The importance of cognitive effects on perception can be largely gleaned from the importance of perception. What do we use seeing, hearing, and other forms of perceiving for? This is not, on its face, a deeply theoretical question. It doesn’t take a philosopher or a cognitive scientist to identify good answers. We use perceiving to know about the world around us, to better understand it, and to act on it and in it. If you cannot somehow perceive the objects and events in your environment, you cannot successfully act upon them. The information we acquire through the senses is integral to action. This connects with the first of two epistemic roles of perception. Successful action and action planning require accurate and reliable information about one’s environment. Thus, again intuitively, perceiving serves to provide us knowledge about the world. Some of this knowledge is codified in systematic ways, forming general principles, explanations, and predictions for worldly events. Perception also importantly serves here: We use it to make many of the observations that allow us to form and test scientific hypotheses. These are intuitive and undeniable points about the role that percep tion plays in enabling everyday knowledge and scientific knowledge and in determining action and action potential. Given these roles and their impor tance, why might cognitive influence on perception matter? Here, too, some common-sense observations provide an intuitive starting point. You use vision and other senses to find out about and, in the best cases know
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
about, the world. Crudely, seeing leads to knowing. Knowing can also lead to seeing; for instance, if you know where your partner is, you can put yourself in a position to see their face. But what about a case in the vicinity, where you believe or expect, say, your partner to be upset and as a result of that background cognitive state see them as being upset? This would seem to threaten the trustworthiness of perception in providing you knowledge about your partner’s emotional state. Or imagine a more positive case, one where because of your experience and knowledge of certain kinds of faces, you are better able to individuate instances of that kind. Here cognition would seem to improve your perceptual ability to make discriminations and, in turn, improve your ability to know about the things discriminated. This general epistemic importance of possible cognitive influence on per ception has an analogue in scientific investigation. If a scientist perceives in ways influenced by her theoretical commitments, then how can we be sure that those observations are providing genuine corroboration of her hypoth eses? Again put crudely, good science requires that the observation tests the hypothesis; this evidential relation may be strained if the hypothesis ante cedently biases the observation. This is especially salient when two scientists disagree about the explanation of some phenomenon and their respective perception of that phenomenon is infected by their distinctive theoretical commitments. It would seem that observation may thus not serve to resolve their dispute and enable a rational choice of theory. Finally, what of the behavioural role of perception? What consequences might cognitive influence on perception have here? Consider activities that are highly cognitively demanding. On one quantification – the “Shannon number” – the number of possible moves in a typical game of chess is 10123.4 Chess games can take a long time, but they have a finite conclusion with one player winning. What if a chess master, given her experience with and understanding of the game, better perceives important patterns in a chess game, especially given the massive array of possible moves? Plausibly, this would in turn better enable her to make successful moves, some of them very efficiently, in that game. Or consider sports that, at an elite level, require extremely rapid response times of their players: returning a 120-mph tennis serve, blocking a short-range penalty shot in hockey, hitting a 95-mph fast ball in baseball. If the elite athlete’s expert understanding and strategy could influence their visual perception of these features of the game, this would well explain why or how they are able to do something that only a tiny portion of the human population can do. Cognitive influence on perception thus could enable and enhance exceptional behavioural performance.
17
18
THOuGHT AND PeRCePTION
Given how intuitive they may seem, some of these cases might encourage us to envisage a model of the mind – a mental architecture – whereby cognition can sometimes influence perception, be it in scientific investigation or every day life. A dominant trend in cognitive science, however, has erred towards a more restrictive position. Cognition – what one believes or expects or one’s goals – does not, or at least not with any frequency, influence sensory per ception. And this is for a basic epistemological reason: A perceptual system responsive only to the external objects and events in the environment, and not the particular perceiver’s learning or thoughts or goals about that envi ronment, is more likely to be successful. And it is more likely to be successful, such a theorist will argue, precisely because it is unbiased by the perceiver and will in turn enable knowledge and fitter behaviours. Whether this kind of the ory is correct is very much a point of debate (and it will be debated through out this book). The point here is that this is another reason possible cognitive influence on perception would be important: It would have theoretical impli cations for which mental architecture best describes the human mind. Some additional philosophical problems that connect with these points of interest are discussed in finer detail in Section 1.5. First, we turn to some traditional theories of, and then more recent possible markers for, the perception-cognition distinction.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
1.3 Reductive and revisionary theories Consider what you do when seeing, hearing, or otherwise perceiving events in the world. Suppose you are working at your desk and you hear your part ner come home for the day: You hear the jingling of keys, then a key in the door, and then the door opening and footfalls into the foyer. How do you respond? Well, whatever you actually do in your home, suppose you pause work for the moment and leave your desk to greet your partner upon their return. This action makes sense only if you believe that your partner has come home (or, at least, you believe that someone has just entered your house). And so it would seem that your perceptual experience – hearing a series of events – quickly led to at least one belief – that your partner was home. This example is not special: We use the information that perception provides to act, to navigate – usually successfully – our world. Perceiving the world is importantly connected to forming beliefs about the world. This connection has led some to theorize perception and belief as deeply connected, if not equivalent. And this connection, and others, has led theorists to question any clear distinction between thought and perception.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
The most reductive theory of thought and perception claims that to per ceive is to have, or form, a kind of belief. The intuition here is the one captured by the example just provided: Perceiving an event leads to beliefs about those events. In fact, this may be an overly theoretical way to speak. The pre-theoretic interpretation may just collapse seeing to believing. After all, in many cases, one does not see or hear or otherwise perceive an event and then after some decision, form beliefs about those events. In many cases, the beliefs seem to form concurrently with the perception. As the idiom has it, we believe our eyes (a lot, if not most, of the time). A view developed by David Armstrong (1968) takes this intuition seriously: To perceive is to believe. Recall that a belief that P is a psychological state that involves the conviction that P is true; to believe that P is to mentally represent P as true. Accordingly, when I see the apple on the table, on this view, I believe that the apple is on the table, with an important qualification. Beliefs are also very plausibly dispositional: One has many beliefs at any time, but only a few of them are in any sense conscious. You believe that 2 + 2 = 4, but you only manifest this belief in relevant circumstances. But processes like visual experience seem to be conscious, occurrent ones (or at least a great lot of them are conscious, even if there is unconscious perception). This forced Armstrong, along with another proponent of the view (Pitcher 1971), to characterize perceptions as acquirings of belief which, they claimed, are con scious mental events. When I see the apple on the table, this is to acquire the belief that the apple is on the table. This theory has been widely criticized and on many fronts. Here are two standard worries, which forced additional qualifications by proponents of the perception-is-belief view. First, consider a case where one is already aware of features of one’s immediate environment. For example, I know (and there fore believe) that the apple is on the table – after all, I put it there. This yields an odd result for the view: It does not seem that there is any relevant belief acquired when I visually perceive the apple now. Second, consider cases of persistent illusion. If one knows that the Müller Lyer illusion is an illusion, then one knows (and therefore believes) that the two lines are of the same length. But because it is a persistent illusion, one continues to perceive the lines as being of unequal length. This yields the counterintuitive result that each time one views the illusory figure, one acquires a belief that is logically contradictory to one’s standing belief that the lines are of the same length. For these reasons and others, most theorists of perception have abandoned the perception-as-belief view. However, there are, as with many abandoned theories, important lessons to be gleaned here.5
19
20
THOuGHT AND PeRCePTION
One such lesson is given by A.D. Smith, who, following a series of analy ses given by Fred Dretske (1969, 1979, 1981), concedes that there is or often can be for any perceptual experience a lack of belief regarding the object perceived. As one of Dretske’s many examples illustrates, one can smell the toast burning without smelling that, believing that, the toast is burning. But Smith urges us to not move from this observation to the claim that a theory of perception can entirely lack mention of belief or other doxastic commit ments. He develops this theme in part by advocating a shift from the analytic philosopher’s emphasis on perception as directed at objects to the phenom enologist’s emphasis on perception as directed at the world:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
So although we may not be doxastically related to every object which we perceive, we are doxastically related to every portion of our sensory fields. There are no pockets of ‘meaningless sensation’ in perceptual conscious ness. . . . To disbelieve that a certain object is present, or to be agnostic on the matter, is to relate to a non-presence in a spatial environment which itself is present, and to which we are perceptually attuned in its presence, and hence is related to doxastically. even when we are wholly idle, the world is constantly ‘on hand’ for us. everything we do, or don’t do, presup poses belief in its continued reality-indeed, its reliability. every negation, every doubt . . . is local: a more or less restricted questioning or cancelling that is performed within what Husserl called the ‘overall thesis of the world’. (Smith 2001: 307)
On Smith’s analysis, perception is not belief, and there is a component of sensory consciousness that can be distinguished from beliefs, whether or not the perceptual state and the belief are about the same objects or features of the world. But, crucially, perception of the kind enjoyed at least by mature humans is done against the background of a doxastic stance towards the world, beliefs about the reality of the world which provide essential con text for any singular perceptual experience we may have. Smith’s point is both one about the informational role of perception and its phenomenology. A perceptual state would serve no informational role if the subject noticed nothing, even if the noticing was that there was nothing (of interest) to be seen. And so even the most “empty” of perceptions will presuppose some beliefs on the part of the subject (for example, and minimally, that there is something to be found, that there is a space to look in, that there is a local surrounds to take in perceptually). And these background beliefs about the
THOuGHT AND PeRCePTION
world figure in perceptual phenomenology, entirely suffusing “all of our sensory fields in such a way as to constitute the phenomenological presence of the world to us in perception” (Smith 2001: 307). What follows, Smith urges, is that a successful theory of perception must include a doxastic component. Another important way that the distinction between perception and cog nition has been collapsed, or at least weakened, comes out of the 20th-cen tury “New Look Psychology” movement, which centred around the research of Jerome Bruner and his collaborators.6 As one critic writes, According to Bruner, perception is characterized by two essential prop erties: it is categorical and it is inferential. Thus it can be thought of as a form of problem solving in which part of the input happens to come in through the senses and part through needs, expectations, and beliefs, and in which the output is the category of the object being perceived. Because of this there is no distinction between perception and thought. (Pylyshyn 1999: 342)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Pylsyhyn is correct in identifying the former two features as, according to the New Look theorists, typical of (if not essential to) perception. But the infer ence from that characterization to the claim that perception and thought are thereby indistinguishable is a problematic overstatement of Bruner’s view.7 A more constructive emphasis, then, is on the putative categorial and infer ential nature of perception and how this points to similarities between per ception and cognition rather than their (in)distinguishability. Bruner writes: Put in terms of the antecedent and subsequent conditions from which we make our inferences, we stimulate an organism with some appropriate input and he responds by referring the input to some class of things or events. “That is an orange,” he states. . . . On the basis of certain defining or criterial attributes in the input, what are usually called cues although they should be called clues, there is a selective placing of the input in one category of identity rather than another. . . . The use of cues in inferring the categorial identity of a perceived object, . . . is as much a feature of perception as the sensory stuff from which percepts are made. (Bruner 1957: 123)
Note a few things from this example. First, Bruner clearly takes the transition from sensory stimulation to a perceptual experience to involve something
21
Copyright © 2021. Taylor & Francis Group. All rights reserved.
22
THOuGHT AND PeRCePTION
like an inference, where the perceiver is presented with a range of features, some clusters of which serve as evidence (“clues”) that the perceived object is one category and other clusters that would serve as evidence that the object is in other categories. The resultant experience is not a bare sense datum – seeing x as a round thing with an orange colour – but involves presentation of the inferred category, in this case, that the object is an orange. Perception, on Bruner’s view, represents things as being in a category or being of a kind.8 And some of the underlying computational mechanisms that take a subject from sensory input to an experience are, Bruner argues, of the same broad inferential kind as those mechanisms that are involved in thought processes. Bruner’s argument was that perception must be this way – categorical and inferential – if it is to enable meaningful perceptual representation. And because humans by and large enjoy meaningful perceptual representation, then perception is this way. Further, Bruner argued that how perceptual sys tems categorize and infer is not purely stimulus driven, but sensitive to the subject’s values, needs, goals, tasks, and so on. Finally note, explicit in the quotation provided earlier, that Bruner is not denying that there is a sensory component (the “sensory stuff”) distinctive of perception. Accordingly, and contrary to recent critics of the New Look, there remain grounds to distin guish perception from cognition. The New Look view of perception has been criticized, though, on a vari ety of other grounds. Variations on many of Bruner and colleagues’ experi ments failed to replicate the same results (this is not uncommon in much of psychological experimentation). And further, the perceptual circumstances available in laboratory testing are massively outrun by the variety of cir cumstances we face when out in the world. Replication worries to one side, this means it is extremely difficult to move from empirical results of the lab to a general theory of the nature of perception. All of that said, there is a similarly broad-reaching theory of the mind that has been gaining steam in philosophy and cognitive science and which traces back to central points of Bruner’s New Look theory.9 According to this theoretical framework, the brain (and certainly including perceptual processes) functions as a “pre diction machine” (Clark 2013). Without going into details of the frame work, perceptual processing proceeds by making predictions, using stored representations of the environment, about the sensory information that the subject will receive at any given time. A prediction is then compared follow ing Bayesian rules of probability to the sensory inputs that actually occur. This will typically involve some mismatch, or prediction error, and the system proceeds so as to minimize this error. Again, skipping many complicated
THOuGHT AND PeRCePTION
details, the result of this prediction error minimization process is perceptual experience. Like Bruner and the New Look, this view identifies a distinctive, inferential mechanism not just for higher-level cognition but also for percep tion. This threatens the perception/cognition distinction, and indeed some proponents of the view are keen to advance this consequence (most notably, Clark 2013, 2016).10 What unites these disparate theoretical views is a commitment to some feature(s) of perception and thought, such that the traditional distinction between them is threatened. Each view, in different ways, suggests that per ception and thought may be closer than standardly assumed. But we have also found that each view (with perhaps only the last as an exception) allows for remaining substantial distinction between the two. And so we should next ask what some of the motivations and markers are supposed to be for distinguishing perception and cognition.
1.4 Perception and thought: grounds for distinction
Copyright © 2021. Taylor & Francis Group. All rights reserved.
In Section 1.2 a list of paradigmatic perceptual modalities was given: seeing, hearing, smelling, and so on. And perception was identified as sensory rep resentation, with an emphasis on perceptual experience understood as con scious and phenomenal. This alone may not suffice to demarcate perception from cognition. For instance, some of thought seems to be conscious and experiential, and some cognitive states seem to be sensory in some sense. If cognition and perception are to be distinguished, what does the work of distinguishing them? Here are a number of (largely compatible) answers to this question. Subject activity Perhaps the most traditional marker for distinguishing sensory perception from cognition concerns the activity of the perceiving or thinking subject. Perceiving, the idea was supposed to be, is a merely passive mental faculty. One can find this idea in Descartes, among other ancient and early modern thinkers. “I cannot doubt that there is in me a certain passive faculty of per ception, that is, of receiving and recognizing the ideas of sensible objects” (1641/1984 Meditations VI: 30). So long as a subject’s sensory organs are work ing properly and she is temporally and spatially proximal to some object or event, then she will hear, see, feel, and so on that object or event. There are two separable claims here: one concerns agency or voluntary control, and
23
Copyright © 2021. Taylor & Francis Group. All rights reserved.
24
THOuGHT AND PeRCePTION
the other concerns the passivity of the neurophysiological system. The first, weaker, claim is just that sensory perception is not something the agent her self does; one does not have immediate voluntary control over seeing, apart from opening one’s eyes and directing one’s gaze. By contrast, to engage in thought, to reason, to make decisions, to form intentions to act, these were all supposed to be active mental behaviours on the part of the subject and, accordingly, ones for which she could be reasonably praised or blamed. The second, stronger, claim is that sensory perception is a passive faculty of receiving information about the world, both involuntary and entirely stimulus-driven.11 Thought or cognition, by contrast, was supposed to be an active faculty, one that is sometimes under voluntary control and that can importantly deviate from one’s immediate environmental stimuli.12 This traditional ground for distinguishing perception and cognition has largely fallen out of favour in contemporary theorizing, at least in its stronger form, since it looks increasingly difficult to maintain the claim that percep tion is entirely passive. Setting to one side the possibility that non-perceptual processes may influence perceptual processing (the main claim of this book), many philosophers and cognitive scientists today think of vision, audition, and mental representation in other sensory modalities to be far more active than the traditional passive view of perception. An array of evidence suggests that sensory perceptual systems, even if not agents themselves, are actively constructive in some important sense. Here are just a few examples. Consider standing on a train platform and watching a train depart away from you. You perceive the train as getting farther away, but you do not perceive the train as getting smaller. And this is in spite of the fact that the stimulus is, relative to you, getting smaller as the train recedes into the distance. This is a standard case of size constancy. Notice that if visual experience were just a matter of passively receiving details of the worldly stimulus, then you should perceive the train as smaller and smaller as the stimulus is getting smaller and smaller (relative to you). This indicates that vision (even if not the perceiver herself) is instead actively processing the received information so as to ensure that what we know in fact happens, namely, an object perceived as constant in size (even as it recedes into the distance). And the reader might think for a moment why this is an extremely important thing for vision to do! Here is another phenomenon, one more difficult to illustrate by descrip tion but one also well evidenced through empirical study. The passive model predicts that one’s goals or tasks should not affect the visual experience one has if the visual scene is held fixed. But research on inattentional blindness
THOuGHT AND PeRCePTION
confounds this prediction. If subjects are given a specific visual task – for example, in a busy visual array, count the number of times that the people dressed in black pass a ball to one another – they will generally miss sub stantial events occurring in the very same visual field – for example, a person with a white umbrella appearing in the middle of the scene. Here again, it seems that perception is not merely passive but is instead importantly sen sitive to the subject’s task, actively responding to some available information at the cost of other available information. The resulting theories vary wildly, from claiming that the computational processes of perception are not merely stimulus driven but inferential or sensitive to other features of the cognitive system, to claims that perception is itself action, to claims that knowledge of how one’s actions may affect one’s experience causally affects or consti tutes one’s perceptual experiences. The lesson for us is a simple one: The traditional passive/active distinction (at least as a hard distinction) is not a viable one for marking off perception from cognition. Perception is, in some important sense, more active than that traditional view allows.13
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Sensory organ activity A second grounds for distinction concerns the need (or not) for activity in one’s sensory organs. Suppose there is some distinctive bodily organ or set of organs for each sense modality. There may be some space for debate about how precisely to match organ to sense modality – for example, are the eyes the sole organ for vision, or should the optic nerves that connect behind each eye be included? But for our present concerns, it can safely be assumed that there is some empirical fact of this matter. That assumption in place, one way to distinguish perception from cognition is to say that the first, but not the second, requires current activity in some sensory organ. Plausibly, visually perceiving, for example, requires current activity in one’s eyes (and presumably other parts of the visual processing system, whether or not we call the latter an “organ”); hearing a sound requires stimulation in the ears, smelling requires stimulation in the nose, and so on. But reasoning or con sidering one’s beliefs or recalling the birth date of one’s friend does not, in any one instance, require activity in any of one’s sensory organs or systems.14 Presence of appropriate stimuli The third possible mark of distinction is related. For perception, it seems both that one needs activation in some sensory organ and that this activation
25
Copyright © 2021. Taylor & Francis Group. All rights reserved.
26
THOuGHT AND PeRCePTION
results causally from the presence of some appropriate stimulus. Famously, philosophers have spilt much ink on cases of hallucination and illusion. To employ a somewhat tired example, one putatively can have an experience as of a pink elephant even though no pink elephant is present. Granting that such an experience is possible (which is no minor concession), it is most certainly not common, and it is not a case where vision (and, broadly, per ception) is functioning properly. Instead, what humans perceive, and how they perceive those things to be, is typically a function of what objects are there to stimulate one’s sensory organs and what features those objects have. This is important: We need sensory perception, as individuals and as an evolving species, to provide information about what is here and now and what features those things have; it makes a difference if the thing behind the brush is growling versus purring. Visual perception, when functioning prop erly, when providing verdical experience of the world, only occurs when the relevant stimulus is present. One’s accurate visual perception of the house occurs only if there is a house in one’s presence. This is not true for many of our thoughts. Even sticking with cognitive states that are doxastic, one can have many of these thoughts without the presence of the things those thoughts are about. I can know that Canada is north of the United States, believe truly that my favourite pizza shop is now closed, predict correctly that tonight’s storm will effectively water my garden. In each case, I can have the thought, and moreover that thought can accurately represent part(s) of the world without any relevant stimulus being present. The same is true for non-doxastic cognitive states, for desiring, intending, evaluating, forming a goal. This suggests another plausible dimension along which perception and cognition may be distinguished; states of the first type, but not the second, causally depend on the presence of appropriate stimuli. For two recent variants on this view in the form of a stimulus-dependence requirement for perception, see (Beck 2018; Phillips 2019). Conceptual requirements Another possible mark of distinction concerns what kind of past experience is required to have sensory perception versus thought. Almost all theorists agree that thought requires concepts. Going back at least as far as Locke, to have a belief, for example, that there is an armadillo crossing the road, one has to have the concept ARMADILLO, among others.15 This should be an intuitive point. If one has no experience with instances of some kind K, then one cannot have beliefs about Ks or about particular instances of that kind.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
This is because, on one plausible basic notion of what a concept is, to have a concept is to have, minimally, an ability to identify and distinguish instances of that concept. If one has the concept ARMADILLO, then one can distinguish an armadillo from other objects, say from racoons and porcupines and tur tles. The intuitive point also manifests behaviourally. When and only when a child has acquired a concept – say BLUE – can she successfully distinguish blue things from other coloured things, sorting them, pointing to them, and so on.16 On this picture, concepts are basic constituents of thought. One can form beliefs about particular armadillos as such, one can use the fact that an animal is an armadillo in inference, one can intend to have an armadillo for a pet, only if one has some rudimentary concept of an armadillo. According to some, the same is not true of perception (Crane 1992). One can see the armadillo crossing the road without any such concept. One can perceive, have a sensory experience of, the blue sky even if one does not yet have an ability to distinguish blue things from non-blue things. If the analy sis stopped here, the distinction that follows is this. Cognition is conceptual in the sense that forming cognitive states requires relevant concepts; percep tion is non-conceptual in the sense that one can enjoy visual, auditory, and other sensory perceptions without possession of relevant concepts. An astute reader may note, however, that this way of marking the dis tinction sits oddly with the way that concepts were introduced earlier. If to have a concept is to have an ability to distinguish things as being of a category, then concepts themselves are, at least in part, perceptual discrim inatory abilities. And if so, it would seem that one can no more form a belief about armadillos than she can visually identify an armadillo without possession of the concept ARMADILLO. This would make perception no less in need of concept possession than thought. Indeed, there is a common way of acknowledging this point, and one which pervades a significant amount of the work of Fred Dretske. Dretske’s suggestion is that it is true that in order to see that some x is an F one must possess the concept of an F. I can see that the thing crossing the road is an armadillo only if I have the concept ARMADILLO. Thus to perceive facts – what Dretske calls epistemic seeing – one must have relevant concepts. But there is remaining space, at least according to Dretske and others, for perceiving without concepts, for non-epistemic seeing. The armadillo-ignorant will still see, if her visual system is working properly and so on, will perceptually register, the armadillo crossing the road. She just won’t see that the thing is an armadillo. If one is compelled by this dis tinction, then one can distinguish thought from non-epistemic perception via the conceptual requirement criterion.17
27
28
THOuGHT AND PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Phenomenology Finally, philosophers from a range of traditions – continental, phenomeno logical, analytic, empiricist – tend to distinguish thought from perception in terms of phenomenological richness. There are distinctive first-person characters, or “feels”, to one’s perceptual experiences: the look of the sunset, the taste of roasted garlic, the sound of a purring cat. In terms made famous by Thomas Nagel, there is something that it’s like, from the first-person perspec tive, to have conscious perceptual experience (Nagel 1974). Different objects of experience feel differently, when experienced, to the perceiver. The smell of roasted garlic is markedly, phenomenologically different from the smell of freshly picked tomatoes. And the same objects of experience can be experi enced through distinct sensory modalities, and those experiences will also be phenomenally different. Watching the tall pines sway in the wind is a very different experience in its first-person qualities than hearing the same event. A traditional way to distinguish cognition and perception on these grounds is to deny that cognitive processes involve the same rich phenomenology. One’s beliefs, intentions, memories, and other thoughts, many claim, either lack this phenomenology entirely or enjoy much less of it. You believe that France is east of the United Kingdom, that 7 + 5 =12, that whales are mam mals. These beliefs, like perhaps most beliefs, lack a first-person phenome nology: There is nothing that it is like to believe that whales are mammals. If this lack of phenomenology generalizes to cognitive states, then there is a clear path to distinguishing perception from cognition: The first, but not the second, enjoys rich phenomenal character. Philosophers have recently begun to question (or revive questioning) this flatfooted way of marking the distinction. Reasoning would seem, at least sometimes, to feel some way from the first-person perspective. Likewise for making decisions and wondering about something. And some non-doxastic cognitive states would also seem to enjoy a phenomenology. Consider an unsatisfied desire – that you get the job, that a relationship not end, that the medical diagnosis wasn’t so grim. It is hard to accept that these mental episodes do not feel some way to you when you have them. The same goes, it would seem, for at least some of our intentions, memories, and other thoughts: There is something that it is like, subjectively, to have them. A pos sible fix here is to grant that some cognitive states enjoy some phenomenol ogy, but maintain that it is comparatively thin relative to visual experience, auditory experience, and other perceptual experiences. And all instances of the latter have a phenomenal character. If that is right, then an appeal to
THOuGHT AND PeRCePTION
phenomenology still provides some, even if imperfect, grounds for marking off perception and cognition.18
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Summary There are five (at least) possible marks of distinction: subject activity, sensory organ activity, presence of appropriate stimuli, conceptual requirements, and phenomenology. As discussed earlier, each of these grounds comes with complications, and some of these grounds may be more viable than others. The hope, nonetheless, is that some combination of these possible marks will provide the reader with an intuitive sense of why, in fact, philosophers and psychologists have tended to populate their distinctive lists for cognitive states and perceptual states in the way they have. (And additional possible grounds for distinction will be mentioned in Section 1.5.) It should provide something more than an ostensive or stipulated understanding of the two broad categories, but instead some substantive grounds for what is on those lists, for why vision, audition, and touch go on the perceptual list, and belief, judgment, and intention go on the cognitive list, and so on. This falls short of providing a distinction amenable to the philosopher’s, typically elusive, essential definition; no set of necessary and sufficient conditions for “perception” or “cognition” is offered here. But it provides a stock of differences to set next to some of the similarities we have already identified. A cautionary note here is in order. Another way to summarize the attempt to distinguish perception from cognition is, simply, as largely a failure. The cluster of distinguishing marks certainly fails to provide a conclusive reason to mark the distinction in some way or other, or to mark it at all. A result may be scepticism about the perception/cognition distinction. Indeed, this scepticism may be partly vindicated by some of the analyses that follow in this book. If what we call “thought” affects what we call “perception” in deep and rich ways – if they are entirely “continuous” – it may follow that the very distinction between the two is blurred. So while it may seem natural or useful to distinguish them, the philosophical and empirical investigation may uncover little solid grounds for the basis of distinction. We are therefore in the somewhat awkward position of having to assume a distinction, with the marks of distinction given here in mind, and then use that distinction to potentially undo itself. Whether this conclusion ultimately follows is a provocative point for consideration that the reader will hopefully take away from considering the analyses and discussions given in the chapters that follow.
29
30
THOuGHT AND PeRCePTION
1.5 Mental content and its importance My visual experience of the apple tree before me, and my belief about the apple tree on the opposite side of the house, both carry information about certain features and objects of the world. Traditionally, this aboutness has raised important questions, or a set of questions, in the philosophy of mind. Mental states like visual experience and belief are intentional in the sense that they are about or directed at things other than themselves. In this respect, they function like representations and are like words and pictures. A com mon way to characterize what some words or a picture is about is to identify their content. If I ask you what is in today’s newspaper, I am not asking you to designate spatial features of that physical object (as I might if I ask you what is in the parcel that was just delivered). I am asking you what the sto ries in the paper are about. Your answer will thus identify the contents of the newspaper. If I ask you what your realist painting is going to be about, you will reply with a report on what events or objects or things it is going to represent, say, a sporting match or a lively café. Again, this is to identify the representational content of the picture. There are important questions about mental content in both traditional and contemporary philosophical literature, some of which prove important to the contents of this book. This final section discusses four of those questions.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Determining the content of mental states What determines mental content? On the face of it, one might think this is an easy question to answer. The content of a mental representation is what ever thing or event caused it. My visual experience of the apple tree is about the apple tree because it is that physical object that caused my visual experi ence. My belief that there is an apple tree on the opposite side of the house is about the apple tree on the opposite side of the house because, perhaps through a mediated and more temporally extended causal chain, that is the physical object that caused my belief. But matters are not so simple. For one, we have lots of mental representations of items that are not physical objects, such as numbers, possibilities, and fictional characters. Second, we can certainly have thoughts, and perhaps perceptions, about things where those things are not the cause of the relevant mental states. My belief about the intelligence of the fictional character Sherlock Holmes is not caused by the fictional character Sherlock Holmes! And a drug-induced hallucina tion of flashing lights is not caused by any such event in the environment.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
Finally, and relatedly, mental states can misrepresent, just like other kinds of representations. We form false beliefs, make inaccurate judgments, and suffer perceptual illusions. All of these are genuine instances of mental states with content, but their content is not straightforwardly given by their causes. A different angle on theorizing content concerns not the causal deter minants or metaphysical constituents of mental content, but instead what, most naturally, we would say a mental state is saying. When one asks what a picture is about, one is not (at least not primarily) inquiring about the causes of the marks on the canvass, nor about some elusive relation between the representation and what is represented. One just wants to know what the picture is, as it were, saying, what it aims to represent. Similarly, to specify the content of a mental state is to identify what it is saying, what aspect or part of the world (or not) the state is pointing to. Because mental representations, like other kinds of representations, can misrepresent, to say what a mental state is about is not to identify its causes. To say what a representation is about is to say how the world would need to be in order for that representation to be apt or good. The most obvious candidate feature that would make a representation good is that it is accurate. Therefore, to specify how the world would need to be for the representation to be accurate is to say what that representation is about. My belief that the spare key is under the door mat will be accurate just in case the spare key is under the door mat. Notice that this speci fies the content of my belief even if the spare key is not under the door mat (my belief could be false). Importantly, then, this method allows for misrepresentation. However, a complication arises once we consider mental states that do not in any interesting sense aim for accurate representation of the world. Belief, memory, and perceptual experience function to accurately represent the world. But desires, intentions, and imaginings do not share this function. For instance, many of our desires concern ways that the world is not but that we want it to be. A traditional way to specify the content of these types of mental states is to specify their success or satisfaction conditions (Searle 1983). My desire for a Caesar salad and a glass of red wine will be satisfied just in case I get a Caesar salad and a glass of red wine. Therefore, my desire is about a state of affairs where just those events occur. This way of specifying con tent subsumes specification of accuracy conditions. My belief that the spare key is under the door mat is successful (and therefore accurate) just in case the spare key is under the door mat.
31
32
THOuGHT AND PeRCePTION
Beginning with the next chapter, this book will be concerned with con troversial possible relations between cognitive and perceptual states and pro cesses. When focusing on particular phenomena, it will often be important to discuss the content of the relevant mental states. The discussion here provides some basic tools for analyzing mental states and their contents, but without getting stuck with a number of additional challenging philosophical problems about content.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The importance of mental content and its variety Both cognitive states, like belief, and perceptual states, like visual experience, are, it will be assumed here, states with content.19 Whether this content is the same in important respects is another controversial matter. Historically, the importance of this question is epistemological. Consider the following simple scenario. You claim to know something, say, that Jones has a new red car. A friend challenges you, “How do you know it is red?” In an ordinary scenario (that is, not in a philosophical thought experiment), if you reply by reporting that you saw the car and saw that it was red, this will satisfy your friend. Visual experience and per ceptual experience generally (depending upon the kind of information in question) tend to justify claims to knowledge. Put most simply, a powerful intuition says that perception provides us with knowledge about the world (or, if one prefers, justification for our beliefs about the world).20 One way this epistemic role of perception is traditionally described is as the given. Per ceptual experience is supposed to be epistemically special in the sense that it gives warrant to beliefs formed on its basis but does not itself require any further warrant or justification. There is no obvious further query that your friend can make if she takes your report of seeing the redness of the car to be genuine. A challenge to this intuitive story, famously laid by Wilfred Sellars (1956), concerns the nature of perceptual content versus the nature of cognitive content (in this case, the nature of the content of belief). One can think about beliefs (which on a standard model, when justified and true, consti tute knowledge) on the model of assertions: They are representations with assertive content (Bonjour 1985). Your belief that the car is red makes an asser tion about an object in the world, namely, that it is red. (Recall how belief, unlike mental states such as desire, have an accuracy function. Identification of assertive content is just another way to specify how the world would need to be in order for the belief to be accurate; whatever P it is asserting
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
has to actually be the case in order for the belief that P to be true). This is a largely uncontroversial characterization of belief content; what about per ceptual content? The challenge goes either perception, too, enjoys assertive content or it does not. And either option spells trouble; thus, the Sellarsian dilemma. In brief, if a perceptual experience – say, with the content “The car is red” – also makes a kind of assertion, then just as the belief with the same content required some further basis or justification, so, too, will the perceptual experience require some further justification. But then its role as the given disappears, and the door opens for a potential regress where no final justificatory basis is forthcoming (since whatever justifies the percep tual experience would then need also to have assertive content, and it would need further justification, and so on). If perceptual content lacks assertive content, the second horn of the challenge goes, then it is not structured in any appropriate way to justify the relevant belief. The simplest way to under stand this is in terms of a very basic inference. The perceptual content, as a mental representation in response to receiving information from the world, would justify the belief if the content were just the same (both of them asser tions that “The car is red”). If the content is not the same (which is the horn of the dilemma we are now considering), then it appears to lack the requisite logical form to do the justificatory work. Sellars therefore concludes that the given, this foundational role for perceptual experience, is a myth. This traditional epistemological problem concerns whether perceptual contents are of the right kind to justify beliefs formed on their basis. The relation of interest moves from perception to thought. The central relation of interest in this book, by contrast, moves in the opposite direction: Does thinking influence perceiving in important ways? But this second question is not irrelevant to the traditional epistemological problem. As we will soon see, if thinking does influence perceiving in important ways, then again the foundational knowledge providing role of perception is potentially threat ened, or at least in need of some revision as theorized. A distinct but importantly related issue concerns the format of mental con tent and whether it provides a viable criterion for distinguishing perception from cognition. Some philosophers and cognitive scientists have argued that perceptual representation takes the form of an icon (Carey 2009). Iconic rep resentations function roughly like pictures. An icon i makes explicit some information about an object or event x, thereby representing x, because the parts of i stand in similar spatial and other structural relations that parts of x stand in. Cognitive content, by contrast, is not formatted this way. Recent theorists have been critical of the success of this approach (Quilty-Dunn
33
34
THOuGHT AND PeRCePTION
2016; Clarke forthcoming). Out of this criticism emerges a related alterna tive format of content criterion. Some argue that the format of perception is analog, while the format of cognition is digital. Here again an appeal to pictures is instructive. A picture of a fly on the wall carries a great deal more informa tion than just that the fly is on the wall. It will carry information about the bodily position of the fly, its shape and movement, the colour of the wall, and much else. The picture is “pregnant with information”. The sentence “The fly is on the wall” itself carries only the information about that event (even if one could draw inferences from it, for instance, that there exists a wall and there is a winged creature on it). Perceptual content is supposed to be analog like the picture, cognition digital like the sentence (Dretske 1981; Evans 1982). Some recent theorists have resurrected and revised this line of thought, arguing that it remains a plausible mark for the perception–cogni tion border (Beck 2019; Clarke forthcoming).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Concepts and perceptual content Recall the discussion from Section 1.4 regarding the need or lack of need of concepts for perceptual experience. There the basic point was that while cognitive states like belief clearly require concept possession (at a bare min imum, concepts of the things about which one has beliefs), perception, or at least some forms of perception (what Dretske calls non-epistemic seeing), can occur without possession of relevant concepts. This problem generalizes to one concerning perceptual content and whether it can be non-conceptual or whether, like belief, it must be conceptual. Content conceptualism makes a similar claim (and content non conceptualism the opposite claim), but about the type of content that cog nitive versus perceptual states can enjoy. This debate is full of complication, but here is one way to capture a central point, given by Richard Heck. Heck is largely motivated to adjudicate between two prominent opponents in the debate, Gareth Evans (1982) and John McDowell (1996), the first arguing for non-conceptualism, the second, conceptualism. According to Heck, Evans understands belief (cognitive) content to be Fregean Thoughts or propositions. This implies that such contents are structured, where the elements of that structure are concepts. And this imposes an important constraint on what it means for a subject to have a mental representation, say a belief, with content that is structured in this way. To entertain the Thought that John is tall, for example, a thinker would have to exercise a capacity to think of John and a capacity to think of a
THOuGHT AND PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
thing as being tall. And to have these cognitive capacities is, on evans’s view, to grasp the concepts of which the Thought is composed. So, in effect, what evans is claiming is that no one is capable of entertaining a given Thought, conceived as structured in a particular way, who does not apprehend it as being so structured: To entertain the Thought that John is tall, one must exercise certain capacities, possession of which constitutes grasp of the concepts of which that Thought is composed; so, in entertain ing this Thought, one must make use of one’s grasp of these concepts, must conceive the Thought as composed of those concepts in a particular way. And it is, on evans’s view, because entertaining the Thought that John is tall presupposes possession of such cognitive capacities-presupposes a grasp of certain concepts-that an ability to entertain that Thought requires the ability, mentioned in the Generality Constraint, to “recombine” its elements. (Heck 2000: 487)
Both parties agree that cognitive content is conceptually articulated in this way, such that to have a belief with some content P (say, that John is tall or that Juan is loud), one must grasp the concepts that articulate P (John: being tall; Juan: being loud). And a condition on genuinely grasping those concepts – Evans’s generality constraint – is that one can recombine them to form new representational contents, say, that John is loud or Juan is tall.21 What they disagree about is whether perceptual content is so structured. This point is elusive, not least because descriptions of perceptual content often take the very same linguistic form as descriptions of belief content. For instance, one might have a perceptual experience with the content “John is tall”. The non-conceptualist will accept this as a “conceptual report” of the perceptual content, but deny that one must grasp the concepts (John: being tall) in order to have that experience. Perceptual content, then, lacks conceptual structure: It is not composed (or need not be composed) of concepts. The conceptualist maintains the contrary claim: Perceptual content is conceptu ally structured. An argument for non-conceptual perceptual content derives from the rich ness of perceptual experience versus the coarseness of our conceptual rep ertoire. Just about any visual experience will present, phenomenally, the subject with an incredibly rich array of shades of colour. But none but a few colour experts possess sufficiently fine-grained concepts of colour to indi viduate each experienced shade. Therefore, visual experience enjoys content that is not composed out of concepts the agent possesses (since she has the experience but lacks the concepts).22
35
36
THOuGHT AND PeRCePTION
Here is an argument for conceptual perceptual content. The argument is Kantian in spirit and has been argued forcefully by McDowell (1996) (see also Brewer 1999). Suppose that we take it as a bedrock, empirical starting point that we do gain knowledge from perception. Perceptual experience provides justification for belief because it must. And our common-sense, pre-theoretical intuitions powerfully suggest this is the case: We gain knowl edge through perceiving the world. Taking on the first horn of the dilemma (that perceptual content has the appropriate kind of content, what we called “assertive”), McDowell and others who give this argument claim that a per ceptual experience e could justify a belief b formed on its basis only if p (like b) is conceptually structured. And because perceptual experience does justify belief formed on its basis, it therefore must be conceptually structured. This argument for conceptualism, like the traditional Sellarsian dilemma, is epistemological. And so one might think that it is separable from archi tectural questions about cognitive effects on perception. This is partly but not entirely true. Concepts are theorized in a number of different ways, but the way they’ve been discussed here is as a cognitive capacity of some sort (perhaps grounded in sensory discriminatory capacities). On the conceptu alist line, perceptual content is structured by concepts and so, just as Kant understood it, experience is cognitive through and through. And even if one denies this robust conceptualism, concepts do structure belief (and other cognitive content), and so if beliefs and other cognitive processes can affect perception, there may be important questions about the role of concepts in those effects (see Macpherson 2015).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The admissible contents of perceptual experience A closely related debate concerns not the structure of perceptual content (though it may well concern the conceptual [in]dependence of perceptual content), but rather what kinds of entities can figure in perceptual content. Assuming that perception is representational, what kinds of things does it represent? A common-sense answer might say that vision represents things like tables and chairs and dogs and cats and mountains and what they are like – their colours, shapes, where they are located, and so on. Audition, too, represents these kinds of things, or at least some of them: the sounds things like dogs and cats make, at what volume, at what distance, and so on. In short, percep tion represents the different kinds of objects that populate our environment and the features that those things have, ranging from how they appear to
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
what kinds or categories they fall under. This common-sense answer, or at least part of it, is not far off from a contemporary and controversial answer to the same question in philosophy of perception. To appreciate the contro versy, though, we first have to give some brief characterization of the more traditional, and still more orthodox, answer. Consider a subject who does not have many concepts, say a young child, but with relatively mature and healthy sensory faculties (not an infant). Can this subject see a horse or a dog or a table if they do not yet have a concept of those things? A defensible answer says that such a per son does see, in the sense of non-epistemically perceives, these things, but she does not see them as horses or dogs or tables. This is the answer given by a sparse perceptual content theorist. According to this theorist, what perception represents is only the basic perceptible features of the envi ronment and not the categories those things fall into or the more com plex relations they stand in. Vision thus represents the colour, shape, location, and rest/motion of objects. Audition represents sounds and cer tain features of those sounds like volume and distance relative to the perceiver. Olfaction represents scents but not the particular kind of object that makes those scents. And so on. Higher-level information – about what kind of thing an object is, or about causal relations between objects, or moral or aesthetic features of objects – is mentally represented, but post-perceptually, at the level of judgment or belief. This traditional the ory about perceptual content remains the dominant one and is motivated by a number of factors. One general motivation is what we can call perceptual convergence: In spite of marked cultural and social differences, there is a statistically normal trajec tory for human sensory perception. Visual perceivers in different regions of the world differ in their concepts, language, and cultural practices, yet they tend to fall into importantly similar patterns of colour discrimination, shape discrimination, motion detection, and so on. This is true even acknowl edging known individual differences. This convergence would seem to be undermined by visual representation of culturally sensitive or specific con cepts (for example, what category an object falls under). Therefore, vision must be representing only the culturally neutral perceptible features of the perceiver’s environment. The sparse content theorist then generalizes to other sensory modalities. There are reasons, though, to favour what may be the more common-sense answer, that we don’t just perceive the basic or low-level properties of objects but also the high-level properties. Recently, theorists have argued that we
37
Copyright © 2021. Taylor & Francis Group. All rights reserved.
38
THOuGHT AND PeRCePTION
perceive natural kind properties,23 emotional properties,24 causal proper ties,25 and moral and aesthetic properties.26 Here, in brief, are two distinct types of argument for these sorts of claims. One argument is phenomenological. Consider acquiring an ability, what some would call a concept, to distinguish a kind of object, say an armadillo. Plausibly, what it is like for you to see armadillos before and after acquiring this ability is different. This phenomenological difference calls out for expla nation. And, some theorists have argued, that difference is going to be best explained as a difference in visual content and, moreover, a difference in the representation of the property “being an armadillo”. Before acquisition of the armadillo-recognition ability, vision did not represent this property. But after acquisition of the ability, vision does represent this property. This high-level property, the theorist concludes, is now part of the content of your visual experiences of armadillos (Siegel 2005). A second argument is behavioural (for lack of a better label). You’ve long known how to recognize bananas by seeing them. When you iden tify a banana, you do so rapidly and reliably, without any sense of rea soning or drawing an inference or making a decision. Furthermore, you do not have any sense of drawing upon memories of previous bananas or applying some kind of conceptual template to the thing seen in order to determine that it is, in fact, a banana. If asked how you know that what you are looking at is a banana, you would report, flatly and incredulously, that you “just see” that it’s a banana. However far these considerations go, they suggest that your banana identification is a visual phenomenon. Because you long ago learned how to distinguish bananas from other things, vision represents the banana-ness of bananas, and this explains the speed, reliability, and non-inferential performance of your banana identi fication. Therefore, contrary to the sparse theorist, perception can repre sent higher-level properties. The reader will note that this debate connects in important ways with some of the issues and problems already discussed. For one, if indeed per ception is sparse, as some theorists have it, then this provides another possi ble distinguishing mark between perception and cognition. Perception only represents basic sensible features. But cognition can represent much more. You can certainly believe that something is an armadillo – this is an admissi ble content for belief – even if perception alone does not give you that infor mation. E.J. Green argues that perception, but not cognition, is restricted with respect to the dimensions to which it is sensitive. Vision, for instance, is only sensitive to “low-level dimensions”. “[S]ize, orientation, brightness,
THOuGHT AND PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
speed, and saturation, are quantitative, and there is strong reason to believe that the visual system is sensitive to their ordering and closeness relations” (Green 2020: 327). If this claim is right – which is allied with the claim that perceptual content is sparse – then this provides yet another distinguishing mark for the perception/cognition border. Second, how and whether perception contributes to knowledge may depend importantly on what kinds of properties are admissible for percep tual content. In fact, there is a way to understand part of this debate as another recapitulation of the worries that ground the Sellarsian dilemma. If perceptual content cannot, with learning, become structured by high-level properties like “being a horse” or “being a banana”, then how do we make the bridge from a perceptual representation of an object to a true belief that the object we perceive is a horse or a banana such that that belief is justified by the perceptual content? Or if perceptual content is high-level, and a prop erty like “being a banana” structures that content, then does this perceptual state not itself require some further justification? This debate also connects in clear ways with the central point of interest for this book. If the recognitional capacities that we acquire, whether we call them concepts or something else, are within the broad category of thought, and the rich content theory is the best theory of perceptual content, then we have identified deeply important ways that thinking affects perceiving. How one person sees an object of a kind – be it a banana, a baseball bat, or a boomerang – may vary substantially, representationally and phenomenolog ically, from how another person sees the same object, assuming those two persons differ in what they know about or what concepts they have of those kinds. Accordingly, questions about perceptual content, whether it is rich in its structure, and how or whether our cognitive and conceptual abilities influence it, will pervade much of the discussion of Thinking and Perceiving.
1.6 Summary This chapter provided our first look at how philosophers and cognitive sci entists distinguish cognitive processes from perceptual processes and why possible influences of the first on the second would be important. Cognitive states are typically supposed to include beliefs, desires, intentions, and other mental phenomena that factor in decision-making and reasoning. Percep tual states are typically supposed to include conscious experiences associated with sensory systems, including vision, audition, olfaction, touch, flavour, and proprioception. Although there is no established consensus, several
39
40
THOuGHT AND PeRCePTION
possible, non-exclusive modes for distinguishing cognition from perception were offered: subject activity, sensory organ activity, presence of appropriate stimuli, conceptual requirements, phenomenology, format of content and richness of content. In spite of these (possible) differences, cognitive and perceptual states are all contentful states. And the nature of the representa tional content of these states bears important epistemological and scientific consequences. Further reading Recent analyses of the perception–cognition distinction, some of them dis cussed in this chapter (and later), include (Burge 2010; Stokes 2013; Shea 2015; Quilty-Dunn 2016; Burnston 2017; Beck 2018, 2019; Mandelbaum 2018; Kriegel 2019a, 2019b; Phillips 2019; Green 2020; Clarke forthcoming). For an analysis distinct in approach from all of these, see Anderson et al. (2012), which offers a critical analysis of the perception–cognition distinc tion on the basis of embodiment and recent research on neuroanatomy. Two seminal analyses of concepts and perception importantly tied to epistemic issues raised by Sellars and Dretkse are Evans (1982) and McDowell (1996). Finally, for recent analyses of perceptual content and its potential richness, see (Siegel 2005, 2011; Hawley and Macpherson 2011; Pautz 2009).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 An exciting development in the philosophy and cognitive science of per ception is a departure from strictly vision-centric studies. Thus, important research is being done on non-visual senses, on non-traditional senses, on interaction between the senses, and on how to individuate the senses. For two recent collections, see Macpherson 2011; Stokes et al. 2014. 2 This thesis will figure importantly in discussions in Chapters 7 and 8. 3 except when such a distinction is needed, the term “perception” will often be used to denote conscious perceptual experience thus understood. If one thinks that perception can be unconscious, then one should interpret the use of “perception” to denote the subset of perceptual phenomena that are conscious and experiential. 4 The game Go has an even more astronomical number of possible moves: 10360. See www.scientificamerican.com/article/how-the-computer-beat-the go-master/#:~:text=Given%20that%20a%20typical%20chess,on%20 how%20to%20program%20a
Copyright © 2021. Taylor & Francis Group. All rights reserved.
THOuGHT AND PeRCePTION
5 For an updated defence of a view of this kind, see Glüer 2009. 6 Sometimes grouped with Bruner and colleagues as later proponents of something like the New Look Psychology is the work of R.L. Gregory (1970) and Irving Rock (1983). 7 Oddly, Pylyshyn acknowledges in a footnote attached to that very sentence (1999: 342, fn5) that Bruner is explicit in denying that his view commits to the claim that perception and cognition are indistinguishable (see, for instance, Bruner 1957: 124), but nonetheless makes the claim anyway and then proceeds to criticize Bruner and the New Look on that basis. 8 This anticipates a view defended by a small number of contemporary philosophers of perception – the rich content view of perception. This philosophical issue will be discussed in Section 1.5 and then again in Section 8.5. 9 Advocates of these predictive coding theories typically credit Helmholtz (1867) and Ashby (1947) as the clear historical precedents for their view, while Bruner’s view seems not sufficiently acknowledged as another important precedent. Nonetheless, the conceptual connection is clear in a number of places. For example, at the close of his “On Perceptual Read iness”, a review of a decade of New Look theorizing, Bruner writes “the person builds a model of the likelihood of events, a form of probability learning only now beginning to be understood . . . one can think of this activity as achieving a minimization of surprise for the organism” (1957: 148). This anticipates prediction error minimization, the core principle for predictive coding frameworks. 10 See also Lee and Mumford (2003), Friston (2005), and Hohwy (2013). For criticism, see Ransom et al. (2017) and Vance and Stokes (2017). 11 See Matthen (2005: 39–54) for related discussion. He attributes the prin ciple of “passive fidelity” to Descartes. It says “the sensory process pre serves the receptoral image; sensory consciousness corresponds to the latter except where it is decayed in the process of transmission” (332). 12 Although to varying degrees and with important differences, this is true of the theorizing of Aristotle, Hobbes, Descartes, Locke, and Hume, among others. See (Newman 1994). The clear exception from the early modern period is Berkeley’s New Theory of Vision (1709), which served as inspira tion for many of the active theories mentioned later. 13 On inattentional blindness, see (Simons and Chabris 1999; Mack and Rock 1998). For a sampling of “active” theories of perception (ranging in method from continental phenomenology to empirical psychology to
41
42
THOuGHT AND PeRCePTION
14
15 16
17
Copyright © 2021. Taylor & Francis Group. All rights reserved.
18
19
20
analytic philosophy), see (Gibson 1979/2014; Dreyfus 1992; Varela et al. 1991; Hurley 1998; O’Regan and Noë 2001; Noë 2004; Matthen 2005; Prof fitt 2006; Nanay 2013). Indeed, even Bruner (1957), as discussed earlier, took perceptual processing to involve inferential activity. The “in any one instance” qualification here is important. Traditional empiricists maintained that all of thought and understanding is grounded in experience of the world, and for some, “experience” denoted perceptual experience – so, no perceptual experience of the world, no knowledge or thought about the world. The distinction being offered here is compatible with this broad empiricism, since it claims only that a person could, in prin ciple, enjoy a cognitive state c at time t, without activity in one’s sensory organs at t (say, in a sensory deprivation chamber). This makes no claims that the person need never have had activity in her sensory organs. To be clear, the armadillo example is Dretske’s (1993) not Locke’s. Locke did not talk about armadillos. Note, importantly, that this is not a point about language, about having, say, the english word “blue” in one’s vocabulary. It is instead, and again, about having an ability to distinguish instances of a kind from non-instances of a kind. This can, in principle and in practice, be done without possessing relevant vocabulary. Dretske uses other terms in other places to mark the same distinction: seeing vs. seeing that; thing-perception vs. fact-perception; awareness of things vs. awareness of facts. See Dretske (1969, 1979, 1993). See also Hanson (1969). For two collections on cognition and phenomenology, see Bayne and Montague (2011) and kriegel (2013). For a recent analysis that argues for a perception–cognition distinction on phenomenal grounds, see kriegel (2019a, 2019b). This is not an innocuous assumption. Naïve or direct realist theories of perception maintain that visual and other perceptual experiences are not fundamentally, or not wholly, representational (and therefore not neces sarily states with content), instead involving direct acquaintance or contact with worldly objects and their features. See (Hinton 1967; Snowdon 1980; Brewer 2011; Martin 2002, 2004). For a line of criticism from psychological science, see Burge (2005, 2011). And for a recent volume of papers, see Haddock and Macpherson (2008). The reader will recall that this role was briefly discussed in Section 1.2.
THOuGHT AND PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
21 For a recent discussion of this constraint and the structure of thought, see Beck (2012). 22 See (Dretske 1981; evans 1982; Martin 1992; Heck 2000). 23 See (van Gulick 1994; Siewert 1998; Siegel 2005; Bayne 2009). See Hawley and Macpherson (2011) for a collection of recent papers and Helton (2016) for a recent overview. 24 Butterfill (2015). 25 Siegel (2009). 26 (Nanay 2016; Stokes 2018; Bergqvist and Cowan 2018).
43
2
MENTAL ARCHITECTURE: THE
CASE FOR MODULARITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
2.1 The modularity of mind: the essentials 2.2 The modularity of mind: arguments 2.3 Summary The previous chapter identified a handful of non-exclusive ways that one might distinguish cognitive states and processes from perceptual states and processes. It also discussed ways that some theorists have challenged the thought/perception distinction, as well as interesting ways that thoughts and their contents might relate to perceptual experiences and their contents. With these distinctions, theories, and concepts in the background, the broad question, finally, is how does thinking influence perceiving? And in what (if any) interesting ways? One of those points of interest concerns mental architecture. As discussed at the close of Section 1.2, if cognition influences perception in certain ways, it might imply negative conclusions for theories that maintain a very strong border between perception and cognition. And if such influences do not occur, those theories would be partly vindicated. This point about archi tecture connects with the other points of interest: with the epistemic and behavioural roles of perception and, in turn, how cognitive influence on
Copyright © 2021. Taylor & Francis Group. All rights reserved.
MeNTAL ARCHITeCTuRe
perception might impact those roles. These roles in fact figure in two argu ments – the argument from stability and the argument from reliability – for the strong, modularist theory of perception (to be discussed in Section 2.2). The basic details of that theory are provided in Section 2.1. The remainder of this section provides some intuitive cases of cognitive effects on percep tion and then some history on theories of mental architecture. Chapter 3 is a contrasting companion to this chapter, outlining an array of cases of alleged cognitive penetration of perception that are supposed to challenge modularity. To begin, consider some uncontroversial cases of cognitive influence on perception. You are working from home and you crave a coffee and a change of scenery. You know that the café around the corner is fairly quiet at this hour and so you pack up your things and head around the corner. This action, driven by your desires and your knowledge, results in a series of rich, distinct perceptual experiences: the smell of roasting coffee beans, the sound of the patrons’ conversations and the espresso machine, the look of the old wooden floor and hodgepodge tables and chairs. Here thought clearly affects perception by, most simply, causing action(s) which effects distinct sensory experiences. This can also occur less immediately and with a much longer causal sequence of events. You are considering a weekend getaway between (oddly) Glasgow and Istanbul. After considerable deliberation over transport costs and time, lodging, sights, and the relevant season weighed against your interests and preferences, you choose Istanbul. The resulting experiences of Istanbul are dramatically different from those you would have had in Glasgow. Here again, these perceptual experiences causally depend upon your deliberate decision-making process and the beliefs, desires, and values that you, as it were, inputted to that process. Although less immediate than the previous example, here again thinking is having a substantial effect on perceiving. These kinds of cases, and everything in between, are not trivial. They are the subject matter of philosophy of action, practical reasoning, and value the ory, not to mention sociologists, economists, and social psychologists. None theless, there should be no controversy that they occur, and they occur often. We act on the basis of what we think, and our actions regularly change what parts of the world are available to us to be seen, touched, heard, and so on. The more interesting question is both difficult to characterize and challeng ing to answer without controversy. With the examples just given in mind, one way to put the more difficult question is this. Does thinking affect per ceiving in a way such that the antecedent thought(s) more directly causally impacts the consequent perceptual experience(s)? In the first example, a set
45
Copyright © 2021. Taylor & Francis Group. All rights reserved.
46
MeNTAL ARCHITeCTuRe
of actions stands in between the background cognitive states (such as a desire for a coffee) and the resultant perceptual experiences (such as the flavour experience of coffee). And in the second case, in addition to more intermedi ate actions, there is a lot more time between antecedent cognitive states and resultant perceptual experiences. The possibility of the cognitive penetrability of perception concerns cases where these causal and temporal gaps are, if not closed, significantly reduced. Put in the extreme, does cognition directly or immediately affect perception? (As will be discussed later in Chapters 4 and 5, this directness condition on the possible phenomenon may be too extreme.) Sometimes this is put as a point about contrast. Suppose two individuals have different beliefs about a category of object, say a cultural icon like the Smurf. As most readers know, the Smurf is a characteristic bright blue colour. But suppose that for one of the individuals, this is not a belief she holds. This individual, a, has never been exposed to Smurfs before, and so she lacks a belief that Smurfs are a characteristic bright blue (for all she knows, these fictional creatures called Smurfs could be a variety of colours). When presented with her first Smurf, under normal lighting conditions, a presum ably would perceive it as bright blue. But would she see it as the same bright blue, say, under dim lighting conditions or when the Smurf is, objectively, closer to greyscale in saturation and luminance? If this seems like a difficult question to answer, consider a different individual, b, who has grown up in a culture where Smurfs populate children’s television, popular culture, and so on. For b, one might think, a Smurf in dim lighting conditions would still visually appear, perhaps even “pop out”, as bright blue. And perhaps even when (by laboratory manipulation) a Smurf figure is closer in luminance to a neutral grey, b will, because of her long-standing cultural belief about the colour of Smurfs, still see that Smurf as blueish. From here one can general ize to a variety of possible cognitive-perceptual relations, varying the back ground cognitive states and perceptual modalities. Do things of significant value – say money – look larger than they, objectively, are? Does the look of a painting differ depending upon what your aesthetic tastes are? Does the sound of a symphony performance differ for the composer versus the musical novice? Does one’s intentions to act affect how one perceives space or distance? These are all examples of how thinking could affect perceiving, and in profoundly interesting and important ways. Both philosophical and scientific thinking has shifted, and many times over, on how to understand this case, or even its mere possibility. These shifts tend to correspond with even broader shifts (or, one might say, trends) in theories of the mind and its organization. The oldest and most dominant
MeNTAL ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(as in, most time consuming) way to think about the mind is as a singu lar entity, “the mind” or “the soul”, often explicitly distinguished from all physical bodies or parts of those bodies. Descartes (1641/1984) championed this view in the early modern period. Indeed, one of his central arguments for a mind/body dualism exploits the following (putative) distinction: One’s physical body is divisible, but one’s mind is not divisible. The latter is sup posed to be true because although there are many possible mental activities – seeing, desiring, believing, doubting – those are all, according to Descartes, activities of a singular mind or soul. On its face, this implies no restriction on the intermingling of mental activities: Seeing can affect believing and believ ing can affect seeing, since the mental architecture is one where, simply, there is one mind capable of these activities, and many others, and nothing in that kind of sparse architecture that denies causal relations between them. Yet those distinctions in mental activities lead quite naturally to a rather different way of carving up the mind or, perhaps we should say, a way of actually carving it up. Absent any commitment to an immutable soul, instead one might take so-called mental activities like seeing and desiring and imag ining to be distinctive mental kinds or faculties. This mental picture traces back, on one reading, all the way to Aristotle (De Sensu and De Memoria 2014; De Anima 2017), but it is present in more articulated fashion in the theories of near contemporaries of Descartes, importantly, in David Hume’s A Treatise of Human Nature (1739). Here is Hume raising a question about how we can achieve certainty that a persisting, external world exists: we ought to examine apart those two questions, which are commonly confounded together, viz. why we attribute a CONTINu’D existence to objects, even when they are not present to the senses; and why we sup pose them to have an existence DISTINCT from the mind and percep tion. . . . we shall carry along with us this distinction, and shall consider, whether it be the senses, reason, or the imagination, that produces the opin ion of a continu’d or of a distinct existence. (Treatise 1.4.2)
Hume distinguishes three mental faculties: sensation, reasoning, and imag ining. (Elsewhere he plausibly distinguishes more, but this will do for pres ent purposes.) Any reader familiar with Hume’s Treatise will recognize this as a familiar strategy. Hume inspects the plausibility that the relevant idea – here an idea of a mind-independent, persisting external world – can rea sonably be acquired through any of these means. And after he concludes
47
Copyright © 2021. Taylor & Francis Group. All rights reserved.
48
MeNTAL ARCHITeCTuRe
that we neither sense it nor reason to it in any logically grounded way, we must have imagined it. The result here, and elsewhere, is scepticism. Hume’s distinction between faculties is thus not innocuous, doing important work throughout his empiricist theorizing. A second important feature of Hume’s faculty theory is how he characterizes the faculty versus its content. Although sensation and reason may behave differently in the overall mental economy (today, one might say “function” differently), in Hume’s view they share the same kind of content. A visual impression, for example, is in the form of an image. Similarly, when one reasons (or attempts to), one tokens ideas that are copies of impressions and are therefore, according to Hume, also images. Accordingly on one reading of Hume, all mental content is imagistic. It is on this point that some more recent faculty psychology departs. Another way to carve up the mind into faculties is to claim not just that there are distinctive mental activities that serve distinctive roles in the overall mind, but that those faculties are about distinctive kinds of things, that is, they have distinctive content. This kind of faculty psychology was brought to a kind of logical terminus in the 19th century by the phrenologist Franz Joseph Gall and then extended by some of his students and collaborators (Gall 1835). On Hume’s view, there is just one faculty of reason, and one can reason (or attempt to) about a wide range of things. Suppose that forming judgments is a way to reason; Humean judgment could thus be about matters ethical, aesthetic, mathematical, legal, historical, chemical, and so on. One faculty, lots of possible subject matters or content. Gall denied this kind of generality and instead identified aptitudes or powers that are specific to, and therefore individuated by, particular subject matters or content types. Thus, Gall iden tified a range of faculties as diverse as a reproductive instinct, a metaphysical sense, a sense of colours, a feeling of property, and an instinct to murder. What makes these the faculties that they are is that they have a proprietary subject matter: A sense of colours is only about colours and, furthermore, operates in a way independent of all other senses (for Gall, there were twen ty-seven, for later proponents of the view, faculties numbered in the forties). Gall’s faculty psychology was part and parcel with his phrenology – the view that mental faculties could be ascertained through measurements and mappings of the human skull – and the latter has been scientifically debunked on many fronts.1 And phrenology aside, Gall’s method of individ uating faculties by their proprietary content was seriously challenged in the 20th century, most notably by Chomsky’s nativist theory of language and by the New Look Psychology of Bruner and others. As the saying goes, though, treasure is user-relative. And Gall’s faculty psychology was resurrected in the 1980s in the form of Jerry Fodor’s Modularity of Mind.
MeNTAL ARCHITeCTuRe
Fodor provides an exemplary overview of the historical background lead ing up to his attempt to redux Gall’s faculty psychology (a much more thor ough, even if controversial, overview than that provided here). He is explicit that he is responding to all of the above: from Descartes to Hume, from Bruner and the New Look to Chomsky’s nativism. He is also explicit about just how much of his proposed theory derives from Gall: we can distinguish four major ingredients of Gall’s notion of a fundamen tal power [faculty]: vertical faculties are domain specific, they are genetically determined, they are associated with distinct neural structures, and – to introduce a new point – they are computationally autonomous. . . . This view of vertical faculties as not merely distinct in the functions they per form, but also relatively independent in the performance of those func tions, will be important later when we turn to consider the notion of a cognitive module. (Fodor 1983: 21)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
2.1 The modularity of mind: the essentials On one of its faces, Fodor’s modularity theory answers the broad question – how does thinking influence perceiving in interesting ways? – as follows: It doesn’t. It will take significant discussion to identify why he comes to this conclusion. What emerges is not only a denial of significant top-down cog nitive effects on perception, but indeed another possible way to distinguish cognition from perception.2 To preview, and on that point, perceptual sys tems are modules in a way that resurrects Gall’s domain-specific faculties, while cognitive systems are not computationally structured in this way. As it happened, this cognitive architecture set a dominant research trajectory in the behavioural and brain sciences for decades to follow.3 The remainder of the book sometimes takes modularity as an explicit foil. This is in keeping with the dialectical situation and strategy that one finds in most of the extant contemporary literature on possible cognitive effects on perception. But there is an alternative strategy, and one that will develop over the next many chapters, culminating ultimately in the book’s conclusion. It is worth briefly highlighting that strategy. The standard view in the philosophy of perception, and in much of the cognitive science of perception, is that modularity is the default theory of the architecture of perception. Call this the default position assumption. This surfaces most clearly in the cognitive penetrability literature. A theorist arguing for a genuine case of the cognitive penetration of perception will typically do so by attempting
49
Copyright © 2021. Taylor & Francis Group. All rights reserved.
50
MeNTAL ARCHITeCTuRe
to show that the case in question violates some important feature of mod ularity (as the reader will find later, the central relevant feature is informa tional encapsulation).4 The debate that ensues then concerns whether that case really does violate conditions of modularity. The point to note at this stage is that for a theory to be granted this default position, to be the theory against which any alternative must position itself, it must be supported by either strong argument or powerful empirical predictions. Put another way, a theory imposes the burden of proof on all competing theories only if it is superior in argument and evidence. The central goal of this chapter is to determine if indeed modularity enjoys this kind of support.5 In his original 1983 presentation of modularity, Fodor identifies nine fea tures that typify a mental module. This introduces a decision point of inter pretation. On the one hand, Fodor is often careful to insist that none of these features is necessary for a system to be a module, even if he also maintains that if a system s has one of these features, then s tends to have them all. So one way to read Fodor is, as he describes, as presenting a “modest” mod ularity concept, where each feature is captured in a pro tanto condition for modularity, and satisfaction of some substantial number of those conditions by an s suffices to show that s is a mental module. On the other hand, he elsewhere says that one of those features – informational encapsulation, to be dis cussed later – is the essence of modularity (1983: 71, 1985: 3, 2001: 58, 63). And further, a careful reading of his analysis suggests that in addition to this feature, some features are more important than others – for example, it is clear (and should become clear why later) that, following the tradition of Gall’s faculty psychology, domain specificity is more central to modularity than a commitment to, by contrast, fixed neural architectures. The strategy here will be twofold. First, take Fodor at one of his words: Informational encapsulation is essential to modularity. Second, and hope fully this further justifies the first part of the strategy, focus on just those features of the nine that imply that cognition does not affect perception in some important way, that is, that perception is cognitively impenetrable.6 A bit of further historical background is in order here. In 1978, the Sloan Institute commissioned an interdisciplinary committee of researchers to give a report on the state of the art of cognitive science as a discipline. Part of the mandate of this committee was to say what unified the research programmes of cognitive scientists, given that all such individuals were formally trained and typically performed their research as anthropologists, computer scientists, linguists, neuroscientists, psychologists, and philosophers – quite a disparate bunch. One clear answer to the question of unification given
MeNTAL ARCHITeCTuRe
in this 300-plus-page report is this: “What the subdisciplines of cognitive science share, indeed, what has brought the field into existence, is a com mon research objective: to discover the representational and computational capacities of the-mind and their structural and functional representation in the brain” (Sloan Report 1978: 6).7 This statement betrays a commitment to what today is called the computational theory of mind which claims, centrally, that the mind is a computer. Fodor’s Modularity was published just a few years later, and its theoretical approach bears the clear stamp of the report and its commitment to the computational doctrine.8
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Domain specificity of modules Perhaps the clearest residue of Gall’s faculty psychology is here: Modules are domain specific. In many ways, Fodor does not maintain the fine grain of Gall’s domain-specific aptitudes or powers – Fodor does not postulate a module for the drive to reproduce or to murder! But for those modules he does countenance, the notion of domain specificity is basically the same. Modules are sensitive to a domain of proprietary input, and likewise deliver output – for example, “hypotheses” about distal properties of the environ ment – just in terms of the information in that domain. Fodor’s emphasis generally is on two kinds of modules: sensory perceptual modules and lan guage modules, which he categorizes as “input systems”. For example, there may be a colour vision module that is sensitive only to (that is, takes as computational input for its processing) certain reflectance properties of the organism’s environment and then delivers colour representations to some later stage of processing. And, importantly, those representations may occur at a level not accessible to conscious experience. For instance, perhaps all of the “sub-visual” modules – including, say, modules for shape and for spatial relations – deliver their outputs to some overall visual mechanism, which only then delivers a conscious visual representation. For this system, and likewise for other sensory modules and language modules, what turns the system on, as it were, is a limited class of stimuli. It is in this sense that modules are domain specific.9 Mandatoriness and speed of modules If your eyes are open and functioning properly, there is sufficient light, and there is a large oak tree in the centre of your field of vision, you will see the oak tree. And you have no choice over the matter. Your looking in the
51
52
MeNTAL ARCHITeCTuRe
direction of the oak may be voluntary, but your seeing it, when facing it and under these conditions, is involuntary. It is in this sense that Fodor identifies perceptual systems as functioning in a mandatory way. By con trast, one’s non-perceptual systems – which Fodor calls “cognitive” – can be deployed with significant freedom.10 You can largely think about, imagine, ponder, evaluate, deliberate over just about anything; at the very least, you can do this in a way that widely departs from what is presently available to be seen, heard, touched, and otherwise perceived. Cognitive processes are not, according to Fodor, mandatory in their operation. With mandatoriness comes speed, since mandatory systems typically do not require a decision to trigger them. As Fodor puts the point: [I]t may well be that processes of input analysis are fast because they are mandatory. Because these processes are automatic you save computation (hence time) that would otherwise have to be devoted to deciding whether, and how, they ought to be performed. . . . Automatic responses are, in a certain sense, deeply unintelligent. . . . But what you save by indulging this kind of stupidity is not having to make up your mind, and making up your mind takes time. Reflexes, whatever their limitations, are not in jeopardy of being sicklied o’er with the pale cast of thought. (Fodor 1983: 64)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Again, by contrast, Fodor emphasizes the time consumption that comes with non-mandatory mental systems, since they tend to involve some minimal amount of decision or choice and therefore require more computational resources. Informational encapsulation: the essence of modularity “The informational encapsulation of the input systems is, or so I shall argue, the essence of their modularity. It’s also the essence of the analogy between the input systems and reflexes; reflexes are informationally encapsulated with bells on” (Fodor 1983: 71). In spite of the rich implications of this brief comment by Fodor, it is not often noted in secondary literature. First, it clearly indicates, contrary to some assumptions about or interpretations of Fodor’s theory (not to mention contrary to other of Fodor’s own words), that informational encapsulation is no mere pro tanto condition: It is what makes a module a module, whatever other conditions such a system may meet. Second, it indicates the strength of informational encapsulation by
Copyright © 2021. Taylor & Francis Group. All rights reserved.
MeNTAL ARCHITeCTuRe
analogy with reflexes. To understand this, let us take a moment to clarify how reflexes work. Think, for the moment, about a reflex as a very simple input–output pair ing. The doctor taps your knee in just the right spot (input), and the lower half of your leg extends involuntarily (output). Now, of course, we know that there is something that happens in between the input and output, but, as a first pass, one can note how for ordinary, non-medically trained individuals, the nature of that process is unavailable. It’s like a black box. Lacking knowl edge of what goes on inside the box, one is entirely incapable of stopping the processing inside it. Now consider someone who is fully aware of the phys ical events that mediate that input and output. This knowledge makes no difference to how that system runs its process; once the input goes in, that process will run its course and the relevant bodily output will be produced. That knowledge cannot, as it were, intervene or override this process once it has been triggered. Importantly, this is what makes it a reflex rather than an action. Understood in Fodor’s computational language, a reflex is thus an (informationally) encapsulated system. And the relevant claim about mental modules – sensory transducer systems and language processing systems – is that they are like that. Since our interest is in perception, consider vision. Fodor’s view is that understood as an input system, the visual processing system is informationally encapsulated. This means that once it has received its proprietary input from the world – that is, information to which the rod and cone cells of the eye are receptive – the system will run its course without input or interference from any systems outside of the visual system. Just like the processing responsible for the earlier reflex, the only information available to the visual system once it has been triggered is whatever information was contained in the input, plus any information stored within that system – thus, the informational encapsulation of visual processing, auditory processing, language processing, and so on. Notice how the informational encapsulation of modular systems serves to explain the other features discussed earlier. A module is domain specific in the sense that it has a propriety class of information to which it is sen sitive and that it will process in a way functionally independent of, encap sulated from, systems sensitive to other domains. (Here again we see just how closely Fodor’s modules resemble Gall’s aptitudes.) And by virtue of this encapsulation, a module will run its processing involuntarily (manditori ness) and quickly (speed). Again, like a reflex.11 An implication of informational encapsulation concerning perceptual sys tems is that visual and other perceptual processing modules are cognitively
53
54
MeNTAL ARCHITeCTuRe
impenetrable. Amongst cognitive systems, Fodor includes central memory (which would presumably include beliefs and other doxastic states), goals, preferences, and motivations. Importantly for Fodor, these systems are not informationally encapsulated, but can exchange information with one another freely.12 Equally importantly, perceptual modules, because they are informationally encapsulated, do not draw on any information from those cognitive systems. An implication of this architecture, and one central to the remainder of this chapter, is that none of those cognitive states or pro cesses directly influence how an individual perceives the world through her senses. Put one way, cognition does not bias perceptual processing. It is in this way that vision and other perceptual systems are robustly, computation ally autonomous. In the last couple of decades, philosophers and cognitive scientists have challenged this implication of Fodor’s modularity and, in turn, challenged that very theory. Before turning to some of the types of challenges that have been made, though, it is worth identifying some of the arguments that Fodor provides for input systems being modular, again with an emphasis on their putative informational encapsulation.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
2.2 The modularity of mind: arguments Fodor’s modularity theory is an empirical one. Furthermore, its content out runs the state of scientific research on the mind, then and now. This is not a criticism, and indeed is standard in scientific investigation: The hypothe ses and theories we draw from empirical data are typically underdetermined by that data. This is all the more plausible in the context of disciplines like the behavioural and brain sciences. The upshot, though, is that the theory is pre dictive and its argument abductive. The theory predicts that as our sciences develop – as we discover more about the structure of the human brain, for instance –we will discover structures that confirm the modularist architec ture. In this respect, it is a scientific theory and one that will stand or fall with scientific discovery. The theory is largely motivated, though, by acknowledg ing both common sense and empirically grounded observations about the mind and then claiming that a modular architecture will best explain those observed phenomena. In this respect, it is a philosophical theory motivated by inference to the best explanation. A few of Fodor’s many arguments are clarified and analyzed here, with an eye towards how they have inspired critics of the theory and how they engage (or are engaged by) attempts to show that cognition affects perception in ways that counter modularity. Not
MeNTAL ARCHITeCTuRe
one of these arguments is particularly effective on its own; the question for the reader is whether they succeed when taken together as components in an overarching abductive argument for modular perceptual systems.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
2.2.1 Arguments from the stability of perception This first argument, like the second one discussed later, is really a combi nation of individual arguments and claims made by Fodor. The basic idea is this. Think of standing on the corner of an intersection of busy city streets at rush hour. Or imagine watching the sunset over the sea while various birds race in sweeping circles, hunting for their evening meal. These contexts are busy, complex, and potentially overwhelming. Yet we handle them with incredible perceptual stability, tracking fast-moving objects, distinguishing individual faces from a distance, discerning more colours than we have con cepts for. Generalizing, in spite of the immense complexity of our immedi ate environment at any one time, we manage to successfully visually perceive (and perceive through the other senses) that environment. Except in cases of poor light or some physical abnormality (say cataracts), the visual system does not (or rarely), as it were, break down. Again, in spite of the massive complexity and variety of what is there to be seen, vision is remarkably sta ble. It doesn’t have, we might say, an overload threshold (or if it does, it is very high). An argument says that this stability would be weakened, if not undermined, if visual processing was subject to substantial influence from other states or processes of the mind, say, what the perceiver believes or dislikes about city traffic, or knows about sunsets, or fears about fast-flying birds. Since vision is, as a matter of observable, empirical fact, largely stable, these top-down influences must be absent. There are at least two fronts along which Fodor attempts to motivate this line of argument. In a number of places, Fodor appeals to an argument concerning the per sistence of visual illusions. The stability of vision is best enabled, the thought goes, by operating much like a dumb reflex. The apparent difference in length of the Muller-Lyer figures, for example, doesn’t disappear when one learns that the arrows are in fact the same size. It seems to follow that at least some perceptual processes are insensi tive to at least some of one’s beliefs. . . . The ecological good sense of this arrangement is surely self-evident. Prejudiced and wishful seeing makes for dead animals. (Fodor 1985: 2)13
55
56
MeNTAL ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The relevant conditional premise is this: If beliefs could influence perceptual processing of the illusory figure, then one would expect the illusion to dis appear (once one believes that the figure is illusory and the lines are in fact of the same length). But the illusion (and other so-called persistent illusions) remains (see Figure 2.1). Therefore, Fodor concludes, at least some of our thoughts do not penetrate visual processing. It is important to note that this argument does not go very far on its own. First, it is questionable that the conditional premise is true. No opponent of modularity makes the bald claim that all beliefs can influence perceptual processing, and this is the claim that would be required to imply the con sequent of the premise. Second, and this is really the same point put differ ently, the persistence of illusion only shows that there are some cases where (some) top-down effects do not occur. But, again, this bears no implication about the possibility of other cases where top-down effects do occur. The existence of persistent illusions would be a counterexample only to a univer sally quantified claim, and that’s not a claim that is on the table. Fodor will reply that the persistence of illusion is best explained by percep tual systems that are encapsulated. Therefore, perceptual systems are encap sulated. For this reasoning to be persuasive, an encapsulation explanation must be compared, and prove superior to, competing explanations. As Jonna Vance frames the dialectic in a recent discussion of predictive processing and illusion, the relevant premise claims that such competing explanations
Figure 2.1 The Müller-Lyer illusion
MeNTAL ARCHITeCTuRe
“lack a principled reason for why synchronic cognitive penetration fails in these cases” (Vance, unpublished ms). A number of recent theorists have in fact offered such principled reasons, and all of them trace back in spirit to an early reply given by Paul Churchland. Churchland suggests first that the Müller-Lyer illusion may be an odd choice for the modular theorist, since one standard explanation for the phe nomenon is a cultural one, where individuals living in “carpentered” envi ronments seem more susceptible to the illusion.14 Churchland floats the idea that this may already suggest culturally sensitive cognitive influence on per ception. And he then concedes that the illusion does indeed persist and that it
Copyright © 2021. Taylor & Francis Group. All rights reserved.
cannot be overridden by any casual, fleeting, “voluntary” attempt to mod ify the character of one’s visual experience. By itself, however, this means relatively little, for the issue is not whether visual processing is in gen eral very easily or quickly penetrated by novel or contrary information: the issue is whether in general it is penetrable at all, where the acceptable means of penetration can include long regimes of determined training, practice, or conditioning. (Churchland 1988: 174)
This last point, concerning synchronic versus diachronic cognitive effect on perception, will figure in later discussion. Note for now that Churchland is pointing to the fact there are existing, non-encapsulation explanations for persistent illusions. This is the important point of Vance’s framing: What competing explanations are there for the fact that your belief about the length of the two lines does not correct your illusory experience of those two lines? Jesse Prinz argues that there is good adaptive reason for this failure of override. When perception and belief conflict in content, as we have in the Müller-Lyer illusion, perception will override belief. Why? Such a mecha nism “would be advantageous, because, otherwise, we could not use experi ence to correct our beliefs” (Prinz 2006). Notice that this in fact subscribes to one of the very basic motivations of modularity, namely, the epistemic role of perception. Importantly, though, such an overriding mechanism would be compatible with the possibility of unencapsulated visual systems. Belief (and other cognitive states) may influence perception when the two do not have conflicting content. More recent, predictive coding theorists go further, arguing that persis tent illusions are a local artifact of the global strategy of the mind. Recall that
57
Copyright © 2021. Taylor & Francis Group. All rights reserved.
58
MeNTAL ARCHITeCTuRe
according to this model, all psychological processes function as “prediction machines”. These predictions are layered in neural networks, with any one layer making predictions about the layer below it, all the way from high-level cognitive processing down to immediate sensory reception. Further, these predictions (or anticipations) are almost always to some degree incorrect, and so the actual information checked against the predicted information produces an error signal. This error signal results in attempts to minimize further error and in the form of feedback from a higher layer to a lower layer. The global goal of the system is to minimize prediction error. Some illusions persist because they result from this globally optimal strategy even if it results in local error. Catharine Howe and Dale Purves (2005) tested this very hypothesis as it pertains to the Müller-Lyer illusion. In brief, they constructed images that comprised natural scenes that could validly contain the arrowhead and arrowtail (found on the upper and lower images in the standard Müller-Lyer figure), as well as other “adornments”. In such images, the adornments fit, as it were, the natural scene. What they found was across the range of test images, when an arrowtail is placed in the same position as an arrowhead (say to the right of centre of the line or shaft which the arrows adorn), the arrowhead has a higher probability (in the sense that there are more natural scene circumstances) of objectively being farther to the right than the arrow head. In ecologically valid circumstances, statistically speaking, the line with arrowtails is more likely to be longer than the line with arrowheads. In the context of the ecologically invalid Müller-Lyer figure, the perceptual system makes an optimal prediction: The arrowtailed line is longer. These same researchers have applied this methodology and explanation to other persis tent illusions (Howe and Purves 2002, 2004; Howe et al. 2005). On this account, to “override” such a strategy synchronically, as Fodor suggests we would expect from unencapsulated systems, would “break the rest of vision in the process” (Lupyan 2015: 559). From that more global perspective, our susceptibility to the illusion is not really a cognitive failure at all. For were the system to overturn the many delicately interlaced layers of intermediate-level processing that deliver this verdict, the result would be failures of veridical perception in many other (more ecologically normal) circumstances. (Clark 2016: 200)
These theorists conclude that the Müller-Lyer and other persistent illusions are illusions in some local sense, but they are symptomatic of an architecture
MeNTAL ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
that is, overall, advantageous. “[T]he illusory percepts are compromises, representing globally optimal solutions” (Lupyan 2015: 558) The conclusion here is not that these alternative explanations are them selves conclusively proved. But they are plausible, and plausible enough that the persistence of illusions lacks the abductive leverage for which Fodor enlists it. And these competing explanations are compatible with, if not committed to, the unencapsulation of perceptual systems. Notice further that these are substantive explanations: They attempt to make sense of the persistent illusions like the Müller-Lyer. Whatever else one says about such attempts, they clearly go further than an explanation that only identifies a negative condition for such explanation (namely, the absence of cognitive influence). Still, Fodor’s point about persistent illusions is an important one, since it suggests that perceptual stability is well explained by a visual module that behaves reflexively, without input from higher-level cognitive processes. He might then couple this observation with a second line of reasoning that con cerns perceptual constancies. It has long been established – both observed and empirically tested – that the visual system comprises constancy mechanisms that ensure that basic perceptible properties like colour, illumination, size, and shape are stably represented in spite of small differences in how those properties are instan tiated in an object or event. For example, a bright red fire engine may be more or less illuminated by sunlight, some of it perhaps shaded by a nearby building. In spite of this, one sees the entire (relevant) surface of the engine as uniformly red. Similar visual constancies are achieved for size, shape, and other properties. Fodor argues that this achievement of visual systems requires modularity. “Central processes”, memory, belief, motivational states, and other cognitive processes involved in reasoning and decision-making, must remain partitioned off from perceptual systems. Proximal variation is very often misleading; the world is, in general, con siderably more stable than are its projections onto surfaces of transduc ers. Constancies correct for this, so that in general percepts correspond to distal layouts better than proximal stimuli do. . . . the work of constan cies would be undone unless the central systems which run behavior were required to largely ignore the representations which encode uncorrected proximal information. The obvious architectural solution is to allow central systems to access information engendered by proximal stimulation only after it has been run through the input analyzers. which is to say that the central processes should have free access only to the outputs of perceptual processors, interlevels of perceptual processing being correspondingly
59
60
MeNTAL ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
opaque to higher cognitive systems. This, I’m claiming, is the architecture that we do in fact find. (Fodor 1983: 60)
There are two separable claims here. First, central processes (for instance, the capacity to form judgments) only have access to the representations that perceptual processing outputs. Call this inaccessibility (Prinz 2006; Rob bins 2017). Second, information from central processes does not inter fere with the computations performed by perceptual processes. Call this non-interference. The inaccessibility claim seems to be that in one’s capacity to access and reason about mental representations of one’s environment, one does not have access to pre-constancy perceptual representations. So your judgment that the portion of the fire engine in the building’s shade is less illuminated than the portion in direct sunlight is not a result of access to any constancy uncorrected perceptual representation of that luminance difference. It is first important to note that we do have access to more than just the colour con stancy corrected representation. Although it is true that we typically see the surface of the fire engine as uniformly red, we can in fact perceptually attend to the lighter or darker aspects of the surface. You will see the fire engine as uniformly red, but you can also see the differences in luminance. (This would prove important if, say, you wanted to realistically depict the scene in a painting.) This is a result of an additional mechanism that enables light ness or luminance constancy. So, two visual constancy mechanisms yield perceptual representations of colour and lightness. So far so good for Fodor’s argument from inaccessibility.15 Again, Fodor’s inaccessibility claim here is that, with respect to the various constancy processing components of the visual system, processes in central systems such as memory or belief do not have access to those processes. As an empirical matter, it is important to note that this is unproven. And indeed there is evidence that such constancy mechanisms are sensitive to background cognition and learning. Researchers have found that constancy may be enhanced by background cognition. For instance, there is evidence that familiarity with object types may influence colour constancy (Olkkonen et al. 2008, 2012), that memory and imagery may influence object constancy (Schendan and Stern 2008; Schendan and Ganis 2015), and that memory of the canonical shapes of objects may influence shape and size constancy (Palmer 1999; see Cohen 2015 for discussion). Fodor may reply, on introspective grounds, that clearly one’s ability to judge colour, illumination, or size is not made in ways sensitive to the
Copyright © 2021. Taylor & Francis Group. All rights reserved.
MeNTAL ARCHITeCTuRe
computational processing of constancy mechanisms; one does not have access to those early neural processes. This is true, but it is trivial. No party to these debates thinks we have access to early stages of visual processing, say what happens in retinal stimulation, or in the information flow from the eyes to the lateral geniculate nucleus (LGN). And importantly, this claim is compati ble with unconscious representations in central systems accessing and inter acting with the processing of constancy mechanisms. Fodor’s observation, even if apt, “shows only that we lack conscious access. It tells us nothing about whether operations within unconscious mental systems are accessi ble to other unconscious systems. For all we know, there may be extensive accessibility below the level of awareness” (Prinz 2006). The inaccessibility claim and line of argument, then, do not seem to take us any further to a fully encapsulated visual system. What of the second, non-interference claim? First, it is important to note that as an empirical matter, current neuroscientific research is far from conclusive on “where” constancy processing occurs in the brain.16 More to the point, what is the principled reason for drawing a sharp line between all central (cognitive) processes and all of visual processing? Again, there is evi dence that constancy corrections in perceptual representations are sensitive to contextual and cognitive factors, even if this is not something we can, at a subject level, consciously access. Therefore, the claim can be established on neither introspective nor empirical grounds. The strength of Fodor’s reasoning is thus, at best, abductive (and he is more or less clear about this). But if the debated point is whether the visual system is robustly informa tionally encapsulated (and therefore modular), one can invoke constancy phenomena to affirm that point only by begging the question. Constancies do achieve stability, but they may do so with influence from cognition. So, it simply does not follow from the existence of these phenomena that com putations in the visual system suffer no interference from cognitive, central processes. 2.2.2 Arguments from the reliability of perception If I were to ask you what perception, say vision, does, you would likely say something in the vicinity of the following. Vision tells us what’s there. That is, it informs the perceiver about her environment; it presents the objects and their features so that she can navigate and act upon those very things. As traditionally put, perception provides knowledge about the world outside of us. Now if you are philosopher or student of philosophy, these intuitive observations lead quickly to a number of deep epistemological challenges.
61
62
MeNTAL ARCHITeCTuRe
Whether perception provides knowledge, and what nature or structure it must take in order to do so, remain some of the central problems for philo sophical theories of knowledge.17 Such problems are not our present concern (though they will surface in later chapters), but the intuitive observations just mentioned make up the materials of this second broad argument of Fodor’s. As a matter of function, visual and other perceptual systems are supposed to reliably provide us with information about the immediate environment; this is what they seem to be for. And given broad behavioural success in navigating and acting upon our immediate environments, it is plausible that visual and other perceptual systems do reliably inform us about what is here and now. Here is one place where Fodor articulates the line of reasoning:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Perception is above all concerned with keeping track of the state of the organism’s local spatiotemporal environment. Not the distant past, not the distant future, and not . . . what is very far away. Perception is built to detect what is right here, right now – what is available, for example, for eating or being eaten by. If this is indeed its teleology, then it is under standable that perception should be performed by fast, mandatory, encap sulated, . . . etc. systems. (Fodor 1985: 4)
Let’s set to one side the teleological component: Motivating modularity and opponent theories on the basis of biological considerations will be consid ered in later chapters. Instead, we can glean an inference to the best expla nation from Fodor here. Perception is generally reliable in its provision of accurate information about the immediate environment. (There is a myr iad of possible descriptions here: Perception is largely accurate or veridical, it tracks or represents objective features of the environment, and so on). This reliability would plausibly be undermined by cognitive influences on perception; Fodor mentions a variety of possible influences, from beliefs to desires to expectations to “utilities”. By contrast, an encapsulated visual system will better ensure that visual experiences accurately present the envi ronment, since this precludes cognitive influences that threaten that accu racy. Therefore, the observed reliability of perception is best explained by modular perceptual systems. It is important to flag here the possible strength of relevant claims, since this will become relevant in the analysis that follows. In some places, Fodor describes informational encapsulation in categorical or absolute terms. Of such
Copyright © 2021. Taylor & Francis Group. All rights reserved.
MeNTAL ARCHITeCTuRe
a system, he writes “its informational resources are restricted to what its pro prietary database contains. That is, the system is ‘encapsulated’ with respect to information that is not in its database” (Fodor 2001: 63; see also Fodor 1983: 62, 66, 71–72). Thus the visual system is entirely blocked from influence by cognitive systems, and a violation of encapsulation would be a violation of modularity (Wu 2013). And there are reasons to think that this is the strength that is needed insofar as one wants to maintain robust functional independence of perceptual systems (Stokes and Bergeron 2015).18 Elsewhere, for example, when providing arguments for encapsulated systems, Fodor describes the phe nomenon in terms of degree. For example, “the claim that input systems are informationally encapsulated is equivalent to the claim that the data that can bear on the confirmation of perceptual hypotheses includes . . . considera bly less than the organism may know” (1983: 69). Similarly, how modularity opposes or rebuts explicit opponents depends upon the strength of the claims made. For example, particular instances of encapsulated visual processing (say, in the case of a persistent illusion) challenge only the claim that cognition and perception are entirely “continuous”. And a claim that vision draws upon less than all of the cognitive information or knowledge possessed by an organism similarly would oppose only a claim of that strength. Such a claim is sometimes attributed to the New Look psychologists but, as indicated in Section 1.3, it is far from clear that they ever made a commitment of such strength (Bruner 1957, for one, explicitly denounced it). So, if the claim made by the opponent of modularity is just that some mental phenomena involve cognitive influences on perceptual processing, then only the categorical encapsulation claim, if true, would undermine such opponents. All of this is relevant in the context of think ing about the epistemology of perception, since such questions concern how much cognitive influence, and what kind of influencing states or processes, would force epistemological worries or revision to a theory of perception. One more specific way that Fodor attempts to motivate the encapsulation of modules on broadly epistemic grounds is by appeal to the redundancy of information. This is a rather difficult argument to parse, so quoting Fodor at length is important: The next of these is a point of principle: feedback works only to the extent that the information which perception supplies is redundant; and it is pos sible to perceptually analyze arbitrarily unredundant stimulus arrays. . . . whereas, the point of perception is, surely, that it lets us find out how the world is even when the world is some way that we don’t expect it to be. (1983: 67)
63
64
MeNTAL ARCHITeCTuRe
So: the perceptual analysis of unanticipated stimulus layouts . . . is possi ble only to the extent that (a) the output of the transducer is insensitive to the beliefs/expectations of the organism; and (b) the input analyzers are adequate to compute a representation of the stimulus from the informa tion that the transducers supply. This is to say that perception of novelty depends on bottom-to-top perceptual mechanisms. (1983: 68)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
There is a variety of ways of putting this point, which is, I think, among the most important for understanding systems. Pylyshyn (1980) speaks of the “cognitive impenetrability” of perception, meaning that the output of the perceptual systems is largely insensitive to what the perceiver pre sumes or desires. (1983: 68)
The first thing to note is that no claim about the effectiveness of top-down feedback (articulated in the first quoted passage) is going to generate a fur ther modal claim about perception (articulated in the second quoted pas sage). Fodor also mentions usefulness and reliability, but here the point is the same. One can easily imagine a creature with ineffective (not useful, unre liable) top-down feedback to perceptual systems that can (to some degree) successfully perceive stimuli (anticipated or not), and the “can” here may be interpreted conceptually, metaphysically, or nomologically. And this point generalizes. No claim about the value of a mental phenomenon entails any thing about the possibility of that phenomenon. For instance, sloppy rea soning is bad. But, sadly, this doesn’t an impossibility make. The modally qualified claims are thus omitted and the argument reconstructed as follows. The argument here concerns the “goodness” of top-down or cognitive effects on perception. Again, Fodor qualifies goodness at least three different ways in the relevant paragraphs: as effectiveness, utility, and reliability. Let us then simply take “goodness” in the following argument to refer to any of those notions, whatever their differences. The reader can substitute any of these terms; the success of the argument does not change. (1) Top-down effects (feedback) on perception result in good perception only if the stimuli are anticipated. (2) If the stimuli are unanticipated, then top-down effects (feedback) on perception will not result in good perception.
MeNTAL ARCHITeCTuRe
(3) Therefore, perception of unanticipated stimuli will be good only if there are no top-down effects (feedback) on perception.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(C) Perception is good (as empirically observed), and so perception is informationally encapsulated (and thus cognitively impenetrable, suf fering no top-down effects).
(1) is taken from the first quoted passage. (2) is the contraposition of (1) (It is therefore not essential, but it may help to make the argument structure a bit more lucid). (3) is supposed to follow from (1)/(2). The conclusion is supposed to follow, but this is in fact only tacit in Fodor’s discussion; he does not explicitly state (C) in or after the passages quoted earlier. Nonetheless, this discussion is embedded in a section entitled “Input systems are infor mationally encapsulated”, the aimed conclusion of which should be obvious. First a worry and a related qualification. The worry concerns the scope of the argument: It cannot generalize to cover all cases of perception. Indeed, it does not even generalize to cover most cases of perception. One cannot generate a broad claim about perception or perceptual processing, as such, from putative facts about a subset of those phenomena. And on one reading, this subset is quite small. This forces an important qualification. There is ambiguity concerning just what would count as an unanticipated (novel) stimulus. It would seem that this cannot be interpreted in one of its strongest senses, such that a stimulus is novel only if it is a token of a type, where the individual has never per ceived any tokens of that type. At the level of types of stimuli – object types like people, trees, tables, and chairs and feature types like colours, shapes, and motion (or determinate instances of those features) – most of what we per ceive is not unanticipated. Sadly, novel stimuli are, well, novel. And the rest of the world, as we say, we’ve “seen it”. Encountering your first armadillo – your first token of that type – was a rare and novel occurrence. If this strong sense of unanticipated (novel) stimulus is the correct disambiguation, then Fodor’s argument is fatally flawed for being so limited in scope. Therefore, if the argument is going to have any remotely broad scope, unanticipated (novel) stimuli must be tokens of types, where the type(s) of stimuli are (or can be) anticipated. In other words, anticipating a type of stimulus S, does not imply that one anticipates the particular token s in one’s present environment. Fodor can add to this that novelty occurs, and occurs regularly, at the level of whole scenes. So even if a scene in one’s perceptible environment is populated by tokens of objects, events, and their features
65
Copyright © 2021. Taylor & Francis Group. All rights reserved.
66
MeNTAL ARCHITeCTuRe
where the relevant types are not novel to the perceiver, the combination of those tokens is nearly always novel. The world changes constantly, and the way that its objects, events, and features are bound in a scene is different from moment to moment. In this sense, much of the world is unanticipated. The question then becomes, would top-down feedback be a hindrance or an enhancement to reliably perceiving unanticipated stimuli (tokens), and their arrangement, qualified in this way. Once we analyze the individual premises of Fodor’s argument, we find good reason to think that such feedback would be epistemically beneficial. Premises (1)/(2) and (3) are not independently plausible. Consider first, premise (1) [or, equivalently, its contrapositive, (2)]. It claims that top-down effects (feedback) on perception result in good perception only if the stimuli are anticipated. First note that this putative “principle” cannot be a general empirical or conceptual point about the utility of feedback. Anyone who has learned a skill has learned this lesson and improved the skill by learning it. Sports are good examples: Returning a tennis serve with a backhand, negoti ating moguls on skis or a snowboard, making a steep and rocky descent on a mountain bike. You anticipate features of the activity at a type level: You know this ski run is steep and involves moguls, the serve is coming momentarily, etc. And so in one sense Fodor’s claim – that feedback can be useful only given some background anticipation – is plausible, but trivially so. One has to have some minimal information or knowledge about an activity type, stimulus type, or situation in order to be able to use any feedback relevant to the type. Importantly, that anticipation only goes so far in performing the skill more effectively or reliably. Why? Because you often do not anticipate what precise form those features of the activity will take, how they will be combined, or the precise tokens of relevant stimulus types. An expert-level skier can still be surprised by aspects of her descent! The tennis player knows the serve is coming, not where it is going or at what speed. And feedback – say from a coach or from one’s inner voice – can make a marked, sometimes essential, difference in enhancing performance in such contexts. In other words, lack of anticipation of precise details of the features of an activity or event (of a type) in these contexts does not undermine the value of feedback in these contexts. On the contrary. Nor can the principle be a conceptual point about perception. It is perfectly conceivable that there could be creatures who better perceive unanticipated stimuli in top-down influenced ways. So the “principle” must be a contingent claim about the psychology of human perception.19 Therefore, a counterexample to (1) understood as a contingent claim takes the following form: a case involving unanticipated stimuli, where top-down
Copyright © 2021. Taylor & Francis Group. All rights reserved.
MeNTAL ARCHITeCTuRe
effects on perception of those stimuli are good-making in some sense (effec tive, useful, reliable). So, consider a perceptual situation where the individual is familiar with stimuli of the relevant type(s) but where there are important features or objects or combinations in that stimulus array that are unantici pated. The expert radiologist has seen many, many radiographic images and knows a great deal about them. But, importantly, she may need to identify or diagnose, and identify or diagnose through seeing, anomalous features in a radiographic image. This, after all, is how tumours are identified in radiograms. And it would not be plausible to say that she expected to see that tumour or even to see a tumour, even if she may have had more or less suspicion given other known features of the patient’s case. In Fodor’s terms, this is not redundant information. What role are top-down or background cognitive states playing here? Plausibly, it is precisely the expert radiolo gist’s familiarity with, her knowledge and understanding of, radiographic images that allows her to quickly and accurately perceive the tumour. Her knowledge will plausibly be useful to how she diagnoses on the basis of such images and how she sees such images. Selective visual attention is tuned to diagnostically relevant features of the visual array and, accordingly, those features are more visually salient, “accentuated”, and the array is perceptu ally organized and attended in a way that is useful in the context. This kind of case will be discussed and defended at length in Chapters 6 to 8 (and cases involving attention and cognitive effects on perception in Chapter 5). For now it is enough to show that premise (1) is dubious: There are plausible cases (many, it will be argued) where perception of unanticipated stimu lus tokens (non-redundant information) is improved (good in all three given senses) by top-down cognitive effects on perception. Such cases are certainly possible, and this is enough to show that (1) [and, equivalently, (2)] is not a mere matter of principle. What of premise (3)? If (1) is dubious [and, equivalently, its contrapositive (2)], then (3) cannot be granted on its basis. But is (3) at least independently plausible? The premise claims that perception of unanticipated stimuli will be good only if there are no top-down effects (feedback) on perception. A counterexample will take the form of top-down affected perception of unanticipated stimuli that, at least, is not “bad” as a result. Consider Fig ure 2.2. Unless you have seen this image before, it will appear as a meaning less array of black and white shapes. Now take some time (say, 15 seconds or so) viewing Figure 2.3. Now return to Figure 2.2. Your experience should be very different. You should now see (even if impoverished relative to Figure 2.3) an organized
67
68
MeNTAL ARCHITeCTuRe
Figure 2.2 “Mooney” image 1
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Source: Reprinted with permission from Teufel et al. (2018); Copyright Springer-Nature
scene, with meaningful features: a person kissing a horse’s nose. This is an unanticipated stimulus if anything is. And in this case, the top-down feed back not only did not hinder the success of your perceptual experience, it seemed essential to it (see Teufel et al. 2018). This illustrates a counterex ample to (3). Additionally, the case of the expert radiologist (and the cases like it, to be discussed in later chapters) are further counterexamples. These are cases where top-down cognitive effects not only are not “bad” making effects on perception, they are improvements to perception, and even when stimuli are unanticipated. Indeed, as it will be argued, these are epistemi cally virtuous cases of cognitive effects on perception.20 A second line of argument from the reliability of perception concerns possible non-doxastic effects on perception. Fodor attributes this point partly to Pylyshyn (1980), who suggests that “a condition for the reliability of per ception, at least for a fallible organism, is that it generally sees what’s there, not what it wants or expects to be there. Organisms that don’t do so become deceased” (Fodor 1983: 68). And again, since perception is by and large accurate, Fodor concludes that it must not be subject to influences such as wishful thinking.
MeNTAL ARCHITeCTuRe
Figure 2.3 “Mooney” image 2
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Source: Reprinted with permission from Teufel et al. (2018); Copyright Springer-Nature
It is certainly true that the influence on perception by non-doxastic states like desire would be epistemically worrying (Stokes 2012; Vance 2014). But that epistemic threat does not, by itself, imply that such influences are impossible. Accordingly, for the abductive inference to be well-motivated, the actual reliability of perception will have to be exceptionally stable. A ques tion thus arises regarding cases of perceptual error and whether any such types of cases might be explained by appeal to non-doxastic influence. Some examples of this possibility are considered in the following section (and again in Section 8.3). And second, many of the plausible instances of topdown effects on perception do not involve non-doxastic states, but instead beliefs, learned concepts, and knowledge. So even if this strand of the argu ment from reliability is successful, it is insufficient to motivate a claim about thoroughly encapsulated modular perceptual systems. Finally, there are places where Fodor’s argument from reliability intimates a transcendental form. One takes as a starting intuition that perception is reliable. An enabling condition for this reliability is informational encap sulation. Therefore, perception must be informationally encapsulated. This mode of argument still appeals in some philosophical quarters,21 but it is not particularly compelling as an argument for a scientific model of the mind.
69
70
MeNTAL ARCHITeCTuRe
2.3 Summary To conclude, the arguments reviewed in this chapter – most generally, the arguments from stability and the arguments from reliability – provide sub stantive representation of the motivation for modularity, and in particular for informationally encapsulated perceptual systems. Recall that informa tional encapsulation is the essence of this strong form of modularity and that perceptual systems like vision are supposed to be modular. Each argument has been criticized on multiple fronts. Even if those arguments might be defended or revised, this critical analysis yields at least the following: It is far from obvious that modularity deserves to be the default position against which any new (or old) theories of the architecture of perception must sit uate themselves. Therefore, although the default position assumption may be common, it is not theoretically justified. It is further commonly assumed that modularity is supported by inference to the best explanation. What we have just found is that some of the phenomena that modularity purports to best explain – for instance, persistent illusions, perceptual constancy, and perceptual recognition of novel stimuli – are in fact not well explained by that theory or are simply too isolated to yield generalizations. To further consider the abductive position that the theory is in, we turn in the next chapter to some of the cases made, and explanations offered, by opponent views.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Further reading The original source for Fodor’s modularity of mind is Fodor 1983. He also offers a précis in Fodor 1985 and a later defence of his version of modularity and a criticism of so-called “massive modularity” theories in his 2001 mon ograph. On the latter, see (Cosmides and Tooby 1992; Sperber 1994, 2001; Carruthers 2006). Pylyshyn’s empirical psychological complement to Fodor’s theory can be found in Pylyshyn 1984, 1999. The latter is a target article from the journal Behavioral and Brain Sciences and so contains many commen taries from theorists in cognitive science. For challenges to modularity in the form of the cognitive penetration of perception, in addition to the many citations in the next chapter, see the recent volume edited by Zeimbekis and Raftopoulos (2015) (which also includes new defences of modularity against charges of the cognitive penetration of perception). An alternative line of criticism of modularity and related views focuses on recent neurological evidence which broadly suggests that it is implausible that the brain takes
MeNTAL ARCHITeCTuRe
anything like a modular structure. The most recent and prominent advocate of that critique is Michael Anderson and his neural reuse theory (Anderson 2014, 2016).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 For an historical overview, see Zawidski and Bechtel (2004). 2 For a recent endorsement of this way of marking the cognition/perception border, see Mandelbaum (2018). 3 Two recent critical discussions of this influence on research in cognitive science are Anderson (2014, 2016). uttal (2001) offers a history of a related research trajectory in psychology and neuroscience, what he calls “locali zation theory”, which centres around the attempt to identify psychological processes with specific brain locales. uttal offers thoroughgoing criticism of this approach and its likely success. 4 Remarkably, Fodor (1985) even laid out three conditions that any case would have to satisfy in order for that case to disprove his theory. And many proponents and challengers have just accepted these conditions as given. 5 Patricia Churchland (1980) offers a similar line of criticism of a related the ory – the computational theory of mind – focusing in particular on what she calls “sentential” theories, which claim that the contents of mental representations take the structure of sentences. Churchland identifies the remarkable support and impact that such a theory enjoyed in research in both psychology and philosophy of mind, but worries that this support may be premised more on presupposition than sufficient argumentation. And for recent criticism of computational theories of vision in particular, see Orlandi (2014). 6 Scholars of modularity will note, then, that not all original features of Fodor’s analysis will be discussed here, and for the reasons just given. For a fuller treatment, see Robbins (2013, 2017). 7 The journal Cognitive Science maintains a free online version of the com plete Sloan report here: http://csjarchive.cogsci.rpi.edu/misc/Cognitive Science1978_OCR.pdf. 8 Fodor was not a member of the relevant committee, but his work was cited extensively throughout the report, and at least one of the committee members – Zenon Pylyshyn – was a regular correspondent and eventual co-author with Fodor, as well as a proponent of a modular theory of mind.
71
Copyright © 2021. Taylor & Francis Group. All rights reserved.
72
MeNTAL ARCHITeCTuRe
9 The concept of domain specificity also features in the Sloan report which, again, guided a great deal of research that follows. “The mind is an array of distinct, domain-specific capacities, each with a specialized cognitive role. This view holds that the complexity of cognitive systems derives primarily from innate structural and functional properties of the brain. These special purpose systems not only would manifest an adaptive specialization of function, but also would permit rapid acquisition from relatively limited and varied experience” (Sloan Report 1978: 202). 10 One possible exception, or at least partial exception, concerns the degree to which our beliefs are the sorts of mental state that are under immedi ate voluntary control. Consider: Look out your nearest window right now. Assuming you can make a clear perceptual observation of the weather con ditions outside your window, you will either believe that it is raining or you will not believe it is raining. And how this belief occurs to you, or that it occurs to you at all, is not a matter of any kind of decision that you immedi ately execute. This is sometimes called doxastic involuntarism. See williams (1970), Alston (1988), Bennett (1990), and Stocker (1982). 11 For a more formalized “definition” of informational encapsulation and its importance, see wu (2011). 12 It is in this way, at least, that Fodor’s modularity is “modest” by contrast to more recent “massive modularity” theories. Massive modularists char acterize much, if not all, of the architecture of the mind as modular, even if in an importantly qualified sense. For examples of such a view (which will not be further analyzed here), see Pinker (1997), Sperber (2001), and Carruthers (2006). 13 Fodor also here hints at a biological-teleological argument for informa tional encapsulation. This argument, and related points for analysis, will be discussed in Chapters 6 through 8. For a criticism of Fodor’s argument(s) from persistent illusions, in addition to those discussed below, see McCau ley and Henrich (2006). 14 For a seminal study and proponent of the “carpentered” view, see Segall et al. (1966). For a challenge to that view, see Jahoda (1971). For a related study and explanation, see Ahluwalia (1978). 15 Thank you to an anonymous reviewer for pressing me on this point. For empirically informed philosophical discussion of an array of perceptual constancies, see Burge (2010), who in fact argues that constancies are central to the distinctive objectivity of perception.
MeNTAL ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
16 For example, on size constancy, some studies suggest a correlation between constancy and activity in visual cortex V1 (Sperandio et al. 2012), and others suggest correlation with activity in area V4 (Tanaka and Fujita 2015). And colour constancy has been correlated with activity in V1 (Mace voy and Paradiso 2001). 17 Some of these issues were outlined in the previous chapter. 18 And it is worth further noting that both the terms “informational encapsu lation” and (its implication) “cognitive impenetrability” suggest a categor ical feature of any relevant system. 19 Again, there is a biological or teleological variant on this argument, which will be taken up in Chapter 6 and later chapters. 20 Predictive coding views (discussed in Sections 1.2 and 2.2.1) claim that the perceptual system is constantly making predictions about the information it will receive from the immediate environment. In such views, percep tion and the rest of the mind/brain function so as to minimize prediction error, and this mechanism pervades all levels of neural processing, with free exchange of information across levels. Importantly, this “cascading” process of top-down effects is supposed to occur, no matter whether the perceivable stimuli are redundant. See Friston (2005), Hohwy (2013), Clark (2013, 2016) and Lupyan (2015). 21 This variant of the argument will be considered again in Chapter 7. For a related argument that takes this form, concluding that perceptual content must be conceptually structured, see McDowell (1996) and the discussion in Section 1.5.
73
3
TOP-DOWN EFFECTS ON
PERCEPTION: THE INITIAL CASE
FOR MALLEABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
3.1 3.2 3.3 3.4 3.5 3.6
Value-influenced perception Doxastic influences on perception Social categories, stereotypes, and perception Intention, action, and perception Language and top-down effects on vision Summary: a stalemate of sorts
Traditionally, theories have varied dramatically regarding how they carve up the mind and how perception is or perhaps is not distinguished from thought. A standard term used to broadly describe this set of questions is mental architec ture. The previous chapter discussed one influential approach to mental archi tecture. The modularity of mind maintains that input systems like vision are modular, centring around an informational encapsulation claim. By con trast, cognitive processes like judgment or belief or decision-making are not informationally encapsulated. The commitment to informational encapsu lation thus promises to both explain various aspects of perceptual systems like vision – its domain specificity, speed, mandatoriness – and to ground a distinction between those systems and cognitive systems.
TOP-DOwN eFFeCTS ON PeRCePTION
For the last few decades modularity has enjoyed default position status in cognitive science and philosophy of mind. What the critical analysis of Chapter 2 revealed, however, is that this status is not obviously well justified. The central arguments for modularity – and in particular, for informational encapsulation of perceptual systems – are not successful, or at least they do not suffice to impose the burden of proof on all opponents of modularity. A second and related angle here concerns whether the modularist enjoys an abductive edge over opponents: Does modularity best explain a wide range of perceptual phenomena? The goal of this brief chapter is to sketch some of the extant literature that encourages anti-modularist answers to that very question. The chapter surveys a variety of empirical studies and interpretations, all of them making the case that there are top-down effects on perception. Some employ talk of “cognitive penetration of perception”; some do not. This is a representative sampling, with the cases typed by the kind of cognitive state or process that is allegedly influencing perception.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
3.1 Value-inf luenced perception It is worth first considering some of the earliest empirically motivated cases for top-down effects on perception, namely, some of those made by New Look psychologists and others (much maligned, as discussed earlier, by mod ularity theorists). One of the first such instances, and many that followed, involved apparent effects on visual perception by values and desires. Perhaps the most famous of these studies is one performed by Jerome Bruner and C.C. Goodman in 1947. Bruner and Goodman’s study targeted size perception and whether back ground values could influence it. The central empirical hypothesis of the study, which seems to have been lost in most of the decades of commentary that has followed, concerned what Bruner and Goodman called “accentua tion”, which they described as a phenomenon where “sought after percep tual objects . . . become more vivid” (1947: 37). The subject pool consisted of 10-year-old children. Subjects were placed before a wooden box with a ground glass screen on which was a near circular patch of light that could be adjusted in size by a knob on the bottom right of the box’s face. Exper imental subjects were asked to adjust the patch of light to match the size of coins of varying values; control subjects performed the same task, but in response to grey cardboard circles that corresponded to the size of the same coins. In all cases, the subjects held the target object in their left hand,
75
Copyright © 2021. Taylor & Francis Group. All rights reserved.
76
TOP-DOwN eFFeCTS ON PeRCePTION
six inches to the left and along the same horizontal plane as the adjustable light patch. The results suggest that as the value of a coin goes up, subjects perceive the coin as larger and larger than it objectively is: For example, an American quarter is on average reported (by the adjustable light patch) to be 37% larger than it objectively is. Bruner and Goodman explain these results as an effect of the value (or desire) for money influencing the size perception of the coins, but not the control stimuli. And in a second round of studies, they found that the effects were further accentuated for poor versus rich children. Now let us note straight away that this is, relatively speaking, a rather old experiment. Accordingly, it is not the most state-of-the-art study on possible cognitive effects on size perception. However, its age is not reason alone for dismissal, and, at the very least, the experimental paradigms used are, in some respects, better than some more recent alternatives if the goal is to isolate genuinely perceptual phenomena. The report method is crucial in this regard. Note that subjects were not asked to give a numerical or otherwise verbal report of the target objects’ sizes. And note further that they were giving their reports – adjusting the light patch to match – while simultaneously viewing the target objects; in this sense, the experiments involved an “online matching” task. These features are important, since they provide some leverage against non-perceptual interpretations of the findings. In other experiments, where subjects are only briefly presented with a stim ulus and then asked to report, or where they are asked to provide a verbal report of a present stimulus, it is equally plausible to explain the results as just indicating what, respectively, the subject remembers about the stimu lus or how she is judging (independent of how she is seeing) the stimulus. Respectively, these are sometimes called the memory explanation and judgment explanation.1 But since Bruner and Goodman’s subjects were matching a patch of light to a present stimulus, it remains plausible that the effect is one on perception and, some have argued, it is less plausible that the effect is just one on memory or judgement. Another alternative explanation of some alleged cases of cognitive effects on perception is that background cognitive states are only affecting where the perceiver directs spatial attention. For example, if one knows which portions of the duck–rabbit image to attend to, then one will see the image as a duck (or a rabbit) (Fodor 1988; Churchland 1988). But this attention-shift explanation looks less apt for Bruner and Goodman’s study, since there seems to be no relevant spatial attention shifts to make: In both experimental and control circumstances, subjects look at the coin or cut-out and make an
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOP-DOwN eFFeCTS ON PeRCePTION
adjustment to the circular patch six inches to the right. It is hard to think of a way that knowing the value or valuing money would change how one spatially attends to the coin or some part of the coin in any relevant way. All of this is important, since it reveals that in spite of its age and whatever other infelicities one may think the study has, its experimental procedure is such that many of the alternative explanations favoured by the modu larist do not readily apply.2 Later studies, and with emphasis on value-in fluenced perception, seem to have taken lessons from these features of the original 1947 study. Briefly, here are two such examples. van Ulzen et al. 2008 employed a methodology explicitly influenced by Bruner and Goodman’s original study but with the improvement of updated technology. Adult participants were asked to adjust a circle on the bottom-right corner of a computer monitor to match circles of varying size in the centre of the monitor. Inside each target circle, subjects would see images of positive, negative, or neutral valence (for example: a flower, an aimed gun, or mushrooms). In these conditions, participants reported (by adjusting the report circle) that circles with nega tive images were larger than (objectively identically sized) circles containing positive or neutral images. Thus the emotional value attached to different images appeared, the researchers concluded, to be affecting size perception. The work of Emily Balcetis and David Dunning is similarly inspired by original New Look studies, centring around what they call “wishful seeing”. For example, instead of just asking subjects to make distance reports verbally, researchers asked subjects to toss a bean bag at a target gift card. When the gift card was substantially valuable, subjects would underthrow the bean bag; when the gift card was of no value, subjects would typically throw the bean bag on or nearer the target. To explain this result, Balcetis and Dunning (2010) hypothesize that subjects visually perceive valuable objects as closer than those of lesser or no value. A later study is partly motivated to pre-empt the judgement explanation mentioned earlier. Here, researchers tested whether desires influenced dominance in cases of binocular rivalry (Balcetis et al. 2012). Binocular rivalry occurs typically in controlled laboratory circumstances when each eye is presented with a distinct, but incompatible, image, for example, the number 4 for the left eye and the letter H for the right eye. In these perceptual circumstances, a perceiver will perceive not a mix of the two incompatible images, but instead just one “dominant” image (say a clear 4). Balcetis et al. then imposed a value structure by informing participants that numbers rather than letters (or vice versa) would be worth points, the accumulation of which could result in some valuable reward. Subjects were
77
78
TOP-DOwN eFFeCTS ON PeRCePTION
Figure 3.1 Example of binocular rivalry stimulus
Source: Reprinted with permission from Balcetis et al. 2012. Copyright Elsevier Publishers.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
then exposed to a series of rivalry stimuli (presenting a single letter to one eye and a single number to the other eye) for 300 ms. Finally, one method of report involved subjects reporting whether a dot was on or off the letter/ number they see, where that dot would be superimposed on either the letter or the number, but not both (see Figure 3.1). This method of indirect report better ensured that subjects were accurately reporting what they see rather than employing deceptive strategies to achieve rewards. The results were sig nificant: The stimuli reported in the rivalry conditions were biased towards stimuli valued (or desired) by subjects.3 As we will see at the conclusion of this chapter, there are extant alterna tive explanations of some of these data. The question will simply be whether those explanations better explain the data than an explanation, as favoured by these researchers, that maintains a genuine cognitive effect on perception.
3.2 Doxastic inf luences on perception A second type of case, and probably the most discussed across various rel evant literature, concerns whether beliefs, knowledge, or other doxastically committal cognitive states influence perception. Here again, we can begin with a rather dated set of experiments and then connect those with some more recent experiments. In a 1965 study, Delk and Filenbaum presented subjects with various shapes cut from uniformly orange paper, some with characteristic colours (a love-heart shape) and some with no characteristic colour (an oval), where the task was to adjust a background to match, in real time, the cut-out as they perceived it. In the experimental conditions, say, when presented
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOP-DOwN eFFeCTS ON PeRCePTION
with a love-heart shape, subjects matched to a significantly more red back ground, by contrast to the control conditions, say, when presented with an oval. Some recent theorists have argued that this is plausibly explained as a case where beliefs about the characteristic colours of kinds of objects (love-heart shapes, apples, human lip shapes) influence visual experience such that those objects are perceived as more red than they objectively are (Macpherson 2012). A more contemporary research programme is very much in the spirit of the original Delk and Filenbaum study. Beginning with a study in 2006, Hansen et al. presented subjects with images of objects with “high colour diagnosticity” and found what they call “memory colour effects”. Subjects were then tasked to adjust such images (e.g. banana images versus uniformly coloured discs as controls) to a (subject-specific) perfectly achromatic grey. For the objects with high colour diagnosticity, by contrast to the control images, subjects tend to adjust the image past the point of perfect grey and into the opposite blue hue range resulting, for example, in a banana image that ranges from three to five times the threshold for discrimination. Researchers hypothesize that for such images, subjects must push the image into the opponent range in order to cancel out the residual characteristic colour; one must make a banana image a bit blue in order to eliminate its enhanced yellow appearance. Later studies found the same broad results for realistic fruit and vegetable images (Olkkonen et al. 2008), for some humanmade objects and icons like the red Coca-Cola icon or the blue Smurf (Witzel et al. 2011), and in a series of online surveys comparing responses to banana images versus control disk images (Witzel 2016). Like some of the studies discussed earlier on possible value-influenced perception, these studies have the advantage of involving simultaneous pres entation of target and report method, and the method involves matching rather than verbal report. This provides some leverage against explanations that claim that the background doxastic states are only affecting memory or judgment. That said, some of the very debate concerns the viability of a judgment explanation of these results, versus one that explains the results in terms of cognitive penetration.4
3.3 Social categories, stereotypes, and perception Another strand of research, focused partly on colour perception, deserves separate mention. (This topic is taken up in more detail in Section 8.3, so the mention here will be brief.) A number of experiments have extended
79
80
TOP-DOwN eFFeCTS ON PeRCePTION
studies on implicit bias and stereotype from social psychology into percep tual studies. In one striking study, researchers primed subjects with either a black male face or a white male face. They were then presented with either a handgun or a (non-gun resembling) tool and then, after a brief mask, asked to identify the item as either a handgun or a tool. When primed with the black male face, subjects more frequently misidentify the tool as a handgun. This result is robust: Even when subjects are aware of this priming effect, they cannot seem to avoid making the racially biased mistake (Payne 2001). Because this paradigm involves a report after presentation of the stimulus, though, it is difficult to adjudicate between an explanation that involves a genuine cognitive effect on vision versus an effect just on a subject’s memory or judgment about the stimulus. A more recent study seems to improve on this possible confound in Payne’s study. Levin and Banaji (2006) asked subjects to adjust a report image to match grayscale images of a typical black male face and a typical white male face. Although the images are identically luminant for each, sub jects tend to match the stereotypical black male face to a significantly darker image. And when researchers created a racially ambiguous face and primed it with either “WHITE” or “BLACK”, subjects matched the very same face to a darker patch of grey in the “BLACK” primed condition. The interpretation favoured by Levin and Banaji, and argued by some philosophers involved in the cognitive penetrability debates, is that beliefs or concepts concerning racial stereotypes are influencing, and problematically biasing, basic colour perception.5
Copyright © 2021. Taylor & Francis Group. All rights reserved.
3.4 Intention, action, and perception A separate line of research concerns whether and how perception involves action in some deep sense, for example, how it may be influenced by one’s intentions to act in one’s visual environment. A good deal of this research has been done by Dennis Proffitt and his many collaborators. Proffitt’s approach to perception is very much informed by the work of J.J. Gibson, who under stood perceptual processes like vision as intimately tied to, indeed as geared towards, action (rather than, as much of the philosophical tradition has it, being for knowledge or justified belief) (Gibson 1966, 1979). A number of experiments task subjects with making spatial judgments on a perceived scene or object in their environment where action-cen tred manipulations affect those judgments. For example, subjects wearing a heavy backpack overestimate the slant of a hill by contrast to subjects
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOP-DOwN eFFeCTS ON PeRCePTION
unencumbered by the backpack (Proffitt et al. 2003). Subjects faced with the daunting task of riding down a steep hill on a skateboard overestimate the steepness of the hill by contrast to subjects who are told that they are to walk down the same hill (Stefanucci et al. 2008). These effects seem to be specific to a particular action and the effort that the subject believes that action to require. For example, when researchers increase perceived effort to walk a certain distance, this affects subjects’ distance perception when they intend to walk that distance, but not when they were intending to throw a beanbag that same distance, and vice versa (Witt et al. 2004). From a neurocomputational angle, Wayne Wu has recently argued that intentions can penetrate visual perceptual processing. Consider spatial con stancy mechanisms. Eye movement can create the same image on the eye’s retina as does an object in motion in one’s environment. For example, when I am looking at my dog, he could either move to my right or I could move my eyes to his left, and the immediate effect on the retinas of my eyes would be largely the same. In spite of this fact, only in the first case do I experience my dog as being in motion. Visual processes thus ensure that objects remain spatially constant when, in fact, they actually are. A prominent model of spa tial constancy mechanisms says that when the perceiver performs a motor action, there is a corollary discharge signal sent for further visual processing, and the visual system uses this signal to make a prediction about the information that will be received from the environment. This prediction is then checked against the actual informational feedback that the visual system receives, and in cases of low error, the perceiver experiences spatial constancy and in cases of high error, spatial inconstancy. Wu argues that intentions to act (in particular, to move one’s eyes) serve as an informational resource for this feature of visual processing such that the model is complete only when we acknowledge that one’s intention to act affects how the system processes both the corollary discharge signal and the feedback from the environment. Accordingly, intention plays a role in whether one visually experiences spatial constancy (Wu 2013). In a second paper, Wu argues that intention influences the sensitivity of selective attention to the target of intention (Wu 2017).6
3.5 Language and top-down effects on vision Another set of studies concerns how linguistic knowledge may affect visual attention and perception. In each of the following three studies, subjects were given a visual search task over a time-limited visual array. In some experimental trials, the visual array would include distractors semantically
81
Copyright © 2021. Taylor & Francis Group. All rights reserved.
82
TOP-DOwN eFFeCTS ON PeRCePTION
related to the target image (Moores et al. 2003). For example, if the target was “motorbike”, or “monkey”, or “grapes”, the search array might include, respectively, a helmet, banana, or wine image. Importantly, the distractor images were semantically related to the target but not visually resembling (helmets don’t look like motorbikes). Methods of report involved both ver bally reporting whether the target was present or making a recency judgment given a pair of images (after the search array is removed). In a second, sim ilar study, experimental trials sometimes included amongst targets images of objects with homophonous names. For example, if the target was a base ball bat image, the search array would include an animal bat image (Meyer et al. 2007). Some relevant results of both studies were these. When present, the linguistically related distractors are better identified and recalled than unrelated distractors and sometimes hinder target identification and slow overall search performance. These behavioural results are corroborated by eye tracking measures in both sets of studies. Semantically related distrac tors (when present) reduce the probability that the first saccadic movement goes towards the target image and the probability that the target remains fixated and that eye movement is far more likely to go to the linguistically related image than unrelated distractors. Finally, these results are further corroborated with electroencephalogram (EEG) measures. The N2Pc event related potential (ERP) enjoys peak onset latency around 200 ms. Impor tantly, this ERP measure correlates with activity in relevant visual cortical areas (extra-striate cortex, including area V4) and is taken to not correlate with spatial attention, but instead with non-spatial selective attention (Hopf et al. 2000; Kiss et al. 2007). The results suggest that this neural activity is sensitive to the linguistic information: Linguistically related images affect the magnitude and onset latency of the N2Pc. How studies like these contribute to discussion of cognition and visual attention will be discussed in greater detail in a later chapter.7 For now, one plausible interpretation is this. Linguistic knowledge stored in long term memory – of relations of meaning and homophony between linguistic items – is modulating how visual attention rapidly selects relevant features or objects in visual search. This effect is plausibly non-deliberate (given its speed and irrelevance to the intended task performance). And this effect would plausibly have an effect on the overall phenomenology of one’s expe rience: what items receive greater perceptual salience and how items in an array are structured into the centre and periphery. This would be either a direct cognitive effect on vision or a rapid effect on vision through the medi ation of selective attention.8
TOP-DOwN eFFeCTS ON PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
3.6 Summary: a stalemate of sorts Whatever their differences, what these various studies and interpretations suggest is that the mind, including perception, is malleable. And part of what shapes perception is cognitive states and processes: what individuals have learned, what they believe or know, what they value, how they intend to act. Such theorists might plausibly claim, further, that a non-modularist theory or alternative better explains the evidenced phenomena. Since background cognitive states like those just mentioned seem to factor in explanations of the relevant perceptual phenomena, vision and other perceptual modalities are not inaccessible to or encapsulated from those cognitive states. Thinking affects perceiving, such theorists might conclude. However, none of the just discussed bases of evidence are sufficient to prove conclusively that cognition influences perception in some interesting way. And although each of these and related studies has factored in recent philosophical arguments against modularity and for cognitive effects on per ception, here, too, there are plausible rejoinders. These rejoinders converge on a few common points. First, many of the experiments are designed such that it is difficult to say whether the behavioural effects are ones resulting from genuine perceptual differences or something post-perceptual. For example, even if subjective reports are quick, there is often time between the visual stimulus and the report such that one might argue that the report is tracking how the subject judges the stimulus, not how she visually perceives it. This is true, for exam ple, in the cases involving action and perception: Did the backpack-wearing subject “really see” the hill as steeper, or rather judge it that way given her intention to act on it? To this rejoinder, one might append an appeal to Fodor’s argument from the reliability of perception. We have reasons to think that perception is by and large reliable, and to ensure that reliability, perceptual processing must be unhindered by non-perceptual processes (for example, what the perceiver wants or intends). And since the exper imental subjects in these studies are often making errors (overestimating the size of an object, mischaracterizing the colour of an object, and so on), the better explanation says that the errors are cognitive, not percep tual. Subjects form false beliefs or misjudge as a result of the background cognitive states, but perception remains uninfluenced and reliable. Both of these rejoinders point to a more general underlying challenge: It is diffi cult to isolate, even through a variety of experimental methods, phenom enal perceptual experience. Anyone at all familiar with this debate will be
83
Copyright © 2021. Taylor & Francis Group. All rights reserved.
84
TOP-DOwN eFFeCTS ON PeRCePTION
familiar with the following rejoinder, really a schema of a rejoinder: Did the experimental subject “really see” the stimulus differently (by contrast to the control subject), or did she instead judge it, remember it, attend to it . . . differently? It would not be an overstatement to say that the debate is at something of an impasse. The reasons here are many and non-exclusive. As is often the case, there is theoretical cross-talk. What counts as “cognitive penetration” or a “top-down effect” is not understood in a singular way across theo rists. Further, and related, past theories are not always charitably interpreted when subjected to criticism. As indicated earlier, some interpretations of New Look Psychology are cases in point. Further, and again related, there is disagreement regarding where to demarcate perceptual experience from other mental phenomena: How do we separate visual attention from visual experience from perceptual judgment? One might respond to all of this in a way familiar to philosophers in a critical metaphilosophical mode: Perhaps we are facing a largely made-up problem – a “philosopher’s problem”. Perhaps the distinctions, models, or theories that undergird the debate are simply too flimsy or ambiguous to ground a genuine disagreement. This response remains on the table, but the hope (or at least the hope that will be embraced here) is that we respond a bit more positively. Grant that no argument or even set of arguments has, to date, been decisive to resolve the debate about cognitive effects on per ception. The response, then, is to look for new angles and approaches. We must look for new ways that thinking might affect perceiving, and this can proceed in a multi-faceted way, by scrutinizing some of the definitions, what falls within the scope of the debate, and what kinds of cases should be con sidered. The next five chapters will proceed in this way. The hope is that they sway the reader in favour of genuine cognitive effects on perception, and deeply important ones. This will amount to a shift from a modular architecture of the mind to a malleable architecture. But of course the reader will be left to assess the success of those arguments and evidence on her own. Further reading There has been a revival of research on the topic of cognitive penetration of perception in the last decade or so. Two survey articles, which discuss both empirical work and related philosophical questions, are Stokes (2013) and Silins (2016). For a recent article critical of empirical attempts to
TOP-DOwN eFFeCTS ON PeRCePTION
provide evidence of cognitive penetration, see Firestone and Scholl (2016), which includes several peer commentaries and replies from the authors. Additional articles on the topic are mentioned throughout this chapter (and others). For two recent collections, see Zeimbekis and Raftopoulos (2015) and Jenkin and Siegel’s special issue of Review of Philosophy and Psychol ogy (2015).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 Or memory interpretation and judgment interpretation. 2 For discussion of these alternative explanations – the memory interpre tation, judgment interpretation, and attention-shift interpretation – see Macpherson (2012), Stokes (2012, 2013). 3 See Dunning and Balcetis 2013 for a review piece. For related work on emo tion and perception, see Stefanucci and Storbeck (2009), Stefanucci and Proffitt (2009) and Clerkin et al. (2009). 4 For criticism of the memory colour effect studies, see Valenti and Fires tone (2019). For discussion of cognitive penetration and colour perception specifically, see Macpherson (2012); Zeimbekis (2013), Gross et al. (2014) and Stokes (2019). 5 See Macpherson (2012) and Stokes and Bergeron (2015). For criticism of Levin and Banaji’s studies, see Firestone and Scholl (2014), and for a rejoinder, see Baker and Levin (2016). 6 The topic of attention and cognitive effects on perception will be taken up in Chapter 5. 7 These very studies on language, attention, and perception are discussed in Chapter 5. 8 For two recent studies and discussion, see Lupyan et al. (2020) and Lup yan and Clark (2015).
85
4
TOWARDS A
CONSEQUENTIALIST
UNDERSTANDING OF
COGNITIVE PENETRATION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4.1 4.2 4.3 4.4 4.5
Definitions and the problem of cross-talk Two definitions and their application Consequences of cognitive penetration Consequentialism Conclusion and summary
Controversy abounds concerning how thought may, or may not, affect per ception. As the discussion of the previous two chapters should make clear, this is no less true in recent, contemporary discussion than it is of much of the discussion in 20th-century philosophy and psychology. Considering for the moment the arguments and cases discussed in the previous two chap ters, those for and against significant top-down effects on perception, here are some observations. First, no single argument is near conclusive. Second – and this is an observation that perhaps requires a bit more digging behind the scenes but it foreshadows the discussion to come – the parties of the debate are not obviously using crucial terms in the same way. What counts as “cognitive penetration” or a “top-down effect” for one party may not for another. And finally, and more broadly, not all parties agree about how and
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
where relevant mental processes or broad kinds should be demarcated in the overall architecture of the human mind (before even considering their possible relations with one another). In the face of these observations, one might simply do (as many have done and are doing) more of the same: find new cases and provide new argu ments for those cases and for one’s antecedent commitments about mental architecture. This might work. And indeed if there is a genuine debate here, it has to work in some sense. An alternative, and the one promoted here, is this: Go back to the drawing board, reconsider what is at stake theoretically and scientifically and why any of these debates and their details should mat ter to us as theorists and, more basically, as humans. It is uncontroversial that cognition influences perception in some ways. The real question is whether cognition influences perception in important ways. This chapter and the next begin to embody this spirit with two related and sometimes intertwined, but also largely separable, novel lines of analysis. The remainder of the book proceeds in similar fashion, taking new angles on debates about thinking and perceiving and, when relevant, attempting to remain sensitive to the existing features of those debates. The present chapter attempts to diagnose and remedy some of the theo retical cross-talk intimated earlier. Here the emphasis is centrally on recent attempts to define the cognitive penetrability of perception and how differ ences in these definitions lead to importantly different verdicts on types of empirical data. The general prescription is that at least some of the persis tent, unbudging debate is symptomatic of diverging essentialist definitions. Therefore, rather than expend further effort on attempting to define the phenomenon (and likely continue the cross-talk), theorists on both sides of the debate should return attention to the consequences of the possible phe nomenon and characterize it in terms of those consequences. Those consequences were given an initial, intuitive discussion in Sec tion 1.2. Recall that the importance of possible cognitive influence on perception can be largely gleaned from the roles that perception plays in our lives. Perception plays an epistemic role, both in terms of the every day knowledge it provides about our surroundings and in terms of its con tribution to rational scientific theory construction and choice. Perhaps less intellectually, it plays a role in everyday behaviour, enabling and guiding our bodily actions. Cognitive effects on perception of the right kind might impact on these roles. Finally, a dominant theory of mental architecture claims that perceptual systems are modular, and this claim is argued on the grounds that perceptual success (in serving the just mentioned roles) is
87
88
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
best explained in modularist terms. Cognitive effects on perception would encourage alternative non-modularist theories; absence of those cognitive effects would support the modularist theory. Chapter 5 focuses on top-down effects on perception and how and where lines should be drawn between “top” or “cognitive” and “bot tom” or “perceptual”. The phenomenon of central interest is attention and how attentional mechanisms should be situated in overall mental architecture, with an eye towards contributing to the debates concerning top-down effects on perception. The general line of analysis is this: The orists on both sides of the debate have been too hasty to dismiss possible and experimental cases as not involving important top-down effects on perception (or as not being instances of cognitive penetration) and on the basis of those cases involving some kind of shift in spatial attention between the background cognitive state (say, a belief) and the affected perceptual experience. The oversight here concerns attention and the var ious forms it can take.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4.1 Def initions and the problem of cross-talk Genuine disagreement, whatever else one may say about it, requires some common ground regarding the target phenomenon or phenomena of inter est and the terms and perhaps definitions that one uses to characterize that target. Put simply, disagreeing parties should at least agree on what it is that they are disagreeing about; else there is a risk that the dispute is, as we say, merely verbal. Or if the dispute is not merely verbal, the source of the disa greement is in any case not what meets the eye. Philosophers, and students of philosophy, are no strangers to these observations. It is not uncommon in philosophy, like other theoretical contexts, that debates persist less as a result of substantial disagreement and more as a result of importantly differing term use, definitions, starting assumptions, and so on. Accordingly, progress is sometimes made not by hammering out more arguments on either side of a debate, but instead reassessing some of the fundamentals of that debate: Are the terms being used univocally across both sides of the debate? Are the targets for explanation being defined in the same ways? Are there dubious assumptions that one party in the debate is making that, if removed, might result in adjudication? Most fundamentally, is there a genuine debate; are all parties genuinely disagreeing?1 The suggestion here will not be that there is no genuine debate regard ing the cognitive penetrability of perception, but rather that the debate has
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
gotten unhelpfully removed, given an overabundance and diversity of terms, definitions, and assumptions, from the core concerns of both sides of the debate. There are many iterations of this problem – call it for simplicity the problem of cross-talk. The primary focus in the analysis that follows will be on distinct definitions of the putative phenomenon and how those definitions yield importantly different verdicts on types of empirical cases. Additionally, some discussion will be offered on other assumptions (and terms) that are made by some, but not all, parties in the debate and that, similarly, thwart theoretical progress.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4.2 Two def initions and their application Defining cognitive penetration as a theoretical task is of little final impor tance to theorists involved in this debate. A definition is instead subservient to answering empirical questions about the mind, namely, whether cogni tion and perception relate in some specially important way(s).2 Getting the definition right is only important to the degree that it assists in performing and interpreting the results of experimentation and explanation. And it is for this reason that an uncontroversial – that is, generally agreed upon – defi nition, or at least characterization, is supposed to be needed. Without this agreement, theorists apply distinct criteria to the same data and come away with diverging verdicts on whether that data provides evidence of cognitive penetration. Two distinct definitions from the literature and applying them to two kinds of cases make this clear. The term “cognitively penetrable” was first coined by Zenon Pylyshyn (1980), and his most recent definition is the one most often cited in the lit erature. Here is its clearest statement: “[I]f a system is cognitively penetrable then the function it computes is sensitive, in a semantically coherent way, to the organism’s goals and beliefs, that is, it can be altered in a way that bears some logical relation to what the person knows” (Pylyshyn 1999). This partial definition identifies a necessary condition for cognitive penetration in the form of a semantic criterion (see Macpherson 2012; Stokes 2013). The criterion as presented by Pylyshyn – call it (SC) – is ambiguous. In some places, Pylyshyn seems to have in mind a logical, inference-supporting relation. A cognitive state like a belief penetrates a perceptual experience only if the content of the belief could support an inference to the content of the resulting experience. This makes the semantic criterion a rationality criterion. One worry here is that, at least as discussed in the literature, it is unclear that worries about cog nitive penetration are worries about a rational relation. Indeed, most theorists
89
90
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
are concerned with cognitive penetration qua non-rational relation – going back to the concerns about theory-ladenness of the 1950s and 1960s, and up through the epistemic circularity worries current in literature today (Siegel 2011, 2013; Lyons 2011, 2016; Raftopoulos 2001a, 2001b). Elsewhere, Pyly shyn ostensibly takes the criterion to require only representational coherence: “This is the essence of what we mean by cognitive penetration: it is an influ ence that is coherent . . . when the meaning of the representation is taken into account” (Pylyshyn 1999: 365, fn3). This weakened semantic criterion requires that one could, perhaps under idealized conditions, identify how the content of the penetrating state, say a belief, affected the content of the resultant perceptual experience. Here again, however, one may worry about the motivation for the criterion thus interpreted, since it imposes a kind of operationalist condition on cognitive penetration. A recent alternative definition follows Pylyshyn by maintaining that cog nitive penetration is not merely a causal relation between cognition and perception, while excluding his semantic criterion.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(CP) A perceptual experience e is cognitively penetrated if and only if (1) e is causally dependent upon some cognitive state C and (2) the causal link between e and C is internal and mental.
This definition accomplishes a few things. First, it makes explicit that the phenomenon of interest involves (at least partly) an effect on perceptual expe rience. This would be a case where the phenomenal character of a sensory experience – what it’s like for the subject to have it – is affected by a cogni tive state like belief. Clause (1) thus ensures that instances where states like belief or memory are affected by antecedent cognitive states do not count as cognitive penetration. Clause (2) maintains that this relation must be an internal one, and the causal chain must involve mental states or processes (but with no restriction on how long that chain is). This clause ensures that instances where one’s beliefs or other cognitive states cause an action of some kind that then causes a (change in) perceptual experience do not count as cognitive penetration. Much could be said about (CP). Suffice it to say that this definition is not, like most definitions, without its problems (Stokes 2012, 2013). Indeed, problems of application of the definition will emerge later. So instead of any extended analysis on those scores, it will prove more instructive to instead identify the ways in which (SC) and (CP) deliver divergent verdicts on dif ferent sets of cases.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Divergent verdicts: the definitions applied to cases of value-influenced perception The first set of cases involves apparent effects of evaluative attitudes on visual size perception. Recall the experiments of Bruner and Goodman (1947) dis cussed in Section 3.1. In both control and experimental groups composed of 10-year-old children, subjects were placed before a wooden box with a glass face. They were to turn a knob, located on the bottom-right corner of the box face, in order to adjust the circumference of a patch of light to match the presented targets. Targets were placed in the subject’s left hand and held six inches to the left of, and on the same horizontal plane as, the adjustable light patch. Target stimuli for experimental subjects were ordinary American coins of values ranging from 1 cent to 50 cents and, as control stimuli, card board cut-out analogues of the varying coins. In both conditions, subjects took as much time as they wished to match the light patch to the respective target stimuli. Control subjects reported the size of the cardboard cut-outs with near-perfect accuracy. Experimental subjects consistently overestimated the size of the coins and by differences (by comparison with controls) greater than 30%. Bruner and Goodman interpret the data to suggest that subjects’ values or desires for money are influencing, “accentuating”, how the coins are perceived. Put simply, the high value of money results in seeing money as bigger than it actually is. Following this early New Look study was a barrage of similarly spirited experimentation, and a number of contemporary theorists, both in philos ophy and psychology, have revived attention to the New Look approach. One very recent example, also discussed in Section 3.1, involves valenced images and size perception (van Ulzen et al. 2008). Here subjects were presented with circles on a computer screen containing either negatively, positively, or neutrally valenced images (e.g. respectively, an aimed gun, kit tens, mushrooms). The task was then to adjust a report circle, located on the bottom-right corner of the computer screen, by pushing or pulling a computer mouse. Leaving out various details, subjects consistently reported circles containing negatively valenced images to be larger than circles con taining images of positive or neutral valence (where the actual size of the image-containing circle is static across trials). On the face of it, these are cases where evaluative attitudes held by subjects (perhaps a fear or aversion to the objects depicted in the negative images) are influencing visual size perception. The question to be asked here is whether these data are to be explained as instances of cognitive penetration or something else. Consider, in turn, the two definitions discussed in the previous section.
91
Copyright © 2021. Taylor & Francis Group. All rights reserved.
92
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
Recall that Pylyshyn’s (SC) requires at least an identifiable representational coherence between the casually antecedent mental state and the resultant perception. In both the Bruner and Goodman studies, and the van Ulzen et al. studies, the antecedent mental state is, plausibly, some orectic attitude (for example, a desire for money in the first case, and a negative affective attitude towards images like guns and spiders in the second). In both studies, it appears that size perception is affected: Subjects see coins or circles with negative images as bigger. However, application of (SC) undercuts this inter pretation: It is far from clear that the semantic coherence criterion is satis fied. Consider the Bruner and Goodman results. Here, Bruner and Goodman conjecture, there is some background desire or value for money. Candidate cognitive states/contents would include S desires that [I have money] or S evaluates that [Money is good]. And the (allegedly) resultant experience is one where coins are experienced as bigger than they objectively are. Put crudely, and begging important questions about the content of experience for the moment, suppose that this experience should be specified as some thing like this: S sees [the coin is size n]. (Here it does not matter how the variable, n, is filled; it simply placeholds a size-specific slot in the content of the subject’s experience. And as per the experimental data, this quanti fication is such that the coin is experienced as significantly bigger than it objectively is.) Here then is the trouble: There is no clear way to specify how contents of the first sort would cohere with content of the second sort. As it is sometimes put in discussion of perceptual content and/or perceptual justification, there is nothing in the first content to “hook up” with the con tent of the (allegedly) resultant experience. One way to put this is in terms of inference: One could not infer [the coin is size n] for any n, from, say, [Money is good]. And this is true no matter the attitude taken towards the latter content. For example, the inference would be no better if the attitude were doxastic rather than orectic. Even on the weaker, non-inferential, inter pretation of (SC), the verdict is the same: The content of the antecedent state (the “meaning” as Pylyshyn puts it) bears no intelligible connection to the content of the second state. Therefore, these data do not provide evidence for cognitive penetration of perception. As noted in the previous chapter, this non-cognitive penetration inter pretation is often coupled with some alternative interpretation, where the latter allegedly better explains the data than the former. The most obvious candidate here is a judgement interpretation. According to this interpretation, we have, at most, evidence for cognitive effects on (other) cognitive states: The subjects make and report judgements on the size of the coins, biased by their
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
values or desires for money. And they do this while perceiving the stimuli veridically (Pylyshyn 1999). What of the second definition, (CP), and its application to the New Look cases? Recall that this definition eschews the semantic criterion, maintaining instead that cognitive penetration requires both an effect on the phenome nal character of perceptual experience and an internal, mental causal link between antecedent mental state (a belief, value, desire, etc.) and resultant experience. This definition has the advantage of incompatibility with some of the alternative interpretations already discussed: If a phenomenon or case meets (CP), then it is not an instance of mere cognitive effects on cognition [by appeal to clause (1) of (CP)] nor an instance of action-guided perceptual change [by appeal to clause (2) of (CP)]. In other words, neither the mem ory interpretation, nor the judgement interpretation, nor the attention-shift interpretation (mentioned also in Section 3.1) look especially apt to capture instances of (CP) should any occur. And plausibly, these New Look cases do meet the conditions of (CP). Taking again the Bruner and Goodman experiment as the example, these subjects have a desire or value for money which affects, directly, the perceptual experience of the size of coins (by contrast with the control stimuli, where no such effect is recorded). Given the online nature of the task – subjects inspect the stimulus as they adjust the light patch to match – the memory interpretation is implausible. Similarly, there is little reason to think that the judgment interpretation is apt, since it would require a consistent mismatch between veridical visual experience of the coins and an erroneous online report of the size of the coins. Subjects give no indication (for example, surprise or confusion) that their reports devi ate in this way from their current experience. Finally, the attention-shift interpretation is implausible, since there seems to be no relevant atten tional difference between controls and experimental subjects, and thus no attentional explanation of the respective differences between these sub jects’ reports. This, anyway, is how an advocate of (CP) might defend its application to the case. The critic may be unpersuaded by this line of reasoning. No matter. The important point for the present discussion is that, as some have argued, appeal to (CP) plausibly yields a pro-cognitive penetration verdict on this set of data. The result: By appeal to one definition, (SC), a set of data is judged not to provide evidence of cognitive penetration. By appeal to another definition, (CP), the same data are judged as good evidence for cognitive penetration. Put simply: (SC) plausibly yields a “NO” verdict, (CP), a “YES”
93
94
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
verdict. This is a clear case of theoretical cross-talk. And no matter one’s side of the debate, this scenario does not bode well for theory adjudication. A second (broad) type of case, and a recent extension of Pylyshyn’s (SC), further reveals the problem. Divergent verdicts: the definitions applied to cases of diachronic changes in perception As already indicated, Pylyshyn offers, at most, a necessary condition for cognitive penetration of perception in the form of his semantic criterion. Something closer to a definition is offered by Fiona Macpherson (2012). Macpherson states the condition negatively: as a description of what condi tions would need to obtain in order for cognitive penetrability of perception to be impossible. Inverting that condition, Wayne Wu describes Macpher son’s definition as follows:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Perceptual experience is cognitively penetrable if it is possible for two subjects (or one subject at different times) to have two different experi ences on account of a difference in their cognitive systems which makes this difference intelligible when certain facts about the case are held fixed, namely, the nature of the proximal stimulus on the sensory organ, the state of the sensory organ, and the location of attentional focus of the subject. (wu 2013: 655)
There are a few features of Macpherson’s definition worth highlighting. First, she endorses a weak version of Pylyshyn’s (SC): Cognitively influenced per ceptual differences must bear an intelligible connection between their con tents. Second, Macpherson consolidates a number of additional features of Pylyshyn’s and Fodor’s analyses on the topic by identifying certain circum stances that allegedly need to be fixed (or held static) in order for a cognitive effect on perception to be relevant to the debate. Note, in particular, the sug gestion that cognitive penetration requires, on this account, that the location of one’s attentional focus is held fixed.3 Now consider cases involving apparent diachronic changes to perceptual uptake and, in turn, cases of visual recognition and visual search. Some recent philosophers of perception have argued that perceptual experience represents high-level properties (see Section 1.5). If vision, for example, is representational, then it represents basic low-level properties like colour and
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
shape. Traditionally, however, properties like being of a natural or artificial kind, being caused/a cause, and being expressive of emotion are understood as the result of post-perceptual cognitive processes like judgment or belief. These high-level properties, tradition has it, are not “picked up” by vision or other perceptual modalities. Both the phenomenology of experience and empirical evidence provide reason to doubt the traditional view. One broad kind of case, experimental examples to be discussed at length in Chapters 6 to 8, involves domain-specific expertise, where deployment of that expertise has a crucial visual or otherwise perceptual element. The composer must be able to distinguish, through hearing, the various instruments that make up the symphony orchestra. The forensic expert must be able to individuate fingerprints through visual inspection. The radiologist must be able to see and then diagnose anomalous features in a mammogram. There are oppos ing ways to understand these cases. First, the expert visually represents the object of expertise just like the non-expert; both experiences would share the same visual content, involving shapes, colours, edges, and rest/motion. The difference between the two is how they make judgements based on those experiences (or what they attend to, if they attend differently; more on this later). By contrast, one might maintain instead that there is a difference in visual representation between expert and non-expert, and this difference occurs by virtue of the differences in knowledge and expert-level skill pos sessed by the first but not the second. The first explanation is one given by a sparse content theorist, the second, a rich content theorist. And the sparse theory tends to comport with modularity, the rich theory, with accounts that favour cognitive penetration as a genuine psychological phenomenon. Consider also cases of visual search where the search array is an especially complex one. Waldo always wears a red and white striped sweater, red cap, and round spectacles. To find Waldo in a Where’s Waldo puzzle, one needs to know this information, since he is always hidden, to some degree, in a maddeningly busy scene. If one does possess this information, it is natural to say that one’s attention is drawn to red and white striped features of the array, and one will quickly see those features (and the objects possessing them) at the relative neglect of other features and objects. Lacking this infor mation, one does not seem to enjoy that same perceptual advantage, and finding Waldo would be largely up to chance. Put baldly, what one knows (or doesn’t) about Where’s Waldo puzzles influences how one rapidly searches the puzzle’s array and how/what one sees in them. But what is the nature of this influence vis-à-vis debates about cog nitive penetration? A unifying feature of these cases is that each one
95
Copyright © 2021. Taylor & Francis Group. All rights reserved.
96
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
involves learning – to recognize features of a mammogram, to spot Wal do-resembling features of a search array. And apparently, or so some argue, this learning affects the phenomenology of perceptual experience. Note that this effect (if it is a genuine effect) takes place over time. This is what one would expect insofar as the learning also takes place over time. So, plausibly, the look of fingerprints, say, gradually changes as one better learns to identify and diagnose on the basis of fingerprints. The question to ask now is whether this diachronic feature of all of the relevant cases, whatever their differences may be, excludes such cases from being genuine changes to visual perception that causally depend in relevant ways upon background cognition. Consider once more the two definitions articulated in the previous section: (SC) and (CP). Begin with (SC) and Macpherson’s extension of the criterion. First note that unlike the previous case, this may be one where there is a seman tic coherence between background cognitive state and resultant percep tual experience. First take the case of visual recognition. Suppose you are learning to become an expert arborist, distinguishing types of conifer trees: pines, spruces, firs. Here you gradually learn that (or, if one prefers, form a belief that) [Pine trees have features F], where “F” placeholds the perceivable shape and colour features typically possessed by pine trees. According to the rich content theory, and simplifying, your experience of pine trees will also gradually change, where eventually you will (in the presence of pine trees) token an experience with the content [There is a tree with features F]. Alternatively, you token a visual experience with the content [There is a pine tree], where the high-level property of being a pine tree is at least partly characterized by certain low-level colour and shape properties, F. Here the coherence is much clearer than in the previous set of cases: One learns about pine-tree looks, and the content of these concepts or beliefs informs, in a coherent way, the consequent perceptual experiences of pine trees. A simi lar story can be told about the basic cases of visual search. Simplifying, one believes that Waldo wears a red and white striped sweater and, as a result of this belief, one enjoys more (and more quickly) visual experiences as of red and white striped objects. The features represented by the visual expe rience cohere with – are made intelligible by – the features represented by antecedent belief. However, an advocate of (SC) will resist this case as one of cognitive penetration, and largely for reasons made clear in Macpherson’s definition. There are two plausible ways to resist this case. First, one might maintain that these cases do not even meet (SC) as baldly characterized, since (SC)
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
requires a coherence relation between a cognitive state and perception (in Macpherson’s terms, the first must make the second intelligible). And, one might maintain, the cases in question do not involve a relation between belief, say, and perception, but instead just some kind of non-cognitive per ceptual change. Fodor gives this reply to Churchland’s discussion (1988) of perceptual adaptation to inverting lenses which, remarkably, can occur in subjects in a week or so (Stratton 1897; Kottenhoff 1957; Taylor 1962).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
For there are, after all, good ecological reasons why you might expect plas ticity of this sort . . . what needs to be kept open for re-calibration is what ever mechanisms compute the appropriate motor commands for getting to (or pointing to, or grasping) a visible object on the basis of its perceived location. Adaptation to inverted (and otherwise spatially distorting) lenses is plausibly an extreme case of this sort of recalibration. (Fodor 1988: 193)
This intra-perceptual interpretation may well be invoked for the visual recognition and search cases, which alleges that the changes in perceptual systems are made by perceptual systems over time. Therefore, no cognitive penetration, because one necessary relatum, a cognitive state, is absent in the relevant causal story. Alternatively, a modularist may appeal to an attention-shift interpretation. For example, what is not static across the Where’s Waldo–savvy perceiver and the Where’s Waldo–ignorant perceiver is spatial attention. The former has beliefs about Waldo’s appearance, and these guide the location of her attentional focus; the latter lacks those beliefs, and so the location of her attentional focus will vary in importantly different ways. In effect, atten tion mediates between background cognition and resultant perception by shifting where the perceiver focuses. This accordingly makes a difference in the visual information that is available for pickup. But this means that the effect on perception is relatively trivial: If you change the (attended) stim ulus, you change the perceptual experience. Same conclusion: no cognitive penetration. If this is the right characterization of the role of either basic perceptual learning or attention in the recognition and search cases, then for similar reasons, it would appear that the case also fails to meet the conditions set by (CP). On the intra-perceptual interpretation, clause (1) – E is causally dependent upon some cognitive state C – is not met, since the changes that take place over time are ones done by, so to speak, the visual system, with no explanatorily
97
Copyright © 2021. Taylor & Francis Group. All rights reserved.
98
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
relevant background cognitive states. And on the attention-shift interpreta tion, the relevant clause is (2): the causal link between E and C is internal and mental. As described so far, the internal, mental link is broken, since one’s beliefs about, say, Waldo and his appearance drive bodily actions – looking here or shifting your eyes there – which only then results in a different visual experience (in contrast to the Waldo-ignorant). This, the sceptic of cognitive penetration will urge, is not structurally different from cases where beliefs, desires, and other mental states cause you to perform some action (say, to walk out of your apartment) that then results in changes in visual experi ence (you see the sunset). And that cognitive-perceptual relation is, again, relatively trivial. The point to note for the present discussion is that applying (SC), Macpherson’s definition, and (CP) is less than straightforward. Each applica tion requires further interpretation of important features or circumstances of the cases. In this case, the relevant circumstance appears to involve either perceptual learning or attention. This further reveals, again, contentious assumptions made by one party of the debate – sceptics of cognitive pen etrability – that are not shared by all members in the debate. Perceptual learning is assumed to be purely perceptual. And attention is assumed to be spatial attention, which is something under the voluntary control of the agent and that stands “in between” cognition and perception. In short, both phe nomena are being interpreted or understood in a rather limited way, further undermining debate-neutral analysis of possible cases. This is more of the same: a problematic instance of theoretical cross-talk.4 One reply to this situation is to conclude that (SC) and its recent exten sion and (CP) and the theorists that advocate these definitions (or some thing like them) target distinct definienda. If the definitions are used in such divergent ways or deliver clearly distinct verdicts on the same observed phenomenon, then perhaps they are simply not talking about the same thing. There is some truth to this reply: It acknowledges what may be a mistake endemic to the cognitive penetration literature. Attempts to define or characterise the phenomenon have erred towards defining cognitive penetration as such and in a way that has lost sight of the supposed con sequences of the (possible) phenomenon. That is to say, there are reasons that philosophers and cognitive scientists began discussing the possibility of something like cognitive penetration, and those reasons all concern the consequences that a phenomenon would have if that phenomenon occurred (and, perhaps, with some frequency). These consequences are,
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
it will now be argued, what is of common interest to both parties of the cognitive penetration debate and no matter what definitions such parties offer. The fix, then, is to understand the target phenomenon in terms of its consequences and, in turn, to abandon essentialist definitions of “cog nitive penetration”.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4.3 Consequences of cognitive penetration There are four central consequences of interest. They concern, respectively, theory-ladenness of empirical observation, the general epistemic role of per ception, the behavioural role of perception, and (modular) architectures of the mind. These are briefly discussed in turn. The thread through these consequences, the big consequence if one likes, concerns how and what we use perception for and, accordingly, how we understand it as philosophers and scientists. As we will see, these are high stakes, both humanistic and theoretical. Traditional empiricist models of scientific inquiry came under fire in the middle of the 20th century. (Empiricism here includes, but is not exclu sive to, logical positivists and logical empiricists.) A number of philoso phers challenged the traditional assumption that scientific observation is theory-neutral in a way that would support rational theory choice (Hanson 1958, 1969; Kuhn 1962; Feyerabend 1958). At least one version of this worry understands the relevant observation as simply perceptual observation. The corresponding worry is epistemic: If perceptual observation is laden with theory (in particular, the theory or theories being tested), then it will not provide a means for rationally choosing or adjudicating between scientific theories. A pair of theories – say, an earth-centred theory vs. a sun-centred theory of our universe – may be equally successful in terms of elegance, parsimony, internal coherence, and other (non-observational) criteria of theoretical suc cess. The plausible empirical method of adjudicating between these com peting theories is to test them against perceptual observations of the world. However, if the observations made by the respective theorists are imbued with theoretical commitment, then those observations fail to provide a neutral form of adjudication. Extending the toy example, the earth-centred theorist might report, upon watching the sunrise, that he sees a sun mov ing across a stationary horizon. And, predictably, the sun-centred theorist reports seeing a horizon moving to expose a stationary sun. Accordingly,
99
100
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
both theorists will report empirical corroboration of their respective pre dictions, and no progress is made in motivating a rational choice between theories. This is, in one characterization, a perceptual phenomenon. Differ ences in perception potentially undermine the rational, theoretical role of empirical observation.5 Although theory-ladenness is typically presented as problematizing the empiricist picture of science and, more generally, the role of perception in scientific theory choice, it is worth noting possible instances of “good” theory-laden observation. Perhaps the cartographer more efficiently (percep tually) identifies the map’s features, the radiologist, a cancerous tumour. This kind of accuracy and efficiency gives the expert in science and medicine an advantage over the non-expert. And this advantage would depend upon the background theoretical and technological understanding of the perceivers. These cases – cases of perceptual expertise – will be discussed in Chapters 6 to 8 as potentially virtuous cases of theory-laden observation. Accordingly, questions about the epistemic valence of theory-ladenness will be postponed until then. For now, the point to be emphasized is that if observation is theory-laden, this is of epistemic import. Theory-ladenness is an important possible consequence of cognitive effects on perception.6 The second consequence is also epistemic, generalizing from the first consequence. Perceptual experience, on the most intuitive picture, provides us with knowledge about the world. But this epistemic role for perception is threatened if there are circumstances where background cognitive states influence perceptual experience that, in turn, influence belief formation. The most obviously pernicious cases are ones that involve a rough causal schema of the form:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Belief that P → Experience that P → Belief that P
Here the causal history of the consequent belief involves a circularity that plausibly undermines reason for that very belief (Siegel 2011, 2013). The problem with this circularity, or anything like it, is clear given the supposed role for perception and action. Cognitive impenetrability theorists are no less clear on this point. As Fodor suggests, the “function of perception is to deliver to thought a representation of the world” (1985: 5). And since this representation is supposed to inform belief and action (about or in response to the here and now), it should track not what the agent believes, wants, or otherwise thinks about the non-present; it should instead track the present environment, here and now. Indeed, and as we have already seen, it is for
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
this very reason that Fodor argues that perceptual systems are cognitively impenetrable, claiming that “isolation of perceptual analysis from certain effects of background belief and set . . . has implications for both the speed and objectivity of perceptual integration” (Fodor 1983: 43; emphasis added; see also Pylyshyn 1980). Whether perceptual processing is isolated (and thus cognitively impenetrable) is an empirical question and up for current debate. But the consequence here is clear: If perception is influenced by background cognitive states in these ways, then the putative objectivity of perceptual representation is threatened. A question concerns the scope of the epistemic consequence. If a belief has a circular etiology (following the schema noted earlier), does this result in a mere case of local unjustified belief, or does some kind of global scepticism follow? Jack Lyons frames the question this way.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Suppose . . . that top-down influence is merely probabilistic in the sense that theory-consonant observations are more likely than they would oth erwise be but that objective facts are still significant determinants of what is observed. . . . Many factors keep our perceptual access to the world from being infallible – poor observation conditions, camouflage, distrac tion, sleepiness, etc. – why should prior beliefs be more than another such factor? (Lyons 2011)
Lyons is suggesting that these other factors are insufficient to motivate global scepticism, so why should something like cognitive penetration motivate global scepticism? A response might go as follows: Factors like camouflage, distraction, and the like are perceptually muddying, but they are epistemically neutral with respect to some proposition P. Belief etiologies like the circular one schematized, however, present a perceptual bias towards a particular P. In some of the alleged cases, the top-down influence is biasing in ways that raise the probability that the subject infers that P. So it is not the confounding of infallibility – the muddiness – that is epistemically pernicious; it is the apparent biasing effect towards experiences that encourage certain infer ences. But this brings Lyons’s point regarding local versus global results to the fore: Are these biasing effects sufficiently frequent to motivate a general epistemic problem, or are they just another example (even if slightly differ ent) of human fallibility? These first two consequences betray a standard view in analytic episte mology, namely, that perceiving is for knowing. This view seems right, but it
101
Copyright © 2021. Taylor & Francis Group. All rights reserved.
102
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
may also be short-sighted. As a variety of theorists have suggested, perception is more fundamentally for behaviour, for acting in and on the world. This third consequence then concerns the role that perception plays in determin ing action and action potential. A relatively extreme articulation of this role of perception derives from the ecological theory of psychologist J.J. Gibson. According to Gibson, the role of senses like vision is not primarily to provide knowledge or rep resentations of the world, but to provide the perceiver with affordances. “The affordances of the environment are what it offers the animal, what it pro vides or furnishes, either for good or ill” (Gibson 1979: 127). And what is offered is, for Gibson, action-able. For a human perceiver, a flat horizon tal surface affords sitting, a door handle affords opening, a tennis racket affords swinging. Importantly, an affordance is then relational, constituted both by objective physical features of an object and the action potential relative to the organism. The door handle would not afford opening to the infant or to a cat. Gibson thus tied perception directly to how it drives action; vision gives you the subject-relative utility or action-ability of the objects in your environment.7 Lessons can be gleaned from Gibson’s account without commitment to its anti-representationalism. We can still reasonably think about perception as involving representation. On a fairly strong account, then, included in those representational contents are the action potential(s) of the object per ceived – what those objects afford. A weaker commitment may not include affordances in perceptual content, but still maintains that the causal link between perception and action is often direct, without any intermediate thought or planning or inference. Perception is, in short, for acting. In this way, Gibson’s account laid the groundwork for resisting what some have called a “sense–think–act” model of the relation between sensation, thought, and action. Perception is commonly cast as a process by which we receive infor mation from the world. Cognition then comprises intelligent processes defined over some inner rendition of such information. Intentional action is glossed as the carrying out of commands that constitute the output of a cogitative, central system. But real-time, real-world success is no respecter of this neat tripartite division of labor. Instead, perception is itself tan gled up with specific possibilities of action – so tangled up, in fact, that the job of central cognition often ceases to exist. The internal representa tions the mind uses to guide actions may thus be best understood as
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
action-and-context-specific control structures rather than as passive reca pitulations of external reality. (Clark 1997: 51)
When you open the door or sit on the chair, you need not, and often do not, see the handle or the chair and then form an intention to open the door or a plan to sit on the chair. You simply see and act, and you do so effortlessly and, as we say, without thinking. We constantly navigate the world by per ceiving it. This is the behavioural role of perception. If cognition could impact on this behavioural role of perception, this would be of significant consequence: It would be an influence on how we act on the world through seeing it. Imagine an unusual door handle. It appears just like a standard round doorknob but works differently. Instead of turning the knob one way or the other, you must first push the knob in, then pull it out towards you, and then turn the knob clockwise. Obviously, lacking this knowledge, you would not be able to open the door. Once you have the knowledge, you will use it to guide your action as you feel and look at the doorknob, perhaps verbalizing in your inner voice something like “push then pull then turn to the right”. Performance here will initially be clunky and involve thought in combination with sensation. But just as with a standard doorknob, you will eventually internalize these instruc tions for action and open the door effortlessly and without thought. Here is one way to think about this change. Your perceptually guided action has changed because, after the acquisition of knowledge about or a concept of the unusual doorknob, your perception of it changed. Another good exam ple which will figure in later discussion is sports. When one first learns a sport, the order of play and the patterns of that play are unfamiliar and therefore unpredictable. Performance follows suit: The beginning tennis player will flail, miss the serve wildly, lose grasp of the racket, and so on. The elite-level tennis player obviously performs dramatically better. After years of training, practices and drills, learning about the game’s possibilities and strategies, the elite player makes child’s play of child’s play, performing effortlessly much of what was once laborious and slow. On one interpreta tion, her visually guided action is markedly improved by her knowledge and expectations about the sport. It would be of importance if the immediate action potential of our envi ronment, as we perceive it, is impacted by what we know and expect and by what categories and concepts we possess. This is the possible behavioural consequence of cognitive influence on perception.
103
Copyright © 2021. Taylor & Francis Group. All rights reserved.
104
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
The first consequence concerns the scientific-theoretical role of percep tion. The second consequence concerns, as we might say, the “everyday” knowledge providing role of perception. This general epistemic conse quence is importantly relevant to the fourth possible consequence of cog nitive penetration, namely, architectures of the mind. As we have already seen, modularity theorists (at least of Fodorian strength) claim that per ceptual systems are informationally encapsulated. A core motivation (per haps the primary motivation) for this claim is that the processing of such systems would be immune to error introduced by the broader cognitive system. Since computations performed by these modules are supposed to be insensitive to what the organism knows, expects, or wants, the result ing perceptual representations more reliably inform the organism about its environment (Fodor 1983: 68–70). The modularity thesis is an empirical one, and one that has set substantial research agendas in cognitive science. So, if it turns out that cognitive penetration of perceptual systems occurs (that is, perceptual systems are unencapsulated relative to cognitive sys tems), then this alleged feature of modules, even if epistemically desirable, is not actual. Although this last consequence is often put this way, it is not exclusively a concern about perceptual processing. If perceptual experience is directly influ enced by background cognitive states, modularity is no less threatened. Here are two reasons. First, although higher-level effects on some components of perceptual processing do not imply the cognitive penetration of expe rience, and higher-level effects on perceptual experience do not imply by themselves that a particular stage of perceptual processing is penetrated, higher-level effects on perceptual experience do imply an effect on percep tual processing at some stage. At least this follows if we assume any kind of physicalism, according to which experience is identified with, constituted by, (metaphysically) determined by, or the output of perceptual processes. Second, as mentioned earlier, the modularity thesis is largely motivated by epistemic concerns. The idea is that since the envisioned encapsulated per ceptual module rigidly performs its computational function and with no interference from extraneous higher-level information – what the organism knows, expects, or wants – the resulting perceptual representations more reliably inform the organism about its environment. And a concern with the reliability or accuracy of perception is a concern with perceptual representa tion or experience, not merely processing. One forms beliefs on the basis of what one sees, hears, and so on. Therefore, modularity theorists have to be concerned with perceptual experiences, not just processing.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
These four consequences are substantial, concerning the epistemology of science, everyday reasoning and rationality, the role of perception in deter mining action, and broad cognitive scientific theories about the architec ture of the mind. Some of the details of such consequences require further analysis, and it is unclear which alleged cases of cognitive penetration imply which, if any, of the consequences. But what is clear is that agreement con verges on these consequences. Fine details aside, theorists on either side of the cognitive penetration debate agree that the importance of the possible phenomenon – cognitive penetration – consists in its bearing these conse quences. And so, it will now be argued, the question – Is perception cog nitively penetrable? – should be revised (or revived) accordingly. We should instead ask: Does cognition affect perception in such a way that one or more of these consequences is realized?
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4.4 Consequentialism The basic prescriptive thesis is this: Any analysis of cognitive penetration should be constrained by its consequences. Therefore, an analysis (or, if one prefers, a definition) of cognitive penetrability will be successful just in case and to the degree that it describes a phenomenon (or class of phenomena) that has implications for the rationality of scientific theory choice, the epis temic role of perception, the behavioural role of perception, mental architec ture. Call this the consequentialist constraint on analyses of cognitive penetration. Note that as stated, the consequences are presented as a list; the logical rela tion between the analysis (or definition) and the four general consequences need elucidation. And indeed the constraint could be satisfied in a number of ways. The remainder of this chapter will articulate three such options and then give reasons for favouring the third. The resulting characterization of cognitive penetration will then be applied to the two cases discussed in Section 4.2 (and also to cases involving attention in Chapter 5). Finally, the two definitions—(SC) and (CP)—will be reconsidered in the light of the consequentialist constraint. The first option is to stick with traditional conceptual analysis. One con structs an essential definition that describes a phenomenon that results in one or more of the relevant consequences. Here the conditions specified in the definition need not make any one consequence explicit, but instead just describe the relevant mental and causal structures. So long as any one satisfac tion of these conditions is co-extensive with a phenomenon that bears at least one of the relevant consequences, then the consequentialist constraint is met.
105
Copyright © 2021. Taylor & Francis Group. All rights reserved.
106
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
There are familiar reasons to doubt both the need for and probable success of an essentialist definition. Contemporary philosophy is rife with examples of controversial, if not failed, attempts at conceptual analysis. So, although it might be ideal for some purposes, we have reason to be sceptical that “cogni tive penetrability” is amenable to this kind of definition. Perhaps more impor tantly, if we can do the work by appeal to the consequences, why bother with the conceptual analysis and corresponding counterexample game? The second option is conjunctive consequentialism. It says that ψ is cognitive pen etration if and only if ψ is a cognitive-perceptual relation that implies con sequences for theory-ladenness and the epistemic role of perception and the behavioural role of perception and mental architecture. In some ways, this is an improvement over the previous option, since it makes the consequences explicit. However, this is no less an instance of traditional conceptual analy sis, and so any worries that applied to this logical feature of the first option will apply here as well. More substantively, the conjunctivist option makes an assumption: Cognitive penetration is a phenomenon that has all (and only) four general consequences. Accordingly, it rules out the possibility that there are distinct phenomena of interest that imply distinct consequences. This deviates from the spirit of the consequentialist proposal. Each consequence matters and, by the very nature of the consequences themselves, for different reasons. And it is not implausible that there are distinct cognitive-perceptual phenomena that imply one consequence but not the other. Finally, a third option is disjunctive consequentialism. It says that ψ is cognitive penetration if and only if ψ is a cognitive-perceptual relation, and ψ implies consequences for theory-ladenness or the epistemic role of perception or the behavioural role of perception or mental architecture. This best captures the spirit of the consequentialist constraint. Theorists on both sides of the debate (or involved in related, but distinct debates) are interested in some relation between cognition and perception that implies some of these con sequences; they are interested in whether cognition influences perception in some important way. Accordingly, the characterization leaves open the possibility that there are distinct phenomena that meet distinct disjunctive conditions. Of course, one may maintain that what unifies the phenomena is simply that they bear one or more consequence; thus, the consequences unify. And accordingly, one may call any phenomenon that satisfies the clus ter condition “cognitive penetration”. A virtue of disjunctive consequentialism is that it makes no commitment regarding unified explananda beyond the appeal to consequences. This makes explicit the most important issues of concern (namely, the consequences of concern to both parties of the
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
debate) but without commitment to a single unified mental phenomenon of interest.8 One does not simply apply the disjunctive analysis to any one case to deliver a verdict. Instead, the inferential procedure remains abductive. For any given case, one considers alternative interpretations and asks what best explains the data. So, one asks, are these data best interpreted in accordance with the, for example, judgement interpretation or some other alternative, or as a phenomenon that satisfies the cluster condition specified by disjunc tive consequentialism? It is important to note that these alternatives are alter native to any interpretation that characterizes a phenomenon as implying one or more of the relevant consequences. So, for example, if a phenomenon is best interpreted as involving a cognitive effect on judgement (but not per ception), then there are no relevant consequent concerns about the epistemic or behavioural roles of perception nor about mental architecture vis-a-vis perceptual systems. The standard alternatives thus preclude the consequences. Making this relationship explicit is another virtue of the account. This will become clearer upon application of the consequentialist account to the two empirical cases types discussed earlier. It should be noted at the outset that the goal here is not to employ the consequentialist account to deliver conclusive verdicts. This is, of course, a final goal, and for any one case or set of data, extended analysis would be required. But here the ambition is more modest, namely, to offer a sketch of the debate-neutral value of the consequentialist line. Accordingly, discussion of both kinds of cases will be brief, with points for further analysis flagged for later chapters of the book. Consider the New Look and New Look–inspired cases involving apparent effects of evaluative attitudes on perception. Here the question is whether this apparent effect is actual: Is perception affected by value in such a way that more than one of the relevant consequences follow? And, recall, if the empirical results are best explained in one of the alternative ways (as advo cated by sceptics of cognitive penetration), then these consequences do not follow. Working through these alternatives, then, begin with the memory and judgement interpretations. In both control and experimental circumstances, these experiments are online: The perceptual target is available while reports (of a match) are made. Accordingly, it is implausible that the reports are just of subjects’ memories. Although more plausible than the memory inter pretation, it also seems unlikely that perception is wholly unaffected (that is, coins are seen accurately), while judgements consistently are erroneous (coins are reported as significantly larger), since size judgements are made
107
Copyright © 2021. Taylor & Francis Group. All rights reserved.
108
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
on the basis of current perceptual experience. If this were the effect, we would expect some kind of confusion on the part of the subjects, since they would, on this (judgment) interpretation, be seeing the two objects (coin and matching light patch in the experimental condition) veridically (where, given the results, the adjusted light patch is significantly larger than the coin) while reporting that the two objects match. No indication of this kind of confusion or surprise shows up in the relevant experiments. There also seems to be insufficient differences between control and exper imental circumstances for an explanation involving overt shifts in attention. In the Bruner and Goodman studies, for example, the task is relevantly the same, with target (coin or cardboard cut-out analogues) and report circle displaced by six inches along the horizontal plane. Any attentional perfor mance that occurs in the experimental case would presumably occur in the control case. Finally, the target stimuli in these experiments – coins, or things like guns in the van Ulzen experiments – are not likely those for which we have a mere perceptually learned response or some kind of percep tual evolution or plasticity. Instead, we learn about these artificial kinds. And this cognitive learning, and the accordant values of those kinds, may affect how we perceive such kinds. Although only a brief treatment, these data look very much like evidence for a genuine cognitive effect on perception. The final question is whether this effect is of relevant consequence. If the brief analysis just provided is apt, then the cases in question do appear to bear the relevant consequences. If non-doxastic states like desires or values influence experience in these ways – by contrast with the alter native interpretations – then both epistemic consequences follow. Percep tual observation in scientific consequences will not be theory-neutral, and the general knowledge-providing role of perception will require theoretical revision. Put neutrally, at the very least, the epistemic role of perception is cast in a new light and will accordingly require new analysis.9 One can also imagine how the behavioural role of perception could be significantly impacted. Shape and size perception factor importantly in how we act upon objects. If the subjects in Bruner and Goodman’s study, for example, were also subjected to a grasping task, there might be a significant difference in their success at grasping coins versus control discs that accords to the evi denced differences in size perception of one versus the other. Finally, conse quences for mental architecture. Strong modularity theorists like Fodor are emphatic on at least this point: The computational processing of perceptual modules is not supposed to be influenced by the organism’s goals and values (in addition to beliefs). And, plausibly, this is precisely what is occurring in
Copyright © 2021. Taylor & Francis Group. All rights reserved.
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
the relevant cases. The consequence then is that any strong form of percep tual modularity is threatened. Consider the second kind of case, involving some apparent diachronic change to cognitive-perceptual systems. Here one’s perceptual experience apparently changes, but over time, as one learns in some way (for example, one learns how to individuate fingerprints or what Waldo’s appearance is). An implicit assumption often made in the current literature is that these cases are not cases of cognitive penetration simply by virtue of their dia chronic nature. But this assumption is reasonably questioned, and applica tion of consequentialism makes this clear. The alternative interpretations most relevant here are the intra-perceptual interpretation and an attention-based interpretation. As discussed earlier, Fodor favours an intra-perceptual interpretation to explain cases of adap tation to inverting lenses. By extension, he might plausibly invoke a similar explanation for cases of expertise and putative-rich perceptual content. Here he can appeal to a story about perceptual learning where, on one gloss, repeated exposure to a stimulus kind improves one’s ability (in some way) to perceive instances of that kind. Thus, repeated exposure to pine trees results in an increased capacity for distinguishing pine trees from other kinds of conifer tree. This explanation can be given entirely, the thought goes, in terms of perceptual systems. Or one may think that what happens in the diachronic cases is simply that one learns how to better attend to relevant features and objects in the environment. How can the consequentialist approach help in these interpretative contexts? Fodor may be right about inverting-lens adaptation: It is plausible that this kind of adaptation could take place merely perceptually, so to speak. And for some other cases of learning, one may be able to adequately explain experiential changes just in terms of perceptual changes. Repeated expo sure to colour chips or flavour testers as one finds in a sommelier kit may suffice, without any category or concept learning, to enhance one’s percep tual capacities to recognize relevant differences in those stimuli tokens. If this intra-perceptual explanation can be generalized to the other diachronic cases, including the wide range of cases of perceptual expertise, then dis junctive consequentialism delivers a clear verdict: Lacking a cognitive influ ence on perception, none of these cases imply the relevant consequences for epistemology or behaviour or architectures of mind. Note, however, that this verdict has nothing to do with the diachronicity of the cases. And note that whether the intra-perceptual explanation can generalize to all cases of expertise is very much an open question.10
109
Copyright © 2021. Taylor & Francis Group. All rights reserved.
110
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
What of attention-based explanations of diachronic cases of visual recog nition and search? The assumption made here is that attention – typically understood as spatial attention – is a kind of intermediate step or gatekeeper between cognition and perception. Your beliefs and intentions (cognition) guide where you look (attention), and this in turn influences which stim uli affect sensory receptors, which, finally, influences what experiences you have (perception). This assumption will be challenged in the following chapter (as three distinct assumptions, in fact), which focuses centrally on how to think about attention and its role in thinking and perceiving. Here is a brief preview of that analysis, highlighting again how consequentialism can encourage progress in the debate. Much of recent philosophical and psychological work on attention concerns selective attention. Focusing on vision, attention can select features or objects in the visual array at the cost of others. Importantly, this can occur independent of spatial attention. That is, even holding fixed the attentional “spotlight”, different perceivers can attend to different features and/or objects within that space. And, crucially, this attentional selection can occur rapidly and in ways sensitive to the perceiver’s behavioural needs and goals. Therefore, it is tempting to think of these attentional mechanisms as part of the visual sys tem and, moreover, as sensitive to background cognition. The first thing to note is that by emphasizing selective attention, we already see a shift in how to think about certain types of cases once consequentialism is applied. The basic condition of disjunctive consequentialism is satisfied by this picture: There is a cognitive-perceptual relation. The question is whether the fact that selective attention plays a role in that relation precludes an important cog nitive effect on perception. And here the answer should ultimately be given (and will be given in Chapter 5) by appeal to the consequences. Here are some summary lessons on the two kinds of cases considered and how they are treated by consequentialism. First, as stated at the start of this discussion, neither kind of case has been given a conclusive verdict by disjunctive consequentialism. Instead, a clear debate-neutral strategy has been countenanced and additional points for analysis have been identified.11 For example, a complete analysis of the diachronic cases should involve some further discussion of perceptual learning and attention, with one eye on how the cases, thus interpreted, do or do not bear the relevant conse quences. Second, it should be emphasized that the consequentialist line does deliver the following clear result: Whatever else one says about either kind of case, it is not the distinguishing feature of the kind that drives the verdict
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
on cognitive penetration. In both the evaluative cases and the diachron ic-learning cases, interpretations are offered whereby the cases come out as cognitive penetration in spite of, respectively, the penetrating state being non-doxastic and the mental effect being one involving learning over time. This is progress: It reveals that certain kinds of cases are still relevant to the debate. And they are relevant to the debate not because they clearly satisfy some extant essential definition, but instead because they have prima facie bearing on the debate-neutral consequences. Consequentialism can claim this result as a virtue.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
4.5 Conclusion and summary One final way to motivate consequentialism is to see how it relates to exist ing definitions of cognitive penetration. Recall Pylyshyn’s (SC). It claims that cognitive penetration must be an inferentially or representationally coher ent relation between penetrating state and penetrated experience. Conse quentialism encourages us to reject this criterion, since it is not motivated by the consequences. The epistemic concerns certainly do not require this kind of coherence. Indeed, in the extreme cases, one can imagine scenarios where background cognitive states affect experience in wholly semantically incoherent ways and, accordingly, the epistemic worries would be all the more pressing. Put simply, at least some of the epistemic concern is that perception may be influenced in ways that are irrational, and so a rational ity constraint on the phenomenon is ill placed. It is even more difficult to see why a behaviourally important cognitive influence on perception would require such coherence. And the same is true for the mental architecture consequence. Focusing again on modularity, the question is whether per ceptual systems are modular in the sense that their information processing is not affected by the processing of independent, cognitive systems. And although it may be difficult for us to understand this kind of situation, perceptual processing could be affected in non-modular ways without this effect being one that is semantically coherent. So here again, SC is simply not motivated by the consequences of concern. Of course one may, as Pyly shyn apparently does, motivate (SC) by appeal to a broad computationalism about the mind. But this motivation is independent of, and insufficiently neutral with respect to, the concerns of existing debate in cognitive science. Consequentialism helps us see this and, hopefully, better enables progress on the topic.
111
112
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
By contrast, the second definition (CP) discussed in Section 4.2 is not entirely uninformed by consequences. As it was originally introduced (Stokes 2012), the definition was constructed to describe a phenomenon that is not well interpreted in any of the standard alternative ways given by critics. And one way to understand these interpretations as alternatives is that if they are apt for a given case, then that case implies none of the rele vant consequences. For example, a case best interpreted as a cognitive effect on judgement implies no relevant consequences for theories of perception and knowledge or for modularity of perceptual systems. So, although the prescription here is to abandon such essentialist definitions, some of them are more informed than others by consequentialist concerns. Put otherwise, some meet the consequentialist constraint; some don’t. The virtue of this consequentialist approach is simple: Once such an anal ysis is in hand, empirical studies can be devised and executed, with a prin cipled basis for interpreting the results. Testing for cognitive penetration becomes testing for phenomena that bear the relevant consequences for the epistemology or architecture of mind. This (re)identifies the phenomenon (or phenomena) of interest as one common to both sides of the debate. And it provides a unified metric for assessing the relevance of particular mental phenomena. The motivation for consequentialism, then, is twofold. It provides a clearer alternative to data interpretation – like the two kinds considered in Sections 4.2 and 4.4– than extant definitions. And second, it is debate-neutral, sensitive to the concerns of all relevant theorists. In sum, consequentialism dissolves the problem of cross-talk. The hope is that it may enable new progress on how we understand important cognitive-perceptual relations, on how thinking does or does not affect perceiving. Further reading Original definitions and partial definitions of the cognitive penetrability of perception can be found in Pylyshyn (1984, 1999) and Fodor (1983). More recently, see (Macpherson 2012; Siegel 2012; Stokes 2012, 2013, 2015; Wu 2013). Although Kuhn’s (1962) tends to garner more attention, the best orig inal source for discussion of theory-ladenness of perceptual observation in science is Hanson (1958, 1969). More recently, see (Churchland 1979; Brewer and Lambert 2001). For discussion of the general epistemology of cognitive penetration, see (Siegel 2012, 2013; Lyons 2011, 2016; Tucker 2014; Vance 2015; Silins 2016).
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 One nice example where this kind of meta-theoretical lens is especially rev elatory concerns the debate between internalist versus externalist theories of epistemic justification. Epistemic internalism says that an agent’s belief that P is justified just in case that agent could potentially access one’s rea son(s) for the belief and, perhaps stronger, identify why or how that reason grounds the belief that P. An epistemic externalist, by contrast, denies the internalist’s access condition and maintains that the agent’s belief that P is justified just in case the process by which the belief was formed is one that typically leads to true beliefs. each view, through increasingly sophisticated definitions, arguments, and examples, attempts to show how their analysis of justification delivers intuitive verdicts on possible cases of belief forma tion while criticizing the opposite view as delivering counterintuitive verdicts on those same, and other, cases. All of this is premised on the assumption that there is a single target phenomenon – epistemic justification – that is the subject of all sides of the debate. kent Bach (1985) challenges this assumption in illuminating ways. He distinguishes a justified believer from a justified belief. A believer is justified by virtue of some epistemically meritori ous action, say, considering her evidence or interrogating how that evidence supports belief(s) formed. A belief is justified by virtue of some fact about the belief itself, say, its being formed in a way that typically results in accu racy. A moment’s reflection reveals that a believer could be justified while holding unjustified beliefs, and vice versa. Bach’s application is to suggest that internalists are primarily concerned with justified believers, and exter nalists, with justified belief. “I think that our distinction captures what is right about both conceptions of justification. They are conceptions of two different things! Taking rationality and responsibility as the marks of justi fication, internalism can maintain that whatever makes a person justified in holding a belief must be available to him. And externalism can maintain that being justified is whatever property a true belief must possess to qualify as knowledge” (Bach 1985: 252). Predictably, this hasn’t ended the debate, but it most certainly introduced new angles for reflection and progress and by forcing reconsideration of the grounds for the debate and whether the parties involved were always arguing about the same thing(s). 2 In like manner, “cognitive penetration” is unnecessary and, indeed, is an unfortunate choice of terms. In this chapter and those that follow, effort
113
114
TOwARDS A CONSeQueNTIALIST uNDeRSTANDING
3 4
5 6 7
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8
9
10 11
will be made to use “cognitive penetration” and its cognates only when explicitly engaging literature that uses these terms. elsewhere, alterna tive terms (e.g. “top-down effects on perception”) will be favoured, and, as is the spirt of the present section of the present chapter, substantive descriptions of the phenomena of interest to various parties in the debate will be favoured over labels like “cognitive penetration”. At bottom, it is the epistemological and scientific-theoretical importance of possible cognitive-perceptual relations that matters, not that there is a singular phenomenon that satisfies the necessary and sufficient conditions for “cognitive penetration” as such. Note further how this analysis bears the mark of the default position assumption discussed in the Chapter 2. Issues of definition to one side, both of these lines of reasoning and the assumptions they rest on are challenged in later chapters. Chapter 5 con cerns attention and the varieties it can take. Chapter 6 begins discussion of perceptual expertise, which concerns how cognition can be involved in perceptual skills and learning. This example is modelled on Hanson 1959: 5–8. For more on theory-ladenness, see, among others, Churchland 1979, 1988; Fodor 1984, 1988; Brewer and Lambert 2001; and Raftopoulos 2001, 2006. The activity of perception was, in one sense, discussed in Section 1.4. See also note 13 in that chapter for a list of “active” theories of perception in both psychology and philosophy. For an extension and development of ideas very much in the spirit of Gibson’s ecological approach, see the work of Dennis Proffitt (2006, 2020). There is precedent for appeal to the epistemic consequences in Fodor’s own work to evaluate certain arguments and empirical evidence for cogni tive penetration (see 1988: 189). And more recently, see Raftopoulos 2001, 2006. Considerations of theory-ladenness, including the possibility of non doxastic influence on perception, is discussed in greater detail in Chap ters 7 and 8. A question to be taken up in Chapters 6 to 8. Again, those points – the role of attention and how to theorize cases of per ceptual learning and expertise – are the subjects of the chapters to come.
5
ATTENTION AND
COGNITIVE INFLUENCE ON
PERCEPTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5.1 Rethinking the role of attention in cognitive effects on perception 5.2 Four questions about attention-involving, cognitive-perceptual phenomena 5.3 Two arguments for attention-involving, cognitive penetration of perception 5.4 Summary: moving beyond cognitive penetration One doesn’t get the duck–rabbit (or the Necker Cube) to flip by “changing one’s assumptions”; one does it by (for example), changing one’s fixation point. Believing that it’s a duck doesn’t help you see it as one; wanting to see it as a duck doesn’t help much either. But knowing where to fixate can help. Fixate there and then the flipping is automatic. [O]ne squints to make things look sharper; one cups one’s hand behind one’s ear to make them sound louder, etc. It doesn’t begin to follow that auditory and visual acuity are cognitively penetrable. (Fodor 1988: 190–191) Apart from changes brought about by shifting attention, however, there is no evidence that voluntarily “changing one’s assumptions about the object” has any direct effect on how one perceives the figure”. (Pylyshyn 1999: 358; emphasis added)
116
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Hints and instructions rarely have an effect, but when they do it is invariably by influencing the allocation of focal attention . . . the evidence pointed to the improvement being due to learning where to direct attention – in some cases aided by better domain-specific knowledge that helps anticipate where the essential information will occur. (Pylyshyn 1999: 364; emphasis added)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
[S]hifting patterns of attention can change what we see. Selective attention is obviously closely linked to perception – often serving as a gateway to conscious awareness in the first place, such that we may completely fail to see what we do not attend to. . . . Moreover, attention – which is often likened to a “spot light” or “zoom lens” – can sometimes literally highlight or enhance attended objects, making them appear (relative to unattended objects) clearer . . . and more finely detailed. . . . Attentional phenomena relate to top-down effects simply because attention is at least partly under intentional control – insofar as we can often choose to pay attention to one object, event, feature, or region rather than another. (Firestone and Scholl 2016: 13)
The central question of this chapter is how may attention count against, versus be neutral with respect to (if not count for), explaining a phenome non as an important cognitive effect on perception. The quotations provided represent a common rejoinder to putative cases of cognitive penetration or to interesting candidate cases of cognitive effects on perception. For instance, Fodor is replying to some examples that Paul Churchland (1988) invokes as instances of the cognitive penetration of perception. Churchland claims that one changes one’s assumptions about ambiguous figures depending on how one intends to see the image, and this results in a change in one’s experience of the very same image. Fodor’s reply is to say that what one does is instead to voluntarily shift one’s attentional focus by fixating on one part of the fig ure at the cost of others, and this intermediate step between one’s cognitive states and one’s visual uptake results in a difference in experience. He, and others such as Pylyshyn, generalize from this case to other cases that appar ently involve attention, and all as a way of resisting explanations that posit significant top-down effects on perception. There are three distinct but related assumptions being made here, all of them important to how far the rejoinder generalizes. The first assumption says that when attention is involved in relevant cognitive-perceptual relations, it is spatial attention. Call this the spotlight of attention assumption. The second assump tion says that when attention is involved in relevant cognitive-perceptual relations, it involves some bodily movement or action done by the perceiver. Call this the attention as act assumption. The third assumption is that, generally,
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
attention is an intermediary process between cognition and perception. Call this the attention as gatekeeper assumption. Not all assumptions are operative in all sceptical rejoinders to possible attention-involving cases of cognitive pene tration, but at least one of them is typically operative in most, if not all, of those rejoinders. And in some cases, more than one assumption is explicit. In the second quotation from Pylyshyn, he is making at least the first and second assumptions. And all three assumptions are explicit in the Firestone and Scholl quotations. The general aim of this chapter is to challenge all of these assumptions and keep attention-involving cases on the table for debate. The more ambitious aim is to make a positive case for attention-involving instances of the cogni tive penetration of visual perception. There are two arguments to motivate either the weaker or more ambitious aim. The first argument involves reject ing all three assumptions, including the third gatekeeper assumption, but proceeds independent of consequentialism. It appeals to a more standard, and probably overly stringent, characterization (or definition) of cognitive penetration whereby the cognitive effect on perception must be a direct one. If the gatekeeper assumption is rejected, and some attentional mechanisms are instead part of the visual system, then cognitive effects on those attentional mechanisms that result in differences in experience are bona fide instances of relevant cognitive effects on perception. The second argument involves rejecting both the spotlight and the act assumption, while remaining neutral on the gatekeeper assumption, and then applying the consequentialism of the previous chapter.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5.1 Rethinking the role of attention in cognitive effects on perception Consider the duck–rabbit image (see Figure 5.1). Suppose one alleges that one’s ability to see the image sometimes as a duck and sometimes as a rabbit is a case of cognitive penetration of perception (Churchland 1988). Since one knows that the image is ambiguous and that it can be seen either as a duck or as a rabbit, one can apply this knowledge (depending upon which animal likeness one wants to see) and accordingly see it in one of two importantly different ways. The sceptic can grant that here cognition is affecting percep tion, but only through shifts in spatial attention – that is, changing the focal space to which one, in this case, visually attends (Fodor 1983, 1984, 1985, 1988; Pylyshyn 1999: 358, 364). Elsewhere this has been called, as a rejoin der to alleged cases of cognitive penetration of perception, the attention-shift interpretation (Macpherson 2012; Stokes 2012, 2014).
117
118
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Figure 5.1 The duck–rabbit image
Appeal to consequences is instructive here. Fodor’s reasoning might con cern either of two possible features of attention shifting. First, shifts in attention are often actions in the philosopher’s sense. Fodor describes the duck–rabbit case in precisely this way: Cognitive states cause action, and those actions affect what information reaches sensory receptors, which in turn affects perception. This bears no important consequences for the epis temology or behavioural role or alleged modularity of perceptual systems or experience. Second, agency to one side, shifts in attention involve a change in the visual spatial array of attention. Here cognition changes the visual spotlight or focus and accordingly changes the information available for perception. Again, Fodor can charge that this implies no consequence for the epistemic or behavioural roles of perception, nor any obvious implication for the alleged modularity of perceptual systems. Instead, it is just a change in the input to those systems, by contrast to an encapsu lation-violating effect on the computational processing of those systems. Bringing these two points together (which correspond to the attention as act and the spotlight of attention assumptions) reveals that attention is being treated as some kind of process or stage that stands in between cognition and perception: Given certain antecedent cognitive states, one
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
actively attends by shifting one’s focus and thereby changes the input to a separate mental system, namely, vision (this corresponds to the attention as gatekeeper assumption). Grant for the sake of argument that if the attention-shift interpretation best explains a case then the sceptic is right, the case in question is not cog nitive penetration. But what is the scope of the attention-shift interpretation? Sceptics like Fodor and Pylyshyn assume that a shift in spatial attention is the only relevant way that attention might be involved in cognitive-perceptual phenomena. This assumption is mistaken. Put another way, insofar as the three assumptions mentioned are operative in the attention-shift interpreta tion, we can ask: How safe is each assumption? Each of them is contentious and, it will be argued later, at least two of them are false. The standard attention-shift interpretation countenances a scenario where a perceiver has certain cognitive states that cause a shift in attention and that shift clearly changes what the perceiver sees. This scenario is captured by a simple, mental causal schema:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(a) Cognitive state → Attention shift → Perceptual experience
For simplicity, interpret the arrows as causal arrows such that the attention-shift counterfactually depends on the cognitive state and the per ceptual experience counterfactually depends on the attention shift. One way to understand attention shifting is as an intentional bodily action. Second, and nonexclusively, attention shifting can be understood as a shift in spatial attention: a change in the spatial visual field attended, to the focal spotlight. Further notice that the schema suggests that the attentional change is neither part of cognition nor part of perception, but something in between. Let’s grant that the sceptic is correct to claim that if a case is best explained in terms of (a), then it is not a case of cognitive penetration.1
5.2 Four questions about attention-involving, cognitiveperceptual phenomena Are all attention-involving, cognitive-perceptual phenomena best described in terms of schema (a)? To motivate a negative answer to this question, affirmative answers for each of the following questions must be given: Q1: Can attention (or attentional mechanisms) operate in a non-agential way?
119
120
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Q2: Can attention change without changes in spatial attention? Q3: Can these non-spatial attentional mechanisms be influenced or partly driven by cognitive states?
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Q4: Do these non-spatial attentional mechanisms influence conscious perceptual experience?
In response to Q1, it should be obvious that attention can be guided without one’s trying or intending to do any guiding. Some objects or events grab attention – think about a snake popping out of the brush, or one’s name mentioned on the opposite side of the room, or a sudden pain in one’s foot. In these cases, attention does the grabbing, and often enough it happens so rapidly that you certainly wouldn’t describe it as something you did. And this comports well with the most dominant, general categories of scientific theo rizing about attention. Bottleneck theories (Broadbent 1958), feature-binding theories (Treisman and Gelade 1980), and competition-based theories (Desi mone and Duncan 1995) are broadly unified by the following thread: Atten tional mechanisms serve to select perceptible features, objects, and events, and often automatically. This suffices to answer “yes” to Q1. We can then ask about the forms that attention, understood at least in part as selection, can take. Q2 asks, can attention change without changes in spatial attention? Atten tion is used, intentionally and non-intentionally, as a way to search, rec ognize, and inspect one’s environment. One clear way this occurs involves focused attention to some region of space in the visual field, for example, the region where one might expect to find the object of one’s visual search. This is one standard account of spatial attention – the “spotlight of atten tion” – and it is the account assumed by the Fodorian sceptic. But it should be unsurprising that attentional selection can instead involve enhanced sen sitivity to features of one’s environment (like colour or shape) or to whole, bound objects in one’s environment. And this sensitivity can be enhanced in behaviourally relevant ways. Over the last few decades, there has been sub stantial empirical research on both of these last two mechanisms of selective attention: on object-based attention (OBA) and feature-based attention (FBA). The earliest work on OBA comes from Neisser and Becklen (1975). Exper imenters asked subjects to view a screen with two distinct, spatially over lapping videos. In one version of the study, one video depicted two sets of hands playing a hand-slapping game, and the second video depicted a group of people passing around a basketball. When subjects were tasked in
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
a way specific to one video – for example, to count the number of basket ball passes – they would miss events taking place in the other video. The videos in this experiment are spatially superimposed, and so if attention was operating just on the basis of a spatial position (or “spotlight”), then subjects should not be blind to changes in those positions. But in fact, robust inattentional blindness occurs. This is best explained by appeal to an attentional mechanism that responds to objects and does so independently of spatial attention (see Scholl 2001). In visual search, attentional selection is sometimes guided by relevant fea tures of the environment. Suppose you are searching a crowded baggage claim area for a missing piece of bright pink luggage. FBA operates in such a way that when scanning this visual array, pinkish features of the array are favoured for perceptual representation at the neglect of other features. Pink ish features enjoy a brief pop-out effect to aid in search. Importantly, like OBA, FBA can operate independently of spatial attention: FBA will highlight relevant features both “within and outside the current spotlight of attention” (Treue and Martinez-Trujillo 2007: 175; see also Scolari et al. 2014; White and Carrasco 2011). So, in answer to Q2: Both OBA and FBA can function, selecting objects or features, without any change in spatial attention.2 Note further that these selection mechanisms can operate in covert, non-deliberate ways. This research can be traced back to some earlier research on attention that marked a crucial change in how psychologists understood attentional phenomena. Q3 then asks if these mechanisms can be modulated by cognition and, Q4, whether these mechanisms function in such a way to influence conscious perceptual experience. Consider again familiar visual search tasks like Where’s Waldo? puzzles. These are overwhelmingly busy visual scenes where one is tasked to find Waldo, who is typified by matching red and white striped sweater and cap and round spectacles. This cognitive-perceptual situation is plausibly interpreted as follows. The practiced Where’s Waldo? puzzler has a number of relevant cog nitive states: a belief that there is an object in the scene with Waldo’s typical features, an expectation to find items with those features, and an intention to find those typical features. FBA then functions in such a way that behaviour ally relevant features (red and white striped items) are rapidly selected at the expense of non-relevant features. This suggests a “yes” to Q3. This cognitively enhanced activity in FBA then results in or amounts to enhanced perceptual representation – pop-out – of some features at the neglect of others. This is an effect on conscious perceptual experience (a “yes” to Q4).
121
122
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
All of this can be usefully described in terms of a competition model of attention. Desimone and Duncan (1995), two of the pioneers of this approach, use examples of simple visual search tasks over arrays like those included in Figure 5.2 to identify two explananda for theories of attention. First, they identify a limit on how much information can be processed by the agent or visual system. Consider parts a) and b) of Figure 5.2. Before a brief presentation of each array (given separately), a subject is instructed that they are to report any black letters. Note that a) contains three black letters; b) contains one black letter. In this scenario, the probability that the subject reports the letter N post-stimulus is much lower when presented with a) than when presented with b). This statistical finding is well explained by positing a limit on the information that can be picked up, that is, there are a limited amount of attentional resources to be allocated. The result is one much like we find in our busy day-to-day environments: When multiple rel evant targets are present, only some can be attended at any one time. Second, Desimone and Duncan identify the selectivity of attentional allocation. Now consider a) and c) and suppose the task is the same as before. Note that both images have the very same black letters, while a) has three (nontarget) white letters and c) has one (nontarget) white letter. In this scenario, differences in accuracy in reporting targets (black letters) are insignificant: Subjects do just as well at identifying target letters in image a) even though it contains more non-target letters than c). This result is explained by positing selection or fil tering as a feature of attentional allocation; irrelevant information is filtered out and so not behaviourally disruptive. Desimone and Duncan’s explanation of these two features – their solution to what they call the “problem of visual attention” – describes attention as involving (or as being) a biased competition. Chris Mole (2011) nicely char acterizes the relevant notion by distinguishing two kinds of competition:
Figure 5.2 Examples of stimuli for visual search task in studies on competition in visual attention Source: Reprinted with permission from Desimone and Duncan 1995. Copyright Annual Review of Neuroscience.
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
races and struggles. In the most familiar kind of race, two or more individuals simultaneously attempt to get from one point to another, and the winner is the one who does this the fastest. What’s common to this and other races is that individuals simultaneously compete to perform some activity best along some dimension of value, but they do it independently of and without inter fering with one another. A struggle is importantly different: It involves the “active suppression of other competitors” who are all attempting to perform the same activity. Attention takes this second form of competition. If atten tion is limited and selective, then it involves a competition to deploy what’s available (given those limits) and thereby favour some available information over other available information (thus making selections). What is selected is the winner and, crucially, there are biases that determine that winner. There are two additional features of Desimone and Duncan’s biased com petition model to be highlighted. First, competition occurs not just at the level of behaviour but occurs at identifiable stages of neural processing. In the simplest (visual) case, a receptive field is the region of sensory space to which a neural cell is responsive; that is, that cell will respond only to stimuli within that space. But some cells, including those in visual cortical areas, have more fine-grained receptive fields, responsive not just to a region of space, but only to stimuli with particular features within that space, say edges with a par ticular orientation or colour. Referring to single neural cell recording studies on visual cortical areas (areas V4 and the inferior temporal gyrus), Desimone and Duncan identify competition within, as it were, these receptive fields. If one were to add ever more independent objects to a V4 or IT recep tive field, the information available about any one of them would certainly decrease. If, for example, a color-sensitive IT neuron were to integrate wavelength over its large receptive field, one might not be able to tell from that cell alone if a given level of response was due to, say, one red object or two yellow ones or three green ones at different locations in the field. (1995: 197)
Competition thus extends all the way down to the receptive fields of indi vidual neural cells. Second, Desimone and Duncan show that this competition, from the observable behavioural level all the way down to the neural level, is influ enced in both bottom-up and top-down ways. Bottom-up, stimulus-driven influences on visual attention are familiar and uncontroversial. How the objects and features of the environment are structured partly determines
123
124
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
what catches our attention. For instance, a novel object or sudden motion is more likely to be attended and at the cost of other stimuli; these are some of the bottom-up biases that determine, on the competition model, what wins the competition for limited resources. More interestingly, Desimone, Duncan, and other collaborators found substantial evidence for top-down influences or biases on the competition for resources and thus, on this model, on what gets attended. For example, in a study by Moran and Desimone (1985), mon keys were primed before a shape discrimination task with a fixation point, where that fixation would instruct the monkey where the target shape would be. It is not surprising that this cue affected a behavioural bias such that the subjects ignored nearby distractor objects. What is surprising is that this cue influenced the receptive fields of individual neural cells in visual cortex.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The target location for a given run was indicated to the monkey by special instruction trials at the start of that run, i.e. the spatial bias was purely top down and presumably required spatial working memory.when target and distractor were both within the receptive field of the recorded cell, the neuronal response was determined primarily by the target; responses to the distractor were greatly attenuated. The cells responded as though their receptive fields had shrunk around the target. (Desimone and Duncan 1995: 203)
The crucial lessons here are that, on the biased competition model, atten tional selection is sensitive to the behavioural needs and task demands of the subject, and this effect appears to be one that occurs rapidly and at the level of neural processing. What wins the competition for attention is not deter mined just by features of the environment, but by features of the agent and what they know, expect, and plan. This view applies readily to visual search cases like Where’s Waldo? puz zles. In this case, cognitive states like belief and intention affect conscious experience by modulating non-agential, spatially independent, selective attention. These cognitive states bias attentional selection. Contrast this with the naïve–Where’s Waldo? viewer, tasked to simply inspect the same visual scene. This subject will not (immediately) form the relevant cogni tive states, will not enjoy FBA-driven search, and, finally, will not enjoy the resulting, enhanced perceptual representation. On the competition model, the naïve subject enjoys no analogous top-down bias on attentional selec tion. Whatever one says about this anecdotal case, what’s crucial to note is
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
that the sceptic cannot invoke the attention-shift interpretation in response. A case of this kind (involving the non-naive viewer) does not involve shift ing (intentionally or not) attention from one space to another.3 We can already see, then, how some of the assumptions made by modularists and other sceptics appear problematic. There the idea was that when attention plays a role in a cognitive effect on perception, it involves an action and/or a shift in spatial attention. But the positive answers just given to Q1 to Q4 reveal that, at least in principle, non-agential selection mechanisms like FBA can both be modulated by perception and have an effect on perceptual rep resentation. All of this suggests that there are possible cases, like the visual search cases just discussed, that are plausibly characterized by an alternative causal schema:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(b) Cognitive state → Non-agential selective attention → Perceptual experience
Now consider schema (b) in light of three recent, related experimental studies. Moores et al. (2003) found that object representations and associated cogni tive representations influence attentional selection in visual search tasks. The basic experimental structure went as follows. Subjects were placed before a computer screen in a dimly lit room, viewing the screen at a distance of roughly 57 cm. Subjects were cued with an instruction for visual search; for example, the target item might be MOTORBIKE. Subjects would then see a fixation cross, followed by time variable presentation (sometimes as brief as 100 ms) of an array of four images. Sometimes the target item was present and in other trials absent. Additionally, in some experimental trials, amongst the visual array was a semantically related but visually non-resembling image (a HELMET image when the target was MOTORBIKE). Subjects would then receive a masking screen or a fixation cross and then might be tasked to report whether the target image was present or absent and/or be asked to make a forced choice concerning which of two items was recently seen. (For illustrations of trial structures, see Moores et al. 2003: 183, 185.) Images semantically related to the search target were better recalled and identi fied, and these images sometimes hindered target identification and reduced recall of unrelated distractor images. These researchers hypothesize that selective attention is being captured in a way independent of the search task but that depends upon information about relations of meaning, stored in long-term memory.
125
Copyright © 2021. Taylor & Francis Group. All rights reserved.
126
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
The same researchers found that semantically related images modulated saccadic eye movement.4 Saccades were more typically directed at images semantically related with the explicit search target to the near exclusion of unrelated distractors. And in trials where a related image was present in the visual display, the probability that the first saccade tracks the target is significantly reduced, and by contrast to control trials with no semantically related image where initial saccades are dominantly directed at the search target image (the mean onset latency of the first saccade was 306 ms). This is rapid, non-deliberate selection of attention that is sensitive to semantic relationships. (See also Belke et al. 2008; Gazzaley and Nobre 2012.) In a second set of studies (Meyer et al. 2007), subjects again performed a visual search over a time-limited visual array. The relevant difference is that in some trials, included in the visual array was an image of an object with an homophonous name. For example, if the target item was a BOW (weapon), on some trials an image of a BOW (ribbon) was present. The related images here are neither visually resembling nor semantically related (the respective terms for a bow weapon and a bow of ribbon stand in no semantic relations). In line with the Moores et al. results, Meyer et al. found that when present, images with homophonous names (shared with the target) more quickly receive attention, are better recalled and identified, and tend to slow overall response time. Here again eye tracking studies corroborate: The first saccade after display onset (onset latency averaging between 211 and 220 msec) was more likely to track the related (homophonous) image than unrelated distractors, and the target was less likely to remain fixated if there was a related image present in the visual array. Finally, these studies have been corroborated by neurological studies using electroencephalographic (EEG) recordings. In these studies, Telling et al. (2010) used measurements of N2pc amplitudes. The N2pc is an event-related potential (ERP) and is a standard measure for fairly rapid activity in response to visual stimuli, typically around or before 200 ms post-stimulus. In brief, what these researchers found was that semantically related (but visually non-resembling) distractor items affected the magnitude and onset latency of this EEG component. This cortical activity is fast, and so it is improba ble that it is attentional selection consistently done by, in any intentional way, the perceiver. Here it is worth stressing some relevant features of N2pc research. First, using magnetoencephalography neuroimaging techniques, researchers identify the N2pc component as largely correlating with activity in extra-striate cortex, including visual area V4 (Hopf et al. 2000). Activity in the latter visual cortical area has long been taken to be modulated by (or
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
correlated with) selective attention (Moran and Desimone 1985). Further more, recent research suggests that N2pc amplitudes are not correlated with shifts in spatial attention (Kiss et al. 2007). Therefore, semantic information stored in long-term memory appears to be modulating rapid neural process ing that occurs, by our best neuroscientific models, in the visual system. Now recall Q3 asks: Can these non-spatial (selective) attentional mecha nisms be influenced or partly driven by cognitive states? What all of these experimental studies converge on – from behavioural to eye tracking to neu rological studies – is that non-targeted but linguistically related images have a significant effect on visual search. The difference made by semantically and homophonously related images requires explanation. The methodology used, including the use of the N2pc EEG component, suggests that the differ ence is not well explained by shifts in spatial attention. Since representation of linguistic relations is not merely (or even) perceptual, even if “low-level” or “fast” or even “encapsulated”, the difference is not explained as a merely perceptual (that is, non-cognitive) one. And finally, the eye tracking and EEG results suggest that the difference is not explained as a mere post-perceptual difference in judgment or memory. Instead the data suggest important effects both at the level of saccadic eye movement and in extra-striate cortical activ ity, both of them part of visual processing if anything is. As Meyer et al. describe it, the “data suggest that there is sufficiently rapid access to conceptual information from distractors for this information to influence the first fixations made during search” (2007: 715). Recall that these first saccades averaged, in these studies and the Moores et al. (2003) studies, an onset latency between 200 and 306 ms. It is not just the fact that the effect is rapid that is important, but the fact that linguistically related distractors are both task-irrelevant and visually non-resembling to the rele vant target images. In this regard, saccadic movements to distractors are not deliberate, but instead involve an “automatic” spread of attention (Meyer et al. 2007: 710).5 This latter explanation comports well with an explanation in terms of selective attention, with an emphasis on feature-based attention. Typically, researchers of FBA focus on low-level features like colour or motion and how selection of these features may be modulated in behaviourally relevant ways. However, the data in these studies suggest that linguistic features (for example, that the image stands in a semantic relation with the search target) are operative, such that features (or objects) associated with those linguistic features are somehow enhanced. When searching for a MOTORBIKE image, features of semantically related images (like those of a HELMET image) are
127
Copyright © 2021. Taylor & Francis Group. All rights reserved.
128
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
deemed relevant by the cognitive system and, accordingly, are selected for attention at the cost of unrelated distractors. In brief, this would be working memory somehow modulating feature selection in favour not of relations of visual similarity to the target, but in favour of linguistic relations. This is an extension of the standard model of feature-based attention, but it gains plausibility from the data for the simple reason that it best explains that data. Even in such a simple array, and time-limited exposure, there is com petition for selection. And one way that this selection can be resolved is, implicitly, by behaviourally relevant biases. There is, of course, a sense in which a task-irrelevant image of a HELMET (when the instructed target for visual search is MOTORBIKE) is not behaviourally useful. (There is also a sense in which HELMET is behaviourally relevant in such a trial by contrast to other competing distractors, say a LOCK and a KETTLE.) Indeed, as the results show, such distractors sometimes hinder performance. But the results nonetheless cannot be denied: Semantically related items affect rapid visual attentional deployment, and this shows up behaviourally, at the level of rapid (non-agential) eye movement, and at the level of neural activity. These effects require explanation, and the most plausible explanation is that information from working memory about such semantic relations is biasing, and there fore determining the winners of, the competition for attentional resources. Bearing all of this in mind, we turn finally to Q4. Q4 asks: Do these non-spatial (selective) attentional mechanisms influ ence conscious perceptual experience? As forms of selective attention, FBA and OBA can affect perceptual salience without the perceiver performing any action to bring about the changes and without any change to or manip ulation of features of the environment. These perceptual changes can take a number of forms. Most simply, a feature or object that pops out is expe rienced sometimes at the exclusion of, or longer than, other stimuli in the array.6 Selective attention also effects changes in perceptual organization. Consider the Rubin goblet (see Figure 5.3). Even holding the stimulus fixed in your spatial attentional spotlight, you experience a phenomenal difference between seeing the image as a goblet and seeing the image as two faces in a face-off. This is a perceptual change that you can deliberately bring about. But it is entirely plausible that selective attentional mechanisms, when oper ating non-agentially, bring about similar changes: changes in what features or objects are figure(s) versus ground and what features or objects are focally versus peripherally represented. Recent theorists have suggested that attention serves to structure or organize conscious experience (Watzl 2011, 2017; Wu 2014: Ch. 4). This
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Figure 5.3 The Rubin goblet
observation traces back to gestalt psychology and continental phenomenol ogy. It is clearly articulated by Sartre. when I enter this cafe to search for Pierre, there is formed a synthetic organization of all the objects in the café, on the ground of which Pierre is given as about to appear. . . . [I]f I should finally discover Pierre, my intu ition would be filled by a solid element, I should be suddenly arrested by his face and the whole cafe would organize itself around him as a discrete presence. (Sartre 1943: 9–10)
Should Sartre find, attentionally select, Pierre’s face, it is clear that Pierre’s face would take up the central, focal position in Sartre’s conscious
129
130
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
experience. More interestingly, it is around this one object or feature – Pierre’s face – that Sartre’s conscious visual experience is then organized. So, it is crucial to note not just that there is a centre or a priority space of attention, but that there are various other features of the perceptual array that are organized by relation to that centre. These components are no less a part of the phenomenology of experience. And the selec tions that attention makes determine how all of these components are structured. The explanation in reply to Q3 regarding the discussed experimental studies extends to a reply to Q4 as follows. The semantically and lin guistically related items receive attention at the exclusion of unrelated distractors or for periods longer than those distractors (as suggested by the eye tracking studies). This is the simplest perceptual effect, showing up in the relevant subjects’ performance on recall tasks. It is also plausi ble that these attentional selections influence the structure of conscious experience in the ways just described. For example, the targeted MOTOR BIKE image might take the centre of attention, with the HELMET image “pulling” dominantly at the periphery of experience. And this experience enjoys a distinctive phenomenology: It feels different to have something peripheral pulling attention from your focus vs. a case where attention is perfectly focalized. These are attention-modulated perceptual differences that causally depend upon the cognitive, linguistic representations stored in long-term memory.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5.3 Two arguments for attention-involving, cognitive penetration of perception A brief reminder of the dialectic is useful here. The central question is: How may attention count against versus be neutral with respect to (if not count for) explaining a phenomenon as cognitive penetration? This question can be approached by asking whether all attention-involving, cognitive-perceptual phenomena are best explained in terms of the attention-shift interpreta tion [causal schema (a)]. To answer “no” to this last question, an affirmative answer must be motivated for questions Q1 to Q4. This has now been done and by appeal to both anecdotal and experimental evidence. This analysis grounds the second causal schema: (b) Cognitive state → Non-agential selective attention → Perceptual experience
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
The very same evidence and considerations that motivated affirmative answers to Q1 to Q4 suggest that we have compelling evidence for actual psychological phenomena that are best described in terms of schema (b). It should be clear by this stage of the analysis that a phenomenon that fits (b) would not, for inclusion of a central role of attention, count against that phenomenon being cognitive penetration. The final step in the analysis is to argue that any phenomenon that is best described in terms of (b) is best explained as cognitive penetration.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5.3.1 An argument from directness Cognitive penetration is often theorized as involving a direct cognitive effect on perception. Pylyshyn (1999) and Fodor (1988) at least implicitly assume this in their claim that cognitive-perceptual effects involving active shifts in spatial attention do not count as cognitive penetration. Cases involving spatial attention shifts fail to satisfy what recent theorists have called a vehicle criterion, which requires of genuine cognitive penetration that perceptual processing must draw information or representations directly from cognitive systems (Zeimbekis and Raftopoulos 2015: 27). More weakly, some have allowed for indirect connections (Macpherson 2012), while also maintaining that the con nection must be internal and mental (Stokes 2013). Attention-shift cases best described in terms of causal schema (a) do not count as cognitive penetration by failure of all of these criteria. But on one plausible analysis, this is not true of phenomena that are best described in terms of causal schema (b). First note that a phenomenon that fits (b) does not involve (any relevant) intermediate action, nor does it involve a shift in spatial attention. If this is correct, then both the attention as act and attention as spotlight assumptions may be rejected. Recall further that neural correlates for selective attention are found in visual areas such as V4 of the extra-striate cortex. Moreover, an observable signature of selective attention mechanisms like FBA and OBA is saccadic eye movement patterns. Many researchers would count these physical activities – V4 activity and involuntary saccadic eye movement pat terns – as part of visual processing and therefore count (some of) selective attention as part of, rather than antecedent to, visual processing. Finally, there is principled reason to carve up the space in this same way. What these selection mechanisms do is change the salience of a perceiver’s immediate environment, in effect changing what the eyes pick up. And this is part of what perception itself does. If this is correct, the gatekeeper assumption should also be rejected.7
131
132
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Here are two additional (and separable) analyses that provide reason to reject the gatekeeper assumption. Chris Mole (2015) argues against the claim that attention serves only as an intermediary stage – a partition – between cognition and perception, at most changing the input to separable perceptual processes. Mole’s emphasis is on the distinction between overt and covert attention. In cases where cognition influences overt attentional action, say by causing one to turn one’s head in the direction of a stimulus, this changes one’s orientation with respect to the world and, predictably, changes the experiences one has. Cases such as these, the sceptic is right, are not controversial and do not involve cognitive penetration of perception. However, to treat all cases of attention in this fashion is, Mole urges, a mis take. Attention can also take a covert form, where attentional selection takes place without an overt bodily movement. There is ample evidence for this claim, but Mole cites several studies performed by Kravitz and Behrmann (2011) that suggest that colour, shape, and object perception, as well as vis ually detected orthographic categories, are not downstream from attention, but instead are richly intertwined with covert attentional selection. Impor tantly, the latter is apparently sensitive to learnt cognitive information. Mole buttresses this analysis with an “integrated competition” model of attention, where competition for mental resources takes place across percep tual, cognitive, and motor processing, with both bottom-up and top-down biases throughout. Mole writes: There is no separate system that uses cognitive information to allocate attention to certain targets, and that operates before any perceptual pro cessing takes place. The system that allocates attention is, instead, the whole of the perceptual, cognitive, and motor system. Integrated compe tition takes place everywhere. The integrated competition theory therefore has no room for the idea that attention stands as a partition between cognition and perception. It entails that some cases of attention-mediated cognitive influence are cases in which the influence of “conceptual recog nition” is, as kuhn said, “built into the nature of the perceptual process itself”. (Mole 2015: 233)
Jesse Prinz (2011) argues that attention is both necessary and sufficient for conscious perceptual experience. Sensory consciousness arises, according to Prinz, at the intermediate level of representation, which lies in between early sen sory processing and later representations that abstract away from egocentric
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
details. The intermediate level “represents whole objects with rich surface details, located in depth and presented from a particular point of view” (2011: 174). Representations at this level of processing can occur with or without conscious awareness. Attention, understood functionally, is the pro cess that makes these representations available to working memory, such that those representations can be reasoned about, examined, acted upon, and so on. And when this happens, those perceptual representations are conscious. Prinz then applies this to a variety of phenomena that we pre-theoretically describe in terms of attention: pop-out, search, monitoring, tracking, vigi lance. Of the first, he describes his thesis as follows (and, like Mole, in terms of competition). “Pop-out occurs when the representation of one stimulus competes with the representations of surrounding stimuli and wins. When it wins, it is processed in a way that makes it available to working memory. This process is the psychological correlate of attention”. Prinz’s view suggests that attention, understood as the process whereby some, but not all, per ceptual representations are made available to working memory, is part of the processing or mechanism of perceptual systems. It is not something outside of whatever system (or systems) produce those intermediate perceptual rep resentations, but instead some component of that system. Again, attention is not a gatekeeper, but part of, for example, the visual system. If any of these analyses is correct, then a phenomenon that is best described in terms of (b) involves a direct cognitive (and certainly an inter nal mental) effect on perception. Since the intermediary attentional selection mechanisms as schematized in (b) are part of perceptual processing (and, as argued, have an influence on perceptual experience), cognitive modulation of those mechanisms is cognitive modulation of perception. So, a phenom enon that fits (b) is cognitive penetration. One rejoinder says that the phenomena discussed here are or involve instances of spatial attention. For example, behaviourally sensitive changes in feature selection could enhance sensitivity just to those spaces where the features are present, and perhaps one would want to call this “spatial atten tion”. However, this would not amount to an instance of schema (a), since “attention” in that schema has been defined in terms of attentional spotlight or focus, sometimes actively controlled by the agent, and this follows the analyses given by Fodor, Pylyshyn, and others. So, if one theorizes some of these selection mechanisms as nuanced forms of spatial attention, and those mechanisms are cognitively sensitive and influence perception, this would just be an additional example of attention-mediated cognitive effects on per ception, not dismissed as an attention shift.8
133
134
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
In a recent paper, Jake Quilty-Dunn argues that no effect of attention on perception is a violation of informational encapsulation, and thus is no threat to modularity. He acknowledges that some theorists (Mole 2015; Wu 2017) have argued to the contrary and abandoned (as has been suggested here) the gatekeeper assumption. Granting this architectural posit, Quilty-Dunn offers this rejoinder:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
If attention is a perceptual process, one might argue, then cognitively driven attention is ipso facto a case where a perceptual process accesses cognition (see also wu 2017). But this claim, depending on how it is meant, is either trivial or false. If the claim is merely that attention (qua percep tual process) accesses information in cognition, then it is trivial; nobody on any side of the debate about encapsulation would deny that attention can be driven by cognitive states. If, on the other hand, the claim is that the independent perceptual processes modulated by attention thereby access information in cognition, then it’s false. There is a clear difference between (i) a process that accesses representation R and (ii) a process that is affected by a separate process that accesses R. The latter process does not access R, and therefore does not have R within its store of information. (2020: 344)
The claim is not trivial, however, if interpreted as Mole and others suggest. It is true that neither side of the debate denies that cognition can drive attention when attention is understood as an intermediate gatekeeper (and, typically for Fodor and Pylyshyn, as overt, spatial attention). It is not true that both sides accept that cognition can drive attention when attention is understood as part of perception (and, moreover, as covert, selective attention of the sorts discussed in this chapter). This is one of the very claims that the modularist denies. Or better, this is one of the very claims that is neglected by virtue of making the various assumptions identified in this chapter. Therefore the alleged triviality of the claim seems premised on a slide back into the very gatekeeper assumption that is supposed to be abandoned. A final, related rejoinder insists that attention and perception are always separate mental processes. Such a theorist would then claim that a phe nomenon fitting (b) is not cognitive penetration for failure to satisfy the directness (or vehicle) criterion described earlier.9 However, here it is important to ask why directness is supposed to be required for cogni tive penetration. One diagnosis is this: A direct cognitive effect on per ception is the best evidence that we could have for a cognitive-perceptual
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
phenomenon of important consequence. Direct cognitive influence on perceptual processing would provide the clearest threat for the modularity of sensory systems and imply important consequences for the knowledge providing and behavioural roles of perceptual experience. Suppose that this is cor rect. It only implies that the directness criterion is an evidentialist criterion in a weak sense: Direct effects are, in principle, the best or most easily identified evidence for cognitive penetration. This alone does not imply that direct effects are the only types of relevant effect nor that evidence for directness is the only evidence of relevance. Indeed, if directness is at most an evidentialist (or operationalist) criterion, then to insist that directness is a necessary condition for the phenomenon is to set the bar for cognitive penetration in a way that is unprincipled and in favour of the sceptic. Instead, it seems, it is not the directness of the cognitive effect that is essential but the consequences. The question then becomes: What does a phenomenon fitting (b) imply about the consequences? And this brings us to a second argument.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
5.3.2 An argument from consequences The second argument remains neutral on the gatekeeper assumption while rejecting the attention as act and spotlight assumptions. Recall that a mod ular theory of perceptual systems maintains that perceptual processing is informationally encapsulated. A phenomenon that fits (b) would plausibly violate this condition, since intermediary selective attention is sensitive to the perceiver’s learned cognitive states. This selectivity of features or objects directly influences the informational processing of perceptual systems. And as hypothesized, this kind of attentional selection may operate non-agentially and independently of spatial attention. Therefore, perception is not informa tionally encapsulated relative to cognitive information. It is instead sensitive to learned individual and cultural differences. This is especially perspicuous in the empirical studies discussed in Section 5.2, where perceptual process ing – evidenced by behavioural, eye tracking, and EEG results – varies with linguistic concepts or knowledge. These results suggest that informational processing is sensitive to something more than only “purely” visual features of the external stimuli. So, even if the causal process runs from cognition through attention to perception, the standard modular cognition/perception architecture is called into question. Here again the point can be usefully put in terms of biased competi tion. In addition to, and in some cases independent of, the bottom-up biases
135
Copyright © 2021. Taylor & Francis Group. All rights reserved.
136
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
on attentional allocation, FBA and OBA are biased in top-down ways. For instance, in the visual search studies by Moores and colleagues, stored infor mation about linguistic relations influence what objects catch attention and resulting behaviour. In the light of this array of evidence – again, behav ioural, eye tracking, and EEG data – it is difficult to maintain the claim that visual processing is fully encapsulated from information stored in memory. Wu (2017) gives an analysis in the same spirit and comes to the same gen eral conclusion as this one, though his emphasis and the mode of analysis are substantively distinct. Wu takes violation of informational encapsulation to be an important marker for cognitive penetration.10 Wu’s central question is whether background cognitive states directly influence the computational processing in visual structures. Wu argues, by appeal to neurological stud ies on macaque monkeys and computational models, that intention modu lates visual attention, and because intention is cognitive and attention is an “aspect” of vision, this counts as cognitive penetration. So, Wu’s analysis centrally concerns penetration of perceptual processing and directness by contrast to the emphasis here on perceptual experience and relevant con sequences. But like the analysis given here, Wu partly emphasizes selective attention and explicitly notes the importance of epistemological conse quences of intention-involving cases of cognitive effect on perception. What, then, of the epistemological consequence (grouping theory-ladenness and general empirical knowledge)? Fodor was clear that modular systems are epistemically preferable. Perceptual processing independent of background beliefs, goals, and intention will, he claimed, more rapidly and reliably deliver objective representations about the immediate environment (Fodor 1983, 1985, 1988). And as these cognitive influences go up, the relevant perceptual independence goes down. Recall that the answer to Q4 was that selective attention can influence what (and for how long) objects or fea tures are experienced, what “pops out” in experience or is salient, and how experience is structured. And recall that selective attention can affect per ceptual salience and structure without the perceiver acting to bring about the changes, nor changing or manipulating input from the perceptible envi ronment. These are genuine perceptual effects. Insofar as these effects are sensitive to background cognitive information, rather than just features of the immediate environment, the independence of perceptual representation is called into question. Thus, some epistemic consequence for perception follows.11 One might object: Surely these effects are too minor to sound sceptical alarm bells. In the empirical studies discussed, for example, there were no
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
massive errors made and no illusions suffered. So why should we be worried about the epistemic role of perception on the basis of evidenced phenom ena that fit (b)? An important lesson can be gleaned from this rejoinder. While it is important to note that some perceptual error or loss in efficiency did occur in the studies on linguistic effects, it is more important to note that the relevant epistemological consequence need not be pernicious. This is typically, and sometimes explicitly, supposed. But there is no principled reason to assume that cognitive penetration, should it occur, is bound to produce epistemic problems. In other words, the epistemic consequence can be stated neutrally. What matters is that cognition affects perception in some important way with respect to the knowledge-providing role of perception. In Fodor’s terms, but contrary to his conclusion, epistemic independence of perception may not be necessary or even especially crucial to objective per ceptual representation. And in fact, this kind of effect could be epistemically boosting, rather than downgrading (see Lyons 2011; Stokes 2013, 2014; Siegel 2013; Vance 2015). One clear way to bring out this possibility is to consider expertise. Subjects in the Moores et al. (2003) and related experiments had to have some min imal linguistic expertise. Absent that expertise, there is no relevant percep tual effect, and accordingly no relevant effect on the perceptual judgments that subjects make about the visual field, nor an effect on performance time or accuracy. Therefore, linguistic expertise influences perception in a way that in turn influences reports and doxastic commitment. There is a rich experimental literature on experts in more specialized domains, as will be discussed in Chapters 6 to 8. For now, a brief example. Radiologists reliably identify abnormalities in rapidly presented (200 ms) chest radiographs. Eye tracking studies corroborate these results by revealing substantially differ ent scan patterns between experts and controls: Expert radiologists make fewer total saccadic eye movements and fixate more quickly on abnormal ities (Drew et al. 2013; Evans et al. 2013). These and many other cases of empirically studied perceptual expertise are candidate examples of epis temically enhancing cognitive penetration. (And, as will be discussed in those later chapters, cases of expertise are behaviourally enhancing. How and whether this amounts to theory-ladenness of any interesting sort will be a further point for later discussion.) Consideration of these cases should at least encourage caution about the assumption that the epistemic con sequence of cognitive penetration must be pernicious, and at most such cases are plausible instances of epistemically enhancing cognitive effects on perception.
137
138
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
This concludes the argument that attention-involving cognitive effects on perception need not take the form of causal schema (a). Instead, there is good reason to think that phenomena described in terms of causal schema (b) are possible and substantial empirical evidence that actual phenomena do fit this very schema. Finally, and in answer to the central question of this chapter, these are plausibly attention-involving instances of cognitive penetration, and this can be argued in a way that conforms to the directness criterion and/or by appeal to consequences.
5.4 Summary: moving beyond cognitive penetration This is progress. The framework offered here moves beyond various assumptions of extant theorists. Simplifying, that assumption is captured by the following conditional claim: If attention is part of the causal explanation of a relation between cognition and perception, then that relation is not cognitive penetration. This conditional is true only if one grants the further assumption that the following causal schema is the only possible one: (a) Cognitive state → Attention-shift → Perceptual experience
But there is an additional important causal schema:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(b) Cognitive state → Non-agential selective attention → Perceptual experience
So the assumed conditional is false. There are other ways that attention could be involved in cognitive effects on perception. This is the first lesson. The second lesson is that, if this analysis has been successful, any actual psychological phenomenon that fits schema (b) is cognitive penetration. And this changes the dialectical space. The question now becomes, in the context of thinking about attention and cognitive penetration, are there any actual phenomena appropriately described to fit schema (b)? The sceptic must motivate a “no” in answer to this question. And he must do this in the face of the mounting evidence from perceptual psychology discussed here. Therefore, the cautious conclusion is that, at the very least, the burden of proof shifts to the sceptics, to those who deny the cognitive penetration of perception. The more ambitious conclusion of this chapter, in the context of thinking about attention and cognitive penetration, is that the modularist’s sceptical position is thereby undermined. Generalizing, the intermediate conclusion
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
motivated by the analyses of the last few chapters is that thinking does affect perceiving. Perception is malleable. How we think, what we think, and what we have learned can influence conscious perceptual experience, sometimes directly and with important epistemic, behavioural, and scientific conse quences. This thesis – the thinking–affects–perceiving thesis, or TaP thesis – will now be extended to a stronger thesis over the remaining chapters. Thinking not only affects perceiving; it improves perceiving.12 Further reading For seminal scientific studies and theories of attention, see (Broadbent 1958; Treisman and Gelade 1980; Allport 1987; Desimone and Duncan 1995) and a philosophical review: (Mole 2017). For recent philosophical discussion on attention: (Campbell 2002; Mole 2011; Nanay 2010; Wu 2011, 2014; Mole et al. 2011; Watzl 2017; Fazekas and Nanay forthcoming). For discussion on attention as a gatekeeper and its possible relation to consciousness, see (Wu 2011: chapter 5 and 6). For some recent work on visual attention and conscious experience, see (Carrasco et al. 2004; Fuller and Carrasco 2006; Block 2010; Nanay 2010). For recent discussion of attention and cognitive penetration: (Mole 2015; Wu 2017; Stokes 2018; Firestone and Scholl 2015; Gross 2017; Quilty-Dunn 2020).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 It would be more accurate, then, to replace “attention shift” in schema (a) with either “focal attention shift” or “intentional/spatial attention shift” (and the schema, again, should be read that way). However, schema (a) has been termed in keeping with the standard terminology in the rele vant literature, which widely discusses the “attention-shift interpretation”. Thank you to an anonymous referee for The Australasian Journal of Philoso phy for emphasizing this qualification. 2 Another way to mark this distinction is in terms of focused vs. distributed attention. Attention can be focused on one region or space. Or attention can be distributed across features or objects within a region of space. Put ting the current points in these terms, distributed attention (to features or objects) can change without any change to the focus (the space) of attention. See Nanay 2016 for an application of this distinction to aesthetic experience.
139
Copyright © 2021. Taylor & Francis Group. All rights reserved.
140
ATTeNTION AND COGNITIVeLY INFLueNCeD PeRCePTION
3 To be clear, the analysis given in reply to Q3 and Q4 regarding cognitive effects on FBA and OBA and the effects that those attentional mechanisms have on conscious experience can be motivated independent of commit ment to the biased competition model of attention. However, something like biased competition seems to be approaching orthodoxy in the psy chology of attention (where debates persist regarding finer details of the approach). And if such a model is accepted, then it becomes very difficult, as we will see in Section 5.3.1, to resist certain cases by appealing to an attention-shift interpretation. 4 Humans produce an average of three to five saccades per second, and saccades take an average of 100 to 200 ms. Saccades can be voluntary, as when one intends to shift one’s gaze to another part of the room. But they also occur involuntarily or “reflexively”. These movements are ballistic: Once a saccade is initiated (voluntarily or reflexively), it cannot be adjusted mid-saccade (e.g. if the target moves). 5 One might object that these studies are just evidence for semantic prim ing. Here are two additional qualifications against that rejoinder. First, the effects here are not all behaviourally enhancing, as priming effects often are (see Moores et al. 2003: 187–188; Meyer et al. 2007: 714). Second, the effects here are not, as priming is often defined, (mere) effects on mem ory. This is precisely the claim to be defended in reply to Q4. 6 It is worth noting also that the New Look psychologists identified this phe nomenon under the term “accentuation” – “the tendency for sought-after perceptual objects to become more vivid” (Bruner and Goodman 1947). 7 In addition to those studies already cited in Section 5.1, see Moore and Zirnsak (2017) for a review piece.
8 Thanks to an anonymous reviewer for criticism on this point.
9 See Mole’s discussion on this point (2015: 233).
10 See also Raftopoulos (2001a) and Cecchi (2014). 11 Thanks to Chris Mole for helpful suggestions on this point. 12 Acronyms are almost invariably disappointing or silly, but there it is.
6
PERCEPTUAL EXPERTISE I:
MENTAL ARCHITECTURE
From worse to better perceivers Perceptual experts are perceptual experts Perceptual expertise as a genuinely perceptual phenomenon Perceptual expertise as genuine cognitively sensitive expertise 6.5 Situating perceptual expertise in theories of the architecture of mind 6.6 Summary: towards the malleability of mind Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.1 6.2 6.3 6.4
A brief interlude: Much of the last chapters has been devoted to making the case for cognitive effects on perception and ones that are probable challenges to modular architectures. These cases have sometimes been described in the terms of cognitive penetration. The prescription now, and the practice that follows for the rest of the book, is to move away from this emphasis. If the analyses of previous chapters were successful, modularity does not deserve the default theoretical position: Neither its arguments nor its explanations are sufficient to justify such a default. Accordingly, novel proposals in the philosophy and cognitive science of perception need not attach their success conditions to undermining or disproving modularity; they need not make
142
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
the case that an interesting cognitive-perceptual phenomenon “counts” as cognitive penetration. To do so is to continue to make that very default assumption and suffer the theoretical restrictions that come with it. To further qualify, though, this is not to indicate that modularity will not figure in the remaining discussion. It will. And this is for good reason. In addition to its continued default status (which should be undermined), advo cates of modularity provide a range of important philosophical questions and conceptual markers and ones that often provide a useful frame, if not foil, for current study and theory. Those advocates advance answers that are restrictive. But this does not bar us from employing some of the same framework in order to weaken those restrictions and develop new ways to theorize thinking and perceiving. The ladder can be kicked away once suffi cient progress is being made. Recall the following passage:
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Perception is above all concerned with keeping track of the state of the organism’s local spatiotemporal environment. Not the distant past, not the distant future, and not . . . what is very far away. Perception is built to detect what is right here, right now – what is available, for example, for eating or being eaten by. If this is indeed its teleology, then it is under standable that perception should be performed by fast, mandatory, encapsulated, . . . etc. systems that – considered, as it were, detection – theoretically – are prepared to trade false positives for high gain. It is, no doubt, important to attend to the eternally beautiful and to believe the eternally true. But it is more important not to be eaten. (Fodor 1985: 4)
When we considered (most of) this passage in Chapter 2, emphasis was placed on how it contributes to an argument for informationally encapsulated perceptual modules from the observed reliability of perception. Now we can emphasize something else occurring here and elsewhere in the modularist’s reasoning, namely, that given the function of perceptual systems to provide accurate information about one’s immediate environment, we should expect those systems to enjoy a computational autonomy that best ensures success ful performance of that function. There is a teleological argument in close proximity. Organisms with computationally autonomous visual systems – ones that perform their proprietary computations in a way encapsulated from information from the organism’s memory, expectations, desires, and other cognitive systems – will prove fitter in their respective environments.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
And this is for the simple reason that the perceptual representations com puted by such modular perceptual systems (by contrast to those computed by non-modular systems in other individuals) will best represent, accurately and without bias, the objects and features of the actual environment. There fore, assuming there ever were non-modular visual (or other perceptual) systems in ancestral species, those systems (or if one likes, traits) would have been maladaptive and, accordingly, selected against. The argument then con cludes that we should expect, across an evolutionary timescale, species that by and large enjoy modular perceptual systems. This argument certainly has force. But that force is enabled by an under lying assumption that, once identified and challenged, greatly weakens the argument. The assumption, as was mentioned towards the end of the previ ous chapter, is that cognitive effects on perception are always or mostly epis temically pernicious. The simple thought seems to be that if an organism’s beliefs or expectations or, worse, its goals, desires, or other evaluative states, were to substantially influence perceptual representation, then the reliability or accuracy of those representations in general would greatly decrease. Call this the pernicious cognitive effects assumption. This chapter and the next two work in tandem. The present chapter pro ceeds as follows. In Section 6.1, some analysis will be given of why or how the pernicious cognitive effects assumption is made. Why is it assumed, tacitly if not explicitly, that cognitive influences on perception are epis temically pernicious? And second, what rejoinders are offered by propo nents of modularity to candidate “good” cases of such cognitive influence? The analysis then shifts to a project that moves in the opposite direction, namely, arguing for a rich variety of cases that appear to be epistemically virtuous: cases where we become better perceivers as a consequence of higher-level, cognitive learning. These are cases of perceptual expertise, and the analysis of such cases divides into two, the first focusing largely on ontology or mental architecture, the second on epistemology. Section 6.2 provides an initial sketch of empirical research on perceptual expertise. Section 6.3 makes the argument that many cases of perceptual experts, as studied in the behavioural and brain sciences, should be understood as genuine perceptual phenomena; Section 6.4 argues that those perceptual effects are genuinely sensitive to cognition or thought. Section 6.5 then brings that analysis back to bear on questions about cognitive influences on perception. Chapter 7 then extends this analysis by arguing that many cases of perceptual expertise are best understood as cases of epistemic vir tue. Second, a teleological analysis is given that grounds the argument for
143
144
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
epistemic virtue. Finally, in Chapter 8, lessons from all of this are drawn for theories of perception, including a novel account of perceptual accuracy and objectivity. The thesis that emerges, and that will be substantiated over the remainder of the book, is a compliment to the TaP thesis. It says that thinking improves perceiving (the TiP thesis). Cognitive learning changes perceptual experience in a rich variety of ways, and the results are epistemically and behaviourally enhancing. This is what perceptual expertise teaches. To make a start on this line of analysis, further consider an opponent view.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.1 From worse to better perceivers The pernicious cognitive effects assumption maintains that cognitive or topdown effects on perception would be, by and large, epistemically bad. See ing or hearing in ways importantly influenced by beliefs or expectations or values would be, in short, a problematic kind of bias on those perceptual representations. What drives this assumption? The first possible culprit is just in the choice of cases for consideration. Modularists like Fodor commonly employ cases of persistent illusion, argu ing that their persistence reveals that knowledge (say that the Müller-Lyer lines are of equal length) does not penetrate perceptual experience. Theorists on the opposite side of the debate make similar choices. In Chapter 3 a survey was given of various kinds of evidence for cognitive effects on perception. Nearly all of these cases are ones where subjects, if not suffering illusions, are making errors in some way – for instance, regarding the size or colour or distance of a perceived object. And so, quite naturally, if these are candidate cases of the cognitive penetration of perception, then one will be inclined to think that any such possible influences will be epistemically negative, taking the perceiver further from a veridical perception of the available stimuli.1 A second possible culprit concerns objectivity and what it means for a perceptual representation to be objective. In much of the relevant litera ture, there is an at least tacit assumption that accurate perception is some how “pure”, latching on to the perceiver-independent features of the world without any person-level input from that individual. As it is sometimes put, accurate perception achieves objectivity with a big “O”. This assumption is problematic, and it will take some time to unpack properly and then criti cize; it will be taken up again in a later chapter.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
A final culprit of sorts concerns a bias towards the negative in philosophy of perception and, less so, perceptual psychology. For instance, illusion and hallucination continue to receive substantial emphasis as a kind of litmus test for both the structure and epistemic nature of perception. A great deal of discussion is centred around perceptual error and misrepresentation: How do theories of perception explain perceptual error? How can perceptual misrepresentation be naturalized? If there is regular risk for error, what is the epistemically responsible stance to take towards perception? The thread common to these various philosophical problems and theories is an empha sis on how perceivers or their perceptual systems fail or could fail. One can identify a similar thread in perceptual psychology. Many designs in behavioural psychology involve experimental tasks that evoke or expose subject error. Subjects may be primed in ways that lead to errors in percep tual report, or ambiguous stimuli may be used, or a conflicting stimulus (perhaps in a distinct sense modality) may be presented in tandem with the target stimulus. Cognitive neuropsychology is largely grounded on lesion studies, where inferences about proper function of sensory systems (identi fied with neurophysiological structures) are derived from experimentation on subjects with damage to a particular neurophysiological structure. The inferential pattern here is to first correlate perceptual errors or task fail ures with lesions of a distinctive type and then from this correlation derive the proper psychological or behavioural function of that neurophysiological structure for “normal” subjects (Coltheart 2001). There are important differ ences here, and as we will see, the psychological literature on perception is much less dominated by emphasis on the negative, but here again there is an identifiable thread: A great deal of research in these scientific fields focuses on how perception goes wrong or performs sub-optimally. Unlike the first two culprits, this last one is a general feature of the broader literature. But it, too, could partly explain why there is a commonly made assumption that cognitive influences on perception would be perni cious; there seems to be a bias towards the negative, or at best the neutral, in theories of perception. This is a curious feature of the literature and one worth remedy in its own right, given the epistemic and normative impor tance of perception to the human condition. Perception is supposed by all parties, at least when not in sceptical moods, to be crucial to how we come to know about and navigate the world around us. In both related and dis parate domains that concern normativity, there is far greater emphasis on the positive, on improvement, on the success cases. In normative ethics, a central theoretical project is to identify how we can become better persons,
145
Copyright © 2021. Taylor & Francis Group. All rights reserved.
146
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
how to do the right thing, how to achieve virtue. In studies on reasoning and decision-making, principles are identified that are supposed to optimize suc cessful inference, planning, and prediction. And, closest to the problems at hand, broader epistemology aims to identify good-making epistemic prop erties of agents or mental acts: reasons, warrant, justification, reliability. So even if explicable, the scarcity of philosophical discussion of the improve ment and optimization of perception, once revealed, is worthy of attention. An exception to the lack of emphasis on positive cases of perception is empirical research on perceptual expertise. This programme of research traces back to work on face recognition. One point of debate in that literature is whether the fusiform face area (FFA), as well as the occipital face area (OFA), in humans is a neural structure devoted exclusively to the recognition of faces (Sergent et al. 1992; Kanwisher et al. 1997). Experimental research led cen trally by Isabel Gauthier in the late 1990s and early 2000s was interpreted to show that the FFA could be recruited in a number of domains other than face perception. For example, some of the first of these studies indicated similar neural activity in FFA in visual recognition of birds and cars and trained visual recognition of instances of a novel, artificially constructed category of object (Gauthier and Tarr 1997; Gauthier et al. 1999; Tarr and Gauthier 2000; Gauthier et al. 2000). This question about the face specificity of the FFA can be distinguished from a second question.2 One can ask, inde pendent of debates about the FFA, whether genuine perceptual expertise is exclusive to the recognition of faces, or can we acquire enhanced capacities for perceptual recognition and discrimination of similarly complex non-face objects?3 Ongoing debates in cognitive science centre around both, but it is the second question that is of central interest here. Researchers investigate “real-world” experts: expert bird watchers, fin gerprint examiners, radiologists, car experts, and tree experts, among many others. And researchers study subjects trained in laboratory circumstances: training novice subjects to recognize birds, cars, and other-race faces, among other categories. Another line of research involves training subjects to recog nize and classify objects of artificial (laboratory) construction. These behavioural studies commonly centre around categorization and identification tasks, where both real-world expert or lab-trained expert sub jects (plus novice controls) perform feature listing, object naming, category verification, and object discrimination tasks, among others, often in response to a visual presentation (of birds, or cars, or lab-created stimuli, etc.). Fol lowing the work of Rosch and colleagues (1976), research on categorization distinguishes a hierarchy of category names: sub-subordinate (e.g. “eastern
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
kingbird”), subordinate (e.g. “kingbird”), family (e.g. “flycatcher”), domain (e.g. “songbird”), basic (e.g. “bird”), intermediate (e.g. “vertebrate”), or superordinate (e.g. “animal”) (Johnson and Mervis 1997: 257). The research by Rosch demonstrated that basic-level categories enjoy a primacy when making initial classifications of a perceived object. When non-experts cate gorize, say, an eastern kingbird, “bird” is the name most readily applied. One important finding of various, more recent studies on expert-level subjects is that this basic-level categorization – sometimes called “the entry level” – can shift with acquisition of expertise. Thus the entry level for the bird expert is commonly at the subordinate level – it shifts to “kingbird” for the relevant perceived object. And, as one would predict, experts perform better on other tasks – feature listing, category verification, object discrim ination – that are sensitive to subordinate-level (or even more fine-grained) categorization. Other tasks involve discrimination at an individual level. For instance, researchers have studied expert fingerprint examiners, where after brief presentation of a fingerprint sample image, subjects can quickly identify the target out of two very similar test images (Busey and Vanderkolk 2005). These same methods are employed for laboratory-trained expertise. The most common experimental paradigm here involves Greebles. Greebles are a “class of non-face, novel, cartoon-like objects that can be classified at mul tiple levels” (Scott 2011: 196). (See Figure 6.1.) While Greebles are not faces, they have parts that protrude from a cylindrical body, where those parts tend to be organized according to a few similar patterns. Each Greeble will then vary to some degree along these patterns (no two Greebles are identi cal). Accordingly, Greebles can be distinguished by two “genders” and five
Figure 6.1 Greebles
Source: Reprinted with permission of von Bubnoff 2014. Copyright Springer-Nature.
147
148
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
“families”. Similar research has been done with “blobs” and “Ziggerins” (respectively, see Nishimura and Maurer 2008; Wong et al. 2009). While importantly different in timescale and other details from real-world expertise, this approach affords better control of various factors relevant to expertise acquisition. Once adequately trained, subjects respond more accu rately and quickly in both categorizing Greebles (for example, by “family”) and individuating a particular Greeble from others and recognize novel, in-category exemplars. This method also allows researchers to identify changes in perceptual processing of stimuli over the expertise acquisition period. Similar methods, and with similar results, have been used for natu ral (non-laboratory created) stimuli, where novices are trained to recognize categories of object like birds or other-race faces. These behavioural meth ods and results are corroborated with a number of additional physiological and neural methods, including eye tracking studies to identify attentional patterns of experts; electroencephalogram (EEG) recordings of event-related potentials; and functional magnetic resonance imaging (fMRI) studies on face-related neural areas, all to be discussed next.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.2 Perceptual experts are perceptual experts Proponents of modularity have not entirely ignored cases of perceptual expertise (or at least cases in the vicinity). They have, however, dismissed them as irrelevant to debates about cognitive effects on perception. Fodor speaks, for example, of the alleged skill that painters have to perform a kind of “perceptual reduction”. As discussed in Chapter 2, the visual system involves mechanisms for achieving constancies with respect to colour, shape, distance, and other properties. You perceive the uniformly red fire engine as being uniformly red in spite of some parts being better illuminated, others more shaded, and so on. To realistically depict such a visual scene, however, the painter must somehow see the inconstancies. Fodor grants that there may be such skills, but dismisses them on the grounds that “because they are highly skilled they may tell us very little about the character of normal perceptual processing” (Fodor 1983: 54). (In that same discussion, he dis misses expert phoneticians on the same grounds.) It is difficult to see how this admission would be acceptable to a strong Fodorian modularist, since it would seem to allow that some perceivers acquire skills that violate the infor mational encapsulation of visual or other perceptual systems.4 But this viola tion, however special it may be, would be a counterexample to modularity. It is therefore unclear how far Fodor intended the rejoinder to go.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Pylyshyn (1999) spends more time considering cases of perceptual exper tise.5 He points to research on master chess players and athletes, amongst others. He resists interpreting these cases as genuine instances of top-down effects on perception, however, by appeal to the attention-shift interpreta tion criticized in Chapter 5. He writes,
Copyright © 2021. Taylor & Francis Group. All rights reserved.
when asked how they do it, experts typically say that they can tell that a stimulus has certain properties by “just looking.” But the research that is available shows that often what the expert has learned is not a “way of see ing” as such, but rather some combination of a task-relevant mnemonic skill with a knowledge of where to direct attention. (1999: 359)
As we will see, and for reasons largely articulated in the discussion of the previous chapter, this dismissal is equally unwarranted. Many of the more recently studied cases of perceptual expertise are not well explained as just involving a cognitively directed shift in spatial attention. The analysis here attempts to motivate an account of perceptual expertise that takes seriously both the apparent cognitive and perceptual elements of the phenomena. In their domain of expertise, perceptual experts per form with more accuracy, greater speed, and more sensitivity to fine-grained details than novice subjects. This is a statistical claim. Whether this amounts to a perceptual epistemic good of some important kind is left an open ques tion, to be addressed in Chapter 7. Before we get to that question, Section 6.3 addresses the question: Are these expert achievements constituted by gen uine perceptual processes? And Section 6.4 addresses the question: Are the perceptual achievements constituted by cognitive expertise in a way that is sensitive to the high-level content specific to the domain of expertise? Both questions will be answered in the affirmative, and whether this amounts to a challenge to modularity will to some degree also be left an open question. The suggestion will be that it does present a challenge to modularity in its strongest form. But, and to stress once more, the success of the analysis and its consequences for theories of thinking and perceiving does not depend upon conclusive refutation of modularity. To assume this is to continue to assume that modularity is the default position. The answers to all questions separate into two architectural claims and an epistemic claim. Perceptual expertise is a genuinely perceptual and genuinely cognitive expert-involving phenomenon. And the effects of expertise on per ception are epistemically enhancing.6 Perceptual experts achieve skills that
149
Copyright © 2021. Taylor & Francis Group. All rights reserved.
150
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
are best described as intellectual virtues. These claims may sound innocuous on their face, but there are sceptics for each. The cases made for cognitive penetration highlight data from perceptual psychology and then argue that these data are best explained as an impor tant cognitive effect on perception. As we have seen, there are a number of ways that theorists resist a cognitive penetration explanation of any such empirical case. First, for a given case (or type of case), one might claim that the reports or behaviour of experimental subjects are better explained as indicating some post-perceptual cognitive effect on judgment or belief. So, background cognitive states influence other of one’s cognitive states (which show up in subjective report), while having no relevant effect on perceptual experience. Call this sceptic 1. Second, for a given case (or type of case), one might argue that the data are better explained in non-cognitive terms. This alternative explanation grants that there is some change or effect on percep tion but denies that this is a consequence of some antecedent cognitive state or process. Instead, the case is typically explained as some intra-perceptual phenomenon, as some kind of purely perceptual learning or perceptual adap tation that requires no explanatory appeal to beliefs or knowledge. Call this sceptic 2. Finally, if one does not resist but grants the possibility (or actuality) of cognitive penetration of perception, the consequences then emphasized are predominantly negative. Theorists on both sides of the debate emphasize the epistemic threat of cognitive penetration. The now-familiar worry is that if perceptual experience is influenced by background beliefs or, worse, non-doxastic states like desire or emotion, then it would not provide suffi ciently objective grounds for perceptual belief. This encourages a founda tional kind of philosophical scepticism. Call this sceptic 3. Sceptics 1 and 2 are sceptical about the reality of certain kinds of non-perceptual effects on per ception. Sceptic 3 captures a general epistemic worry about non-perceptual effects on perception, should they actually occur.7 The three types of sceptics just identified have clear analogues with respect to perceptual expertise. One might deny that relevant evidence from percep tual psychology is evidence for a genuinely perceptual phenomenon (sceptic 1). Or one might deny that such evidence is genuinely expert-involving, in the sense that it is actually sensitive to cognitive or conceptual content of the relevant domain (sceptic 2). And finally, one may grant that there are genu ine instances of perceptual expertise – properly called, contrary to sceptics 1 and 2 – but conclude that these kinds of higher-level effects on sensory perception are generally epistemically pernicious or otherwise worrisome (sceptic 3). Thus, the perceptual expert is not, epistemically, a perceptual
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
expert. Each of these three sceptical positions is further developed, and rebutted, in the sections (and chapters) that follow.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.3 Perceptual expertise as a genuinely perceptual phenomenon Sceptic 1 resists the claim that perceptual expertise is a genuinely perceptual phenomenon. There are a number of non-exclusive reasons for this scep ticism, and ones now well familiar to us given previous discussions. Most familiar, one may hold a strongly modularist theory of perception, accord ing to which perceptual processes operate in a way that is informationally encapsulated and therefore cognitively impenetrable. Although there may be conceptual space to allow for the compatibility of genuinely perceptual exper tise and a modular account, the default reading of the latter is that perceptual systems are not subject to substantial changes that result from cognitive or conceptual learning.8 Second, one may commit independently to a sparse theory of perceptual content. Recall that vision, according to such a theory, only represents shape, colour, illumination, and rest/motion. An intuitive way to characterize some instances of acquisition of perceptual expertise is as involving changes in recognitional capacities: The subject that learns to dis tinguish birds at the subordinate level (“kingbird”, “cardinal”, “crow”, etc.) is able to recognize instances of these categories as such. And one might be tempted to construe this perceptually: Acquiring the capacity to recognize cardinals as cardinals involves acquiring a visual capacity to represent the property “being a cardinal”. This construal is, however, incompatible with the sparse content theory. Perception, says that theory, does not represent “being a cardinal” any more than it represents “being a bird” (or “being a pine tree”, etc.). Perception instead only provides information about the shape, colour, and illumination of, say, the bird. Sceptic 1 can then couple either or both of these theories with a pre-perceptual or a post-perceptual explanation of the performance of experts. The post-perceptual explanation goes as follows. The expert bird watcher and the non-expert control subject enjoy the same perceptual expe rience of birds – the same perceived colours, shapes, illumination, etc. However, the expert better cognizes features that are distinctive of birds of particular subordinate categories and, consequently, behaves in ways sensi tive to these features. The expert and the novice see the same, but the expert does more cognitively with what she sees: forming judgments, drawing infer ences, applying concepts or beliefs to the visual scene. The pre-perceptual
151
Copyright © 2021. Taylor & Francis Group. All rights reserved.
152
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
explanation grants that the expert and novice enjoy distinct experiences, but this is only because the expert knows how to direct her attention voluntarily to the relevant elements of the visual array. Accordingly, any effect on visual experience depends on active differences in spatial attention, which changes the input to visual processing (in the expert vs. the novice).9 The case against sceptic 1 takes the form of an inference to the best expla nation. The analysis will thus proceed by laying out a variety of behavioural and neural-physiological evidence and conclude that this variety of evidence is best explained by the hypothesis that much of so-called perceptual exper tise is genuinely perceptual. The first strand of behavioural evidence concerns “automaticity”. Not only do experts more rapidly perform categorizations or other forms of recogni tion (categorizing a bird image at the subordinate or sub-subordinate level or matching a fingerprint), but they do so in ways that they often cannot carefully describe. For example, expert radiologists often report a sense that there is something anomalous in a medical image before they can point to the anomaly. When asked to describe their phenomenology (once identi fications or categorizations are successfully made in a task), experts report that the relevant object or feature is “highly salient” or just “pops out”. And they regularly invoke visual terms, reporting that they “just see” the relevant object or feature immediately.10 These reports and the speed of performance suggest that the expert expends little or no deliberate cognitive effort and that her performance is non-inferential. Automaticity is studied using interference effects. One relevant method is the composite task. Subjects are presented with both the bottom and top half of an object of expertise, with an intermittent mask. (See Figure 6.2.) The task is to determine if the designated half (this is sometimes varied between top and bottom from trial to trial) of the second image (the “test image”) is the same as the relevant half of the first image (the “study image”). The two halves of the test image can be congruent or not (congruence condition) and aligned or not (alignment condition). (The study image may also be congruent/incongruent and/or aligned/misaligned, but as mentioned later, this affects novices, not experts.) What these studies typically show is that at test, the irrelevant object half that is to be ignored interferes with expert performance. For example, performance degrades in experts when, in the test image, the bottom half of a Greeble is incongruent with the (target) top half. Further, this effect is modulated by the alignment of the test image: The magnitude of these interference effects in experts is highest when the object halves of the test image are aligned. This effect is well substantiated
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Figure 6.2 Examples of composite task structure for faces, Greebles, and musical notes Source: Reprinted with permission from Richler et al. 2011. Copyright SAGE Publishers.
153
Copyright © 2021. Taylor & Francis Group. All rights reserved.
154
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
for faces as well, about which all subjects are experts, so long as the faces are “within race”.11 The standard explanation for this effect is that experts rely on holistic processing of objects of expertise and these tasks require them to attend only one part of the whole. And moreover, the holistic process would seem to be most strongly triggered in cases where the object is intact as the relevant whole (the aligned condition) (Richler et al. 2011). It is worth noting that novices do suffer interference, but in ways impor tantly different from expert interference. Novice interference effects are context-sensitive, while expert interference effects are automatic and context-insensitive. For experts, manipulations to the study image (e.g. align ment) does not modulate interference. And a randomization manipulation of experimental context (where the study images are sometimes aligned and sometimes misaligned) does not modulate interference. For novices, inter ference occurs when the study images are misaligned but the alignment of the test image has no significant effect. Furthermore, this interference effect can pervade an entire set of randomized trials: When the experimental con text is one where the study images are sometimes misaligned, novice per formance suffers interference across the set of trials. All of this suggests the following distinctive explanations for interference in novices versus experts. For novices, if misalignment is introduced at the study image or established as a contextual feature of the experiment through randomized trial struc tures, subjects will thereby employ strategies to attend to more of the image, searching for anomalies. This deliberate and flexible “wider attention” strategy then extends to the test image such that interference occurs (for instance, when the to-be-ignored half of the test image is incongruent). So, the process that suffers interference in the novice is a context-sensitive atten tional strategy. By contrast, it appears that the expert process is inflexible and automatic, and it involves holistic processing. What the expert “should” do is selectively attend to only the half of the test image that is to be matched with the half of the study image. If she did this, then incongruence of the to-be-ignored half would have no interference effect. But in fact, such inter ference effects do occur in the expert and are modulated by alignment. It therefore appears that the expert cannot “turn off” an automatic, holistic perceptual strategy, and this is why interference is strongest when the two test images are aligned as a whole. And this is also why, in misaligned con ditions, experts are “released” to some degree from an interference effect. (See Richler et al. 2009, 2011; Gauthier and Tarr 2002; Wong et al. 2009; Gauthier et al. 2003.) Interference effects in composite tasks for experts suggest that holistic processing co-varies with, or becomes more dominant in, acquisition of
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
expertise. Moreover, experts rely heavily upon configural information in an object of expertise: on the spatial relations between features as much or more than the features taken independently. These are standardly the orized markers in face perception. In the part–whole task, subjects are pre sented with a study image (a face) and then presented two test images and asked which feature is the same as the study image (the eyes). Subjects do significantly better at this task when the test images are wholes (two whole faces) than when the two images are isolated parts (two pairs of eyes). (See Figure 6.3.) Studies also compare sensitivity to spatial changes (the space between the eyes) versus sensitivity to feature changes (the size of the eyes). (See Figure 6.4.) The results suggest that although subjects are sensitive to both, they show greater sensitivity to the spatial changes. All of this sug gests a greater sensitivity to local and global spatial relationships in faces. This kind of configural processing has been well-evidenced, using the same methods for both real-world and lab-trained experts, for example, for cars, fingerprints, and Greebles.12 Another common method used involves inversion of target images. A famous example is the Thatcher illusion. In spite of one’s expertise with faces and, further, one’s familiarity with famous faces, it is more difficult to detect substantial featural changes – complete inversion of the eyes and mouth – in a face that is perfectly inverted. These changes are horrifically obvious in an upright face (Thompson 1980). Researchers have found the same effect in a variety of non-face objects, for example, for cars (Curby et al. 2009; Rossion and Curran 2010), dogs (Diamond and Carey 1986), Greebles (Gauthier and Tarr 1997), and fingerprints (Busey and Vanderkolk 2005). Importantly, the magnitude of this effect varies with the relevant expertise. For example,
Figure 6.3 A part–whole task
Source: Reprinted with permission from Bukach et al. 2006. Copyright Elsevier Publishers.
155
156
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Figure 6.4 Spatial change vs. feature change task
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Source: Reprinted with permission from Bukach et al. 2006. Copyright Elsevier Publishers.
performance on dog breed identification declines sharply for dog show judges when images of dogs are inverted, but novices show little to no differ ence in performance on upright versus inverted images of dogs. This again encourages the following explanation: The inversion-task method thwarts the expert’s holistic processing, and thus experts do worse than novices in these tasks. Another kind of behavioural study involves expertise-expertise interference, which suggests competition for cognitive resources in experts between face recognition and recognition of objects of expertise. For example, car experts (but not novices) less successfully reidentify target faces when a task involves rapid presentation of both car and face images. A relevant ques tion is whether this is a genuine effect of expertise, where visual resources for, say, car expertise compete for and exhaust the similar resources needed for face recognition, or whether this is simply a consequence of objects of expertise grabbing attention and thus distracting from faces. Researchers use a rapid serial visual presentation (RSVP) method to test the two hypotheses. In one such experiment, subjects consisting of car experts and non-experts are presented with a pair of target faces, followed by a fixation cross, and then a sequence of twenty images (McKeeff et al. 2010). In the experimen tal condition, the RSVP sequence alternated between face images and car
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
images (F/FC condition). In one control condition, the sequence contained face and watch images (F/FW condition). In both conditions, the face images in the sequence included one of the two target faces, with the rest of the face images being “distractors”. Subjects then reported, in either condition, which of the two target faces appeared in the sequence. Car experts, by contrast to non-experts, performed more slowly in the F/FC condition. Thus task-irrelevant car images seemed to undermine face recognition, but only for car experts. Performance between subject groups in the F/FW condition showed no significant difference. One might be tempted to infer here that the difference is explained in terms of attention. However, the results from a second pair of control con ditions suggests otherwise. In these trials, the targets were a pair of distinct watches, with an RSVP sequence of either watches and cars (W/WC condi tion) or watches and faces (W/WF condition). Again subjects were either car experts or car non-experts (no subjects were watch experts). The important result is that in the W/WC condition, the car experts perform significantly better than the novices (while there is no significant difference between sub ject groups in the W/WF condition). Therefore, it looks implausible that face recognition interference in car experts in the F/FC condition is explained by dominant attention (to cars), since this explanation would predict that in the W/WC condition car experts do equally poorly (since their attention would be drawn to the task-irrelevant distractor images of cars). But the opposite result obtains: Car experts do better in this condition. Similar interference effects have been found in experts in visual search tasks and in inversion tasks (see McGugin et al. 2011; Gauthier et al. 2003, respectively). One final set of behavioural studies concerns visual short-term memory (VSTM). As it is standardly theorized, VSTM is a short-term (or working) memory system that encodes only visually acquired information. VSTM is limited both with respect to how long memories are available for use and how much information it may store. Different models explain this limited capacity differently: Some claim that there are a limited number of “feature slots” available in VSTM, others that VSTM can only handle a limited number of complex, bound object representations. One well-evidenced feature of VSTM is that it can store “more” (in either sense) memories for faces than for other similarly complex objects. And here again we find an exception for objects in a domain of expertise (Curby et al. 2009; Curby and Gauthier 2010). Car experts show a VSTM advantage for objects of expertise where, despite the complexity of car images, VSTM capacity increased to the capacity normal for very simple objects (e.g. three to four coloured circles). This capacity increase
157
158
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Figure 6.5 Results of visual short-term memory studies on car experts vs. car novices
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Source: Reprinted with permission from Curby and Gauthier 2010. Copyright Wiley-Blackwell.
is orientation sensitive: When car images are inverted (just as for faces), the VSTM advantage for car experts is eliminated. (See Figure 6.5.) Curby and Gauthier suggest that “extensive experience with a category can result in a domain-specific increase in VSTM performance for complex objects, perhaps because experts can more efficiently encode and represent complex objects in VSTM” (Curby and Gauthier 2010: 195). Thus, the expla nation favoured by these researchers is a perceptual one. Perhaps experts perform better because, by virtue of more efficient, holistic processing of objects of expertise, the category-specific complexity of an object (a subor dinate or sub-subordinate category of car) is already represented by percep tion. When this representation is given to VSTM, less resources are needed to store or use the representation, since perception has already done the work of representing the object (or bound features) as being within a cate gory. By contrast, the non-expert does not enjoy holistic or category-bound perceptual representations of cars. This would explain why the non-expert has lesser capacity for car representations in VSTM: These representations exhaust cognitive resources given the feature complexity of cars, where those features have to be extracted through attention for a given task. These varied behavioural measures have been correlated with ERP ampli tudes. Experimenters have long used EEG recordings to study neural cor relates of face perception. The ERP component N170 has a typical onset latency between 150 and 200 ms post-stimulus and has been taken as a
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
standard index for face perception; this component responds at higher amplitudes at this temporal scale for faces than for non-face objects. More recently, researchers find that the N170 component is greater in amplitude in response to objects of expertise and for both real-world experts (dog experts and bird experts; Tanaka and Curran 2001) and for lab-trained expertise for both real-world objects (birds; Scott et al. 2006) and artificially constructed objects (Greebles; Rossion et al. 2002). Similar results are found for interfer ence effects. For example, in trained subjects (but not in untrained novices), the N170 response is reduced for faces when the task involves concurrent presentation of objects of expertise (Greebles), but with no similar interfer ence in N170 amplitude (for faces) when the concurrent object is equally complex but not an object of expertise (Rossion et al. 2004).13 Evidence for face selectivity in cortical areas FFA and OFA typically pro ceeds by dissociation inference: Performance in object recognition dissoci ates from performance in face recognition, and deficits in the latter can be correlated with damage to FFA and/or OFA (with no damage to other areas of visual cortex). Using fMRI, researchers find that these same areas show enhanced activity for objects of expertise, for expert subjects in contrast to novice control subjects. This has been shown for real-world experts, for example, car experts and bird experts (Gauthier et al. 2000) and for learned expertise with Greebles (Gauthier et al. 1999). Further fMRI studies reveal adaptation in specifiable neural populations for learned, artificial objects (Folstein et al. 2013). Eye tracking studies have been performed on a range of expertise, from radiologists to highly trained visual artists, revealing eye movement patterns that differ from, and result in better performance by comparison to, novices (Vogt and Magnussen 2007). For example, radiologists more efficiently scan radiographic images, making fewer total eye movements, and fixate more quickly on relevant abnormalities. By contrast, novices deploy deliberate, time-consuming visual search strategies with shorter and more numerous eye movements (Kundel and La Follette 1972; Kundel et al. 1978.). Rapid per formance success co-varies with radiological expertise: In one study, half of difficult-to-detect tumours are fixated by the most experienced radiologists in approximately 1 second of exposure (which is insufficient time to visually attend the entire mammogram) (Kundel et al. 2007). And when forced to make a diagnosis after a mere 200 msec of exposure, before a voluntary eye movement can be made, experts make diagnoses (lesion or no lesion) at 70% accuracy, well above chance (Kundel and Nodine 1975; Evans et al. 2013, 2016).
159
Copyright © 2021. Taylor & Francis Group. All rights reserved.
160
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Now to bring this diversity of evidence together with an explanation. We know that there is some important difference between expert and novice at the level of mental representation; the question is whether this is importantly visual representation. Not one of these pieces of evidence is by itself conclusive. Accordingly, the abductive argument relies on the convergence of the array of evidence, concluding that a perceptual expla nation far better explains that array than the non-perceptual explanations available to sceptic 1. Begin with the convergence of behavioural data. It is important to re-emphasize that these experimental paradigms suggest that the expert often deploys a rapid and non-deliberate method of task performance. The timescale of the RSVP paradigm employed in the expertise-expertise inter ference studies is highly suggestive in this regard: Images are presented as rapidly as seven or more per second, and only experts suffer diminished facial recognition when the series includes images from their domain of expertise. This kind of automaticity and competition with other similarly expert-requiring representation all suggests that the expert enjoys some dis tinctive visual representation. This is further corroborated by the sensitiv ity to configural, holistic perceptible features, as evidenced by composite studies, part–whole and spatial change studies, and inversion effect stud ies. Finally, the VSTM advantage is most efficiently explained by visual rep resentation: The expert enjoys category-rich visual representation, freeing cognitive resources to enable increased short-term memory load. The sceptic might attempt to deflect this explanation by maintaining that these experimental manipulations are in fact only affecting post-visual judg ment or pre-visual voluntary attention. However, even before considering the additional neural and physiological evidence, these explanations look strained. Again, the point here is not about any single behavioural study, but about the convergence of data from the array of studies. In brief, the scep tic faces at least these challenges. She must somehow explain, as non-visual, cases of interference with face recognition, the latter of which is an uncon troversial visual capacity. And recall that these interference studies employ control paradigms that appear to undermine a mere attentional explanation. She must explain expert sensitivity to the organization of visual properties (for example, how objects of expertise are rapidly and non-deliberately con figured as perceptual wholes and how their inversion undermines holistic recognition) as non-visual. And she must explain rapid performance on per ceptual tasks (such as the RSVP tasks) as, again, a matter of post-visual judg ment or pre-visual voluntary attention. But this performance occurs on too
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
rapid a timescale to be explained as a judgment on the basis of the scene or as a deliberate shift in spatial attention. The explanatory challenge for sceptic 1 increases when we add to these behavioural data the results of the various neurological and physiological studies. There is clear evidence for changes in the relevant ERP components and in relevant neural areas FFA and OFA. These changes co-vary with acqui sition of expertise, and moreover appear to differentially activate (suffer interference) when tasks involve both faces and objects of expertise. Finally, the activation (as evidenced by the EEG recordings of the N170) occurs in fairly early visual processing, between 150 and 200 ms post-stimulus. Eye tracking results indicate important differences between experts and novices in pre-voluntary attentional selection, time to fixate relevant objects of inter est, length of saccades, and number of fixations.14 As researchers hypothe size, this indicates that experts enjoy a rapid sensitivity to holistic features of the visual array, corroborating the behavioural evidence for the same inter pretation. Importantly, these features are diagnostically relevant to the task, and diagnostic success co-varies with the degree of expertise. These kinds of data – EEG/ERP, fMRI, and eye tracking data – are stand ardly taken by our best vision sciences as measures of, or as identifying cor relates for, visual representation. This may not be conclusive, but sceptic 1 has to explain these data away – as not providing evidence for relevant visual differences between expert and non-expert. And moreover, that sceptic has to explain away the robust correlation between behavioural performance and neural-physiological measures. For example, the sceptic would have to claim of the fMRI data either that when relevant neural activity occurs in correlation with face perception that this is a genuine perceptual phenom enon, but when the same pattern of neural activity occurs in experts in response to objects of expertise, it is non-perceptual, or that in neither case is it perceptual. The first option is saddled with an unexplained asymmetry; the second is flatly at odds with what vision science says about facial recog nition. Further, the sceptic would have to claim (contrary to vision science) that rapid saccade and fixation patterns in experts amount not to distinctive visual perception in experts, but instead only to differences in the judgments made by, or voluntary spatial attention of, those experts. Given the timescale of the neural-physiological markers and the convergence of these data with the behavioural data, these forms of explanation lack plausibility and explan atory leverage by contrast to the perceptual explanation proposed here. Finally, a point about phenomenology. Subjective reports of experts indi cate visual phenomenology, and the automaticity and speed of performance
161
162
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
indicate non-inferential psychology. And the various studies discussed here suggest that experts enjoy greater sensitivity to holistic, complex configu ral and spatial relations of objects of expertise (and, again, on a timescale corroborated by the neural-physiological studies). This implies a difference between expert and novice in visual phenomenology: a sensible difference in how those objects are configured or organized and how they enjoy greater salience in the relevant contexts. These are differences in what it’s like for the expert versus the novice. One might resist any such isolated case and the explanation offered. But taken together, the convergence of this evidence grounds a strong case for a visual perceptual explanation. The concluding architectural claim 1 is that, con trary to sceptic 1, perceptual expertise is a genuinely perceptual phenom enon. This explanation takes seriously the perceptual in perceptual expertise.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.4 Perceptual expertise as genuine cognitively sensitive expertise The second question is whether this kind of expertise, with the perceptual features just described and defended, is sensitive to high-level cognition, or is it just a general improvement to perception? This is a question about mental architecture, but one of epistemic importance to theories of percep tion. There is an important difference between perception improving as a consequence of training and in ways that are sensitive to domain-specific conceptual content versus just improving as a developmental or biological matter. Sceptic 2 attempts to motivate the second answer, claiming that the relevant phenomenon is not genuinely expert-involving: High-level or cog nitive learning is not needed to explain the various data. What motivates this claim? Perceptual systems develop, learn, and adapt to perceptible kinds after repeated exposure and by consequence of evolutionary advantage. Thus, one might attempt to explain perceptual expertise as a purely intra-perceptual phe nomenon. Sceptic 2 grants that the phenomenon is genuinely perceptual, but maintains that it involves some kind of purely perceptual learning, or adap tation or, most simply, perceptual development.15 Accordingly, there is no need for explanatory appeal to what expert subjects know or believe or what they have learned about the specific content of the domain of expertise. Contrary to sceptic 2, the architectural claim defended in this section is that perceptual expertise is genuine, content-sensitive expertise. The per ceptual effects described in Section 6.3 non-trivially depend upon cognitive
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
learning: information about categories, diagnostic detail, goals and tasks within that domain. It should be emphasized here, and will be further clari fied and justified later, that this claim is neutral with respect to the particular means or mechanism whereby expert learning or cognition affects visual perception of experts. Instead, architectural claim 2 embraces an explana tory pluralism, and that pluralism opposes a disjunction of possible claims made by sceptic 2. Broadly, sceptic 2 denies that cognition or higher-level mental content can have a significant effect on vision (and perhaps perception generally). For different theorists, this will come to different claims. Some will claim that cognition does not penetrate perceptual experience or perceptual process ing. Some will claim that there are no top-down effects on visual processing (synchronically or diachronically). Others will claim that there is no topdown modulation of rapid visual attention or visual fixation or saccades. The conclusion of this section, architectural claim 2, can be interpreted as denying each of these sceptical claims but without committing to a sin gle means by which relevant cognitive-perceptual effects are achieved. Here again the inference structure is abductive, ranging over behavioural and neural-physiological studies. First, given objects of expertise like birds and Greebles, one might be tempted to claim that expertise advantages all depend on stimuli that are configurally face-like, and it is already well-established that humans have some kind of perceptual advantage for faces. This claim is easily dispelled. Many objects of expertise are radically, configurally distinct from faces: cars, radiological images, trees, fingerprints, musical notation. So whatever one says about the range of expertise, it is not plausible that experts are just transferring a general face perception sensitivity, after repeated exposure, to face-resembling objects. This does raise, however, an interesting set of questions about the nature of expertise training. Real-world, “naturally” acquired expertise nearly always involves training of techniques, loaded with semantic content, categories, and high-level con cepts. This is no less true of in-lab expertise training. A study by Tanaka et al. (2005) highlights various important features of expertise training, empha sizing both that explicit auditory feedback is required, while mere exposure is insufficient to generate expert performance and that category-specification must occur at a grain finer than the basic level (for instance, in bird training, subordinate category terms and family-level terms – “eastern screech owl” or “wading bird” – are needed for significant expertise effects).16 Similar features of training have been observed in experts trained in non-laboratory
163
Copyright © 2021. Taylor & Francis Group. All rights reserved.
164
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
contexts. For example, radiologists receive explicit training, where success admits of degree and mere exposure provides no diagnostic advantage (for instance, x-ray technologists regularly exposed to but with no training in diagnostics [“reading”] for radiographic images, perform below chance on some measures) (Nodine et al. 1999). One should not infer from this, though, that the expert is just learning how to label things. First, as one can glean from the studies just discussed, not all labels make a difference in performance (basic-level labels do not). These results have been correlated with EEG results. The ERP component N250 has a typical onset latency between 230 and 330 ms post-stimulus and is standardly correlated with familiar or learned objects. For example, this activity in visual cortex spikes, at this temporal scale, in response to one’s own face. What Scott et al. (2008) found was that this ERP activity results from training only when subordinate category information is learned, not with mere exposure or basic category training. For example, in car expertise training, changes in N250 response occurred only when subordinate-level category training was involved (information about car make or model). This suggests that subjects get the “familiarity response” in visual cortex only when they have learned categorical, domain-specific information (and again not as a consequence of mere exposure).17 The Scott et al. studies also corroborated something that anecdotal evi dence already suggests, namely, that expertise is stable across time and across novel task performance. Within their domain of expertise, experts can read ily identify novel exemplars of objects in relevant categories and make fine discriminations between individuals within categories. Scott and colleagues trained subjects for car expertise using subordinate category vs. basic category only vs. exposure only training structures. They measured subjects’ perfor mance pre-training, immediately post-training, and one week post-training. Subjects trained on subordinate categories are the only subjects to make sig nificantly successful discriminations, and this performance persists through the one-week post-training measures. This performance correlates with an ERP effect: Only the subordinate category training results in increased N250 response, and this also persists through the one-week post-training meas ures. This suggests a lasting and stable perceptual sensitivity to features that are learned as distinctive of members of fine-grained categories. Further, these behavioural and neurological results generalize. Lab-trained experts successfully discriminate untrained (not previously viewed) exemplars within fine-grained categories of cars and birds (Scott et al. 2008) and Gree bles (Gauthier et al. 1998). And this shows in further EEG results: Both the
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
N170 and N250 responses generalize to untrained exemplars of the trained categories. Finally, although expertise generalizes to within-domain stimuli, it typi cally does not transfer to or from other perceptual domains. One might be tempted to claim that these effects just derive from a general improvement of solely perceptual skill (say, sharpened attention or enhanced visual acu ity) or a pre-expertise superior perceptual skill, but various studies suggest otherwise. First, there is ample evidence that low-level perceptual learn ing rarely generalizes to other stimulus types.18 Additional studies suggest the same for perceptual expertise. Nodine and Krupinski (1998) compared performance between expert radiologists and non-experts on complicated visual search tasks over non-radiographic images (Where’s Waldo? images and Hirschfield’s “Nina” drawings). The radiologists perform no better than non-experts in these tasks. Therefore, the expert’s skills appear not to trans fer outside the domain of expertise.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
* Again this diversity of data can be unified in an explanation. Here the question is whether perceptual expertise is genuine expertise: Is it a partly learned, cognitive phenomenon? Sceptic 2 claims not: The effect is simply a development or low-level enhancement of perceptual systems. But this posi tion poorly explains the array of studies and effects just discussed. Expert training, in and outside of laboratory circumstances, requires more than mere exposure to stimulus types (in contrast to many examples of lowlevel perceptual learning, such as colour discrimination). And it requires explicit uptake of domain-specific categories and information, where this training is effective and the neural-physiological markers evident, only when the learned categories are fine-grained. Expertise is stable across time, once acquired, and generalizes to previously unperceived, novel exemplars of a category. Finally, these skills do not transfer to similarly complex perceptual tasks outside the domain of expertise. Plausibly, expertise effects such as these would not occur if vision were a strictly passive process of sensory reception. Instead, the simplest, most parsimonious explanation is that the differences in perception between experts and novices depend in substantial ways on the specific conceptual content that is acquired over the cognitive learning period. The concluding architectural claim 2 is that perceptual exper tise is a genuinely cognitive phenomenon, dependent on the conceptual information of the domain of expertise. This explanation takes seriously the expertise in perceptual expertise.
165
166
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.5 Situating perceptual expertise in theories of the architecture of mind For all that has been said, the architectural claim (or conjunction of claims) is a broad one: Many instances of perceptual expertise, such as those just dis cussed in detail, involve cognitively sensitive perceptual changes and differ ences. But no claims have been made about what more precise structure or mechanism is involved across the array of cases. Thus, some cases of exper tise may involve cognitive penetration of perception, some (perhaps non-ex clusively) may involve high-level perceptual content, others differences in mechanisms of selective attention. This pluralism is deliberate. There is little reason to think that all instances of perceptual expertise take the same struc ture. While there is a convergence of behavioural and neural-physiological evidence across many domains of expertise, there are also important differ ences across those domains. Some such domains require visual search as the dominant task, others involve object categorization, others still some kind of perceptual matching or reidentification of an individual stimulus. It is plau sible, then, that the relevant cognitive-perceptual effects might be achieved differently across different domains. The claim meant to take broad scope across these varied domains, argued here and contrary to a disjunction of claims made by sceptic 2 (or distinct sceptics), is simply that many cases of perceptual expertise are both richly perceptual and importantly cognitively sensitive. The importance of perceptual expertise, explained here as a cognitively sensitive perceptual phenomenon, does not depend for its theoretical value upon cases of perceptual expertise “counting” as cognitive penetration of perception. For that matter, that importance does not depend upon a defini tive refutation of the modularity of mind. That said, we should ask how this analysis situates in that debate and, accordingly, what its possible implica tions are for modular architectures. Recall how proponents of the latter view have treated related instances of expertise. Fodor discounted painters and expert phoneticians, granting that they may enjoy perceptual representational capacity distinctive to their domains, as too “highly skilled” to teach anything about “normal perceptual processing”. This dismissal is far too casual. Bear in mind that on one plausi ble and literal reading of Fodor’s theory of modularity, informational encap sulation is the “essence of”, and therefore necessary for, modular perceptual systems. Fodor’s admission regarding experts therefore comes to this. Either the expert has somehow developed a skill that puts her (or her relevant
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
perceptual systems) outside the realm of “normal” human perceivers or she has not. The first would amount to a remarkable (and rather pervasive) super-human, perceptual transcendence, and the second to straightforward counterexamples to the encapsulation necessary for perceptual modules. Fodor can’t have both, and he should want neither.19 Pylyshyn (1999) takes cases of perceptual expertise (and learning) less casually, but the casualty is the same. Recall that he suggests that empir ically studied perceptual experts do not enjoy cognitively sensitive per ceptual representation, but instead have developed (presumably rapid) strategies for directing spatial attention in behaviourally relevant ways. The chess master or elite tennis player “knows where to look”, as the famil iar idiom has it. This marks two of the three assumptions criticized in Chapter 5 – the attention as act and attention as spotlight assumptions. Further, Pylyshyn is explicit about the attention as gatekeeper assumption, describing spatial attention as the “interface between vision and cognition” (1999: 360; see also Firestone and Scholl 2016: 13). So, we have seen this kind of attention-based strategy for dismissing cases of cognitive penetra tion before. It grants that the knowledge base, say, of the expert is affect ing perception, but only through the mediation of agent-directed, spatial attention. For the very same reasons given in the foregoing analysis, this strategy is implausible. Recall that the cases of visual expertise discussed in the previous sec tions often involve rapid behavioural responses, both in terms of identifying or categorizing a target and in terms of eye movement patterns. Indeed, experts describe their performance as automatic and in visual terms such as “pop-out”, “just seeing”, and “salience”. And those behavioural meas ures are corroborated by neural changes recorded using both fMRI and EEG measures. None of this precludes a role for attention in the expert’s perfor mance. But just as we saw in the previous chapter, it is implausible that the attention involved is anything like an overt shift in spatial attention. It is far more plausible that the mechanisms involved are selective ones like featureand object-based attention, which can certainly be shaped by an agent’s past learning (think Where’s Waldo? puzzles) but are not in any relevant sense agent-directed or an action. And recall, finally, that there are ample reasons to characterize these attentional mechanisms as part of vision rather than as an interface or gatekeeper between vision and cognition. If this is correct, then Pylyshyn’s attention-based strategy for debunking cases of expertise fails, and it fails because it relies on three problematic assumptions which, once again, we have reason to abandon.
167
Copyright © 2021. Taylor & Francis Group. All rights reserved.
168
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Therefore, instead of providing a way to maintain encapsulated modules in the face of instances of perceptual expertise, Pylyshyn’s attention-based strategy points to a possible argument against that very modularity. This is the argument from directness given in Section 5.3.1. At least with some instances of expertise, the cognitive learning enjoyed by the expert and the domain-specific information stored in the memory of that expert modu late selective attention. And these attentional mechanisms are best theorized as part of, rather than a gatekeeper to, vision. This is a direct cognitive (and internal mental) effect on perceptual processing. Ergo, this is cognitive penetration, and is a violation of the encapsulation required of perceptual modules. Just as we saw in that discussion, however, one may dig in their heels and insist that attention stands as a partition between cognition and per ception. That rejoinder can be faced with the second argument, an ana logue to the argument from consequences given in Section 5.3.2. Recall that the basic thought behind the consequentialism of Chapter 4 was that what really matters to both parties in the cognitive penetrability debates are the consequences of the possible phenomenon. Focusing on the consequences for architectures of mind, what does the analysis given here imply? (The epistemic and behavioural consequences of perceptual expertise are the subject of Chapter 7.) Whatever one says about attention and how it situates with respect to cognition and perception, the kinds of experts discussed here seem not to enjoy informationally encapsulated visual processing. The arguments given in Sections 6.3 and 6.4, respectively, concluded that the expert enjoys distinctive visual representation and that those distinctive perceptual phenomena depend non-trivially upon the cognitive informa tion the expert has learned in her domain(s) of expertise. The arguments were both abductive, relying on a rich variety of evidence: subjective report, behavioural study, eye tracking data, and neurological data. The conver gence of these data strongly suggests that perceptual experts perceive dif ferently because of their cognitively acquired expertise. If one accepts these conclusions and their supporting evidence, then it is implausible to main tain the informational encapsulation of perceptual systems. And this is true no matter whether the phenomena involve the “mediation” of attention. If it is the consequences that matter, then this is cognitive penetration by virtue of bearing deep implications for architectures of the mind. Ergo, modularity fails. Things can be, and should be, spun in a more positive light. The real spirit of the consequentialist approach is to return emphasis to the questions that
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
matter most to us as theorists of the mind, and as humans with minds. And the spirit of this chapter and those that remain is to shift away from offering theoretical proposals, the success of which depend upon proving cognitive penetration or disproving modularity. “Cognitive penetration” and its cog nates are just terms, and unfortunate ones at that. If they mark something, they mark possible cognitive-perceptual phenomena that would have deep import for how we theorize the structure of the mind and how those struc tures put us in contact with reality; they mark phenomena of interest for theories of the architecture and the epistemology of mind. An emphasis on perceptual expertise sheds new light on both questions. This chapter focused on the first, on how careful attention to perceptual expertise should change how we think about the structure of the mind. The next chapter focuses on the second, on how analysis of those same phenomena should change how we think about the epistemology of perception.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
6.6 Summary: towards the malleability of mind This is theoretical progress. Perceptual expertise is an important, and importantly overlooked, way that thinking can influence perceiving. That is the broad conclusion of this chapter; it is further evidence of the TaP thesis. And the conclusion of the following chapter will be that this same phenomenon is a remarkable (and again remarkably overlooked) way that an individual’s thoughts and cognitive efforts can enhance her perceptual skills in epistemically important ways. This is the TiP thesis. Oversimplify ing, these are the descriptive and normative components of a theory of how thought may affect perception, of how the mind is richly malleable. And the approach is novel insofar as it is positive in its emphasis, moving away from the dominant emphasis on negative or neutral cases in philosophy and (much of) the science of the mind. We move, then, to how thinking improves perceiving. Further reading Kanwisher et al. (1997) is in many ways the seminal paper on the FFA and face perception. Gauthier et al. (1999) presents a clear counter to Kanwish er’s view. For reviews on perceptual expertise, see (Bukach et al. 2006; Curby and Gauthier 2010; Scott 2011). For a new volume on expertise and skill, see (Fridland and Pavese 2020), and for discussion of much of the contents of this chapter see Stokes (2020a).
169
170
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 For that matter, even the choice of term “cognitive penetration” suggests something negative or problematizing. 2 Opposite Gauthier and her collaborators’ claims about FFA/OFA is the work of Nancy kanwisher’s lab at MIT, which maintains that these neural areas are genuinely face-specific. In addition to the landmark kanwisher et al. (1997), see (Grill-Spector et al. 2004; kanwisher and Yovel 2006; Mckone et al. 2007). 3 For an early example of the (then) unorthodox answer to this question, see Gauthier and Logothetis (2000). 4 Put another way, such cases would present a problem for the argu ment from constancies for informational encapsulation (which was a sub-argument to the argument from the stability of perception, discussed in Section 2.2). Cases of expertise would seem, by Fodor’s admission, to either be counterexamples to that argument (since they can “override” the constancy mechanisms that are supposed to be encapsulated), or any case of expertise is simply outside the scope of that argument. 5 It is worth noting, however, that Pylyshyn’s analysis predates the bulk of research on perceptual expertise as such, much of it to be discussed in what follows. 6 To be clear, these claims are existentially quantified: Some perceptual experts (with emphasis on some of those domains that have received extensive empirical study) enjoy cognitively effected perceptual enhance ment, and in many cases this is an epistemic good. 7 This sceptic thus commits to the pernicious cognitive effects assumption. 8 whether the phenomenon is one that genuinely involves cognitive learn ing, sensitive to conceptual content of the relevant domain(s), is the ques tion to be addressed in Section 6.4. 9 As indicated earlier, this is the explanation that Pylyshyn favours for some empirically studied instances of expertise (1999: 358–359). 10 Pop-out selection is an aspect common to both bottom-up and top-down visual attention. One well-studied example of the latter is the feature-based attentional selection mechanisms (FBA) discussed in the previous chapter. Recall that FBA can proceed independent of spatial attention, selecting features, for example, colours or shapes, that are behaviourally relevant. A familiar example: Suppose you are searching a crowded baggage ter minal for your luggage and your luggage is bright pink. FBA makes this
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
11 12
13 14
15
Copyright © 2021. Taylor & Francis Group. All rights reserved.
16 17
visual search markedly easier by selecting pinkish items in the visual array such that those pink features will enjoy greater salience – pop-out – in your visual experience. Issues concerning face perception and race are discussed in Section 8.3. See (Gauthier et al. 1998; Freire et al. 2000; Barton et al. 2001; Leder et al. 2001; Gauthier and Tarr 2002; Busey and Vanderkolk 2005). For review, see Bukach et al. (2006). For similar findings on car experts, see Rossion et al. (2007). Note, then, that explanations of some of these phenomena will plausibly involve attention but not voluntary spatial attention, but instead behav iourally sensitive (and typically involuntary) feature- or object-based atten tion (as was discussed in Chapter 5). See also Note 10. For extended discussion of perceptual learning and how it engages some of these questions, see Connolly (2017). Perceptual learning will also be discussed in Section 7.3. See also Gauthier and Tarr (1997) and Scott et al. (2008). Similar results are found in research on elite athletes. As one would predict, elite athletes are exceptionally good at recognizing patterns of play in their respective sport. But one would be wrong to infer that this is a matter of mere exposure to game scenarios, and moreover would be wrong to think that it’s just bare pattern recognition. Instead, expert performance in pat tern recognition is especially sensitive to genuinely possible play sequences or configurations for the sport. even after regular exposure, if a sequence of player movements is random – not one that would make sense or be allowed given the rules of the sport – then pattern recognition rapidly declines (Allard et al. 1980; williams et al. 2011). And expert advantages are highest in ecologically valid contexts and decline in ecologically invalid contexts. Further, those advantages occur on a sliding scale, from in-field contexts/tasks, down to video presentation tasks, down to tasks involving presentation of static slides. So while exposure does matter for pattern recognition, it is insufficient to produce expert performance. Perceiving the relevant patterns, sometimes extremely rapidly, appears to be cognitively modulated, sensitive to knowledge of and memories about the game. This informs the kinds of training interventions that are used and used most effectively. Interventions that involve some cognitive content (e.g. both explicit instructions and guided discovery methods) are more effective than interventions that lack that content (e.g. mere discovery or uncoached
171
172
PeRCePTuAL eXPeRTISe I: ARCHITeCTuRe
Copyright © 2021. Taylor & Francis Group. All rights reserved.
practice) (Smeeton et al. 2005). Narrative about strategy, patterns of play, and so on make a difference not just to how the athlete plays but to how accurately and quickly she responds perceptually to relevant visual stimuli. 18 Ahissar et al. (1998), Fiorentini and Berardi (1980), and Poggio et al. (1992). 19 To foreshadow, perceptual expertise is, or at least this will be argued in Chapters 7 and 8, pervasive in human perception. Statistically speaking, it is not special or abnormal. Many of us become better perceivers in dis tinctive domains of activity, skill, and knowledge. Does this mean that we, in such instances, transcend our “normal” (modular) perceptual systems? No. It means that our perceptual systems are not modular, at least not in the strongest of senses, but instead are malleable and may improve in person-sensitive ways.
7
PERCEPTUAL EXPERTISE II:
EPISTEMOLOGY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
7.1 7.2 7.3 7.4 7.5
An epistemology of perceptual expertise: initial sketches Expertise and epistemic virtue Perceptual improvement and biological function Objections and qualifications Summary: thinking improves perceiving
The traditional picture of perception, traceable at least to Aristotle, is one of a passive faculty that provides purely stimulus-driven (re)presentation of the subject’s present environment. Although there are important departures from this tradition, some recent models of perception in cognitive science still capture this same spirit. Most dominantly, modular theories of percep tual systems maintain that stimulus-driven, proprietary input is processed by, say, vision in a way that is informationally encapsulated from beliefs, goals, and other cognitive representations in the overall mental system. One explicit motivation for modularity of this sort is that this kind of functional independence would better ensure reliable, accurate perceptual representa tion. It gives us, as it is sometimes put, objectivity. The analysis given in Chapter 6 (in addition to those of previous chapters) suggests a widespread challenge to this kind of mental architecture: Cases of perceptual expertise
Copyright © 2021. Taylor & Francis Group. All rights reserved.
174
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
are candidate counterexamples to the encapsulation required of strongly modular perceptual systems. One may then reason that any such top-down cognitive influence on perception will be epistemically problematic, since this undermines the independence of (and, thereby, the reliable accuracy of) perceptual representation. This is the challenge of sceptic 3 and is the pernicious cognitive effects assumption. It is worth identifying how this argument relates to the argument from reliability of Chapter 2. That argument went as follows. Perceptual rep resentation can be largely accurate only if perceptual systems are informa tionally encapsulated. Perceptual representation is (or must be) accurate. Therefore, perceptual systems are informationally encapsulated, that is, modular. Depending upon how the second premise is motivated (and thus whether it is put as a descriptive claim or a claim about necessity), the argu ment can take either an empirical or a transcendental form. One might think that observation tells us that perception is largely accurate. Coupling this with the first conditional premise, one gets to the modularist’s conclusion by inference to the best explanation: An encapsulated visual system would better ensure that visual experiences accurately present the environment. Or one might think that we simply must avoid the sceptical results that follow if the necessary condition for reliable perception is not met, and so one takes for granted that perception must be accurate. Put another way, we start with the assumption that perception provides knowledge, and to do this it must be reliably accurate. Either way, the modularist’s conclusion follows. One person’s transcendental argument is another person’s dogma. Indeed, one might think that this is just a way of ignoring the philosophical sceptic, rather than a substantive rebuttal. And on the empirical end of things, we have good reason to doubt that perception is sufficiently and consistently accurate to motivate a conclusion about encapsulated perceptual systems. One might think here of familiar motivations for Descartes’s scepticism or of the discussion of negative cases of error, illusion, and hallucination that still populate philosophical and psychological theorizing of perception.1 Sceptic 3, then, does not take for granted that perception is sufficiently accurate. And he takes seriously the case made against encapsulation argued in the previous chapters, acquiescing that perceptual systems are not encapsulated. When this last claim is then coupled with the conditional – that perceptual representation is largely accurate only if it is encapsulated – one lands on a worrisome philosophical scepticism after all. Counter to this sceptical challenge and the pernicious cognitive effects assumption, the conclusion here is that perceptual expertise is epistemically
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
enhancing, not pernicious or downgrading. These are instances where think ing improves perceiving. In many ways, this is just a straightforward exten sion of the two architectural claims, and their broadly empirical motivation, of Sections 6.3 and 6.4. There the case was made that perception can actively change – neurophysiologically, representationally, phenomenologically – and in ways sensitive to domain-specific information, that is, to high-level cognitive learning. If successful, this provides the groundwork for a general epistemic claim. The present chapter proceeds as follows. In Section 7.1, some basic epis temic characterizations are given to perceptual expertise understood in terms of the architecture defended in the previous chapter. Section 7.2 then turns to the most promising way to think about the epistemology of per ceptual expertise, namely in terms of epistemic virtue. Section 7.3 couples that analysis with a functional-teleological analysis of perceptual systems. Section 7.4 considers some rejoinders to the epistemology (and underlying architecture) of perceptual expertise offered.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
7.1 An epistemology of perceptual expertise: initial sketches For a number of epistemological theories, the epistemic goodness of percep tual expertise comes easy. What the data show, if they show nothing else, is that experts perform more accurately, more rapidly, with less cognitive effort, and in ways that present advantages for working memory. Expertise is therefore an epistemic good simply because the expert is moving closer to an optimal cognitive stance on the world (or a part of it), where she can bet ter acquire behaviourally relevant category and diagnostic information. This may not be the most richly normative epistemic model, but it stresses the informational connection between the cognitive background and perceptual performance of experts. The expert, whether in laboratory or real-world cir cumstances, undergoes a laborious training regimen, sometimes spanning years. And it is as a causal consequence of this deliberate, highly cognitive training that the expert enjoys perceptual experiences, of domain-specific objects, features, and patterns, rich in information that is missing in the novice’s experiences. This information is behaviourally relevant but not in idiosyncratic or egocentric ways. Whether the expert is diagnosing a cancerous tumour in a mammogram or making a fingerprint match, there are clear and standard conditions for accuracy. What’s remarkable – epis temically remarkable – is that these informational patterns are picked up,
175
Copyright © 2021. Taylor & Francis Group. All rights reserved.
176
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
seemingly automatically, by the perceptual systems of experts. Therefore, on marks of both accuracy and efficiency, the expert is cognitively optimal (or nearing it). This basic characterization can be extended in line with a variety of epis temological theories. Three brief sketches are offered in this section before turning to a more substantive treatment in terms of epistemic virtue. Epistemic internalism, understood as a broad category of theory of epis temic justification, maintains that a subject’s belief is epistemically justified just in case the justifier or grounds for belief is some internal mental state(s) of the subject. Put strongly, this requires that the subject has some evidence base – say a prior belief that q – which stands in some appropriate evidential relation with the belief that p, and she recognizes how or that the former belief is a reason for the latter. Alternatively, one must at least have access, in principle, to the justifying belief or reason. Thus one could, perhaps if appro priately prompted, identify one’s reason (say, the belief that q) for believing that p, and it is that potential access that gives one (or one’s belief) the status of being justified. The most straightforward application of this theory to perceptual expertise, understood as analyzed in the previous chapter, is to the perceptual experiences that ground the beliefs or judgments formed on their basis. When the radiologist confidently identifies a tumour in a mam mogram, or a forensics expert makes a fingerprint match, they do so on the basis of their respective visual experiences. Plausibly, given the automaticity of this performance (for instance, how rapidly an expert can successfully identify regions of interest in a radiogram), these experts do not always iden tify their perceptual grounds for judgment (at least not in any explicit way). However, it is plausible that they sometimes have the relevant access required of internalism, and in at least two ways. First, at least in those cases that are not extremely rapid (for instance, excluding forced-choice cases where radi ologists can successfully report an anomaly in as little as 200 msec but not be able to point to it), experts can identify what they saw and why or how it is relevant to the judgment made. One might think, for instance, of a case where a colleague asks how they “know” that it is a tumour in such-and such location. Second, it is also plausible that such experts can identify and report, if prompted, features of their training, background knowledge base, understanding of relevant technology, and so on. And the account given here has it that these are crucial influencing factors to the perceptual experiences that experts enjoy. Therefore, they also have epistemic access to the etiolo gies of those experiences to some degree, and this would plausibly serve to provide internalist justification for the judgments or beliefs formed.2
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
A related, and more provocative, epistemic treatment of these cases focuses not just on the perceptual experiences as justifiers for resultant belief or judgment, but on the epistemic evaluability of the experiences them selves. Traditionally, experiences are taken to potentially justify consequent beliefs but are not themselves the sorts of mental state that are justified. But the situation changes if we accept the mental architecture of expertise given here. Recall that the pair of claims is that perceptual expertise is a genuinely perceptual phenomenon, and further that it is richly, causally sensitive to the domain-specific cognitive states and processes of the perceiver. In short, one’s expert, partly cognitive training influences one’s perceptual experiences in the relevant contexts. The reason this forces a departure from the traditional view is that there is a clear sense in which the perceptual experiences of the expert are her doing: They depend upon what she has done in training to become an expert. And the accuracy of those experiences, and the success of the behaviours they lead to, vary accordingly. (By contrast, the traditional and modular views of perception assume that perceptual experiences are in no relevant sense the perceiver’s doing. Accordingly, on such views, percep tual experiences seem not to be good candidates for epistemic evaluation; they cannot be justified or unjustified.) If we accept the cognitive-perceptual explanation of perceptual expertise given in the previous chapter, then for at least some cases, it will make good sense to characterize the experiences themselves in epistemically valenced terms. One recent epistemological account that does just this is given by Susanna Siegel (2017, 2018). She describes the account as follows: The Rationality of Perception thesis says that perceptual experiences can, in themselves, manifest an epistemic status. They are not merely an ena bling condition for other mental states to manifest such a status. Nor are they merely contributors to determining the epistemic status of beliefs as either well or poorly justified, though they play this role as well. The conceit of The Rationality of Perception is that processes that occur within a subject’s own mind and are of her own mental doing are para digms of rationally appraisable processes. If such processes culminate in perceptual experiences, then those perceptual experiences are appraisable as well. (Siegel 2018)
Siegel (2017) provides a detailed model of, and proof of concept for, the rational evaluability of perceptual experiences themselves. The claim is that
177
Copyright © 2021. Taylor & Francis Group. All rights reserved.
178
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
perceptual experience could be rational (or “reasonable”) not in the more specialized sense that it involve, or result from, conscious deliberation, but in the broad sense that the etiological route to that experience be rationally appraisable and, thereby, the experience itself can be “evaluable as ration ally better or worse” (Siegel 2017: 18). Inference plays an important role in Siegel’s account of perception: When the inputs to inference are good and the inferential structure sound, then beliefs in response to that inference are rational. On this account, inference is a mental process, one that involves a response to some informational state (a belief, a supposition, an experience) that results in a conclusion. Importantly, Siegel argues against the more tra ditional model that requires that an inference involves “reckoning”, which involves identifying some kind of support relation between the informa tional state and the conclusion drawn. “If perceptual experiences can arise from inference, then the Rational ity of Perception is true” (19). The argument for the rationality of percep tion then proceeds by identifying plausible cases of inference understood in Siegel’s way (for instance, categorizing someone as being kind on the basis of their appearance), where one could achieve the relevant response to the rel evant informational state(s) but without being able to identify the epistemic support relation that moves from the informational state to the conclusion (that the person is kind). If the argument succeeds, this opens up conceptual space for perceptual experiences being the mental response to an inference. Siegel then extends this model of inference to perception and provides an account of the implications for the epistemic valence of inferentially modu lated perceptual experiences. What the architectural analysis given here does is supplement Siegel’s epistemological model (which is defended as a conceptual possibility, not as a descriptive claim about human psychology) with a range of empirically grounded cases. Moreover, cases of expertise are good-making cases, while Siegel’s central emphasis is admittedly on cases where things go epistemically wrong. For the very reasons Siegel gives, experiences of perceptual experts contribute in a positive way to their rational standing as epistemic agents. The most standard opposing alternative to epistemic internalism is, as many readers will know, some form of reliabilism. In ecologically valid cir cumstances (when objects aren’t inverted, misaligned, and so on), experts statistically, significantly perform better, and thereby form true beliefs with greater frequency, than novices. And this is true across the many and widely diverse domains of expertise studied. Put in old familiar terms of process reli abilism, perceptual expertise involves an enhancement to the already reliable process – perception – of forming beliefs about the world (Goldman 1979).
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
This leaves open precisely how perception is improved in these ways – for example, by enjoying richer contents or being subject to top-down atten tional effects – while maintaining that the epistemic advantage is at least partly a perceptual one. Put most simply, expert perception is optimally reliable in the relevant domain. These epistemological sketches could obviously be more fully detailed, where decisions on various choice points would result in a rich variety of epistemic accounts. What each of the approaches lack, however, is a suffi ciently robust emphasis on the role of the expert, qua agent, in the epistemic successes of her expertise. The most natural and illuminating way to high light the agent is to turn to theories of epistemic virtue.
7.2 Expertise and epistemic virtue
Copyright © 2021. Taylor & Francis Group. All rights reserved.
An archer’s shot can be assessed in several ways. It can be accurate (successful in hitting the target). It can also be adroit (skillful or competent). An archery shot is adroit only if, as the arrow leaves the bow, it is oriented well and powerfully enough. But a shot that is both accurate and adroit can still fall short. Con sider an adroitly shot arrow leaving the bow with an orientation and speed that would normally take it straight to the bull’s-eye. A gust of wind then diverts it, but a second gust puts it back on track. This shot is both accurate and adroit, but it fails to be apt. A shot’s aptness requires that its success be attained not just by luck (such as the luck of that second gust). The success must rather be a result of competence. (Sosa 2015b)
Sosa’s favoured archer example illustrates the general view that he has defended now for several decades, a virtue-based account of knowledge, of justified belief, and rebuttals to the philosophical sceptic. Archery, a per formance with an aim, can be assessed along three dimensions. Success at archery will enjoy “accuracy: reaching the aim; adroitness: manifesting skill or competence; and aptness: reaching the aim through the adroitness manifest” Sosa 2007: 23). The successful archer is herself to be credited for such performance. Sosa suggests that this generalizes to any performance that has an aim, and most certainly to cognitive performances. Analogously, according to virtue epistemology, agents themselves, rather than their doxas tic states or commitments, are the primary targets for epistemic evaluation. It is because this style of account attributes the epistemic advantage to the expert as an epistemic agent that it so fruitfully illuminates an epistemology of perceptual experts. The claim to be defended is that some perceptual experts exhibit remarkable instances of intellectual virtue.
179
180
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
There are multiple variations on this broad category of epistemologi cal theory. However, a feature common and central to such theories is the emphasis on intellectual virtue. Agents are virtuous if and to the degree they exhibit intellectual virtues. And a cognitive process is virtuous just in case it is or results in some relevant excellence in performance. Here are three more substantive characterizations of virtue, meant to include intellectual virtue in particular: [A] competence [virtue] is a disposition, one with a basis resident in the competent agent, one that would in appropriately normal conditions ensure (or make highly likely) the success of any relevant performance issued by it. (Sosa 2007: 29)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
[I]ntellectual virtues are innate faculties or acquired habits that enable a person to arrive at truth and avoid error in some relevant field . . . Aristotle defined “intuitive reason” as the ability to grasp first principles and “sci ence” as the ability to demonstrate further truths from these. Some con temporary authors add accurate perception, reliable memory, and various kinds of good reasoning to the list of intellectual virtues. (Greco 2002: 287) A virtue, then, can be defined as a deep and enduring acquired excellence of a person, involving a characteristic motivation to produce a certain desired end and reliable success in bringing about that end. . . . [A] moti vation is a disposition to have a motive; a motive is an action-guiding emotion with a certain end, either internal or external. . . . The definition is broad enough to include intellectual as well as the traditional moral virtues. (Zagzebski 1996: 137)
These definitions represent and highlight different features of distinctive theories. For instance, Greco’s emphasis is on reliable faculties like vision or deductive reasoning. Zagzebski’s emphasis is on character traits, like open-mindedness or conscientiousness. This leads to two opposed types of virtue theory: virtue reliabilism (with which Greco and Sosa are typically associ ated) versus virtue responsibilism (with which Zagzebski and others are typically associated).3 The putative opposition between these two theories is that the reliabilist characterizes virtue as an excellence that anything with a function
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
may exhibit. Thus the archer, and her arrows, could in principle have virtues. The responsibilist, by contrast, places greater emphasis on the individual, such that virtues are more closely tied to the self and to motivation. This is not the space to adjudicate this debate, and many of its details are not of great consequence for providing an account of perceptual expertise (even if they may be of great consequence for other theoretical questions). And indeed, some have argued that it is an unproductive mistake to assume that one of these virtue theories is “the correct” one.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
[O]ne way to be an excellent thinker is to reliably get the truth: to have relia ble faculties of vision, induction, deduction, and the like. Virtue-reliabilism explains the widespread intuition that good vision and memory and skills in critical thinking are cognitive excellences. Another way to be an excellent thinker is to possess virtuous motivations and perform virtuous actions: to have the character traits of open-mindedness, intellectual courage, and conscientiousness. Virtue-responsibilism explains the widespread intui tion that when it comes to active inquiry, we admire people who act appro priately and care about getting the truth. (Battaly 2008: 651)
Battaly’s diagnosis highlights a relevant feature each of the opponent views. Virtue-reliabilism places emphasis on the kinds of faculties that interest us: high-level, domain-specific cognitive learning and how this influences perceptual faculties. And virtue-responsibilism places emphasis on how the agent herself acts and how the motives of and results of those actions are praiseworthy. All of this background in place, then, how can we think about perceptual expertise as epistemic virtue? We can first note that perception is already truth-conducive. It is a faculty virtue that leads to intellectual flourishing. It is very probably required to be a good thinker. What the expert has done is to become a better perceiver and in ways in line with the epistemic norms and theoretical (or otherwise cultural) aims of her domain of expertise. For the statistical reasons here discussed, perceptual experts significantly enhance the accuracy and utility of relevant perceptual faculties. To be clear, if one does not think that perception is an intellectual virtue in general (as some virtue-responsibilists would maintain), one is motivated nonetheless to say that the perceptual performance of the expert is a virtue, since it issues from the training and cognitive goals that that agent has and pursues. In either case, the expert herself is to be credited as developing a perceptual skill that is
181
Copyright © 2021. Taylor & Francis Group. All rights reserved.
182
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
an exceptional case of intellectual virtue. And the fact that it is a perceptual phenomenon here is crucial, since this enables optimal performance in the relevant domain. The radiologist has to be able to identify an anomaly in a visual array, the fingerprint examiner must be able to visually recognize an individual fingerprint, the composer must distinguish the sound of the viola from the violin in the symphony performance, and so on. Virtue in such domains, or some of the highest virtues, in some sense must be perceptual. There are some additional features of virtue theory, again represented in the brief passages provided, that illuminate how to think about the episte mology of perceptual expertise. Intellectual virtues are, or at least can be, domain-specific in the relevant sense; they “enable a person to arrive at truth and avoid error in some relevant field (Greco 2002: 287; emphasis added). This aligns with the cases of perceptual expertise under consideration. In each case, such performance is a skill highly specific to some field or domain by contrast to some general perceptual or intellectual advantage. Virtues are often dispositional, as highlighted in the definition offered by Sosa. And indeed, the expert’s skill deploys dispositionally. This deployment is largely automatic and results in exceptional frequency of success in the appropri ate circumstances. Relevant diagnostic details for the expert radiologist, for example, enjoy visual pop-out and salience, and this aids judgment and diag nosis. And in ecologically invalid circumstances, when the expert should in some sense turn off her learned perceptual sensitivity, she cannot. This is why laboratory-manipulated circumstances often hinder expert perfor mance. But by the same token, expert performance is performance. The expert radiologist’s visual uptake of relevant diagnostic information is not a matter of mere sub-personal visual processes, nor is it some pre-expertise visual acuity. Instead, the expert has acquired, through concept and category-rich cognitive training, through deliberate activity, a skill. The expert radiolo gist performs better visually because of what she has done, because of her actions, as a responsible epistemic agent. In this regard, perceptual expertise is a perceptual skill and one that is improved by agent-driven, accuracy enhanc ing training. It is in this intellectual virtue – an “acquired excellence” to use Zagzebski’s terms – that the epistemic value of expertise consists. Perceptual expertise is an epistemic virtue.
7.3 Perceptual improvement and biological function To think of perceptual expertise as an intellectual, and thus epistemic, virtue is to commit to the view that perception can improve. We can become better
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
perceivers, and in ways sensitive to what we learn, the concepts we acquire, the specialized training we undergo. That vision and other perceptual sys tems can improve is not an entirely new thesis, but it remains unorthodox (at least in the form advanced here). It therefore requires some further elabo ration. This will be done through the lens of considering perceptual systems in the context of human biology, evolution, and development. The func tional-teleological analysis that emerges further grounds the virtue-theoretic epistemology just given. A promising way to begin is to ask: Do perceptual systems like vision have a biological or teleological function? A positive answer to this question has been argued by a number of recent theorists, and the present analysis will follow their important leads. The short answer is that visual perceptual sys tems (and at least some other perceptual systems, like audition and touch) have a representational function. It is successful performance of this func tion that provides standards or norms for the value of a perceptual system and (in part) the fitness of the individual that possesses that system. Under standing the nature of this function and its implications for thinking about human perception requires a longer answer. In ordinary circumstances, we might ask what some object or thing is for? My child might ask me what the bird’s wings are for. Or he might ask a ques tion that can generate a similar kind of response, asking me what a heart is. It is plausible that I might respond by saying, respectively, that the bird’s wings are there in order to allow the bird to fly and the heart is for pumping blood to the rest of the body. The answers are at least teleological in the sense that they designate a purpose for the things queried and, in doing so, specify or individuate the thing by identifying its function. Whether the teleological must take talk of “purpose” as central or literal in some Aristotelean sense, and moreover whether the teleological is necessarily biological, are contro versial questions in the philosophy of biology. Those controversies to one side, it is worth noting that as a matter of fact, we do ask similar questions and give similar answers not just concerning natural kinds but also concern ing artefacts. In the very same way, my child may ask me what a carburettor is for or what a hammer is. And, respectively, I would reply that the carburet tor is part of the engine; it’s there in order to mix oxygen and fuel to create combustion, and a hammer is for driving nails. Here again, the questions are answered, and the things individuated, by identifying a relevant function. To engage in this kind of description of an object, as we ordinarily do, is to engage in a minimal version of functional analysis, a mode of explanation that is central to biological theory. As defined by Robert Cummins (1975),
183
Copyright © 2021. Taylor & Francis Group. All rights reserved.
184
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
such a mode of explanation, for any given system S, proceeds by analyzing S into its components (and components of components) and then identifying what each component does or how it contributes to some relevant effect or product of S. One example that Cummins favours is the functional analysis of an assembly line in a factory. Each point in the assembly line (which could be occupied by a human worker, a machine, a robot) has a distinct task to perform vis-à-vis the ultimate production of some product. And successful production can be explained by appeal to the capacities of each of these indi vidual points in the line and how they are organized as part of the overall system. Once these parts of the analysis are in place, the functional explana tion comes naturally: “[W]e may pick out a certain capacity of an individual exercise of which is his function on the line. Of the many things he does and can do, his function on the line is doing whatever it is that we appeal to in explaining the capacity of the line as a whole” (1975: 760). The function of the individual component is given by identifying what that component does to contribute to the overall effect of the system (in this case, some intended product of the factory). And the norms for successful performance of that function will be given by how, or how well, a given performance (or type of performance) contributes to the relevant, overall effect. Now consider biological systems, starting first with the whole organism. The relevant effect of the organism is that it survive and reproduce. Relevant how? The organism that survives and reproduces contributes to its evolution ary lineage. And organisms that survive longer have a greater probability of reproducing and thus contributing to that lineage. Which animals survive longer? Typically, those that are more fit. And what makes one organism more fit than another is, most basically, that it performs better along three of the four Fs: fighting, fleeing and feeding, so that it has a better chance of (and perhaps with greater frequency) participating in the fourth “F”, repro ducing. Organisms that are more fit in these respects thus have a selective advantage: The traits that they have that contribute to this fitness are selected for – get passed on – since they typically, but not necessarily, survive longer and reproduce more. Our question concerns how to think about biological functions of traits and how they contribute to the fitness, and thus the rele vant “effect” of an organism, surviving and reproducing. First consider a somewhat hackneyed example: a functional analysis of a heart. In the style of functional analysis just given, the function of the heart is identified as follows. The most basic relevant effect for the organism of which the heart is a part is surviving and reproducing. Just like the point in the assembly line, the heart contributes to this effect in some way, and
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
whatever that way is, is its function. The heart’s contribution is distributing blood throughout the organism’s body. This is what it does in the overall sys tem to contribute to survival and reproduction, and so this is its function. This can then be given a more teleological flavour by employing an etiological analysis. In its most basic form, this analysis says that the function of X is Z means that (a) X is there because it does Z and (b) Z is a consequence (or result) of X’s being there (Wright 1973). The heart is present in the organism (say, a human) because it distributes blood, and the distribution of blood in that organism is a consequence of the heart being a part of that organism. Paul Griffiths (1993) combines these two approaches to identify the “proper function” of biological kinds (what things are “really for” rather than what they may, perhaps accidentally, do in a system).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
[T]he biological fitness of a type of organism can be explained by Cum minsesque functional analysis. An organism’s fitness is a measure of its overall capacity to survive and reproduce, relative to the capacities of competing types in the population. . . . The analysis of this capacity will reveal a number of ‘fitness components’. Fitness components are those effects of traits which enhance the fitness of their bearers. They are the Cummins-functions of those traits relative to the overall capacity of the animal to survive and reproduce (fitness). The proper functions of a trait are those effects of the trait which were components of the fitness of ancestors. They are the effects in virtue of which the trait was selected, the effects for which it is an adaptation. (Griffiths 1993: 412)
And now we have a teleological account. The proper biological function of the heart is to distribute blood because this is the effect which contributed to the overall effect, namely fitness – its overall capacity to survive and repro duce – on the organism. This is what the heart was selected for, and so it is there because it does that. How can this kind of analysis be extended to perceptual systems? Peter Graham defends the claim that “human perceptual systems – especially visual systems – have producing reliably accurate perceptual representations as a biological function” (Graham 2014: 13; see also Graham 2012). Impor tantly, this claim has separable components. The idea is not just that the human visual system has the proper function of delivering representations of the environment, but that its function is to do so accurately and reliably. The line of reasoning mirrors the one given regarding hearts. The visual system
185
186
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
contributes to the overall fitness of the human organism by contributing to success in survival and reproduction. And it does this by providing rep resentations of features and particulars in the organism’s immediate envi ronment. On Wright’s classic analysis, the visual system is there because it delivers these representations, and those representations are a consequence of our having the visual systems that we do have. Graham reasons that visual systems that represent accurately and in reliable ways will better contribute to this fitness and therefore enjoy a selective advantage. Organisms rely on their perceptions to navigate their environments. Accu rate representations are better guides. Just as white fur helps the [polar] bear because white matches its environment, accurate perceptions help countless creatures because accurate perceptions match their environ ments. The accuracy of perceptual representations – especially visual rep resentations in humans – plays a role in the functional analysis of how organisms with perceptual systems are able to survive and reproduce. (Graham 2014: 22)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Tyler Burge has given a similar analysis of perceptual systems, but with a subtle difference. On the one hand, he maintains that “perceptual systems, and perceptual states, have the representational function of representing veridi cally, hence reliably” and further that this kind of functional analysis, com mon in biological explanation, allows us to explain the presence of the perceptual systems and perceptual states in an animal as a causal consequence of their enabling the animal to success fully represent the environment. Being capable of representing success fully in favorable, natural circumstances is a consequence of the system and states’ being present in an animal. (Burge 2003: 512; emphasis added)
On the other hand, Burge argues that there is a “root mismatch between representational error and failure and biological function” (Burge 2010: 301). The relevant distinction is between some system or trait having a representa tional function versus its having a biological function, and the relevant claim is that accuracy of representation falls under the first, not the second, for perceptual systems. Evolution does not care about “veridicality per se”. Accu rate representation is a relation between a representational system (or the representation that it produces) and what it represents. It is a semantic value.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
Contribution to fitness is, Burge urges, a “practical” value “grounded in bene fit of its effects for survival and reproduction”. Accordingly, accuracy “as such”, “in itself”, qua non-practical semantic value, cannot be a biological function. It is worth taking a further moment to identify some of Burge’s motivation for this claim, since it, and Graham’s criticism thereof, provide insights for how to think about cases of perceptual expertise. Imagine a skittish or eas ily frightened kind of animal; sparrows come to mind, or rabbits. Suppose that the animal has an avoidance mechanism that “functioned to increase strength and agility – in avoiding the predator – even in cases in which the animal engaged in avoidance behavior, because of a misrepresentation as of a predator, when no predator was present” (Burge 2010: 302). Here we have an example of an organism with a system that regularly misrepresents while, by virtue of that inaccuracy, contributes to the biological fitness of that organism by better enabling it, through physical exercise, to avoid pred ators. Therefore, Burge concludes, “failure of accuracy need not be failure to realize any biological function” (Burge 2010: 302).4 Graham’s response is at least twofold. First, Burge seems to have overlooked the possibility that a mechanism or trait can have more than one biological function. Your heart has only one biological function, but your tongue has a number of biolog ical functions, including those that contribute to eating and speaking. The skittish animal’s avoidance mechanism may have the biological function of prompting predator-avoiding exercise, while also having the biological func tion of accurately representing the environment, and failure or success in fulfilling the one does not entail failure or success in fulfilling the other. This leaves logical space to assess whether the detection mechanism is suc cessful vis-à-vis its function to accurately represent. And second, Graham suggests that even with such an animal, accuracy matters. Burge is right that the imagined animal (and there are surely actual examples) delivers misrepresentations in the form of false positives (and this contributes to the animal’s fitness in spite of that inaccuracy), but it must also deliver enough accurate representations in the form of true positives (rather than false neg atives), namely, in cases where a predator is actually present. Why? Because “false positives are often pretty cheap. But the low cost of false positives does not diminish the high cost of false negatives” (Graham 2014: 27). The rep resentational system is thus successful, Graham concedes, not by being reli able, but by being “effective”, by delivering accurate representations “often enough” (2014: 24–27).5 Graham’s concession then is that some traits with a biological representa tional function can succeed by representing accurately often enough, even
187
188
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
if they are not thereby reliable. But he urges that these cases are exceptional and that most such systems contribute to fitness (at least in part) by being reliable. What then of Burge’s claim that evolution is insensitive to veridical ity per se, that nature does not care about accuracy as such? It is first worth noting, again, that Burge agrees that perceptual systems “have the representa tional function of representing veridically, hence reliably” (2003: 512, emphasis added). And further, that “biological structures that underlie perceptual and cognitive systems evolved and were selected for, . . . They were selected for because they yielded results that were good enough to further fitness. Evolution does not care about veridicality” (Burge 2010: 302–303). Graham’s account is entirely consistent with all of this, excepting the last claim regard ing veridicality per se. How does he suggest that we resist that latter claim? Simply by redrawing our attention to how a Cummins-style functional anal ysis would apply in this context. He writes, “If biological functions are the capacities and effects of traits that contribute to meeting needs, albeit con tingently given their role in the organism in its natural habitat, then many capacities are biological functions even if they don’t contribute to fitness as such, in themselves” (Graham 2014: 23; emphasis added). Therefore, as a con tingent matter, perception has the biological function of representing (and at least as one function, reliably accurately), since it contributes to the relevant overall effect of surviving and reproducing; it contributes to fitness. If instead we follow Burge’s line and insist that to count as a biological function, a trait must be selected for as such or in itself, we incur dubious results. [e]volution also does not care about coloration per se; it does not care about pumping blood per se; it does not care about sharp teeth or long legs per se; it does not care about oxygen diffusion or photosynthesis per se. evolution only cares, per se, about contributions to fitness and repro duction. It does not follow from any of this that evolution did and does not care, as a matter of fact, about coloration, pumping blood, oxygen diffusion, sharp teeth and long legs. It does not follow from any of this that coloration, pumping blood, oxygen diffusion, sharp teeth and long legs cannot enter into the functional analysis of an organism’s ability to survive and reproduce. (Graham 2014: 22–23)
Applying Burge’s reasoning to the heart (among the other examples Gra ham gives) forces a conclusion that the heart lacks the biological function of pumping blood, since evolution does not care about pumping blood per se
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
or as such. Burge’s veridicality-per-se claim overgeneralizes: Evolution cares about almost none of the fitness contributions, per se, that we attribute to traits, but that should not imply (contrary to Burge’s inference regarding percep tual systems) that such traits do not have the correlative biological functions. Now to apply this general analytical approach to perceptual improvement. Human perceptual systems have a representational function, and fulfilment of this function contributes to survival and reproduction. There are thus selective pressures on how perception represents. And the value or success of a representational system (and the representations it produces) will be deter mined by how it serves that role, that is, how and how effectively it contrib utes to the overall effect of the whole organism – survival and reproduction. Graham and Burge agree on this much, and, further, they both maintain that there are natural norms for any given function. In this context, a norm is a standard for possible performance that is to some degree adequate for fulfilling a function. And a natural norm is “a level of performance adequate to fulfill a function . . . independently of any individual’s having a positive or negative attitude toward the function or the norm” (Burge 2010: 311). What, then, are the natural norms for the representational function of perceptual systems like human vision? Most obviously, a standard for fulfill ing this function is accuracy: reliably producing veridical representations.6 Accordingly, a perceptual system like vision can fulfil its function better or worse (and, to foreshadow, more or less virtuously) by producing accurate representations of the environment with more or less frequency. This already identifies space for perceptual improvement. Most obviously, an individual (or her visual system) will improve with respect to this norm through the course of perceptual development. Supposing we begin our sensory lives with the Jamesian “blooming, buzzing, confusion” of sensation; as our bod ies grow and our brains develop, visual experience takes a more and more organized form, representing objects, scenes, and their features with finer and finer grain. This should be uncontroversial and is the most straightfor ward way that a visual system could improve in line with the norms of its representational function. In addition to perceptual development, we enjoy perceptual learning. Elea nor Gibson defined perceptual learning as “[a]ny relatively permanent and consistent change in the perception of a stimulus array, following practice or experience with this array” (Gibson 1963: 29). As Kevin Connolly (2017) characterizes this definition, a bit of learning is perceptual just in case it is a long-term change, the change is genuinely perceptual (rather than, say, in inference or judgment), and it results from some repeated exposure (thus
189
190
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
“practice” in some minimal sense) to the relevant class of stimuli. Perceptual learning is well studied and can take a variety of forms (Goldstone 1998, 2003; see also Connolly 2017, 2019). Perhaps most obviously, by virtue of repeated exposure, we learn to perceptually differentiate, as we might do in learning to distinguish phonemes in a new (and relevantly different) lan guage or to discriminate determinate shades of colour in the same determi nable hue. Second, we learn to unitize items previously perceived as disparate, as we do when we learn to see words as words, rather than as a string of individual letters. Third, as a result of exposure to a stimulus type, percep tion imprints: forming a “detector” or a reliable sensitivity to a stimulus pat tern or configuration or a feature of a stimulus. On one instance-based model of this phenomenon,
Copyright © 2021. Taylor & Francis Group. All rights reserved.
every exposure to a stimulus leads to an internalized trace of that stimu lus. As more instances are stored, performance improves because more relevant instances can be retrieved and the time required to retrieve them decreases. Instance based models are supported by results showing that people’s performance in perceptual tasks is closely tied to their amount of experience with a particular stimulus. (Goldstone 1998: 591; Logan 1988)
Imprinting can occur as a result of mere exposure, but also in ways that are behaviourally relevant or sensitive to a task. Finally, perception adapts or learns, whereby attentional selection becomes more sensitive to features that are behaviourally relevant and less sensitive to those features that are not (Nosofsky 1986; Livingston and Andrews 1995; Goldstone 1998). This type of perceptual learning has been shown to be sensitive to categories of stim uli and, accordingly, to training. Many relevant studies on elite-level athletes provide evidence for this kind of attentional weighting. Researchers use various occlusion paradigms to determine what bodily cues are most relevant to successful performance, for instance, where certain parts of an opposing player’s body are blocked from sight. In one study, elite-level goalkeepers can, with significantly high frequency, successfully predict the direction of a penalty kick based on a view of the hip region only (Causer et al. 2017) (see Figure 7.1).7 These perceptual learning effects are often early in visual processing (Peterson and Gibson 1994; Sekuler et al. 1994; Fahle and Morgan 1996) and automatic (Shiffrin and Schneider 1977; Palmeri 1997). Some of these dimensions of perceptual improvement are ones that concern accuracy, say, accurately differentiating a pair of colour chips or
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
Figure 7.1 Example of a spatial occlusion task: whole-body vs. hip-only conditions
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Source: Reprinted with permission from Causer et al. 2017. Copyright PLoS One.
unitizing words correctly. But others are not improvements in accuracy, but instead in the type of information picked up and how efficiently. In the case of stimulus imprinting, one is able to rapidly recognize an instance of a pat tern or feature type. In the case of attentional weighting, it isn’t (necessarily) that individuals with the relevant exposure or training better enjoy verid ical perceptions, but that rapid attentional mechanisms are more sensitive to relevant information, and sometimes without any effort or intention on the part of the perceiver. Recall that on the convergent analysis of Graham and Burge, the success of a representational system is ultimately determined by how it contributes to survival and reproduction. Accuracy is therefore one natural norm for successful fulfilment of this function. But we now see that there are other ways a representational system (and its representa tions) may contribute to survival and reproduction and how improvement along such dimensions would be improvements in line with the function of those systems. Increased sensitivity to categories, patterns, and feature types, achieved rapidly and efficiently (that is, with little if any effort on the part of the perceiver) and in ways that enable and enhance fit behaviours, are ways that perception can successfully fulfil its representational function and con tribute to overall fitness. Therefore, such improvements are improvements in line with the natural norms of perception. Some of the improvements to perception just outlined are ones that occur simply as a matter of development. Others are improvements that require
191
Copyright © 2021. Taylor & Francis Group. All rights reserved.
192
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
particular kinds of experience and exposure, say repeated exposure to words in the same language or to many colour samples within the same hue, per haps without explicit feedback or conceptual influence (say, learning all the relevant colour terms). Others appear to be perceptual improvements that involve more than mere exposure, but instead training of a domain-specific kind, as in the case of the elite athlete and attentional weighting. Here the individual, to perform at an elite level, must be exposed not only to rel evant visual events, but she learns which components of those events are most relevant to her behavioural performance and in ways sensitive to the actual structures and patterns of organization of the sport (Mann et al. 2007). Therefore, some cases of perceptual improvement are, we might say, non-agential in the sense that the individual need not try to do anything or engage in any high-level cognitive or semantic learning to trigger the per ceptual learning mechanisms. But in other cases, the perceptual changes depend on what the agent does and has done, what she learns about the domain, and how she thinks about it. Many cases of perceptual expertise fall into this second category, where the etiology of the perceptual changes crucially involves domain-specific, content-rich learning. Understood now in terms of epistemic virtue coupled with the teleological analysis of perception, the account goes as follows. Vir tue epistemologists often talk of virtues in terms of tools (for instance, Sosa’s “Epistemic Normativity”, 2007). A hammer is a tool, and insofar as it can improve in line with its proper function – driving nails – it can have more or less virtue. The hammer, of course, does not improve itself, but it can be improved by its user(s) (its weight can be adjusted, its handle made grippier, etc.). Understood as a trait or a tool, the function of perception is to provide useful representations to the perceiver. Accuracy is perhaps the central mode of utility, but, as we saw earlier, not the only one. A perceptual system can, just like the hammer, be improved in line with its proper function (in line with the norms that are given by that function, as described). Sometimes these improvements are ones that occur without the agent’s doing. Increased sensitivity to distinct categories of stimuli, simply as a consequence of natu ral sensory development and/or repeated exposure, is a kind of non-agential improvement. The virtue is thereby attributed to the perceptual system, if not the agent (just like the hammer). But many cases of expertise are differ ent. Here the agent has, deliberately and laboriously, undergone training in a specific domain: in radiology or fingerprint examination, in tennis or foot ball, in ornithology or visual depiction. As a consequence of this training, her perceptual systems perform in exceptional ways within that domain.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
And those levels of performance near maximally satisfy the identified norms for perception, thus fulfilling the representational function of perception in optimal or near-optimal ways (in that domain). The important epistemic difference between this case and the cases of mere development or expo sure is that the agent is herself responsible for the relevant etiology and, accordingly, for the perceptual improvement. The epistemic virtue is there fore attributable to the agent herself (not just, say, her visual system); she has improved the tool of perception through her actions. This grounds the account given in Section 7.2 – that perceptual exper tise is an exceptional kind of intellectual virtue – by explaining perceptual improvement with a teleological analysis. It is worth taking a moment to identify and review some of the improvements that the expert makes to her perceptual systems. If we accept the mental architecture of perceptual expertise defended in Chapter 6 – the claim that some cases of perceptual expertise are genuinely, cognitively sensitive perceptual phenomena – then all of the following per ceptual improvements can be identified in cases of experts. The most obvi ous improvement is in accuracy. Relative to a domain, perceptual experts, by contrast to novices and laypersons, perceive and accordingly perform more accurately. But this is not all. Experts are more perceptually sensitive to behaviourally relevant patterns and organizational features of a stimulus or stimulus array. And this is true for many of the examples of expertise discussed. One especially clear case is elite athletes, whose response time and successful performance co-vary with presentation of genuine patterns for the sport in question. For instance, in studies on skilled versus unskilled basketball players, the former enjoy superior accuracy of performance just in contexts where they are presented with sequences and scenarios that could actually occur in the sport (Allard et al. 1980; see also Williams et al. 2011). The expert athlete thus enjoys enhanced visual representation of how events are organized in her visual array, where that enhancement is sensitive to the patterns that lead to optimized behavioural performance, with less dis traction from irrelevant features or patterns that are not, as it were, ecolog ically valid for that sport. This last phenomenon is found in other domains. Eye-tracking studies on radiologists, for example, suggest less distraction from diagnostically irrelevant features of a stimulus array: Such experts make fewer but longer saccades and fixate less on features of a radiogram that are of no consequence to identifying an anomaly (Kundel et al. 2007; Drew et al. 2013). With less visual distraction comes heightened sensitivity to category-specific and diagnostically relevant information. The forensics
193
Copyright © 2021. Taylor & Francis Group. All rights reserved.
194
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
expert enjoys greater visual sensitivity to features of a fingerprint that will be relevant for making a match; the car expert uses holistic visual information specific to the whole make/model type rather than a feature-based strategy, and enjoys an advantage (for cars) in visual short-term memory. Finally, many experts enjoy more rapid visual representation; they see things and features, and behaviourally relevant ones, more rapidly than novices and laypersons. Some of the perceptual improvements of experts involve improvements in accuracy in a traditional sense. Put flatly, the expert better sees what is to be seen in a scene. But the other improvements just discussed – improved sensitivity to behaviourally relevant patterns, organizational features, cat egory- and diagnostic-specific information, less distraction, speed – are not improvements in accuracy or veridicality in this straightforward sense. They are instead improvements in seeing what is relevant to a task and with increased speed and efficiency. If we broaden our notion of perceptual accu racy (as will be discussed in Section 8.4), then these are still improvements in accuracy; it is just that this accuracy is partly context-sensitive relative to the domain of expertise in question. The perceptual expert better perceives what matters – what is behaviourally relevant – for performance in her domain. Sometimes this means perceiving more, sometimes this means per ceiving less, sometimes this means perceiving a whole rather than its parts. Here again, these kinds of improvements to the visual (or other percep tual) system are in line with the natural norms of perception. The expert’s visual system is, within a domain, fulfilling its representational function in near-optimal ways. And, once more, because she is partly responsible for this improvement – her perceptual advantages non-trivially depend upon the domain-specific, cognitive learning she underwent – she, qua agent, is to be credited as epistemically virtuous. This is what is special about perceptual expertise: The perceptual expert is responsible for enhancing the intellectual virtues of the perceptual representational system. Thus, the epistemic virtue of some experts goes all the way down to their sensory perceptual systems. And so it is no surprise that in some specialized domain or for some specialized task, it is rational to, as we say, call on the experts. The lesson here is that one makes this call not just for what the experts know but for what they see.
7.4 Objections and qualif ications The claims about perceptual improvement and epistemic virtue, as well as some of the underlying mental architecture of perceptual expertise, can be
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
further qualified by considering a few possible objections. Chapter 8 then turns to some more formidable challenges to and consequences of the pro posed account of expertise, finally teasing out some substantial revisions for theories of perception, all through the lens of cognitive effects on perception. The first pair of worries concern the nature of the argued epistemic vir tue. In turn, the first worry concerns how the agent is herself to be credited with epistemic virtue, and the second worry concerns how that virtue partly consists in improvements to perception. The final worry concerns the tele ological analysis given in the previous section and how that comports with the cultural sensitivity of the relevant examples of perceptual expertise. The first worry goes like this. The expert’s performance, as evidenced by some of the experimental studies discussed, deploys automatically. For instance, in rapid forced-choice scenarios the expert radiologist appears to respond visually to anomalies before she can perform an action or make a judgment. This is further supported by the evidence that the expert enjoys rapid attentional selection of relevant features – visual “pop-out” or salience of diagnostic, categorical features. So, it is not as if, in at least some of these cases, the expert is herself deploying some relevant strategy to make a diag nosis or categorize an object. And further, in experimental circumstances like those involved in composite tasks and inversion tasks (as discussed in Sections 6.3 and 7.2), the expert should execute a feature-based strategy to better ensure success, but she doesn’t (and accordingly, such manipulations interfere with the success of performance). This is because the perceptual strategy deployed by her visual system is largely automatic and she cannot readily override it (at least not in time-limited cases). It is for these reasons that it is inappropriate, the worry says, to credit the agent herself for her cog nitive success. Epistemic virtue is properly attributed to an agent who is, to some appropriate degree, responsible for the relevant performance. Because the expert is not responsible for her perceptual performance, she does not (at least perceptually) exhibit epistemic virtue. This worry is useful, since rebutting it further reveals how remarkable some cases of perceptual expertise are. It is first useful to repeat the point that some virtue epistemologies count a variety of faculties that function largely automatically as virtues, including perception but also intuition and memory. So, in that theoretical context, the fact that a mental process type is to some degree automatic does not by itself undermine the potential virtue of such a process. More to the point, and more neutrally, intellectual vir tues are supposed to deploy dispositionally. Contemporary theories of epis temic virtue of nearly every stripe, whatever else their differences, tend to
195
Copyright © 2021. Taylor & Francis Group. All rights reserved.
196
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
countenance, if not define, virtues as dispositions. This is plausibly a matter of pedigree: Virtue epistemologies are rooted in traditional virtue theory, notably Aristotle. And Aristotle defined virtue in terms of hexis, which is commonly translated to mean “disposition” (Nichomachean Ethics 1105b 25–6). Aristotle was also explicit that these dispositions are not static or passive, but instead are skills that manifest in the right circumstances and, crucially, are the result of habits, practice, and experience. The contemporary notion of intellectual virtues follows suit: They are dispositions that manifest in the right circumstances and that are either the result of or can be improved by one’s activities, habits, and practice. Therefore, the automatic perceptual performance of the expert is no less a virtue because of its automaticity. Virtues are supposed to be dispo sitions, and once well-established they should deploy rapidly, if not auto matically. Indeed, with this automaticity comes efficiency, since the agent enjoys greater freedom of cognitive resources for other elements of the task. This is also why, as noted in Section 6.2, the expert’s performance declines under certain experimental manipulations like inversion cases: These are ecologically invalid circumstances for the expert and thus not ones where her dispositions aptly apply. By the same token, we should not infer from automaticity that the agent is not herself responsible for her performance or its cognitive successes. The conclusion of Chapter 6 was that for some perceptual experts, their perceptual advantages are genuinely cognitively sensitive. The etiology of the perceptual representation of the expert (within the relevant domain) involves the rich conceptual and categorical learning undergone in the domain, be it forensics or football. These are genuine cog nitive effects on visual perception. And they are effects for which the expert is most certainly responsible, since she is responsible for the laborious and cognitive training undergone. So even if some of the performance successes are ones that perception achieves automatically, the expert is credited for these successes, and so the intellectual virtue is hers. The second worry concerns how or whether some of the improvements discussed, and the accordant epistemic virtue, is properly understood as perceptual. At least some of the discussed cases involve changes in atten tional selection. For instance, the perceptual learning type called atten tional weighting involves enhanced sensitivity to relevant patterns within a domain. An expert goalkeeper, for instance, better and more rapidly attends to genuine patterns of the game (how the opposing players are spatially organized) while ignoring irrelevant or ecologically invalid patterns. Fur ther, the worry continues, at least some of the eye-tracking measures only
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
provide evidence for differences in attentional patterns. The radiologist vis ually perceives differently, but this is because she attends to different ele ments of the radiogram. Therefore, at least some of these cases are not evidence for an epistemic virtue that consists in genuine perceptual advan tages; they are instead advantages in attention. The expert simply knows better where to look. The first thing to note in reply is that there is something right about this worry. It is absolutely correct that some cases of expertise, even so-called perceptual expertise, plausibly involve superior attentional strategies. Expertise often consists, at least in part, in knowing where to look, what to listen to, and so on. This is entirely compatible with the present analysis of percep tual expertise and its virtues. Recall that no claim is made about all cases of expertise, or even all cases of perceptual expertise. And as emphasized in Section 6.5, a pluralism about explanation of cases of expertise is preferable. Given the variety of domains in question and the kinds of cognitive and perceptual activities required, it is implausible that all cases of expertise will enjoy a singular, unified explanation. It is also implausible that all cases of perceptual expertise, so-called, enjoy genuine changes to the visual or oth erwise perceptual processes of those experts. Some types of experts, as the worry identifies, may just know how to better attend to the space or array in question. Others still may just know better how to interpret or judge what they see. These admissions would challenge the present analysis of percep tual expertise only if the concluding claim were a categorical one, that all instances of perceptual expertise involve cognitive effects on perception. But the conclusion of the analysis is instead that some perceptual experts are genuine perceptual experts. The more focused reply is to consider some of the cases of expertise and perceptual improvement that involve attention and ask whether the inclu sion of attention undermines the claim that the performance success or vir tue is a genuinely perceptual one. Does the worry rightly challenge these cases? Again, it will depend on the case in question, but at least for some of the cases discussed, there is no plausible challenge here, and for reasons that we have already seen. In Chapter 5 (and again in Section 6.5), putative cogni tive effects on perception were defended against an opposing attention-based explanation. The idea there was that if cognition merely effects an overt shift in spatial attention, which then changes the input to perception, then this is an effect on perception but is rendered trivial. Such cases are not structur ally unlike cases where cognition drives some substantial bodily action (say, going to the other room to see what is making an annoying noise), which
197
198
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
then changes what one sees. And we don’t count those cases as important cognitive effects on perception. The same goes for attention-involving cog nitive effects on perception, the opponent concludes. In rebutting this explanation of all attention-involving cases of cognitive effect on perception, a handful of lessons were drawn, some of them use fully applicable here. Recall that a central component in the argument in Chapter 4 was to defend the claim that some instances of attention-involving cognitive effects on perception take the form of schema (b): (b) Cognitive state → Non-agential selective attention→ Perceptual experience
And this is contrary to the modularist’s assumption that all such cases would instead take the form of schema (a):
Copyright © 2021. Taylor & Francis Group. All rights reserved.
(a) Cognitive state → Attention-shift → Perceptual experience
This claim was defended by appeal to a variety of recent philosoph ical and scientific research on attention and, in particular, on selec tive attentional mechanisms such as feature-based attention (FBA) and object-based attention (OBA). Importantly, those mechanisms can oper ate independently of spatial attention or changes in spatial attention, and moreover operate covertly, without any effort on the part of the perceiver. So contrary to the description of the opponent, some cases involving visual search and visual recognition, reliant on FBA and/or OBA (as is plausible in many cases of expertise), are not cases relevantly analogous to action-mediated cognitive effects on perception. They do not involve overt, agent-driven shifts in attention. Just as was concluded in the analysis from that chapter, the attention as spotlight and atten tion as act assumptions should be rejected. Attention sometimes, but not always, takes these latter forms.8 Some of these selective attentional mechanisms are plausibly part of, rather than separate or prior to, visual processing. To re-emphasize the same points as were made in Chapter 5, the saccadic eye movement patterns revealed by eye tracking in relevant studies is standardly taken by visual scientists as a physiological marker for visual processing. And the same goes for activity in visual cortical activity (V4 of extra-striate cortex) correlated with those selec tion mechanisms. What these mechanisms do is determine which stimuli in an array are selected for further processing and thus, typically, which of those features or objects are perceived. This can again be put in terms of
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
biased competition (Desimone and Duncan 1995; Mole 2011, 2015). Atten tion involves a competition for resources, where only a subset of what is visually available will win this competition and thereby be visually attended. The competition is influenced by both bottom-up biases (say, brightness or novelty of a stimulus) and top-down biases. Some cases of perceptual expertise involving attention are instances of top-down bias. Background concepts, knowledge, and skill influence which features or objects win the competition and thereby are visually perceived. And in such cases of topdown or cognitive influence, the influence is not on something separate from perception: The influence is on attentional mechanisms that are part of perception. The final assumption of the modularist – the attention as gatekeeper assumption – is, once again, to be rejected. Contrary to the sec ond worry, then, some cases of expertise involve epistemic virtue at the level of perception, and this is true not in spite but because of the inclusion of selective attentional mechanisms. Finally, one might worry as follows. The experts discussed over the last two chapters enjoy, many of them, expertise specific to particular domains, laden with theory, culture, and technology: radiology, fingerprint examina tion, ornithology, Greebles! Some of these domains are, at best, hundreds of years old, and some of them not much more than a handful of decades old. Accordingly, performance in these domains cannot ground any adap tive or fitness advantage in any strict biological sense. These domains of human activity and relevant theories and technology are simply too new to situate, as it is sometimes put, on an evolutionary timescale. So the percep tual improvements made by the expert in line with those domains cannot themselves introduce or ground new biological functions or norms for per formance of those functions. The visual system of the radiologist surely does not have the proper function of accurately representing tumours, nor the forensics expert of distinguishing fingerprints. The spirit of the worry is apt, but its force is disarmed with the following distinction. Distinguish (perceptual) improvement on a broad, evolutionary timescale from (perceptual) improvement on an individual or generational timescale. The first involves millions of years; the second may involve as little as days and as much as years or decades (depending upon the emphasis on individuals or on populations of individuals). The broad claim made about perceptual systems of humans falls on the broad, evolutionary timescale. That claim, following the work of Graham, Burge, and others, is that human perceptual systems like vision have a representational function. Further, the function is to reliably provide accurate representations. Or, put in terms
199
Copyright © 2021. Taylor & Francis Group. All rights reserved.
200
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
of norms: There are norms for adequate fulfilment of the representational function of systems like vision, and at least one of those norms is reliable provision of veridical representations. Systems like vision have this function by virtue of contributing to the fitness of the human organism – to the over all effect of surviving and reproducing – and any general fitness-enhancing adaptations with respect to this function and its norms would take place across evolutionary time. The case made for perceptual improvement, both in instances of basic perceptual development and learning, as well as expertise, is made against this background. However, no claim is made that an individual, or a popu lation, or generation of individuals possesses perceptual systems that enjoy distinctive biological functions (say, a radiology-specific function). Instead, the claim is that the individual or group improves in line with the already generally established biological function of perception. One norm for ful filling this representational function is accuracy. As the individual develops and learns, she enjoys more accurate perceptual representation. As indi viduals train within a domain of expertise, some of them become experts and enjoy near-optimal accuracy in perceptual representation. Addition ally, individuals (and groups) can better perceptually represent by enjoying improved sensitivity to patterns, to organizational features, to category- and diagnostic-specific information, with less distraction and more quickly and efficiently. These are improvements that the perceptual expert can herself make. It is true that these improvements are improvements, in part, rela tive to the inter-subjective goals and technologies of the domain in ques tion. They are context-sensitive in important respects, and so the success of these changes is conditional on more local norms specific to the relevant domain. But it is also true that they are improvements relative to the general representational function of perceptual systems and the norms that govern adequate performance of that function. Therefore, the virtue of the expert’s improvements is grounded in the general function of perception, without any commitment to domain-specific functions of perception. The individ ual improves her perceptual systems by improving (through development, learning, training) how those systems produce representations that better fulfil the general biological function of perception. It is this that further explains the epistemic virtue of the expert.9 Here again consideration of less controversial biological functions and their analysis proves instructive. Suppose for simplicity and argument that we can specify biological functions for each of the following human traits as follows. A proper function of the legs is to locomote; of the hands, to
Copyright © 2021. Taylor & Francis Group. All rights reserved.
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
grasp; of the arms, to throw; of the feet, to stabilize and jump.10 And the proper function of the heart is to distribute blood; of the lungs, to dis tribute oxygen. Specification of these functions will imply norms for their fulfilment. And, accordingly, for any given individual, a trait of that indi vidual may satisfy the norm(s) in better or worse ways. Now consider how an individual may, through her own actions and experience, improve the performance of such traits in line with their respective norms. Consider a decathlete. Decathlons involve ten individual sporting events, including 100-meter dashes and foot races of other lengths (including hurdles), long and high jump, discus and javelin throw, and so on. Decathletes are, as one would predict, exceptionally fit and versatile athletes. And in achiev ing this fitness and versatility, they improve each of the listed traits in line with their respective norms. Briefly, the decathlete’s legs locomote more quickly and efficiently, his hands grasp with greater strength and agility, his arms enable better movement for throwing, his feet better stability for jumping. And the improvements are not limited to appendages or muscles. The decathlete, like many athletes, will enjoy better heart and better lungs. For example, the resting heart rate of such athletes is often very low (thus efficient), and the VO2 max (which registers maximum oxygen uptake) is often high. These are genuine improvements that individuals can make as a result of intensive training and practice. And they are exceptional improve ments in one important regard: Elite athletes such as these are achieving optimal performance in line with natural norms of the relevant muscles and organs. But such improvements do not require distinctive biological functions relative to such individuals: There are no decathlete-specific bio logical functions of the legs or lungs. The relevant functions for these indi viduals are the same as they are for you and me. It is just that the decathlete has improved her, or her body’s, ability to satisfy the norms for those very functions. These improvements are not exceptional in another, more general regard: All humans can improve their bodies, and in doing so, improve how various traits perform their biological functions. This is not just a point about human development. It is, of course, true that as humans mature through infancy and childhood, their bodies grow and develop. And with these develop ments typically come improvements in how the body performs. But this is also true for the adult body. Human bodies enjoy a remarkable amount of malleability. And because of this malleability, whether one is an athlete or a yogi, one can improve, through training and experience, how certain parts of one’s body perform, from head to toe.
201
202
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
Generally, the brain is no different. It, too, is malleable. The human brain is, functionally and structurally, plastic. Neuroplasticity has long been known regarding crucial developmental stages; as the human brain matures, it develops at all levels, from basic cellular levels, to synaptic organization and connectivity, to functionally relevant cortical areas (Brown and Sherring ton 1912; Kolb and Gibb 2011). And these changes occur both as a result of environmental differences and of individual experience and learning. But neither is this just a point about human development. More recent brain research indicates that the human brain is permanently plastic, enjoying the same range of changes and as a result of environmental changes and triggers (Pascual-Leone et al. 2005; Kanai and Rees 2011). Crucially, recent research indicates that these changes occur in the adult human brain as a result of training and learning (Draganski and May 2008; May 2011). And moreover, research indicates that visual cortical areas also enjoy such plasticity.11 Much is to be learned about the details and further nature of this broad neuroplas ticity.12 But we can combine these observations with what else we know to motivate a provocative conclusion to this chapter.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
7.5 Summary: thinking improves perceiving We know that humans can improve their bodies through experience and learning, and that this is behaviourally advantageous in the relevant domains, and that it often involves improvements in line with more basic biological functions. We know that perceptual experts enjoy enhanced performance in their domains of expertise. We know that experts exhibit physiological dif ferences relative to novices (eye movement and fixation patterns, evidenced by eye tracking studies) and neurological differences (activity in FFA and OFA, evidenced by fMRI study; N170 and N250 responses, evidenced by EEG recordings). And we know that these differences co-vary in significant ways with the kind and extensivity of the domain-specific training undergone. Finally, we know that the human brain is pervasively malleable, plastic not just through normal stages of maturation, but through adult life, and in ways sensitive to individual experience. Why, then, should perceptual systems like vision be the lonely outlier? That is, why should the visual system (and other perceptual systems) be the one aspect (or one of the few) of our lives and bodies that we, as perceiving agents, cannot in substantial ways change? And further, why should it be the aspect that we cannot improve? The more plausible conclusion, given all that we know, is that we can improve percep tually. We can, as a result of our mental and bodily action, become better
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
perceivers. We can become perceptual experts. Thinking does not just affect perceiving; thinking improves perceiving. Further reading Aristotle’s Nicomachean Ethics is the clear historical precedent for virtue the ory, both in ethics and epistemology. And Sosa (1980) is the pioneering contemporary source for the topic in epistemology. In addition to those authors listed throughout the chapter, Battaly (2008) and Turri and Alfano (2019) offer useful surveys, and Fairweather and Zagzebski (2001) and Steup (2001) are two, among many, collections on virtue theory in epistemology. For a survey on perceptual skill, see (Stokes and Nanay 2020). The classic sources for the functional and etiological analyses of the kind employed in this chapter are, respectively, (Cummins 1975; Wright 1973). And for a recent overview on heritability, see Downes and Matthews (2019), and for contemporary teleology, see Forber (2020). For a rich overview of biological theories of mental content, see Neander (2018).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 For some recent examples, see (Smith 2002; Johnston 2004; Antony 2011; kalderon 2011; Macpherson and Platchias 2013; Macpherson and Batty 2016). 2 Precedent for epistemic internalism can be found clearly in Descartes (1641/1984). Classic examples of contemporary internalists include Ginet (1975) and Chisholm (1977, 1988). 3 Of the first: (Greco 1993, 2009a, 2009b, 2010; Sosa 2007, 2009, 2015a). Of the second: (Zagzebski 1996, 2001; Montmarquet 1993). 4 Burge’s motivation, in another sense, is to resist what he calls “deflation ary” or reductive theories of perceptual representation. Those theories identify the function of perceptual systems with their carrying information about the environment, encouraging a reductive kind of naturalism (Burge identifies Millikan 1984, 1989; Dretske 1981, 1986, 1988, 1995 as his central targets for criticism). But Burge wants to defend the explanatory autonomy of the psychological sciences, and so he takes such accounts to be foils. He writes: “The explanatory content and goals of theories of perception and belief are not the same as those that underwrite biology. explaining the way
203
204
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
5
6 7
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8
9
veridical and non-veridical representational states arise, given proximal stimulation, is a different explanatory enterprise from that of explaining any states in terms of their biological functions – their contributions to fit ness. So biological explanations cannot reduce explanations whose point is to explain accuracy and inaccuracy of representational states. . . . The fact that biological functions of sensory systems are relatively close to rep resentational functions makes psychology possible. The fact that biolog ical functions are not the same as representational functions helps make psychology independently interesting” (2010: 303). It is important to note this feature of Burge’s analysis, since it reveals (whatever the success of the analysis) that the distinctions drawn and applied are substantive inso far as they bear (or would bear if successful) important implications for a range of theories of perception. It is also important to note a related broad category of theory that neither Burge nor the present analysis given here commits to, namely, “teleological theories” of mental content. In some form or other, such theories naturalize content by attempting to explain it in terms of biological function. To identify just a few examples of views that fall under this broad category: (Stampe 1977; Dretske 1981; Millikan 1984; Papineau 1984; Matthen 1988; Neander 1995). Graham attributes this point, in part, to Millikan (1984). For discussion of selection pressures, accuracy, and reliability, see (Godfrey-Smith 1991, 1992; Stephens 2001; Sterelny 2003). For Graham, this norm follows trivially, since he claims that the biological function of perception is to reliably produce accurate representations. For recent reviews on expertise, perception, and sport, see (ward and wil liams 2003; Mann et al. 2007; williams et al. 2011). In peer commentary on Pylyshyn’s (1999) defence of the cognitive impen etrability of perception, Paul Sowden invokes cases of expertise and sug gests that Pylyshyn’s claim that experts simply know where to direct spatial attention is far from granted. And he suggests, on neurological grounds, that it is questionable whether attention can even be clearly partitioned from early vision (which is a central to Pylyshyn’s claims). The analysis given here is in some ways an updated version of Sowden’s critique and updated importantly by the twenty-plus years of relevant empirical research. See Sowden (1999). There is a familiar worry in the vicinity, namely, that commitment to domain-sensitive changes in perception is commitment to theory-laden
PeRCePTuAL eXPeRTISe II: ePISTeMOLOGY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
perception. And the latter is supposed to present an important worry for the epistemic role of perception in both ordinary and scientific contexts. This worry will be addressed in Chapter 8. 10 Again, a trait may have more than one biological function, and in these instances such traits plausibly do involve additional functions. 11 See (Schwartz et al. 2002; karmarkar and Dan 2006; Schwartz 2007; Maya-Vetencourt and Origlia 2012; d’Almeida et al. 2013). 12 In a debate on theory-ladenness and perception, Paul Churchland (1988) invokes descending neural pathways and neural plasticity as evidence for cognitive penetration of perception. Fodor rejects the suggestion: “Heaven knows what psychological function ‘descending pathways’ subserve” (1988: 194). whatever heaven knows, we now clearly know a great deal more about such pathways and about the pervasiveness of neural plasticity.
205
8
CONSEQUENCES OF
PERCEPTUAL MALLEABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8.1 8.2 8.3 8.4 8.5 8.6
The generality of perceptual expertise Theory-laden perception and scientific investigation Theory-laden perception, implicit bias, and epistemology Objectivity, accuracy, and perceptual content Expertise and the admissible contents of experience Summary
The conclusion of the last two chapters was that thinking not only changes but improves perceiving. Cases of perceptual expertise are genuine and important cases of cognitive effects on perception, and indeed ones that are epistemically virtuous. Such individuals are to be partly credited for the etiology of their perceptual skills, where that etiology involves spe cialized cognitive learning. The virtue of this achievement was further explained in terms of individual improvements made to the fulfilment of the representational function of perceptual systems, where such individuals perform better with respect to basic accuracy, but also with respect to sensi tivity to patterns, organizational features, category- and diagnostic-specific information, less distraction, and speed. This is not to invoke expert- or domain-specific teleological functions of perception, but simply to claim
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
that the expert, within her domain, better satisfies the general norms of perceptual representation. If this analysis and the conclusions it supports are correct, then a ques tion naturally arises. How special are perceptual experts? This question is important. It provides an angle for answering the question central to this book: Does thinking influence perceiving, and in some important way(s)? If the perceptual expert is a rarity, then perhaps that question is not answered entirely in the negative, but a “not very much” or “not significantly” answer is implied. By contrast, if some degree of perceptual expertise is a per vasive human or biological phenomenon, then the answer to the central question is very much different. Thinking might influence perceiving, to some degree, significantly or pervasively. So, how common is perceptual expertise? Section 8.1 provides an answer to this question. It concludes with the following suggestion. Although perceptual expertise is an intellectual vir tue, and therefore an important achievement, it is a pervasive human phe nomenon. The remainder of the chapter proceeds as follows. Section 8.2 addresses the epistemology of science, and in particular the issues concern ing theory-laden observation. Section 8.3 generalizes to the epistemology of perception, extending the account of epistemic virtue of the previous chap ter. Section 8.4 addresses concerns about perceptual objectivity and percep tual accuracy. Section 8.5 applies some of the conclusions of these analyses to questions about admissible contents of experience. Each of these sections proceeds by considering a worry that one might have about the account given here, coupled with an attempt to disarm that worry and derive more positive theoretical consequences.
8.1 The generality of perceptual expertise One might be tempted to say that perceptual experts are unusual, given the emphasis in the last two chapters on a subset of experimentally studied expertise types. Much emphasis was placed on experts within highly spe cialized fields, including radiology, fingerprint examination, and elite-level sports, among others, as well as emphasis on subjects trained in the lab to recognize artificial, lab created objects. So, the opponent line of reasoning goes, because these experts are so specialized in their range of expertise or are performing in ecologically invalid circumstances, we can make no interesting generalizations from them (even if the analysis of the last two chapters is granted – that these are instances of cognitively influenced per ceptual phenomena).
207
Copyright © 2021. Taylor & Francis Group. All rights reserved.
208
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
This conclusion is short sighted and in a number of ways. First, although the emphasis has been on a relatively small number of highly specialized domains of expertise, this is not a genuine idiosyncrasy, but instead a symp tom of the phenomenon and how it is studied. All the cases of empirically studied expertise are highly specialized. And this comports well with the analysis thus far: What is needed for expertise to be genuinely perceptual, at least in many of the discussed cases, is more than mere perceptual improve ment or exposure to relevant stimuli, but content-sensitive training and cognitive learning. And that content, or the content that is effective must, it seems, be fine-grained, whether one is learning to recognize birds or diagnose cancer. Accordingly, to study bona fide cases of perceptual exper tise, researchers must identify by various experience and performance-based thresholds a select group of experts as experimental subjects (typically con trasted with novices or naïve subjects for that domain). Furthermore, some kinds of expertise are easier to study than others, and perhaps all types are more easily studied in the lab than in the field. For instance, it is far more feasible to use brain scanning technologies in domains for which perfor mance requires a visual search or categorization but, say, no bodily action in response (as in the case of sport). And so, by some measures, studying the radiologist is more feasible than studying the athlete. This, however, is changing. With improvements in eye tracking technologies, for example, research on athletes, studied in the field (or something very close to it), has increased and improved dramatically. These differences in methodo logical limits are just that: They indicate that some types of expertise are currently more amenable to study, not that they are idiosyncratic instances of expertise. And moreover, these limits should not be overstated. Recall the wide variety of expertise types discussed in Chapter 6 in particular, most of them studied using behavioural and neural-physiological methods of various kinds. And this is only a sample of the rich extant psychological literature on perceptual expertise. Therefore, the domain specificity of the experts studied, and the ways that they are and can be studied, just comes with the territory. It comes with studying experts. But this alone does not suffice to preclude the generality of the phenomenon. It is also worth noting that while attempts to study expertise in (more) ecologically valid contexts are important, this is not the only approach of value. In-lab studies have distinctive value. As discussed in Chapter 6, researchers have had significant success training subjects to expert-level per formance both in response to real-world objects (birds and cars, for exam ple) and artificial objects (Greebles). This allows researchers to track, in
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
better controlled ways, behavioural and neural-physiological changes during the expertise acquisition period. And recall that those changes are similar in kind – even if lesser in degree – to the changes that non-lab-trained experts enjoy, whether it is behavioural success or neurological effects that show in EEG recordings.1 Importantly, some of these training regimens for some domains rapidly show lasting expertise effects, sometimes within weeks. So while it is improbable that one could become an elite-level expert in some (or most) domains this quickly, at least for some domains it seems that apt training regimens show relevant effects early and for a significant num ber of initially naïve subjects. This suggests that, generally, expertise is not grounded in some unique or exceptional potential, exclusive to some small percentage of the human population. Further to this, recall that there is little evidence that perceptual experts have some pre-expertise perceptual advantages, such as visual acuity. And moreover the perceptual advantages of such expertise, once acquired, are domain and content sensitive.2 Perceptual experts are indeed special, statistically speaking. That is, the experts in some of these studies, such as some of the relevant radiology studies, are some of the highest performing in a class of competent subjects. They are also special, as argued in the previous chapter, epistemically speaking. Experts exhibit intellectual virtues, and some of them at the level of perception. However, they are not special humans. Nor have they transcended normal human perceptual abilities. They simply have acquired a very highlevel perceptual skill as a result of their achievements within a domain. Finally, note that humans are, like most if not all other animals, habit-forming creatures. What distinguishes us from other species is the array of activities around which we can form habits and how it can pervade all aspects of our lives. Careers or professions often demand specialization. Our non-work life – the thing we sometimes call “life” when concerned with how it is or is not balanced with “work” – also involves substantial spe cialization, even if diverse and regularly changing as we age, move, gain/lose interests, and so on. To live a good life, we tend or try to surround ourselves with the things, activities, and hobbies that motivate, excite, and compel us. And with all of this comes regular and guided exposure to distinctive classes of stimuli. Finally, these habits we form and the activities that such habits involve are often perceptual in non-trivial ways. The reader can no doubt think of many examples of such activities, but here is a relatively short list: Playing chess, watching hockey, doing cross-stitching, returning a tennis serve, spotting and identifying birds, distinguishing wine characteristics,
209
210
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
distinguishing instruments or even particular musicians in a piece of music, tasting various characteristics in food we are cooking, distinguish ing scents in perfumes, tracking an animal by identifying and following its tracks, hunting mushroom types by identifying specific trees where those types are prone to grow, inspecting a mountain face for avalanche dan ger, spotting decay in a dental x-ray, listening to a motorcycle engine to make a diagnosis, spotting a mistake in a string of computer code, taking photographs, painting, drawing, doing calligraphy, sculpting or potting, gardening, identifying birdsongs.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
This is but a small sample of perception-involving activities that offer poten tial for specialization and, accordingly, potential for expertise. We are all of us capable of being experts at something, perhaps many things. Given everything that has been said, it is plausible that many instances of such expertise – at least at the very highest level – will or can be perceptual in nature and in just the ways defended over the last two chapters. Therefore, there is good reason, in spite of the opponent’s rejoinder – which said that expertise is too specialized to generate broad claims about perception – to think that perceptual expertise is a pervasive human phe nomenon.3 It is not an isolated or rare phenomenon (like, say, the MüllerLyer illusion). And recall the architectural claims argued in Chapter 6, namely, that perceptual expertise is a genuinely perceptual and genuinely cognitive, expert-involving phenomenon. Chapter 7 then went on to defend such cases as exceptional cases of intellectual virtue. Bringing all of these lessons together, we have reason to think that human perception can change in culturally and cognitively sensitive ways. This conclusion – that thinking can affect perceiving, and indeed improve it – has significant consequences.4
8.2 Theory-laden perception and scientif ic investigation Let us consider Johannes Kepler: imagine him on a hill watching the dawn. With him is Tycho Brahe. Kepler regarded the sun as fixed: it was the earth that moved. But Tycho followed Ptolemy and Aristotle in this much at least: the earth was fixed and all other celestial bodies moved around it. Do Kepler and Tycho see the same thing in the east at dawn?” (Hanson 1958: 5) There is a sense, then, in which seeing is a “theory-laden” undertaking. Obser vation of x is shaped by prior knowledge of x. (Hanson 1958: 19)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
These passages will be familiar to many philosophers of science. It is a hypothetical example and early articulation of theory-laden observation.5 Dif ferences between theoretical motivations and views notwithstanding, N.R. Hanson (1958, 1969), Thomas Kuhn (1962), and Paul Feyerabend (1958) all suggested that theory can influence observation. Take “theory” to denote one’s (or a group’s) doxastic commitments, to include most obviously one’s beliefs, but also what one accepts, predicts, and expects and where this is sensitive to some theoretical or scientific community. What mode of obser vation might theory “load” or affect in some important way? The Hanson quote suggests that theory might affect observation all the way down, as it were. The rising sun appeared differently to Tycho versus Kepler: They enjoyed distinctive, conscious visual experiences as a consequence of their distinctive theoretical commitments. This is the version of theory-ladenness that will concern us here. But first some less problematic versions of the possible phenomenon must be addressed. Theory might influence the technology we use to aid observation and how we use that technology (or that we use it at all). Theory might influence observational judgment or interpretation. So even if perceptual appearances are static across competing theorists, somewhere between experience and the cognitive uptake or use of those appearances, theory intrudes. Theory might influence how we report our perceptual observations. This is some times called “semantic theory loading”. And finally, theory might influence the meta-standards that a scientist uses. Here the methods for collating, quantifying, and generally sharing data betray theoretical commitment in some way. There is thus a range of possible types of theory-laden observa tion, from perceptual experience all the way up to statistical analysis. Since initial presentation, some of these types of theory-ladenness have lost their bite as philosophical worries. For example, one might think it obvious that statistical analysis or meta-standards in a scientific discipline are theory-laden: They are in many ways just extensions of, or ancillary hypotheses of, the discipline, qua theory. And if some version of an underdetermination thesis is true, then loading observational interpretation with theoretical concepts or commitment is a needed way to start to fill the evidence-to-hypothesis gap.6 But nonetheless, traditionally, challenges are supposed to follow from non-perceptual forms of theory-ladenness. Identi fying those challenges and how they are (dis)solved is instructive, since this leads the way to addressing the challenge that is supposed to come with the perceptual form of theory-ladenness.
211
Copyright © 2021. Taylor & Francis Group. All rights reserved.
212
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Central to criticisms laid by both Hanson and Kuhn were the logical positivists and empiricists. As with many named philosophical “views” or approaches, there is no unified theory that answers to these terms. However, a claim that was espoused in some form by many theorists that fall under such terms was that the logic of scientific explanation is to be devoid of sub jective or psychological factors. These contextual features – facts about the scientist’s psychology – surely play a generative role as a mode of discovering explanation, but they occupy no privileged place in explanation proper, as the positivist understands it. Naturally, then, observation was supposed to be entirely theory neutral, providing raw sensory confirmation or disconfirma tion of the premises that structure the explanatory argument. Theory-laden ness, of some of the variety of types presented here, was a deep challenge to this model for scientific explanation and for the role of observation in the scientific enterprise. More generally, perhaps, theory-ladenness challenged epistemic foundationalism. If theory biases perception, or observational interpretation, or linguistic reports of those observations, this threatens the grounding role that observation is supposed to play for empirical investiga tion. That role is supposed to be one of providing a neutral and basic source of justification for belief and confirmation or disconfirmation of empirical hypotheses. Further generalizing, theory-ladenness of various forms may pose a general epistemic worry, namely, that observation may not track or represent reality in some strongly objective way. The solutions or dissolutions to these challenges, in brief, are as follows. Historically, the challenge to logical positivism and empiricism has been met with an abandonment of those philosophical approaches. Hanson and Kunh led that charge on some fronts, and others such as Quine (1951) and Pop per (1934) led the charge on others. Put bluntly, logical positivism became “dead, or as dead as a philosophical movement ever becomes” (Passmore 1967: 57).7 The solution, of sorts, to the challenge for foundationalism and the general epistemic challenge is more complicated. One thing that seems to unify the targets for the second and third challenge is an assumption that observation achieves, when accurate, a genuine, mind-independent objectiv ity. Some think that theory-ladenness is a threat, then, insofar as it is at odds with objectivity understood in this way. Objectivity is a contested notion in philosophy of science and epistemology, and the terms “objective” and “subjective” are used in a number of disparate ways in both theory and everyday language. So it is worth pausing for some terminological clarification. Most simply, the objective world is supposed to be the mind-independent world. Objectivity in this sense is contrasted with
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
the subjective awareness and experiences had by individuals. The tree before you is objective; your visual experience of the tree is subjective. A scientific thesis or process of investigation is objective if and to the degree that it accu rately reflects or tracks the mind-independent world. In the context of scien tific investigation, the strongest way to characterize a notion of objectivity is as achieving a kind of a-perspectivism: a “view from nowhere” (Nagel 1986). On this characterization, a theory or observation will be objective if it rep resents the facts of the world absent any subject’s or group’s perspective on those facts. This notion – call it Objectivity* – entails further that contextual factors such as social, political, and other values do not influence theory or observation if they are to achieve objectivity. And finally, idiosyncratic fea tures of any one individual should not bias the scientific process or its out comes. To accept that theory can influence observation at various levels is to accept that there is no view from nowhere, or at least not one achievable by human science. It is further to accept, according to many theorists, that values play some legitimate role in the scientific process. At the very least, epistemic values such as accuracy, consistency, broadness of scope, simplicity, and fruit fulness in generating research partly determine how science rationally pro ceeds (Kuhn 1977). One need not accept, however, that scientific investigation thereby becomes problematically subjectively biased or wholly relativistic. We can abandon Objectivity* while keeping a substantive objectivity. Since Hanson’s and Kuhn’s seminal work on the topic, a number of addi tional theorists have developed notions of objectivity that neutralize relevant perceived threats from theory-ladenness. One particularly powerful example is the contextual empiricism of Helen Longino, which in fact extends a point made in the earlier empiricism of Karl Popper (who himself sometimes flirted with Objectivity*). Popper claimed that “the objectivity of scientific statements lies in the fact that they can be inter-subjectively tested” (1934: 22). Shared norms and methodological standards for a scientific practice ena ble a kind of inter-subjective objectivity, and this can be so without Objec tivity*. Longino writes: There are several reasons to treat observation in the sciences as social. In the first place, such observation simply doesn’t or can’t consist in the per ceptual and sensory experiences of one individual. If the point of scientific inquiry is to explain observed regularities, we want assurance that an alleged regularity actually is one. This means that we treat the descriptions of obser vations as intersubjectively verifiable. . . . [This] grounds the principle of repeatability of observations and experiments. . . . Furthermore, negotiation
213
214
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
among members of an experimental group is frequently required to settle what the results of an experiment are, what counts as a genuine result, and what is rejected as an artefact of the experimental setup. (2002: 100)
Longino, and others,8 identify social and structural features of science as a collective enterprise. To accept that science is a socially sensitive enterprise is to abandon the pure Objectivity* of logical positivists, the scientific view from nowhere. Instead, scientific investigation is constituted by a collective of individual psychological agents. This social structure – including shared epistemic values, methodological standards of practice and experimentation, mutual goals amongst the community – provides a system of inter-subjective checks and balances. Therefore, there is subjectivity in science but these structures ensure that, when proceeding well or rationally, the subjective effects are not idiosyncratic or personally biased. And it is by virtue of intersubjective objectivity that theory-ladenness in its many varieties need not undermine the rationality of science.9 This provides grounds for neutralizing any worries that may arise regard ing “higher-level” theory-ladenness – including technology-aided observa tion, observation reports, meta-standards, and so on. But what, finally, of theory-laden perceptual observation? Is there not some epistemic threat that comes with theory-ladenness at this low or basic level? Does it amount to scientific relativism? The first thing to note is that the same checks and balances that neutralize higher-level theory-ladenness will go some way to neutralize possible pernicious effects of theory-laden perception. As Long ino emphasizes, scientific “observation simply doesn’t or can’t consist in the perceptual and sensory experiences of one individual”. To find regularities, to confirm or disconfirm hypotheses, and so on, many observations of many individuals must be held up against the inter-subjective norms and standards of the discipline. While there may be exceptional cases, the epistemic con text and structure of science neutralizes the effects of theory-laden percep tion. In fact, we can go even further and by returning to the seminal analysis given by N.R. Hanson: Causes certainly are connected with effects; but this is because our theo ries connect them, not because the world is held together by cosmic glue. The world may be glued together by imponderables, but that is irrelevant for understanding causal explanation. . .
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The difference between generalizing the repeated occurrence of con tiguous, propinquitous, asymmetric event pairs and understanding the ‘causal’ structure of a natural phenomenon is like the difference between having a visual impression of a lunaroid patch and observing the moon. . . [e]xperience and reflexion have given us good reason to expect a Y every time we confront an X. For X to be thought of as a cause of Y we must have good reasons for treating ‘X’, not as a sensation word like “flash” . . ., “bright”, . . . or “red”, but rather as a theory- loaded, explanatory term like “crater”, . . . “pendelum” . . . “discharge”. (Patterns of Discovery 1958: 64–65)
This is, again, a challenge to traditional accounts of scientific explanation. But it is not an invocation or implication of a sceptical worry. And to infer the second from the first is to mistake the nature of the critique made by Hanson and others. In this context, what was misguided about the strong empiricist model of science was that it assumed that scientific explanation is an entirely Objective* matter. Scientific explanation on that account is a purely logical relation between claims, devoid of input from the scientist and her context of investigation: no psychology, no sociology, no imagi nation or creativity. These contextual features play a role in the discovery of explanation, but they occupy no proper place in explanation. Naturally, then, perceptual observation was supposed to be theory neutral. Hanson (and Kuhn) was challenging this broad philosophy of science but not, it is crucial to emphasize, in the sense that he inferred that science was thereby epistemically doomed, or even that it was non-objective. Instead, and this is perfectly salient in the Hanson passages just quoted, the inference we are urged to make concerns the kinds of theoretical frame works and backgrounds that are inseparable from scientific explanation. Hanson’s profound suggestion is that this may go all the way down to per ception and the observations that we employ to support and challenge theo ries. And moreover, he is here suggesting that this is a good thing. Without these conceptual and theoretical tools and influences on observation, scien tific understanding is impoverished, if not lost. Hanson’s claim about causal understanding could not be clearer in this respect: To understand causal structure, the scientist needs more than mere observational terms and lowlevel sensation. She needs theoretical concepts. All the better, more opti mal, if these are somehow laden into the scientist’s fine-grained, perceptual observations.
215
216
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
If the analysis of perceptual expertise given here is correct, then it is just an instance of the same kind of phenomenon – of theory-laden perception – and one that involves more efficient, more sophisticated, and more diagnos tically useful perceptual observation. This is not to abandon the objectivity that concerns philosophers of science, epistemologists, and modularists (even if it is tantamount to abandoning modularity). But it is to abandon, once more, the Objectivity* of the radical empiricism of the early and mid 20th century. Paul Churchland, in the context of defending scientific real ism, made similar observations some two decades after Hanson and Kuhn: The function of science, therefore, is to provide us with a superior and (in the long run) perhaps profoundly different conception of the world, even at the perceptual level. This last position . . . is commonly indicted on grounds that it throws objectivity to the winds. Thus framed, this is an insensitive criticism. The manner in which the varieties of empiricism provide for objectivity (com parison of theory with some stable and theoretically unbiased level of per ceptual knowledge) is at the same time an analysis of what objectivity is. If the manner of provision is suspect, then so is the analysis. (Churchland 1979: 2–3)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The combined consequence of these analyses for objectivity is this. We must acknowledge that standards for accuracy within a scientific (or other) domain are not objective in a strict mind-independent sense. Those stand ards are instead sensitive to the inter-subjective, behaviourally relevant needs of trained individuals performing within the domain. In science, and in scientific observation, the view from nowhere is a rare (perhaps impossible) exception, not the norm.
8.3 Theory-laden perception, implicit bias, and epistemology A second set of concerns departs from the observation made in Section 8.1 that we humans are habit-forming creatures. We specialize in particular domains and sometimes this specialization amounts to distinctive percep tual skills. In other words, there is wide potential for perceptual expertise across the human species. If this is true, then what is to guarantee that nei ther of the following occur? First, what about “bad theories”, where by virtue of contextual circumstances, we specialize in a domain that distorts our
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
perceptual representation of relevant objects, features, or events? What if the contextually sensitive learning that we achieve concerns items of pseudosci ence or nonsense, and perception is thereby perniciously laden with those beliefs and concepts? Second, if cognitive effects on perception are allowed, what is to prevent that set of influences from including non-doxastic states, such as personal values and desires? Regarding the first concern, this is a genuine risk. If indeed the mind is malleable in the ways defended over the last several chapters, and all the way down to perception, then there is no way to block, as it were, problematic influences on perception. One can imagine so-called experts in activities that are not at all truth-conducive, and this would seem to dredge up some of the most worrisome features of theory-ladenness. For example, one’s beliefs in some pseudoscience – astrology is a favoured example of philosophers of science – might influence one’s perceptual observation, apparently confirm ing those very beliefs, and now one has gone in a vicious epistemic circle. However, in the context of philosophy of science, if we accept underdeter mination or theory-ladenness as coming in at the gap between hypothesis and evidence, then this is just a risk that comes with the territory. To accept it is to abandon the ideal of Objectivity*. It is to accept that science too is sus ceptible to substantial error and to philosophical scepticism. But further, and again, the risks of scepticism and relativism are neutralized by the norms and standards of some broader epistemic community. At least in practice, the astrological community is part of a broader epistemic community, and the inter-subjective intellectual standards of the broader community will provide a check against theory-laden observation should it occur as a result of the narrower community. Thinking more generally about the epistemology of theory-laden per ception, we can say the following. The conclusion of the previous chap ter was that some cases of perceptual expertise are instances of genuinely theory-laden perception. And further, cases of perceptual expertise are cases of intellectual virtue, where the virtue resides in the expert’s cognitively sensitive perceptual skill. It follows that some cases of theory-laden percep tion are epistemic virtues. What do we say of these cases of risk? We accept a familiar Aristotelian point: All virtues have opponent vices. That is, if the worry is that perception can be affected in problematically biased ways, then the worry is apt. Perception has, in the form of expertise, a potential to be an epistemic virtue. But although the malleability of perception can amount to an epistemic virtue, it can come to a vice if in the wrong hands, as it were. This is just to admit that we are human and science is a human enterprise.
217
Copyright © 2021. Taylor & Francis Group. All rights reserved.
218
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
It can get better and it can get worse, and we as individual perceivers can get better and/or worse. The potential for improvement or degeneration can go all the way down to perceptual observation. Here is an especially important and timely example. A pivotal discovery of 20th-century social psychology was of implicit bias: unconscious and invol untary attitudes or associations that are often inconsistent with the subject’s explicit, conscious, and well-considered beliefs about the same subject mat ter. Thus, disturbingly, many human beings display degrees of implicitly rac ist, sexist, homophobic, or otherwise prejudicial biases, even while explicitly disavowing those same biases (Greenwald and Banaji 1995; Gawronski and Bodenhausen 2006). Philosophers have begun to analyze this phenomenon (and, importantly, to acknowledge that implicit bias is a pervasive feature of the academic philosophy profession).10 There is some precedent for thinking that implicit bias is at least partly perceptual. For example, a great deal of work has been done on implicit bias of race, where subjects are more likely to make rapid judgments biased by one racial prime versus another. For instance, subjects are more likely to misjudge a tool as a gun when primed by a typical black male face (in contrast to a typical white male face) (Payne 2001, 2006) and in first-person shooter games subjects are more likely to shoot an unarmed black character than an unarmed white character (Correll et al. 2002).11 These responses are rapid and often occur along with real-time presentations of stimuli (as contrasted with post-perceptual reports). It is not at all implausible (which is not to say it’s proven) that some of these socially influenced biases are grounded in perceptual differences between those who suffer the bias and those who do not. And understood this way, we have a plausible instance of vicious theory-laden perception. This is for the simple reason that racial beliefs, whether implicit or explicit, are beliefs in a kind of theory. That is, it is highly plausible that racial categories are (at least partly) socially constructed, rather than tracking some mind-independently objective facts about humans. And further, if the several decades of social psychological research on racial bias and stereotype tells us anything, they tell us that such factors can play a systematic and devastating role in all facets of society, from education, to economic standing, to the justice system. A full analysis of the possibility of perceptually grounded racial bias will not be given here. But suppose for the moment that some racial biases are sometimes resident in perceptual experience and that this is an epistemic vice opposite the epistemically virtuous cases of perceptual expertise. How might we think about neutralizing this threat? This may be an instance
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
where top-down effects on perception are in fact needed in order to undo the societally grounded, top-down effects already done (namely, those that enable the implicitly biased perceptual representations). One such suggestion concerns face perception. Although we are all expert face perceivers, most subjects tend to be expert discriminators only of faces that are “within” their own race. However, recent studies have shown that subjects can be trained to recognize faces across races and reduce behaviours indicative of racial bias. Across a range of experimental approaches and subject populations, the cross-race effect (CRE) is empirically well established. While individuals within a race are exceptional (relative to other similarly complex stimuli) at recogniz ing, identifying, and recalling individual faces from within their “same race” (SR), subjects fail or frequently err along the same measures with respect to “cross-race” faces (CR) (see Meissner and Brigham 2001 for a review). Until quite recently, the two standard categories of theory focused on explain ing the phenomenon from putatively opposed angles: Perceptual expertise models claim that the phenomenon is a dominantly perceptual one, while social-cognitive models claim that the phenomenon is a dominantly social psychological one. Briefly, the perceptual expertise model explains the CRE as a matter of contact with individuals of “other” races, where when this contact is low, the perceptual learning mechanisms that enable expert-level face recognition do not sufficiently activate. The social-cognitive model, by contrast, explains the CRE as symptomatic of how individuals within a race think of SR members as being part of their ingroup and think of individuals in a distinct race as members of a relative outgroup. This categorical thinking discourages successful individuation of CR faces. These models have enjoyed inverse success and challenges. For instance, the contact hypothesis of the first model fails to explain substantial findings and data; mere exposure to CR faces seems to be only part of the story. And the second model, by contrast, struggles to explain the positive effect that perceptual exposure does have in some studies, and its prediction that salience of racial category should affect face perception is born out in only some experimental conditions (Young et al. 2011). Accordingly, more recent models have combined the perceptual exper tise approach with the social cognitive approach. Indeed, such models begin with the plausible observation that the perceptual and social-cogni tive models are not obviously conceptually or experimentally opposed. It hypothesizes that the CRE is a mix of perceptual and cognitive phenomena. Accordingly, the hybrid approach aligns with the discussion of expertise over the last few chapters. It hypothesizes that exposure is a part of, but not
219
Copyright © 2021. Taylor & Francis Group. All rights reserved.
220
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
the whole, explanation. And moreover it suggests that how subjects cognize racial and other social categories, whether held pre-experiment or experi mentally induced, influences CR and SR face perception in significant ways. For example, when subjects are primed with non-racial ingroup vs. outgroup categories (being from the same versus a different university) for SR faces, performance on composite tasks results in an advantage for the faces arbitrar ily primed as ingroup. These results are analogous to those discussed earlier in Chapter 6, suggesting that configural facial processing can be manipulated by social-cognitive factors (Hugenberg and Corneille 2009). Finally, hybrid models identify motivational factors as operative under a variety of condi tions. For instance, when white subjects are told that CR black faces are ones of an individual with power (say, the individual has a prestigious graduate degree), recognition of those faces is comparable to neutral SR faces (Shriver and Hugenberg 2010). Researchers suggest that this manipulation induces a motivation to individuate and thereby mitigates the CRE. The CRE in face perception is socially problematic, to understate the mat ter. Prejudice and aggression from police and other members of American society towards black males is a statistically grounded fact. But it is not the only example of a possible consequence of implicit (or explicit) bias and the CRE that may sometimes underpin it. Another example concerns eye witness testimony. The robustness of the CRE predicts that jurors will be prone to error in identifying faces from an outgroup race. This prediction is, sadly, born out. In one research study, one-third of wrongful convictions are identified as the result of cross-race eyewitness testimony (Scheck et al. 2003). This is a complicated phenomenon, but there is a clear sense in which the CRE (or at least some instances of the phenomenon) is a vicious case of theory-ladenness, where a bad theory about social or racial categories can affect perception and in deeply consequential ways. The positive side to this is that the effect is both perceptual and cognitive, and it can be mod ulated with the right social-cognitive interventions and perceptual training. Researchers are untangling the phenomenon and the best methods of inter vention. As these interventions become more and more clear, a vicious case of theory-ladenness can be overcome. As with the various theoretical and practical domains discussed in these last few chapters, with the right meth ods, we can make perceptual experts out of non-experts. It is worth emphasizing that this is no mere academic or theoretical point. If part of the problem with racial bias – both implicit and explicit – is failure to perceive important distinctions amongst individuals of “the other” races, then this kind of training is promising. It potentially undoes the “otherness”
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
of individuals within a race that is not one’s own and, perhaps, would elim inate some of the perceptually grounded biases that many possess. And the research suggests that similar phenomena occur along a range of social cate gories – including age, gender, sexual orientation, religion – and that they are similarly malleable (see Cross et al. 1971; Rule et al. 2007, 2010). In one sense, this would be cognitive penetration not as misperception, but as perceptual training, and indeed training of an epistemically and ethically improving kind. Here the malleability of the human mind has engendered culturally sensitive problems, but it also affords possible solutions to those problems.12 On to the second worry that began this section, which said that if theory-laden perception is granted, then what about the possibility of non-doxastic influence on perception? In some ways, we may have just iden tified such instances, where although through experimental manipulation, motivation or value seems to affect facial perception. Fodor argued that cases of “wishful seeing” would clearly be at odds with modular perceptual sys tems. Here again he gives an argument that appeals to reliability. Referring to Pylyshyn’s original notion of cognitive impenetrability (1980), Fodor writes,
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Pylyshyn’s point is that a condition for the reliability of perception, at least for a fallible organism, is that it generally sees what’s there, not what it wants or expects to be there. Organisms that don’t do so become deceased. (Fodor 1983: 68; emphasis added)
Broadly speaking, Pylyshyn’s point is correct, but not in the way the Fodor thinks it is: The point does not suffice to support modularity. Nor does the admission that non-doxastic states may influence perception undermine the account of thinking and perceiving – the TiP thesis – argued over the last many pages. The first thing to note is that the discussed cases of implicitly biased face perception discussed are not straightforward cases of wishful seeing as Fodor envisages. These subjects are not just seeing what they want to be there. In the CRE, subjects suffer a lack in expert face perception as a result, typically, of implicit racial biases rather than explicit, racially charged values or desires. This is precisely what makes implicit bias, generally, such a com plicated phenomenon: The biases that many individuals suffer are incon sistent with their explicitly held values and beliefs about social categories. And in cases where experimenters induce a value (as in the case where a cross-race face is of an individual explicitly categorized as being powerful),
221
Copyright © 2021. Taylor & Francis Group. All rights reserved.
222
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
the motivation induced is to make an identification, again, not to see what one wants to see. Second, an admission that value may influence perception is just a further admission that value may influence our attempts to mentally represent the world in some objective way. It is, once more, to abandon Objectivity*. But as many have argued, and as we have already seen, many such values are enhancing of successful representation of the world. Most obviously, the epistemic values identified by Kuhn (1977) and others – including accuracy, consistency, and simplicity – are con ducive to reliable perceptual and cognitive representations of one’s envi ronment. And additional intellectual values, such as open-mindedness, conscientiousness, perseverance, a desire to understand, are certainly compatible with representational success. Indeed, if some virtue episte mologists are correct, these are the very trait virtues that contribute to an epistemically virtuous life. They are values that do not undermine, but enhance reliable perceptual and cognitive processes. So it is far from a foregone conclusion that value-laden perception would be epistemically or adaptively worrisome. What, then, of clearly problematic non-doxastic influences on percep tion? What about cases of an organism simply seeing what she wants to see? Surely these would be problematic? Yes. They most certainly would be. And for the very reasons that Fodor indicates. “Perception is built to detect what is right here, right now – what is available, for example, for eating or being eaten by” (Fodor 1985: 4). A person who sees what she wants to see, in some straightforward and frequent way, will not perceive the world reliably. Such a phenomenon will also not be common, and again for reasons that Fodor indicates. Frequent wishful seeing does not result in “seeing what’s there” and “[o]rganisms that don’t do so become deceased” (1983: 68). Perceiving in this way, “generally”, would surely be maladaptive. Wishful perceivers would suffer fitness disadvantages. And so, in the evolutionary long run, they would be rare. Is any of this reason to think that such desire-influenced (or other problematic non-doxastic influences) perception is impossible as a conceptual or empirical matter? Absolutely not. At most, Fodor can conclude that wishful seeing is not a “general”, that is, frequent, occurrence. This is what observed reliability of perception ensures. Complete informational encapsulation of perception with respect to non-doxastic influence would require perfect or near-perfect reliability of perception. And that, alas, we do not have.13
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8.4 Objectivity, accuracy, and perceptual content Traditional philosophy of science to one side, there is another implication of the TiP thesis to be discussed. It concerns the relationship between the accuracy and success of perceptual content and the objectivity of perceptual content. The TiP thesis says that thinking improves perceiving. And in cases of expertise, this takes the form of intellectual virtue. In some of these cases, the virtue inheres in perceptual skills, where the agent is credited for the cognitive learning that partly constitutes the etiology of those skills. Some of these improvements take the form of enhanced accuracy: The forensics expert more reliably accurately distinguishes individual fingerprints, the elite goalkeeper better identifies relevant cues on the body of the striker, and so on. Other improvements are not captured by a straightforward or tradi tional notion of perceptual accuracy. Some perceptual experts see differently and in ways that optimize performance within their domain. They suffer less distraction from irrelevant perceptible details, recognize relevant details more rapidly, and enjoy enhanced sensitivity to organizational features and patterns. Generally, they better pick up information relevant to tasks of cate gorization, individuation, and diagnosis. Experts thus see better in a variety of ways: They sometimes see more, they sometimes see less, they sometimes see more quickly, and of course they do often see more accurately. Notice that the values of some of these perceptual advantages – of perceptual suc cess – are partly tied to particular, and often theoretical, domains of human culture – forensics and football, for example. Does this mean that perceptual accuracy and the like are dangerously relativistic? Is accuracy just in the eye of the beholder? This is what some theorists would have us believe: Relativism or worse is a foregone conclusion if we opt for the malleability of the mind, rather than “the fixed structure of human nature” (Fodor 1985: 5). The relevant argument starts with the conditional claim: Perceptual representation can be largely accurate only if perceptual systems are informationally encap sulated. Perceptual representation is accurate. Therefore, perceptual sys tems are informationally encapsulated, that is, modular.14 The analyses and criticisms of much of this book suggest that perceptual systems are not informationally encapsulated and thus not modular in any robust sense. In the good cases, thinking improves perceiving in sometimes direct and oth erwise importantly consequential ways. And with this potential for virtue comes the potential for epistemic vice, where problematic cognitive states
223
Copyright © 2021. Taylor & Francis Group. All rights reserved.
224
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
and processes can bias perception. The case made for these architectural and epistemic claims will not be reworked here. However, we can ask about the consequences of coupling the conditional claim with the malleability theses argued. Once more, the claim was that perceptual representation can be largely accurate only if perceptual systems are informationally encap sulated. Since perception is being theorized as malleable and so, plausi bly, informational encapsulation is being denied, one might be inclined to make a kind of Moorean shift, from Fodor’s modus ponens to a countering modus tollens. The conclusion would then be that perceptual representation is not largely accurate. And indeed this result is supposed to discourage denying the consequent, since it would seem to lead to relativism or phil osophical scepticism. The rationale for the relevant conditional claim is by now a familiar one. The idea is that informational encapsulation of perceptual systems ensures their functional independence, and this in turn ensures that none of the person, as it were, gets into the perceptual representations that result from those systems. Substantial cognitive influence on perception would under mine the accuracy and general success, given its subjectivity, of perception. And so even if the visual system proceeds inferentially and is informed by both innate and learned principles of visual organization, these are all sub-personal processes, importantly distinct from what the perceiver believes, values, expects, and so on. One way to be compelled by this line of reasoning is to assume that subjective influence necessarily introduces bias and where that bias results in unreliability and/or inaccuracy; this is the per nicious cognitive effects assumption. The analyses of perceptual expertise suggest otherwise: The subject is partly responsible for her perceptual skills, her activities and learning partly constitute the etiology of those skills, and those skills are epistemic virtues. And recall further that various arguments that reliable perceptual accuracy requires informational encapsulation proved inconclusive at best.15 What is left to motivate the claim that perceptual representation can be largely accurate only if perceptual systems are informationally encapsulated? To maintain this reasoning, it would seem that one must make an important assumption: The accuracy of perceptual content is determined by, or just is, Objectivity*. Standards for accuracy – what makes a perceptual expe rience accurate – are given entirely by reality, by the stimulus, absent any subjective influence or determinant. This assumption would secure the con ditional. However, it smuggles in a notion of objectivity that should not be the metric for the accuracy or success of perception. We have already seen
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
how Objectivity* is problematic as a notion of objectivity in science and for cognitive states like theoretical belief. The view from nowhere is not a plau sible phenomenon in our scientific and doxastic stances towards the world. It is no less problematic as the relevant notion in the context of perceptual content. Accepting this may come with theoretical revision, but not the rel ativistic or sceptical revisions forewarned. Recall from Section 1.5 that a plausible way to specify the content of a perceptual experience, e, is to specify the accuracy conditions for e. Sup pose one is looking at a crystal glass of red wine, on a wooden table, in afternoon sunlight. One therefore identifies how the world would have to be – what objects, property tokens, and events would have to be present and visually available to the perceiver, given her location – in order for e to be accurate. In this case, the accuracy of e would presumably require that there is a transparent but reflective stemmed glass, it contains a red liquid, is placed on a brown, flat surface, and so on. This (incomplete) list of objects and properties describes what would make e accurate and thereby says what e is about. Given this description, one might be tempted to think that there is a single set of accuracy conditions for any distinct perceiver that might be in these very same perceptual circumstances. Fix the basic sensory capacities of the perceiver, her viewing conditions, and all points in the visible scene, and you fix the accuracy conditions for that perceptual circumstance. Accuracy conditions, in other words, are an Objective* mat ter. However, if the analysis here is correct, then this temptation should be resisted. The accuracy and success of perceptual representation can be, to some non-trivial degree, perceiver and context-relative. But again, we can accept this without falling into a thoroughgoing perceptual relativism or worrisome scepticism. In other words, we can keep objectivity even if we give up Objectivity*. Perceptual accuracy is not just in the eye of the beholder. But neither is perceptual accuracy an Objective* matter. Perception is a process of objectifi cation in Burge’s sense. When one has the visual experience as of something being round or red, this involves the attribution of those features to some object outside of and independent of oneself. The instances of roundness or redness are experienced as being features not of one’s subjective experience but of objective reality. Importantly, this is compatible with error: The objec tivity of perception in this sense does not entail that perception is always accurate. Nor does it entail that accurate perception is Objective*. This is because perception is a capacity for sensory representation by an individual. And so its representations are context and perceiver-dependent. Burge writes:
225
226
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
The objectivity of perception should not be denigrated because it is not the context-independent objectivity of theoretical science. Perception is inevitably from the perceiver’s perspective. It is constitutively marked by egocentric representational frameworks. Such frameworks take their ori gins (spatial or temporal, for example) to be particular to the perceiver or some part of the perceiver. They mark the perceiver’s time or place. More over, such framework origins are essentially associated with motivational or perspectival matters for the individual. (2010: 401)
With the egocentric aspects of perception comes the possibility for error but also the possibility for improvement. The suggestion here, which goes beyond Burge’s commitments, is that our notion of perceptual accuracy and success should reflect this; it should reflect the role that the individual sub ject plays in the perceptual process. If we abandon Objectivity* as the metric of objectivity for scientific inves tigation, we should certainly not take it as rigidly determining the standards for perceptual accuracy. Instead, perceptual accuracy is partly determined by inter-subjective objectivity. That is, the determination of perceptual accu racy, those standards for perceptual accuracy, are sensitive to the features and details of the world that suit diagnostic, categorical, and other behav iourally relevant needs of the individual. And here it is important to note that the collective cognitive norms and values that determine the epistemic valence of higher-level knowledge seeking activities (like science) will also play a role in determining the epistemic valence of perception. So while this is not strict mind-independent objectivity, neither is it wholly egocentric or idiosyncratic: It is accuracy within a specialized domain. This is precisely what prevents a thoroughgoing relativism. It is not that a perceptual experi ence is accurate or not just relative to the perceiver in question. It is instead that perceptual experience is sometimes accurate (or, as will be discussed, successful) relative to the epistemic and/or scientific community in which it occurs.16 Suppose we have two distinct perceivers viewing a football player kick ing penalty kick after penalty kick. The first perceiver is a sports doctor, and he is watching this player perform her penalty kicks with the goal of assessing injury recovery. The player has recently suffered a hamstring strain, and the doctor wants to determine how or whether she is favouring the injured leg. The second perceiver is a goalkeeper, and she is watching for predictable reasons: to determine which direction the penalty kick will travel.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Suppose further (hypothetically) that both perceivers are watching from the very same viewing position. Suppose finally that both perceivers are experts within their domain. Now there is an obvious sense in which their percep tual experiences will be largely the same in content. Both will see the player kicking the football. But there is a more interesting sense in which they might be different. The sports doctor is likely to be sensitive to how evenly the player is distributing weight between feet, how she plants and drives off of her right versus her left foot, and other cues that aid her in making a medical assessment. The goalkeeper, as we learned in Section 7.3, will be especially perceptually sensitive to the positioning of the player’s hips as a reliable cue to which direction the kicked ball will travel. How do we say which of these two perceivers, given these differences in their experiences, is more accurately or better perceiving the penalty kicks? We don’t. Neither is better than the other absent mention of their respective domains; they are only better (or worse) relative to those domains. If the domain is one of medical diagnosis, then the first perceiver’s perception is better than the second. And vice versa if the domain is one of defending against penalty kicks. The quality of the per ceptual representation depends, then, partly on us: on what is behaviourally relevant in the context of perception. In this case, a component of accuracy is in the eyes not of the beholder but in the eyes of the community of the beholder. The two perceivers are members of, respectively, the community or domain of sports medicine and of playing football. The example illus trates how even when viewing the same events, by virtue of their distinctive domain-centred goals, two perceivers can perceive those events differently while both can be accurate and useful. This is paradoxical only if one insists on some singular, Objective* standard(s) for perceptual accuracy. Again in the context of scientific inquiry, Churchland (1979) discusses a related case. It involves two distinct groups of perceivers. The individuals in each group are entirely alike, except that one enjoys perceptual experience laden with beliefs about the more contemporary kinetic theory or explana tion of heat, and the other, perceptual experience laden with beliefs about the now-outmoded caloric theory or explanation of heat. Churchland’s related point is that the second group of perceivers are in an epistemically inferior position not because their experiences are conceptually influenced (this is no less true of the community of kinetic theorists), but because their related theoretical beliefs lack the same consistency, coherence, and explanatory power as their kinetic theory peers. And to assume that either is perceptually flawed just by virtue of theory-laden perception is, Churchland reasons, to assume that perceptual accuracy is or is determined by Objectivity*.
227
Copyright © 2021. Taylor & Francis Group. All rights reserved.
228
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
If the standards for success are partly inter-subjective in the way just described, is this really an account of perceptual accuracy? How one answers this question is a point of decision, and not an obviously crucial one. One may answer “yes” and claim that whether a perceptual experience is accurate is (sometimes) partly determined by the behaviourally relevant information in the context and, further, sometimes by the goals, needs, and standards specific to a domain of inquiry or community. This is to accept that perceptual accuracy is not a matter of some singular, Objective* facts about the stimuli. It is instead partly a matter of what features and pat terns are relevant to the task or goal within that domain, which could be individuating, categorizing, diagnosing, or acting on an object or event in some way. And this is not a mere matter of more is better. The goalkeeper’s perception of the penalty kicker is more accurate because it is optimally sensitive to a relatively small subset of the features that are visibly available; in this case, less is better. And this is crucial given how rapidly she must make a defensive movement. This kind of relativism about accuracy conditions is encouraged by a recent account of perceptual confidence given by John Morrison (2016). Morrison suggests that, for example, when environmental circumstances change – say, as a room gradually transitions from dark to well lit – not only doxastic confidence but perceptual confidence changes. In a case like this, visual experience assigns varying degrees of confidence about visible details in the scene. You begin with little to no confidence about the details of an object in the dark room and move gradually to a high degree of confidence that the object in the lit room is, say, a pumpkin on a chair. These assignments of confidence are, on Morrison’s account, perceiver-relative and part of the content of the experience. And this comes with a consequence for theories of perceptual accuracy: For a perceiver and a perceptual circumstance, there are no purely objective facts about what counts as a fully accurate experi ence. Instead, accuracy depends in part on “interest-level considerations” (2016: 39–40). Those interests can be shared or not: Sports doctors will share interests when observing the injured penalty kicker, and those inter ests may not align with those of goalkeepers in the same context. This does not imply that the sports doctor and goalkeeper disagree, and it does not imply that either’s experience is more accurate than the other. Instead, accu racy is determined by the standards of evaluation that come with the respec tive domains or communities. And relative to their respective domains, the experiences of the sports doctor and the goalkeeper are distinct in content but equally accurate.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Some theorists will resist this, insisting that if two individuals have sim ilarly well-functioning visual systems, are viewing in the same perceptible circumstances, and attending similarly, then the accuracy conditions (and thus content) will be the same for their experiences. If not (say the two perceivers are attending differently), then the accuracy conditions (and thus content) will be distinct for the two perceivers. The sports doctor visually selects distinct features and feature combinations in the scene than the goal keeper. Whether either experience is accurate is determined by whether those feature combinations are in fact instanced in the visible environment. Accuracy, this theorist concludes, is thus not relative to perceiver or domain; accuracy is not determined by behaviourally relevant factors or interest-level considerations. It remains a robustly objective matter.17 This may be correct, but one plausible reply is that it reveals that accu racy does not exhaust what makes for valuable, indeed successful, percep tual representation. Accuracy is instead subsumed by success. Recall the previous discussions of the fundamental behavioural role of perception (especially in Section 4.3). Contrary to a whole lot of 20th-century epis temological theorizing, we perceive very rarely for the sole intellectual purpose of knowing about the world. We do perceive to know, but we do so typically in order to act, to do things, to engage the world. And we do the latter even when we are not aiming for, or even employing as an intermediary, propositional knowledge. Expert perceivers are successful along a variety of measures that align with aims for action, performance, and worldly engagement. So even if perceptual sensitivity to patterns and organizational features, less susceptibility to distraction, and the other listed features of perceptual expertise do not determine perceptual accu racy, they do plausibly determine perceptual success. And either way, such perceptual skills are virtuous. And, as it has been argued, this virtue is grounded in facts about the world and facts about the agent and the epis temic community of which she is a part. A distinct but related analysis about perceptual content of biological organisms in general is given by Mohan Matthen (2005). Matthen argues against a kind of content universalism: where for any environmental situation, there is a singular correct perceptual content that any organism may have. The alternative, pluralistic realism, maintains that organisms from different bio logical taxa may represent different distal features, and all of these features may nonetheless be real in an action-relative sense. Matthen’s view, and the one suggested here, thus extends the Gibsonian tradition of action-oriented theories of perception. Objects and features in the environment have action
229
230
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
potential, and the value or success of perception is thereby partly determined by how it selects or represents that action potential (and, in turn, how it determines action). Another recent proponent of such a view is Michael Anderson, whose view takes the action orientation and embodiment of per ception as central.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Because perception is both active and in the service of action, much of the information to which organisms are attuned is not objective informa tion of the sort one might need for model-building, but rather relational information that is more immediately useful for guiding action in the world. It is the overall job of the organism’s brain and nervous system to manage various organism – environment relationships. Perceptual sys tems keep the organism in touch with the values of these relationships: the closeness of the obstacle, the support of the surface, the passability of the gap. (Anderson 2016: 7; see also Anderson 2014)
To miss this is to overlook what may be the most fundamental importance of perception – its role in behaviour. What, finally, does all of this imply for perceptual content? If the content of an experience e is given by the accuracy conditions for e, and we take the first answer to the question just provided, then accuracy so construed will determine the content of e. Accordingly, the sports doctor and the elite goal keeper will enjoy perceptual representations with overlapping content: They both have experiences that are about the player kicking the ball towards the goal. But their experiences will also have distinctive contents, and in spite of viewing the same events. For example, the sports doctor’s experiences will have contents that include how the player is distributing her weight between feet; the goalkeeper’s experiences will have contents that include how the player’s hips are positioned. Alternatively, suppose we maintain a more traditional account of percep tual accuracy, denying a role vis-à-vis accuracy to some of the features to which the expert is especially sensitive. These features nonetheless contrib ute, it was argued, to perceptual success, and largely by contributing to the action potential to which perception is giving us access. And so they provide success conditions for experience. And success conditions (which again sub sume accuracy conditions) are plausible determinants of perceptual content. To say what a perceiver’s experience is about, you specify how the world would need to be in order for her experience to be successful. This bears the
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
same intuitive result: The sports doctor’s and goalkeeper’s experience of the same events are partly distinct in content.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8.5 Expertise and the admissible contents of experience Perceptual experts are visually sensitive to patterns, gestalts, and organ izational features of objects of expertise. And indeed in some cases, they seem to be sensitive to – to just see – categorical properties, such as being a particular type of bird, or car, or Greeble. These properties seem not to be the standard low-level properties – colour, shape, rest/motion – to which vision is traditionally supposed to be sensitive. Does this mean that to accept the account of perceptual expertise – and the TiP thesis it is used to largely motivate – we must admit natural and artefactual kinds as constituents of perceptual content? The closing suggestion will be that it does not force this conclusion, but that it does provide materials for an argument for rich con tent of experience. As an alternative to discussion of perception of natural kinds, a productive way to begin this brief analysis is to consider aesthetic properties. The term “aesthetic” may bring to mind concepts of beauty, or concepts of (percep tual) experience, or both. And this would indeed accurately reflect a tradi tion going back to Hume, running through Baumgarten (who in fact coined the term “aesthetics”), and up through Kant’s Critique of Judgment (1790/1987). In this tradition, the dominant emphasis has indeed been on beauty. And much of that tradition has taken “the aesthetic” and its related terms to refer to something experiential, typically with explicit emphasis on perceptual experience. A more recent approach keeps the second intuition while shift ing at least some of the emphasis away from concepts of beauty and instead to substantive aesthetic terms, concepts, and properties.18 Frank Sibley (1959) offers an analysis of aesthetic terms and the concepts they refer to and extends this to a thesis about aesthetic properties. He begins with two lists: lists of non-aesthetic terms and lists of aesthetic terms. Included on the first list are: “red”, “hard”, “square”, “curved”, “sharp”. On the second list: “unified”, “graceful”, “balanced”, “serene”, “vivid”. Sibley suggests that terms on the second list refer to taste concepts. Taste concepts require for their proper use something more than mere perceptual capaci ties. Thus, any individual with “normal” perceptual capacities can correctly recognize something as red or hard or square. But to recognize something as unified or graceful or balanced, one must enjoy some additional aesthetic sensitivity. Sibley is famously elusive about what precisely this “something
231
Copyright © 2021. Taylor & Francis Group. All rights reserved.
232
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
more” comes to, but we can glean some lessons from what he does say. Regarding aesthetic concepts, one additional claim Sibley makes is negative: Using aesthetic (or taste) concepts is non-condition governed. There is no discrete set of rules or conditions or procedures that one can employ to cor rectly apply a concept like “being graceful”. And this dovetails with a thesis about the metaphysics of aesthetic properties. Sibley maintains that aesthetic properties emerge from combinations of basic physical descriptive properties, such that any change in aesthetic properties is a result of changes in more basic physical properties and the first do not reduce to the second. We might say that Picasso’s Guernica is vivid and dynamic. And we might justify this aesthetic attribution by identifying a subvenient physical description: It is, say, the sharp and angular figures composed in a tight, complex arrangement that makes the piece vivid and dynamic. However, the latter aesthetic prop erties are not simply a summation of those more basic physical properties. Again, aesthetic properties are, in some sense, emergent properties. An additional, though minimal, thesis about aesthetic properties says that they are nothing more than particular organizations or patterns of proper ties; aesthetic properties (or at least some of them) are gestalt properties. Sibley is correct – there is no single physical description that guarantees that some thing or event is dynamic, or balanced, or serene. But there is, for each of these properties, a distinctive gestalt or cluster of gestalts that character ize such aesthetic properties. Or, stronger, those organizational or gestalt properties just are the aesthetic properties. There is nothing more to being balanced than an organization or gestalt of basic features. And this is what one recognizes when one experiences something, a dance performance, say, as being balanced. Similarly for what some distinguish as artistic proper ties. Impressionist paintings are typified by a number of perceptible features: highlighting of natural light and reflection; a regular (but not categorical) use of lighter colours; identifiable quick, short strokes of paint; an emphasis on a scene rather than any one figure or group of figures; use of angles and com position creating a candid rather than posed depiction of people and events. To know impressionism is to know and respond to some cluster of these features. It is to enjoy a perceptual sensitivity to the impressionist gestalt. The difference between spotting and not spotting the balanced dance per formance or the impressionist painting is plausibly a phenomenological one: Recognizing, in experience, an event as balanced is phenomenally distinct from not so recognizing it. Further, this is plausibly a sensory phenomeno logical difference; in this case, it is at least partly a difference in one’s visual experience. Finally, the perceptual difference is plausibly one in high-level
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
or rich properties. It is not that the impressionist spotter (by contrast to the impressionist-naïve) perceptually represents distinctive colours or shapes or edges, but instead differently perceives how those basic perceptible proper ties are organized. She perceives the impressionism in the artwork. And this is rich perceptual content. (See Stokes 2018.) Now, Sibley characterized this special sensitivity as an aesthetic sensitivity, as taste. But if we consider the situation through the lens of perceptual exper tise, we see that there need be nothing particularly special vis-à-vis some elusive taste capacity. What’s special is simply what one has learned, and sometimes in very domain-specific and sensitive ways. The ballet instructor is especially adept at perceiving aesthetic properties in dancers, like being balanced, serene, and unified. And this is a consequence of learning and training. The art historian can well spot impressionist works and their fine details because she has learned a great deal about that artistic movement and is thereby highly sensitive to the organizational gestalts that typify works in that genre. Such individuals are, in the very ways discussed over the last many pages, perceptual experts. And this expertise may partly consist in the capacity for rich perceptual content.19 Does this mean that all perceptual experts are sensitive to aesthetic properties or that such accomplishments require rich content? No and no. Instead, it means that aesthetic experts are a subset of experts more gener ally. And some instances of expert accomplishment plausibly involve rich content, but this is by no means a condition on expertise or perceptual expertise. What the art historian and the radiologist share, if they achieve a level of perceptual expertise, is an enhanced perceptual sensitivity to pat terns, configurations, organizational features, to the gestalts of objects and properties specific to their domains of specialized training. Insofar as the distinctive perceptual representations that these experts enjoy do not involve (always or only) changes in low-level contents – in visual perception of colour, shape, edge, rest/motion – then they sometimes involve differences in rich contents. Just as there are patterns or gestalts that typify gracefulness, there are patterns or gestalts that typify a tumour in a mammogram or a pair of matching fingerprints. But none of this requires the more controversial commitment to perception of natural or artefactual kinds. Configurations and patterns are clearly perceptible; essences and conventions are not clearly perceptible.20 The argument here concludes only with a claim about the first – about rich content – without a claim about the second. Perceptual content can therefore be rich and in a way sensitive to background cognitive learning, without admission of natural kinds. And what’s novel is that this
233
234
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
lesson can be partly learned, and motivated, by considering cases of aesthetic experience. A lesson can be gleaned in the opposite direction as well, one concerning philosophical aesthetics. Perhaps it is this sensitivity to patterns, configurations, and gestalts that is the “something more” that Sibley claimed characterized competent appli cation of aesthetic concepts. That is, a taste concept in his view need not be elusive or extra-perceptual. Instead, it is a subset of expert-level percep tual skill; Sibley’s taste concepts require some expertise within the aesthetic domain. And if correct, this further illustrates the pervasiveness of percep tual expertise and the positive ways that thinking can improve perceiving. Although he sometimes described them in this way, Sibley did not intend for aesthetic sensitivity, for taste, to be a rare or exclusive perceptual or cog nitive capacity. And when understood as a subset of perceptual expertise, we see why it need not be. Just like the expert radiologist, aesthetic exper tise requires domain-specific, cognitively rich training. And the sensitivity acquired is best understood as skilful and, in some cases, virtuous. But this is an achievement that humans can make. We are, all of us, potential per ceptual experts.21
Copyright © 2021. Taylor & Francis Group. All rights reserved.
8.6 Summary This chapter generalized further from empirically studied cases of percep tual expertise. There is wide potential for cognitive effects on perception, from good to bad. Four concerns were then addressed, concerns regarding theory-ladenness and science, important cases of social and racial bias on perception, perceptual content and perceptual accuracy, and the admissible contents of experience. In each case, there is a call for important revision, but in each case the consequences are not as devastating or revisionary as the critic would have us believe. Science survives theory-ladenness and may even be improved by it (even if not in all cases). With the possibility of epistemically virtuous cognitive effects on perception comes the possibility of opponent vice. One such possibility are cases of perceptually rooted bias and stereotype. But here, too, there was reason for optimism; the CRE in face perception, for instance, is malleable. And if we can correct a percep tual ground for racial stereotypes, then perhaps new strategies for social justice arise. Finally, a malleable mental architecture encourages us to give up characterizing perceptual accuracy in terms of Objectivity*. It encourages an account where behaviourally relevant factors, and sometimes domain- or
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
task-specific ones, partly determine the success and thereby content of per ception and perhaps enrich that content. It turns out that the traditional emphasis on pure mind-independent objectivity is misguided in yet another way. There is a lot more of us, as individuals, in what we perceive. Further reading For discussion of scientific knowledge, value, and objectivity, see (in addition to the many sources listed earlier) especially (Kuhn 1977; Churchland 1979; Longino 1990; Laudan 1996). For review of empirical research on implicit bias and stereotype, see (Greenwald and Banaji 1995; Gawronski and Bod enhausen 2006). For more philosophically oriented discussion, see (Brownstein 2019; Beeghly 2020). For a review of the CRE in face perception, see Meissner and Brigham (2001), and for a review and analysis of alternative explanations of the effect, see Young et al. (2011). For a helpful overview of questions about perceptual content, see Siegel (2016).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Notes 1 See discussion in Section 6.4, and in particular Gauthier et al. (1997), Tanaka et al. (2005), and Scott et al. (2008), for studies on Greebles, birds, and cars, respectively. 2 Recall the discussion in Section 6.4. 3 Indeed, although the line of thought will not be pursued here, it is plausible that there are analogues throughout the biological world. Although in less diverse ways, many animals are experts in detecting behaviourally relevant objects and features in their environment. And the ones that do this excep tionally well are, all else being equal, fitter. It is entirely plausible that some of this “expert” performance occurs at the level of the animal’s senses, whether it is seeing a ripe fruit, smelling an appropriate mate, or hearing a dangerous predator. 4 To be clear, these consequences really only require the weaker claim. That is, the consequences to be discussed follow even if the reader only grants that the particular cases of perceptual expertise are to be theorized, archi tecturally and epistemologically, as argued. 5 Recall that theory-ladenness was discussed briefly in Chapter 4 as a possi ble consequence of cognitive penetration of perception. 6 See Duhem (1914), Quine (1951) for the two classic sources of this notion.
235
Copyright © 2021. Taylor & Francis Group. All rights reserved.
236
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
7 The qualification made by Passmore is important. As he goes on to say of the movement in the very next sentence, “But it has left a legacy behind”. As we will see, there is still residue of the positivist movement in some of the theories currently under consideration. 8 kuhn 1977; Hacking 1983; Cartwright 1989; Franklin 1986; Longino 1990; Laudan 1996; Mayo 1994; Brewer and Lambert 2001; van Fraassen 2002. 9 There are outstanding worries in the vicinity. kuhn and Feyerabend also worried about incommensurability between scientific theories. If two sci entists or, more commonly, two groups of scientists differ sufficiently in their theoretical posits and concepts, methodology, and other commit ments, they may be stuck in a terminal position of cross-talk, incapable of communicating (and thereby incapable of rationally disagreeing) about their theories and empirical findings. No suggestion is made here about a solution to these problems of semantic or methodological incommensura bility, since they are orthogonal to the central points at hand. 10 For a philosophical review piece, see Brownstein (2019). For two recent volumes of newly commissioned papers on the topic, see Brownstein and Saul (2016a, 2016b). 11 See Gendler (2011) for discussion. 12 For a related epistemological discussion, see Siegel (2017: Ch. 10). For dis cussion of face perception and cognitive penetrability, see Cecchi 2018. 13 Recall the research discussed in Chapter 3. 14 Recall the discussions in Chapters 2 and 7. 15 See critical discussions in Chapters 2 and 6. 16 Notice, then, how this mirrors the situation in the context of scientific inquiry as outlined near the end of Section 8.2. 17 Thank you to an anonymous reviewer for enforcing this rejoinder. 18 See Zangwill (2019) for review. 19 It’s worth noting, further, that this analysis is not far off from the one that Hume gave in his attempted solution to the problem of taste. For Hume, we resolve the tension between the apparent subjectivity of aesthetic judg ment and the apparent objectivity of aesthetic judgment by emulating the perceptual and cognitive characteristics of the ideal aesthetic judge. The ideal judge, in turn, provides a standard of taste. The related point here is that the ideal judge, even if idealized, is a perceptual expert. See Hume (1757/2000). 20 It is for this reason that kinds are perhaps odd candidates for rich contents, and on a variety of theories of kinds. On a realist account of natural kinds,
CONSeQueNCeS OF PeRCePTuAL MALLeABILITY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
what makes an instance k a member of some natural kind K is some under lying essence or microstructure of k. The k before you is a pine tree by vir tue of some underlying structure that k has. And these are not perceptible features of an object, pine trees or otherwise. Alternatively, if one is a con ventionalist about kinds, then the conventions (rather than some under lying microstructure) are what cannot be perceived. Or perhaps one does not think that things like pines are kinds, but instead are species (under stood as individuals or classes or sets). Here again, there will be criteria for being a member of a species that exhaust the perceptible, macrostruc tural features of any perceived biological thing, for example, the capacity to successfully interbreed or the sharing of a common ancestry. Indeed, this was a point anticipated even by Darwin in The Origin of Species: “adaptive characters, although of the utmost importance to the welfare of the being, are almost valueless to the systematist. For animals belonging to two most distinct lines of descent, may readily become adapted to similar conditions, and thus assume a close external resemblance; but such resemblances will not reveal – will rather tend to conceal their blood-relationship to their proper lines of descent”. For related analysis, see (Siewert 1998; Stokes 2018; Cutter ms). 21 For further analysis of relevant issues in aesthetics and philosophy of per ception, see (Nanay 2016; Seeley 2020; Stokes 2014, 2018).
237
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONCLUSION
This book has been about thinking and perceiving and, centrally, how the first might interact with the second in important ways. The last several chap ters have made the case for the rich malleability of the mind. First, thinking affects perceiving (the TaP thesis). This thesis was motivated by consider ing alternative ways to characterize the cognitive penetrability of perception (Chapter 4), starting with distinct, extant definitions and then moving to a consequentialist characterization of the phenomenon or phenomena. The basic thought there was that what matters most to both parties in the debate is that cognition may affect perception in ways important for the epistemic or behavioural roles of perception or the theoretical architecture of the mind. And the theoretical and scientific task then becomes one of identify ing cognitive-perceptual relations that do bear some of those consequences, irrespective of whether they satisfy the conditions of this or that extant defi nition of cognitive penetrability so called. This approach was then partly employed in the context of thinking about attention-involving cognitive effects on perception (Chapter 5). The standard treatment of such cases is to discard them as irrelevant to the phenomenon of interest because they involve an overt, often agent-directed, shift in spatial attention. That shift can be driven by cognition and can affect perceptual
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONCLuSION
experience, but it does so only by changing the input to perceptual systems. But this, it was argued, is dramatically short-sighted. Studies of attention have revealed that attention may operate in a variety of ways that depart from the evoked attention-shift interpretation. Relevantly, there are well-studied attentional mechanisms that can operate covertly and involuntarily, sensitive to behaviourally relevant object types or features, and independently of spa tial attention. And further, there are empirical cases where these selective attention mechanisms appear sensitive to background cognitive states and processes and in turn affect conscious perceptual experience. These cases are understood as genuine cognitive effects on perception by appeal to con sequentialism. Or, as was argued, they are plausibly interpreted as involving direct cognitive effects on perception, since the attentional mechanisms in question are not separate gatekeepers to perceptual systems, but indeed are part of those very systems. This provides evidence for the TaP thesis even if we maintain the stringent definitions of cognitive penetration given by the modularist. Chapter 6 marks the most important shift in direction of the book, which then persists through Chapters 7 and 8. It marks the shift to the claim that thinking not only affects perceiving but that it can improve perceiving (the TiP thesis). Before summarizing some of the details of those chapters and drawing out final conclusions, it is important to resituate that analysis with the critical analysis of modularity given primarily in Chapter 2. Recall that the central discussion of that chapter concerned the details of the modular theory of perception, focusing primarily on Fodor’s original articulation. The central commitment of that theory, argued in that chapter and stated explicitly by Fodor himself, is to the informational encapsulation of perceptual systems. Accordingly, the chapter proceeded to articulate and criticize two broad arguments for informational encapsulation – the arguments from the stability of perception and from the reliability of perception. These arguments break further into multiple sub-arguments. None of these arguments are suc cessful.1 Therefore, none of these arguments may contribute successfully to the presumed master argument for modularity, which is supposed to be an inference to the best explanation. And absent inclusion of those various arguments, it is difficult to see what the abductive argument for modularity is supposed to be. The importance of this negative conclusion cannot be overstated. In many quarters of cognitive science, it remains a default assumption that percep tual systems like vision are modular in Fodor’s sense. And in the cognitive
239
Copyright © 2021. Taylor & Francis Group. All rights reserved.
240
CONCLuSION
penetration and related literature, it is at least implicitly (and often explicitly) assumed that the modular theory is the one to be disproved.2 Anecdotal and empirical cases are regularly presented as counterexamples to the alleged informational encapsulation of visual and other perceptual systems. Theo rists on either side of this and related debates behave as if, to employ one of Fodor’s favourites, modularity “is the only game in town”. This, recall, is the default position assumption. If the critical analysis of Chapter 2 is successful, then it reveals that this is, at best, a sociological fact. For a theory to be the default, and for it to be the one against which new theories or explanations must position themselves, it should be proved (to some degree). It should at least be supported by strong arguments and powerful empirical predictions. The conclusion that surfaces from that analysis, coupled with various portions of the analyses in chapters that follow, is that modularity in fact is supported by neither. Therefore, it should enjoy no such theoretical pride of place. One might reply to all of this with something like the following thought. But surely modularists like Fodor and Pylyshyn were right that vision and other perceptual systems are largely functionally independent. Surely they were right that “at least some of the background information at the subject’s disposal is inaccessible to at least some of his perceptual mechanisms” (Fodor 1983: 66). Surely it is not true that at any given time, all of one’s back ground beliefs, goals, expectations, and other cognitive states influence all of perceptual processing! This rejoinder is correct and unfortunately com mon, but also irrelevant. This is indeed how the modularist often presents his opponent. But as was also discussed in Sections 1.3 and 2.2, it is not a genuine opponent, neither in the form of the New Look psychologist nor the cognitivist.3 It is instead an uncharitable, and unfortunately popularized, story about those theoretical approaches.4 All of this shifts the dialectic. First, it undermines the default assumption to modularity in the science and philosophy of perception. This should open doors for new, and more fruitful, experimental and theoretical approaches. When a scientific discipline realizes that it is unnecessarily committed to limiting assumptions, remarkable things can happen. Chomsky’s supplanting of Skinner’s behaviourism, for example, comes quickly to mind. And second and related, it reveals that the field is level at best. The modularity theory has no grounded advantage over genuine theoretical opponents. Stronger, those theories need not be characterized as opponents to modularity, since they need not present as opposition to an ungrounded theory. Accordingly, the merits of the positive explanations of various phenomena can be con sidered independently. And they can, crucially, be considered independently
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONCLuSION
of whether they meet some criteria set by the modularist himself on what “counts” in the relevant theoretical space. This brings us back, finally, to the proposed malleability architecture. The thesis that thinking improves perceiving (the TiP thesis) was argued across Chapters 6 to 8. Marking the dialectical shift casts the analyses of these chap ters in a partly distinct, more positive light. Even if claims and arguments for modularity are considered and challenged in the context of perceptual expertise, the explanations given for the latter phenomenon can be consid ered on their own merits. They need not, in other words, be understood as counterexamples to the presumed default theory. And further, if they are positioned in this way, the explanations for the empirical cases of perceptual expertise are clearly better than the modularist alternatives. If anything, the burden of proof shifts inversely relative to its standardly assumed place. That dialectical shift is not the only important shift that these chapters introduce. They also introduce novel emphasis on good cases of perception. Against the preponderance of emphasis in philosophy of perception on bad or normal cases of perception, the emphasis on perceptual expertise is an emphasis on cases where individuals improve perceptually, and sometimes in near-optimal ways. The hope is that the results of this emphasis encour age other philosophers of mind and perception to follow suit. A final way to motivate that prescription is to recall, broadly, some of the details and the conclusions of the previous three chapters. Chapter 6 provided a review of a range of substantial empirical literature on perceptual expertise, including a variety of behavioural, neurological, and eye tracking experimental methods. The two claims that result from this review are that perceptual expertise is a genuinely perceptual and genu inely cognitive, expert-involving phenomenon. This conjunction of claims, it was argued, best explains that range of empirical data. Those data suggest that perception can actively change – neurophysiologically, representation ally, phenomenologically – and in ways sensitive to domain-specific, cogni tive learning. The chapter also considered how the modularist might address cases of perceptual expertise. Put simply, it is faced with a dilemma: ignore such cases of perceptual expertise as simply too special to generalize from or give up the theory. Later analysis (Chapter 8) reveals just how untenable the first option is. The remaining option is to give up modularity. The architectural claims regarding perceptual expertise defended in Chap ter 6 make a strong case for the thesis that thinking richly affects perceiving. Chapter 7 goes further, providing an analysis of the epistemology of those same cases in aims of supporting the thesis that thinking improves perceiving.
241
Copyright © 2021. Taylor & Francis Group. All rights reserved.
242
CONCLuSION
The most illuminating epistemology of expertise focuses on the agent herself and how she has improved her perceptual skills through specialized, concept-rich training. Because of this training, the expert perceives more accu rately within her domain, but also enjoys greater and more rapid perceptual sensitivity to patterns, organizational features, category- and diagnostic-spe cific information, and with less distraction. These skills, for which the expert is partly responsible and so to be credited, are improvements partly relative to the inter-subjective goals and technologies of the domain in question. More generally, they are improvements relative to the general representa tional function of perceptual systems and the norms that govern adequate performance of that function. The virtue of the expert’s improvements is therefore grounded in the general biological function of perception. Chapter 8 asks how far the analysis of the previous two chapters gener alizes. How common are perceptual experts, and what does this imply for theories of mind? Although the achievements of perceptual experts are laud able and their success is highly specialized, they are not super-humans. They are instead merely individuals who, through specific training, practice, and learning, have acquired domain-enhanced perceptual sensitivities. We are, all of us, potential perceptual experts. And many of us, quite probably, are already perceptual experts in some domain(s) or other, given the fact that we are, all of us, habit-forming creatures. With this virtue, comes a risk, namely the possibility for vicious theory-laden or otherwise cognitively influenced perception. But if indeed perception is malleable in the ways argued here, such risks may be neutralized with appropriate cognitive and perceptual training. This was the suggested recourse to the problematic case of implicit and explicit bias, which may in some cases reach all the way down to face perception. If we can train individuals to perceive cross-race faces as indi viduals rather than as in “the other” category, this could potentially go some way in the battle against devastating racism and other prejudices. The consequences of the TiP thesis are not only epistemic. The thesis will also force, it was argued, a revision to traditional theories of perceptual content. Determinants of perceptual content are not Objective*; they are inter-subjectively objective. They include facts about the environment, but also facts about the perceiver’s epistemic community, which can be very broad or quite narrow. Perceptual success, including accuracy, is determined in part by the task or goal of the perceiver, which can be specific to a domain. The expert goalkeeper’s task is to move her body in such a way as to block the striker’s penalty kick. Success in this instance is maximized by enhanced visual sensitivity to the positioning of the striker’s hips. This is part of the
Copyright © 2021. Taylor & Francis Group. All rights reserved.
CONCLuSION
perceptual content of the goalkeeper’s experience. Accordingly, two individ uals can both enjoy successful, accurate experiences of the same event, even if the content of those experiences is distinct. A sports doctor viewing the same event in order to make a diagnosis of recovery progress would plausi bly enjoy an experience with slightly distinct content (but importantly distinct, given her different goal or task). Enhanced perceptual sensitivity of this kind – to patterns, gestalts, and organizational features – is to enjoy rich perceptual content. The radiologist sees not only the colours, edges, and shapes of the mammogram but also how various of these features are arranged in diagnostically relevant ways. The ballet instructor sees not only the colours, edges, shapes, and motion of her pupils but also how those features are organized in ways that are balanced or serene or graceful. In both cases, the perceptual achievement is one to be credited to the perceiver: The expert radiologist and expert ballet instructor are partly responsible for the etiology of their perceptual skills. They enjoy rich perceptual content, which is relevant to their goals and tasks, because of what they have learned in their respective domains. Think ing thus improves perceiving by enriching its content. To enrich one’s perceptual content is to enrich how one is in contact with the world. Contact with the world is, it was suggested in the first pages of this book, of fundamental human importance. We make sensory contact with the world. We also make cognitive contact with the world. The domi nant, indeed default, orthodoxy in cognitive science has it that the first – sen sory contact – is largely a matter of biology. How individuals see, hear, and otherwise perceive the world is determined by how their sensory systems work, and these systems work in a way largely independent of personal-level cognition – what the individual has learned, the concepts she has acquired, the expertise she may possess. The claims that thought affects perception, and that thought can improve perception, imply that this orthodoxy is both mistaken and limiting. There are empirically studied cases best explained as genuine and consequential cognitive effects on perception. Cases involving visual attention are especially compelling in this regard. And, more impor tantly, there are cases of personal-level perceptual improvement. If percep tual expertise is a genuine and pervasive phenomenon, then we can change how we perceive the world for the better, and in many cases to a degree that achieves epistemic virtue. Overturning orthodoxy thus opens important new avenues for empirical research, on how our sensory contact with the world changes as a result of our being and learning in the world. This means a partial reconstruction of
243
244
CONCLuSION
some standardly theorized mental architectures, and with that reconstruc tion may come new angles and methods of study. But this is not all. It reveals a humanistic insight that the default orthodoxy does not allow. Namely, it reveals a much closer connection between us as individual persons and the world as we perceive it. If thinking can improve (or hinder) our perceptual experience of the world, and partly through our own efforts, then we are genuine agents in how we make contact with the world. And this agency runs much deeper than traditionally supposed. The conclusion is not philo sophical scepticism or radical relativism. The conclusion is, though, that our contact with the world is very much our own.
Notes
Copyright © 2021. Taylor & Francis Group. All rights reserved.
1 Recall, further, that additional variations on some of these arguments are considered, and criticized, in Chapters 6 to 8. 2 For one clear example of how pervasive and persistent this strategy is, and on both sides of the debate, see Firestone and Scholl (2016). 3 Recall from Section 1.3, for example, that although Pylyshyn (1999) notes that Bruner explicitly states that he maintains a cognition/perception dis tinction, even if arguing that the first may influence the second, Pylyshyn goes on, at great length and over the course of years of publications, to criticize Bruner and the New Look as if they make that very claim. 4 Indeed, such characterizations sometimes even take a fictional narrative form. See Fodor (1985).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
EPILOGUE: THE ARGUMENTS
FROM UNDERSTANDING
The analysis in all of the foregoing chapters is in a mode familiar to contem porary philosophers. We might call it, and some do, scientifically informed analytic philosophy of mind. Its aim is almost entirely descriptive, and where it enters into a prescriptive mode, it is a meta-theoretical one: pre scribing how descriptive theories or research might better proceed. This epi logue departs from that mode, and so its appeal will vary dramatically across readers. The discussion departs, though, by making an undeniable descriptive observation. It is far from conclusive which mental architecture – and in particular how thinking may (or may not) influence perceiving – is correct. The science is not there yet. And insofar as a theory of mental architecture is an empirical one, the philosophy is not there yet. And if we take underdetermination seriously, we may never reach a conclusion if we assume that there is a singular discovery that we are attempting to identify. We may be then in a position of decision rather than discovery. Or put more weakly, it may be that some important choice points in theorizing the structure of the mind are not going to be made on descriptive or observational or even straight empirical grounds alone. Suppose that this is correct and this is the theoretical position we find ourselves in. If so, we need some other criteria
Copyright © 2021. Taylor & Francis Group. All rights reserved.
246
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
or considerations to assist in making those theoretical choices. This epilogue argues that those criteria should concern understanding: how we understand ourselves and how we understand ourselves in the world. Accordingly, and again, the appeal of this argument will vary across readers. Some will con sider the argument to be pragmatist in nature. Others might say it is human istic. Others might even cast it as existential. It is certainly impressionistic in nature: It is not a full or sustained argument, but a sketch of one. It is an epilogue meant to offer some hopefully provocative suggestions and direc tions for future research. Whatever the label, the final argument is that we achieve greater understanding and self-understanding if we accept a mallea ble architecture and the pair of theses argued in this book: Thinking affects perceiving and, in some cases, thinking improves perceiving. Consider a personal anecdote. When I tell non-specialists or undergrad uate students about the book I’m working on (this book) and generally my research on cognitive influence on perception, the overwhelmingly most common response takes forms such as these: “Of course different people see the world differently”; “It seems obvious that what a person knows or believes or thinks can change how she perceives things”; “How could we all see the same given that we are all different?” And so on. Now, nobody likes to have their work diminished to triviality or shrugs of disinterest, so these responses evoke a standard form of counter-response. I will reply by employing a strategy common in much of the cognitive penetration litera ture (and, more broadly, in literature on perceptual content). You have to tell the layperson what needs to be “held fixed” and then attempt to isolate the phenomenon that is of interest. Thus, if we are considering two perceivers and whether their different cognitive backgrounds can genuinely affect (or not) how they, say, visually perceive something, I will probably say things such as: Assume that the two perceivers are in the same viewing conditions. And assume they enjoy equally well-functioning visual systems. Assume fur ther that both perceivers have their attention fixed on the same stimulus and indeed the same features of that stimulus. Finally, we are interested not in the immediate judgments that the two perceivers might make, nor what they might report on what they see, nor how they might remember their visual experiences, but only in their conscious visual experience in that moment. So, their experience before they judge it, interpret it, form beliefs about it, and so on. These last few qualifications often result in suspicion or confusion. And so I persist by saying something like: The very look of some part of the visual array has to be different for the two perceivers (holding fixed all of the mentioned features of the case) in that moment. There! It’s a kind of
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
possible cognitive effect on that, on conscious visual experience, that is supposed to be controversial, I conclude. Now in some instances, the individual finally gets it, and she may even see how there is some genuine space for interest ing debate here. But others (or in some cases, the same individual) remain suspicious or puzzled, and perhaps because of all of the qualifications that had to be made. And this suspicion, it seems, betrays a certain feature of this way of describing perception: It is deeply unnatural as a description of our actual lived, perceptual experience. It is an unnatural way to describe how we experience our own mental lives. Diagnosing what seems unnatural about this picture, at least from the perspective of the ordinary perceiver, is useful. First, it describes human perceivers as enjoying clearly discriminable mental states. It assumes that we can always, or at least a great deal of the time, distinguish our visual experi ence from the beliefs, judgments, and other mental states that are supposed to be post-perceptual. It assumes further that the relevant visual experience is discriminable, for the subject, from (putatively) pre-perceptual attention. All of this is to assume that these differences are (always or most of the time) real or important ones from the first personal perspective. And when you ask an actual person about her own sense of her mental life (such as one of my undergradu ates!), you find that these descriptions and assumptions do not comport well with that experience. There is a second, and related, temporal component that exacerbates this mismatch. The “holding fixed” technique attempts to isolate the phenomenon of interest – perceptual content and how it may vary with different cognitive backgrounds. And as indicated, sometimes the technique works in illustrating that targeted phenomenon. But it does so at a cost: isolating a phenomenon that may be alien to most perceivers. And this for the following reason: To finally isolate the phenomenon that may be dis tinct between two (cognitively) distinct perceivers, you don’t just fix view ing conditions, attention, and so on, you fix time. You describe the visual experience as if it is a mostly static snapshot of the environment or stimulus so that you can say: “There! that’s the thing – say the colour appearance or shape appearance – that is or is not different between the two perceivers!” The problem? We don’t experience the world through a series of static per ceptual snapshots. We experience the world, or at least this is what ordinary reflection would suggest, through a perceptual process. None of this is to suggest that the distinctions in mental states or kinds should be abandoned. It is not to express a thoroughgoing scepticism about mental architecture; this would be an odd claim to make in the context of the present book. It is, though, to claim that a certain way of characterizing
247
Copyright © 2021. Taylor & Francis Group. All rights reserved.
248
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
how various mental states and processes interact is at odds with how we seem to experience the world in the ordinary course. This is another symp tom of the continued dominance of both modular and computational the ories of perception. To isolate the visual (and other perceptual) phenomena that are supposed to be informationally encapsulated and therefore func tionally autonomous, we have to make a whole lot of “fixing” assumptions, all the way down to construing perceptual experience as a static snapshot. We then consider pairs of perceivers with relevantly different cognitive backgrounds and ask if their respective snapshots are the same or different. In familiar classical computational language, the verdict is that the outputs of the two visual systems are the same. Many opponents of modularity (this one included, in previous work) follow suit and attempt to identify cases where the perceptual phenomenon is isolated by these very same means, and where the two snapshots differ in some relevant way. But we have been here before: This is just the default position assumption under a different guise. In earlier criticisms of this assumption, the worry was that the assumed default position – modularity – is insufficiently supported by argument and evidence. Insufficient, at least, to hold a privileged position in scientific and philosophical theories of mind. Here the issue is a humanistic one. That assumed position encourages a picture of mental life that is extremely hard to understand, let alone accept, when we consider how mental life actually strikes us. The static snapshot characterization, then, is ecologically unsound. We do not perceive the world in isolable moments. We instead perceive the world, and more generally make mental contact with it, in a dynamic, ongoing process. That process takes time, and freezing it for introspection is difficult at best. For this reason, the characterization is not just ecologically unsound but, as it were, laboratory unsound. One of the most difficult features of “testing for” the possible top-down effects on or cognitive penetration of perception is attempting to isolate, even in highly controlled experimental circumstances, the phenomenon that is supposed to be of interest. Even with well-designed experiments, the results are often compatible with explana tions in terms of genuine cognitive effects on perception and with a variety of opposing explanations, say, in terms of cognitive effects on non-perceptual states like belief or judgment. It is scarcely easier to “fix” all of the relevant features of a perceptual context in the lab – to isolate the perceptual snap shot – than it is in ecologically valid perceptual contexts. The lesson is the same: Perceiving is a process, not a series of snapshots.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
This is hardly a lesson when we are not steeped in theory. Perceiving, like all other mental phenomena, is part of our lives. Indeed, on many theories of the self, our mental lives just are, in some sense, who we are. In that context of observation, it is not controversial that perceiving is a process. Life, and mental life, is a process. And perceiving with it. The lesson is thus surpris ing, if it is, only in distinctive theoretical contexts. If mental architecture is theorized as classically computational, as the modularist has it, then percep tual representations (like all mental representations) are atomic outputs of computational processes. But the modularist is not alone in assuming something like the static snap shot account. Much of the discussion on perceptual content in the foregoing chapters, which partly reports on related extant literature, at least tacitly assumes that perceptual representations have clearly and sharply delineable contents. I have a visual experience as of some singular event: The red phone is sitting on the table. The parsley is to the left of the knife on the cutting board. The cat is licking its paws. Perhaps such sparse determinations of content are offered for ease of exposition, as we say. But this is not the whole explanation. There is epistemological ground for characterizing per ceptual content in this way. The still-dominant epistemic value theorized by contemporary epistemologists is knowledge. And although there have been important deviations – for example, to theorizing procedural and social knowledge – by “knowledge” most epistemologists still have in mind propo sitional knowledge. And to know that P, is to know some proposition expressed by a truth evaluable, descriptive sentence. For instance, one might know that: The red phone is sitting on the table. The parsley is to the left of the knife on the cutting board. The cat is licking its paws. From here the story is by now a familiar one. These are all of them propositions that one plausibly knows on the basis of perception. Set to one side important worries about exactly how perceptual experiences of the relevant events enable knowledge of the corresponding propositions (about how perceptual experience must accordingly be structured, about the nature of epistemic justification, and so on). Whatever story one gives, it standardly starts by describing similar (or identical) contents between perceptual experience and resultant knowledge. One knows that the parsley is to the left of the knife on the cutting board because one has a visual experience as of the parsley to the left of the knife on the cutting board. The dominant value of propositional knowledge thus encourages, at least on the surface, an emphasis on perceptual experience as static snapshots.
249
250
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
In the context of epistemology and its dominant emphasis on the value of knowledge, thinking of perceptual content in this way may be ineliminable. But this does not mean that it is exclusive of other epistemic values, and the ways of theorizing perception that may come with those values. Another epistemic value, not incompatible with the value of knowledge, is understand ing. Indeed, some have argued that understanding should supplant knowledge as the central value in epistemology (Kvanvig 2003). But we can acknowledge the value of understanding without commitment to this more controversial claim. Our question then becomes, simply, what is understanding, and how does the analysis given over the last several chapters engage with its value? Jonathan Kvanvig describes understanding as follows: understanding requires the grasping of explanatory and other coherence-making relationships in a large and comprehensive body of information. One can know many unrelated pieces of information, but understanding is achieved only when informational items are pieced together by the subject in question. . . . [u]nderstanding requires, and knowledge does not, an internal grasping or appreciation of how the vari ous elements in a body of information are related to each other in terms of explanatory, logical, probabilistic, and other kinds of relations that coher entists have thought constitutive of justification. (2003: 192–193)
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Similarly, Wayne Riggs writes: The kind of understanding I have in mind is the appreciation or grasp of order, pattern, and how things “hang together”. understanding has a mul titude of appropriate objects, among them complicated machines, peo ple, subject disciplines, mathematical proofs, and so on. understanding something like this requires a deep appreciation, grasp, or awareness of how its parts fit together, what role each one plays in the context of the whole, and of the role it plays in the larger scheme of things. (2003: 217)
And Linda Zagzebski traces the notion back to Plato and Aristotle, contrast ing their central emphasis on understanding with the Cartesian central emphasis on knowledge and certainty: [O]ne does not understand a part of a field without the ability to explain its place within a much larger theoretical framework, and one acquires
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
the ability to do that by mastering a skill. . . . One gains understanding by knowing how to do something well, and this makes one a reliable person to consult in matters pertaining to the skill in question. . . . understanding is not directed toward a discrete object, but involves seeing the relation of parts to other parts and perhaps even the relation of part to a whole. It follows that the object of understanding is not a discrete proposition. (Zagzebski 2001: 241)
There is a striking convergence in these characterizations of the notion of understanding. To our purposes, there is an equally striking convergence with the suggestion that we think of perception not as a series of static expe riences, but instead as a dynamic process, and further a convergence with the analysis of perceptual expertise and accordant malleability architecture. As acknowledged, a singular emphasis on propositional knowledge as our epistemic goal encourages an account of perceptual content in terms of static snapshots. Suppose instead that we identify understanding as an alterna tive (perhaps additional) epistemic goal. The themes that run through the characterizations of understanding given by Kvanvig, Riggs, Zagzebski, and others are these. First, understanding does not target single propositions as knowledge does. We can, of course, accurately say that someone understands that P, and in doing so we are using the term in a way that approximates (or just is) the sense of “knows” that exercises analytic philosophers. But the sense of “understand” of interest here is different. It targets a body of information or elements specific to a particular domain. The domain could be practical or theoretical. Second, understanding is an achievement, and one involving skill. The skill will, accordingly, be specific to the domain in question. Finally, understanding involves a sensitivity to or grasp of how the body of information in that domain fits together. To understand a domain is to appreciate, in a way that allows for explanation, the patterns, coher ence, structure, relations, and so on between the elements that comprise the domain. Notice, then, that achieving understanding is typically not some thing that could be done in a snap. It is a process. Many domains where we achieve understanding are ones that involve, in some essential way, sensory perception. The reader should not have to reflect long to identify examples. This book is populated by them: radiol ogy, forensics, football, to name just three. To understand in domains such as these, one must enjoy a sensitivity to important patterns in the domain and a grasp of how elements in the domain are organized and how they cohere. The radiologist understands the structure that the mammogram dis plays and how an aberration deviates from the good health of that structure.
251
Copyright © 2021. Taylor & Francis Group. All rights reserved.
252
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
A forensics examiner uses bifurcations in a fingerprint to determine a match with another print, a detail of which most of us are entirely unaware. An experienced goalkeeper will recognize patterns in the actions of the oppos ing players, noting changes in those patterns as they develop in real time. It should be clear that achieving understanding in these ways takes time. This is partly for fairly trivial reasons: Some of these domains are ones where the patterns range over time in addition to space. The offensive pattern rec ognized by the goalkeeper will often emerge only after several movements by several opposing players, and this takes time. And some individuals who achieve understanding in such a domain will take longer than others. For some, however, such understanding can sometimes be achieved extremely rapidly. Experts can sometimes achieve understanding just by looking. And sometimes in a blink of an eye. Begin with face perception, a domain relative to which we are most of us perceptual experts (the important qualifications discussed in Sec tion 8.3 notwithstanding). All the evidence suggests that face recognition is not a mere summation of or sensitivity to individual facial parts or features. It is instead holistic and involves exceptional sensitivity to the organization or pattern – to the whole that is distinctive of a face. And of course that’s not all. Many of us, even upon a single viewing, can recognize and identify specific faces. Face perception thus involves understanding, and some of the time, understanding achieved rapidly. We might call it perceptual understanding. This extends to other forms of expert perception as they have been ana lyzed and explained throughout this book. It is clear that the expert achieves better understanding of her domain. And when she does so, again, she rec ognizes and grasps the patterns, coherence, structure, relations, and so on between the elements that comprise the domain and stimuli in that domain. Sometimes this will involve knowing where to look, how to analyze or make sense of the elements, how the patterns support certain inferences or conclusions, and so on. Understanding in this way involves an ongoing and active engagement with the objects and events of the domain. But just as face perception can be aptly characterized as involving perceptual under standing, so can the achievement of some perceptual experts. What the expert radiologist is able to do, in the best of cases, is to visually recognize and grasp the patterns, organization, and coherence of stimuli within her domain. She can sometimes “just see” an anomaly and how it situates in the overall structure of the lung. The elite goalkeeper immediately perceives and predicts the formation of the offensive players so that she can respond
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
accordingly. This reveals another epistemic consequence of the expert’s achievement. As a result of her cognitive learning, training, and practice, she has improved perceptual skills and enriched perceptual content. In some of these cases, understanding thus extends all the way down to perceptual experience. These observations motivate the first of two arguments; call it the argu ment from understanding. Recall the theoretical decision space that we plausibly find ourselves in. Our current scientific situation, and probable long-term future scientific situation, is one that underdetermines the choice for theory of mental architecture. This book has made the case that modular archi tectures should be abandoned and replaced with a malleable architecture that accepts that thinking affects, and sometimes improves, perceiving. This architecture also abandons the view that attention is always a mechanism or process separable from perceptual systems like vision. And it encourages us to think about perception as a process that can be improved, partly through the cognitive efforts of the agent, to the point of intellectual virtue. But it is not as if the empirical data secure this mental architecture. If I am anywhere near correct, then the proposed malleable architecture – and in particular, the TiP thesis – best explains a broad array of that data. But as we all know, abduction ain’t deduction, and there are other horses in town. To further leverage the choice in favour of a malleable architecture, we can appeal to understanding. A theory of perception that allows for improvement, indeed, potentially virtuous improvement, better comports with an undeniable observation, namely, that we do achieve understanding of the world around us, and often in highly specialized ways. Understanding involves a sensitivity to pattern and structure and a grasp of the coherence of those very organizational fea tures. This is precisely what the expert achieves, and sometimes all the way down to the level of perception. Rapid understanding – perceptual under standing – is only accommodated by a theory that allows for perceptual improvement and expertise. That is a descriptive point. The meta-theoretical corollary of this point, which comes in the form of a prescription, is this: We should favour a theory of perception that favours our achievement of understanding. That is, understanding is an important epistemic goal, and one that is highly desirable for a host of reasons that are hopefully clear. We should favour a theory that allows space for this goal to be achieved some times perceptually and that encourages us to think about the most impor tant mental contact with the world – sensory contact – as a process that is malleable and improvable as a consequence of our cognitive contact with the
253
254
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
world. The improvements made by perceptual experts (and experts more generally) could not fit better with this ambition: What the expert achieves is a heightened perceptual sensitivity to the very features of coherence, pattern, and organization that constitute understanding within a domain. Therefore, we better understand how we do achieve this understanding, and how we can achieve it, if we opt for the mental architecture argued over the course of these pages. The second argument extends the first, turning from understanding of the world and our contact with it to understanding of ourselves. Call it the argument from self-understanding. Here is an observation that philosophers of widely varying commitments have agreed on: Human beings seem, perhaps universally, to believe that they are individual selves. Applied first to himself in his Meditations, Descartes took his own existence as a thinking thing to be first knowledge. Locke agreed with the claim that the self is constituted by mind and thought, and accordingly identified a criterion for persistence of per sons that centred on a continuity of consciousness of that self. Hume famously denied the existence of a self qua substance, but in nearly the same breath accepted that we have a natural compulsion to identify ourselves as selves (even if we lack any rational grounds for doing so). He even goes on to give a causal-generative explanation, a kind of error theory, for our maintenance of the notion of “self” as an organizing principle. One finds a very similar sen timent in some Buddhist philosophies. Speaking of the 6th-century thinker, Candrakīrti, Evan Thompson writes One individuates oneself as a subject of experience and agent of action by laying claim to thoughts, emotions, and feelings . . . and thereby enacts a self that is no different from the self-appropriating activity itself. Again, the self isn’t an object or thing; it’s a process – the process of “I-ing” or ongoing self-appropriating activity. (2015: 363; see also Ganeri 2007)
Thomas Reid, critical of his early modern contemporaries, thought that even if the self is monadic and unanalyzable, it is a notion that inextricably grounds how we reason and behave. “The conviction . . . needs no aid of phi losophy to strengthen it; and no philosophy can weaken it” (Reid 1785). In a similar vein, phenomenologists take self-consciousness to be foundational to consciousness of events outside of oneself. Thus Sartre claimed, “This self-consciousness we ought to consider not as a new consciousness, but as the only mode of existence which is possible for a consciousness of something (Sartre 1943: 20).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
There is a lot to debate here and scores of pages, articles, and books to show for it. And we could find some space for important individual differ ences across the human population. But let us grant this fairly general obser vation: We, each of us, think of ourselves as individual persons, and this is central to how we live our lives. Important to that self-conception, and as theorized by the widely varying thinkers just mentioned and many others, are the following further observations. First, our notion of self is or involves a notion of our own minds: our consciousness, our thoughts, our perceptual experience. Second, to think of oneself is to think of oneself as an autono mous agent, where one bears some responsibility for the events in one’s life. Finally, and related, we conceive of our waking lives as a continuing process. This, too, was part of Reid’s common-sense observation: When I reason out a decision, I maintain the conviction that the agent that acts on my decision will be me, and “there can be no memory of what is past without the convic tion that we existed at the time remembered” (1785). We live through our lives. We gain a better self-understanding, on each of these marks, if we conceive of our minds as richly malleable and improvable. The mental architecture proposed over the course of these pages main tains that thinking affects and sometimes improves perceiving. What you know, what you have learned, what intellectual efforts you have put forth can enhance how you perceive the world. This connects in deep ways with the common intuition that who we are is, at least in part, our consciousness, our mind, our psychology. And the TiP thesis implies, then, that you have an important role to play in who you are. You can change and improve not only your cognitive states, but your perceptual processes and content. And this further grounds an important intuition that we are autonomous agents, responsible for important aspects of our lives. If the proposed malleable architecture is accepted, this responsibility concerns not only our actions and thoughts but our perceptual experiences, our sensory contact with the world. As we persist as persons, we can influence our very own perceptual processes. The performance of those processes is not a mere consequence of biology, but a consequence of the active, cognitive process of being in the world. The patterns and features to which we are sensitive are partly a function of what we have done. In this sense, we are active participants in the contact that we make with the world, where the cognitive contact we have and have had informs the sensory contact we have and can have. Thus the understanding that one achieves of the world is partly a function of who one is. And understanding oneself, in these richly agential ways, reveals how and when we can be credited for the worldly, perceptual understanding we
255
256
ePILOGue: THe ARGuMeNTS FROM uNDeRSTANDING
Copyright © 2021. Taylor & Francis Group. All rights reserved.
enjoy. It reveals how we can, as individual selves, achieve virtue in perceiv ing the world. As you and I move through the world, we will see it differently. And these differences are partly determined by who we are, as selves, and partly deter mine who we will be. We should prefer, then, a malleable mental architec ture that holds out this promise: It makes possible this enriched self-making and the enhanced self-understanding that comes with it. It reveals how, once more, our contact with the world is very much our own.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
GLOSSARY
Accuracy conditions: These are the conditions required for a mental state to be correct. Your visual experience of a cup on the table is accurate just in case there is a cup on the table before you. To identify these accuracy conditions is, some think, to identify the representational contents of perceptual (and other mental) states. Action potential: Objects in one’s environment offer or encourage a range of possible actions for those who perceive them. This potential is relative to the perceiver (e.g. a doorknob offers a potential action to an adult human that it does not to an infant human) and on some views is included in one’s perceptual content. Admissible contents of perceptual experience: Tradition maintains that per ceptual experience (by contrast to cognitive states) is sparse with respect to its content. Vision, for instance, only provides information about shape, colour, size, and rest/motion properties. Rich content theorists argue instead that vision and other perceptual modalities can represent (admit of) more than this: kind properties, emotional properties, aes thetic properties, or causal properties. Aesthetic properties: Verdictive aesthetic properties include beauty and ugli ness. Substantive aesthetic properties include features such as being
Copyright © 2021. Taylor & Francis Group. All rights reserved.
258
GLOSSARY
graceful, being garish, and being balanced. while not unanimous, most theorists maintain that such properties are, in some sense, experience-based. when you attribute gracefulness or garishness to x, you experience x as being graceful or garish. Architectural claim 1: Regarding some perceptual experts, this claim says that such individuals enjoy distinctive perceptual representation (by contrast to naïve control subjects). Perceptual expertise is thus a genuinely per ceptual phenomenon. Architectural claim 2: Regarding some perceptual experts, this claim says that the distinctive perceptual representation that such individuals enjoy is causally sensitive to the concept-rich cognitive learning undergone within the domain of specialization. Perceptual expertise is thus a cogni tively influenced perceptual phenomenon. Argument from consequences: An argument that concludes that some cases of attention-involving cognitive influence on perception are genuine cases of cognitive penetration of perception. The argument involves rejecting the attention as act and attention as spotlight assumptions, focusing on empirically studied cases of cognitive influence on selective attentional mechanisms, where those influences imply debate-neutral consequences (including epistemic consequences). Argument from directness: An argument that concludes that some cases of attention-involving cognitive influence on perception are genuine cases of cognitive penetration of perception. The argument involves rejecting the attention as gatekeeper, attention as act, and attention as spotlight assumptions, focusing on empirically studied cases of cognitive influ ence on selective attentional mechanisms. Because those mechanisms are plausibly part of perception rather than a mere gatekeeper or input to perception, cognitive influences on their processing are cognitive influences on perception. Arguments from reliability of perception: A cluster of arguments that is sup posed to motivate the modularity of perceptual systems like vision. By and large, perception provides accurate representations of the environ ment, and does so with great frequency. This reliability is best explained, the argument concludes, by informationally encapsulated perceptual systems. One specific way that Fodor deploys this argument is by appeal to successful, reliable perception of unanticipated stimuli. There is a fur ther teleological variant on this argument (see Chapters 6 to 8).
Copyright © 2021. Taylor & Francis Group. All rights reserved.
GLOSSARY
Arguments from stability of perception: A cluster of arguments that is sup posed to motivate the modularity of perceptual systems like vision. By and large, perception is stable and in spite of a constantly changing array of perceptible stimuli. This stability is best explained, the argument con cludes, by informationally encapsulated perceptual systems. Two specific ways that Fodor deploys this argument is by appeal to persistent illu sions and to constancy mechanisms. Attention as act assumption: This assumption, made by some defenders of mod ularity, says that when attention is involved in relevant cognitive-perceptual relations, it involves some bodily movement or action done by the per ceiver. Attention on this view is an overt, voluntary action. Attention as gatekeeper assumption: This assumption, made by some defend ers of modularity, says that attention is an intermediary process between cognition and perception. Attention-shift interpretation: See Pre-perceptual explanation. Automaticity: Performance for perceptual experts, within their domain of spe cialization, sometimes appears to occur dispositionally, without a feeling of effort and in a way that cannot be “turned off” in laboratory-induced circumstances (for example, when stimuli are presented in ecologically unsound ways). Behavioural role of perception: Perceiving is, at least in part, for acting. Vari ous theoretical commitments capture this role, from claiming that per ceptual content includes the action potential of an object to claiming that perception sometimes directly causes action without intermediary planning or decision-making. An important possible kind of cognitive influence on perception would involve hindering or enhancing this role of perception. Belief: See Doxastic states. Biased competition: A theory of attention that says that perceptual atten tion is or involves the resolution of a competition for limited cognitive resources. These biases can be bottom-up (say, when some feature of the environment abruptly changes) and top-down (say, where the perceiver has been primed with an expectation about relevant stimuli). The “win ner” of the competition is what gets attended. Cognitive penetrability: A consequence of the putative informational encapsu lation of sensory perception is that such systems compute their proprie tary input without influence from cognitive states or processes (such as
259
Copyright © 2021. Taylor & Francis Group. All rights reserved.
260
GLOSSARY
belief or reasoning or concepts). Vision and other perceptual representa tion is therefore supposed to be cognitively impenetrable. Some argue, contrary to this position, that such cognitive influence is possible; they argue that cognition can penetrate perception by directly or otherwise importantly influencing it. Composite task: A type of experimental task used to study perceptual experts. Subjects are presented with both the bottom and top half of an object of expertise, and the task is to determine if the designated half of the “test image” is the same as the relevant half of the “study image”. Results of these experiments suggest that experts are more sensitive to objects of expertise perceived as wholes, where they struggle or fail to ignore the irrelevant, distractor object half. Computationalism: A theory of mind committed to the central claim that the mind is a computer and so its states and processes should be under stood as involving rule-bound operations over symbolic representations. Conceptual vs. nonconceptual content: Cognitive content is uncontrover sially conceptual. If one believes that “Hawks are birds”, then one grasps the concepts “HAwk” and “BIRD”, and the content of one’s belief is structured by those concepts. Perceptual content, some main tain, is not or need not be conceptual in this way, while others maintain that perceptual content, like cognitive content, is or must be concep tual in this way. Consequentialist constraint: This prescribed constraint says that an analysis or definition of cognitive penetrability will be successful just in case and to the degree that it describes a phenomenon (or class of phenomena) that has implications for the rationality of scientific theory choice, the epistemic role of perception, the behavioural role of perception, mental architecture. Constancy mechanisms: Vision and other perceptual systems correct for changes and differences in distal stimuli so that those features continue to appear stable. A uniformly coloured object will thus typically appear uniformly coloured even if it is illuminated differently in some places relative to the perceiver, and an object will continue to appear to be the same size even if it changes distance relative to the perceiver. Human constancy mechanisms include those for colour, size, shape, luminance, distance, and location. Cross-race effect: An empirically well-established phenomenon of face percep tion. while individuals are exceptional relative to other similarly complex
Copyright © 2021. Taylor & Francis Group. All rights reserved.
GLOSSARY
stimuli at recognizing, identifying, and recalling individual faces from within their “same race” (SR), subjects fail or frequently err along the same measures with respect to “cross-race” faces (CR). Default position assumption: It is commonly assumed (for instance, by those who defend cases of cognitive penetrability and those who deny those cases) that modularity is the established theory of the architecture of cognition and perception, against which all novel theories or analyses must position themselves. Disjunctive consequentialism: A cluster analysis inspired by the consequen tialist constraint. It says that ψ is cognitive penetration if and only if ψ is a cognitive-perceptual relation, and ψ implies consequences for theory-ladenness or the epistemic role of perception or the behavioural role of perception or mental architecture. Domain specificity: Mental modules are supposed to be domain specific in the sense that they are sensitive to a proprietary set of stimulus types and deliver output – for example, “hypotheses” about distal properties of the environment – just in terms of the information in that domain. Doxastic states: A mental state that involves a commitment on the part of the subject who has the mental state to the truth of the state’s proposi tional content. Belief is the paradigmatic example; it is a state or attitude towards some proposition, P, where one possesses a conviction that P is true. Other fairly standard candidates are judgments, opinions, expecta tions, and memories. EEG/ERP: electroencephalogram is a non-invasive method of measuring electrical activity in the brain. One particular method of eeG recording involves measuring event-related potentials (eRPs), where electrical activ ity in the brain is correlated with the presentation of particular types of stimuli. Two particular eRP measures discussed in relation to perceptual expertise are the N170 and the N250. Epistemic role of perception: Both theoretical tradition and common sense have it that perceptual experience provides one with knowledge about one’s immediate environment. One possible implication of cogni tive influence on perception is that this role may need to be revised in some way, since such cognitive influence might hinder or enhance the knowledge-providing capability of perception. Epistemic virtue: Agents are virtuous if and to the degree they exhibit intel lectual virtues. And a cognitive process is virtuous just in case it is or results in some relevant excellence in performance. The relevant
261
Copyright © 2021. Taylor & Francis Group. All rights reserved.
262
GLOSSARY
excellence for an epistemic virtue is typically some kind of truth or accu racy conduciveness. Epistemological claim: Perceptual expertise is epistemically enhancing, not pernicious or downgrading. These are instances where thinking improves perceiving. This is an extension of architectural claims 1 and 2. Etiological analysis: This form of analysis says that the function of X is Z means that (a) X is there because it does Z and (b) Z is a consequence (or result) of X’s being there (wright 1973). Eye tracking: Researchers use this technology to measure saccadic eye move ments, fixation patterns, and other stimulus-related activities in the eye. These measures are sometimes used to corroborate behavioural and/or neural measures. FBA and OBA: Feature-based and object-based attention are two forms of selective attention. These modes of attention involve heightened, and often rapid, perceptual sensitivity to (and selection of) instances of par ticular feature or object types (say, to a colour or a shape). FFA: The fusiform face area is a region in the human cortex traditionally theo rized to be responsible for face perception. The occipital face area (OFA) is similarly theorized. Research on perceptual expertise suggests that these areas may more generally be responsible for perception of features or objects of expertise. fMRI: Functional magnetic resonance imaging. This technique measures brain activity by detecting changes in blood flow in the brain. Research ers then correlate these changes in activity with mental and behavioural functions. Functional analysis: As defined by Robert Cummins (1975), such a mode of explanation, for any given system S, proceeds by analyzing S into its components (and components of components) and then identifying what each component does or how it contributes to some relevant effect or product of S. Functional-teleological analysis: This mode of analysis combines Cummins style functional analysis with the attempt to identify the (biological) pur pose of some trait or feature of an organism. The proper function of (visual) perceptual systems, it was argued, is representation of objects and features in one’s immediate environment. Holistic processing: A common category of explanation of performance of perceptual experts is that they process objects from within their
Copyright © 2021. Taylor & Francis Group. All rights reserved.
GLOSSARY
domain of specialization as wholes, where they are as or more sensitive to how the parts are configured as a whole than to individual features of the whole. Many experimental methods involve manipulations that appear to undermine holistic processing and thus undermine expert performance. Implicit bias: unconscious and involuntary attitudes or associations that are often inconsistent with the subject’s explicit, conscious, and well-considered beliefs about the same subject matter. Many human beings display degrees of implicitly racist, sexist, homophobic, or oth erwise prejudicial biases, even while explicitly disavowing those same biases. An important question is whether and to what degree some of these biases are perceptual phenomena. Informational encapsulation: Fodor (1983, 1985, 2001) identifies this as the essence of modularity. If a module m is informationally encapsulated, then m cannot, during the course of its processing, access or compute over information found in other components of the overall system. Interference effects: Broadly, these are the results of experimental manipula tions of the performance of perceptual experts that somehow undermine or disrupt their high-level performance. This may involve presentation of stimuli of expertise in ecologically invalid ways or presentation of distractor stimuli. Inter-subjective objectivity: In philosophy of science, some worried that subjective commitments and mental processes might influence sci entific investigation in ways that undermine its objectivity. In reply to this worry, some grant that scientific investigation is constituted by a collective of individual psychological agents. However, this social struc ture – including shared epistemic values, methodological standards of practice and experimentation, mutual goals amongst the community – provides a system of inter-subjective checks and balances, an impor tant inter-subjective objectivity. In the context of theories of perception, the suggestion (Chapter 8) is that this notion of objectivity should give standards for accuracy or success of perceptual representation. Intra-perceptual explanation: Sceptics of cognitive penetration or other impor tant cognitive influence on perception sometimes deploy this category of explanatory strategy. It says that an apparent cognitive effect on percep tion is instead an effect of perception on perception. For instance, some instances of perceptual learning seem to be changes or development just
263
Copyright © 2021. Taylor & Francis Group. All rights reserved.
264
GLOSSARY
within vision and not a consequence of one’s beliefs or other cognitive processes. Inversion effects: This is an experimental manipulation used to disrupt per formance in perceptual experts in their domain of expertise. It involves an inversion of stimuli of expertise, which appears to interfere with the holistic perceptual processing of those stimuli (for experts, but not for novices). Judgment interpretation: See Post-perceptual explanation. Malleability of the mind: A broad alternative to modularity, which claims that most or all types of mental processing can be changed as a result of an indi vidual’s experience and learning. And this includes how cognitive learning influences, and in some cases improves, sensory perceptual processes. Memory interpretation: See Post-perceptual explanation. Mental architecture: This concerns or is a model of the mind that distinguishes mental functions and processes by broad category (e.g. cognition versus perception) and theorizes how instances of those categories interact. Mental content: Many philosophers and psychologists theorize mental states (including both cognitive states like belief and sensory states like visual experience) as being representational or intentional: They are about objects, events, and features. To identify what mental states are about is to identify their content. Modularity of mind: A mental architecture that maintains that some mental functions or systems are functionally independent from all other func tions or systems. Such systems are essentially informationally encap sulated and typically fast, mandatory, domain specific, and biologically hardwired (among a few other typifying features). Natural norms: A norm is a standard for possible performance that is to some degree adequate for fulfilling a function. Natural norms are standards that obtain independent of individuals’ attitudes concerning such func tions or standards for successful performance. Neuroplasticity: The human brain is, functionally and structurally, plastic. This neuroplasticity has been well empirically established regarding crucial developmental stages; as the human brain matures, it develops at all levels, from basic cellular levels, to synaptic organization and connectiv ity, to functionally relevant cortical areas. More recent research indicates that the human brain is permanently plastic, enjoying the same range of changes and as a result of environmental changes and triggers.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
GLOSSARY
New Look Psychology: An approach in behavioural psychology associated foremost with Jerome Bruner, beginning with experimental research in the 1940s. Broadly, this approach suggested that sensory perception could be influenced in important ways by social and cultural beliefs, val ues, and concepts. Non-transfer: Perceptual performance of domain-specific experts appears not to generalize or transfer to similar, complex perceptual tasks. For exam ple, although expert radiologists are exceptional at identifying anomalies in radiographs, they may enjoy no general advantage in visual search tasks outside of their domain. Objectivity: The objective world is the mind-independent world. Objectivity in this sense is contrasted with the subjective awareness and experiences had by individuals. The tree before you is objective; your visual experi ence of the tree is subjective. A scientific thesis or process of investiga tion is objective if and to the degree that it accurately reflects or tracks the mind-independent world. A point of controversy concerning sensory perception concerns what degree of objectivity is required for perception to be accurate or successful. Objectivity*: This is an especially strong notion of objectivity (see Objec tivity), which maintains that science (and perhaps other human intel lectual activities and attitudes) should be purely mind-independent, entirely devoid of any subject’s or group’s perspective on relevant facts. Objectification: Perceptual processes like vision objectify in the sense that they involve attributing features (red, round, to the left) to external worldly objects (Burge 2010). Online matching task: An experimental paradigm where subjects are asked to report on a visually present stimuli (say, a coin or a circular disk) by matching a simultaneously visually presented object of report (say, a cir cular patch of light that can be adjusted by the subject). Perceptual constancies: See Constancy mechanisms. Perceptual experience: Conscious sensory representation in one or more sen sory modality (vision, audition, and so on). Such mental states are con scious, at least, in the sense that they have a phenomenology; there is something that it’s like, for the perceiver, to be in them. Perceptual expertise: As understood in empirical psychology, perceptual experts are individuals with a high level of training in a domain of
265
Copyright © 2021. Taylor & Francis Group. All rights reserved.
266
GLOSSARY
specialization who perform with high levels of accuracy and where that performance depends non-trivially upon sensory perception. Perceptual learning: Any relatively permanent and consistent change in the perception of a stimulus array (type) following practice or experience with this array (type) (Gibson 1963). Pernicious cognitive effects assumption: This assumption says that if an organism’s beliefs or expectations, goals, desires, or other evaluative states were to substantially influence perceptual representation, then the reliability or accuracy of those representations in general would greatly decrease. Persistent illusions: Visual and other perceptual illusions that continue (as illusory) in spite of one’s knowledge of the illusion. The Müller-Lyer illu sion is one paradigmatic example, where in spite of one’s knowledge that the two lines are of identical length, they still visually appear as distinct in length. Some modularity theorists make regular appeal to these phe nomena in arguments for modularity. Phenomenology: This is one possible dimension along which perception might be distinguished from cognition. In this usage, it concerns the first personal, subjective aspects of experience – what it’s like to see, hear, taste, and so on. Traditionally, perceptual experience is supposed to enjoy rich phenomenology; cognitive states are supposed to enjoy less or no phenomenology. Post-perceptual explanation: Sceptics of cognitive penetration or other impor tant cognitive influence on perception sometimes deploy this category of explanatory strategy. It says that an apparent cognitive effect on per ception is instead an effect on some mental state that comes after per ception, such as belief or judgment. In the text, two examples of this type of explanation were the memory interpretation and the judgment interpretation. Pre-perceptual explanation: Sceptics of cognitive penetration or other impor tant cognitive influences on perception sometimes deploy this category of explanatory strategy. It says that an apparent cognitive effect on per ception is instead an effect on some mental state that comes before per ception and thus involves a change to the input to perceptual systems. For instance, one’s beliefs may influence spatial attention (pre-visually), which then influences input to vision. In the text, this was also referred to as an attention-shift interpretation.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
GLOSSARY
Predictive processing models: According to this theoretical framework, pro cesses in the brain (including perceptual processes) function by mak ing predictions, using stored representations of the environment, about the sensory information that the subject will receive at any given time. A prediction is then compared, following Bayesian rules of probability, to the sensory inputs that actually occur. This will typically involve some mismatch, or prediction error, and the system proceeds so as to minimize this error. In the case of perception, the result of this prediction error minimization process is perceptual experience. Presence of appropriate stimuli: This is one possible dimension along which perception might be distinguished from cognition. At least when func tioning normally (when one is not, say, hallucinating), perceptual processes like vision seem to require the presence of an appropriate external stimulus (the object or event that one sees). This seems not to be required for cognitive states like belief or desire. Problem of cross-talk: In the context of the cognitive penetrability of percep tion, distinct theorists rely upon distinct definitions of the target phe nomenon and, as a result, interpret empirical findings in importantly different ways. This kind of talking past one another thwarts theoretical progress. Rationality of perception: According to some recent epistemological theories, sensory perception is rationally assessable, such that its etiology can be epistemically good making or epistemically pernicious. Representational function: Visual perceptual systems (and at least some other perceptual systems, like audition and touch) have a representational function. It is successful performance of this function that provides standards or norms for the value of a perceptual system and (in part) the fitness of the individual that possesses that system. Rich perceptual content: See Admissible contents of experience. Sensory organ activity: This is one possible dimension along which perception might be distinguished from cognition. Perceptual states like visual expe rience, for example, seem to require current activity in relevant physiolog ical systems or organs (e.g. the eyes); cognitive states such as belief, by contrast, seem not to require such current activity. Sparse perceptual content: See Admissible contents of experience. Spotlight of attention assumption: This assumption, made by some defend ers of modularity, says that when attention is involved in relevant
267
Copyright © 2021. Taylor & Francis Group. All rights reserved.
268
GLOSSARY
cognitive-perceptual relations, it is spatial attention. Changes in atten tion are thus changes in the focal spotlight of attention. Subject activity: This is one possible dimension along which perception might be distinguished from cognition. while cognition seems to be active and (often) voluntary, one might maintain that perceptual processes like vision are inactive and/or involuntary, instead driven just by external stimuli. Success conditions of perception: Accuracy is the most common standard for perceptual success: An experience e is successful if it accurately rep resents the environment. But it is plausibly not the only condition for such success: e might be successful because it is sensitive to patterns and organizational features, less susceptible to distraction, sensitive to the action potential of an object or event. This array of success condi tions provides an alternative method for determining perceptual content (including but not exclusive to accuracy). TaP thesis: Thinking affects perceiving and in ways that are both theoretically and humanistically important. How we think, what we think, and what we have learned can influence conscious perceptual experience, sometimes directly and with important epistemic and scientific consequences. Theory-ladenness: Traditional philosophy of science through the middle of the 20th century maintained that scientific observation was neutral with respect to one’s theoretical commitments and beliefs. This assumption was challenged in the middle and later 20th century, where some phi losophers suggested that such observation may be biased by theory. One possible implication of cognitive influence on perception is this theory-ladenness. TiP thesis: Thinking improves perceiving. Cognitive learning changes percep tual experience in a rich variety of ways, and the results are epistemically and behaviourally enhancing. Top-down influence: Another commonly used term to refer to apparent cog nitive influence on sensory perceptual processes. Belief affecting visual processing and experience would be a top-down influence. Underdetermination: A thesis in the philosophy of science that says that for any given observational evidence or data, there are a number of theories consistent with that data. Accordingly, there is no straightforward impli cation from data to theory; in this way, the data underdetermine one’s choice between competing theories.
GLOSSARY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Understanding: An epistemic value that some argue to be as important as (but not necessarily exclusive of) the value of knowledge. It involves the appreciation or grasp of how items in a body of information cohere, what patterns or other organizational features they enjoy, and how parts relate to a whole in that context. Visual short-term memory (VSTM): A short-term or working memory system that encodes only visually acquired information. Perceptual experts show VSTM advantages for objects of expertise.
269
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Ahissar, M., Laiwand, R., kozminsky, G., Hochstein, S., 1998. Learning PopOut Detection: Building Representations for Conflicting Target-Distractor Relationships. Vision Research 38, 3095–3107. https://doi.org/10.1016/ s0042-6989(97)00449-5 Ahluwalia, A., 1978. An Intra-Cultural Investigation of Susceptibility to “Per spective” and “Non-Perspective” Spatial Illusions. British Journal of Psy chology 69, 233–241. https://doi.org/10.1111/j.2044-8295.1978.tb01653.x Allard, F., Graham, S., Paarsalu, M.e., 1980. Perception in Sport: Basketball. Jour nal of Sport and Exercise Psychology 2, 14–21. https://doi.org/10.1123/jsp.2.1.14 Allport, D.A., 1987. Selection for Action: Some Behavioral and Neurophysio logical Considerations of Attention and Action, in: Heuer, H., Sanders, H.F. (eds.), Perspectives on Perception and Action. Hillsdale: Lawrence erl baum, pp. 395–419. Alston, w., 1988. The Deontological Conception of epistemic Justification. Phil osophical Perspectives 2, 257–299. Anderson, M.L., 2014. After Phrenology: Neural Reuse and the Interactive Brain. Cambridge, MA: MIT Press. Anderson, M.L., 2016. Précis of After Phrenology: Neural Reuse and the Interac tive Brain. Behavioral and Brain Sciences 39, 1–22. https://doi.org/10.1017/ s0140525x15000631
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Anderson, M.L., Richardson, M.J., Chemero, A., 2012. eroding the Boundaries of Cognition: Implications of embodiment. Topics in Cognitive Science 4, 717–730. https://doi.org/10.1111/j.1756-8765.2012.01211.x Antony, L., 2011. The Openness of Illusions. Philosophical Issues 21, 25–44. https://doi.org/10.1111/j.1533-6077.2011.00196.x Aristotle, 2014. De sensu and De memoria, trans. Ross, G.R.T., Cambridge: Cambridge university Press. Aristotle, 2017. De Anima, trans. Reeve, C.D.C., Indianapolis and Cambridge: Hackett Publishing Company, Inc. Armstrong, D.M., 2002. A Materialist Theory of the Mind. London: Routledge. Ashby, w.R., 1947. Principles of the Self-Organizing Dynamic System. The Jour nal of General Psychology 37, 125–128. https://doi.org/10.1080/00221309 .1947.9918144 Bach, k., 1985. A Rationale for Reliabilism. The Monist 68, 246–263. https://doi. org/10.5840/monist198568224 Baker, L., Levin, D., 2016. The Face-Race Lightness Illusion Is not Driven by Low-Level Stimulus Properties: An empirical Reply to Firestone and Scholl (2014). Psychonomic Bulletin & Review 23. https://doi.org/10.3758/ s13423-016-1048-z Balcetis, e., Dunning, D., 2010. wishful Seeing: More Desired Objects are Seen as Closer. Psychological Science 21, 147–152. https://doi.org/ 10.1177/0956797609356283 Balcetis, e., Dunning, D., Granot, Y., 2012. Subjective Value Determines Initial Dominance in Binocular Rivalry. Journal of Experimental Social Psychology 48, 122–129. https://doi.org/10.1016/j.jesp.2011.08.009 Barton, J.J.S., keenan, J.P., Bass, T., 2001. Discrimination of Spatial Relations and Features in Faces: effects of Inversion and Viewing Duration. British Journal of Psychology 92, 527–549. https://doi.org/10.1348/00071260116 2329 Battaly, H., 2008. Virtue epistemology. Philosophy Compass 3, 639–663. https:// doi.org/10.1111/j.1747-9991.2008.00146.x Bayne, T., 2009. Perception and the Reach of Phenomenal Content. Philosoph ical Quarterly 59, 385–404. https://doi.org/10.1111/j.1467-9213.2009.631.x Bayne, T., Montague, M., 2011. Cognitive Phenomenology. Oxford: Oxford uni versity Press. Beck, J., 2012. The Generality Constraint and the Structure of Thought. Mind 121, 563–600. https://doi.org/10.1093/mind/fzs077
271
Copyright © 2021. Taylor & Francis Group. All rights reserved.
272
BIBLIOGRAPHY
Beck, J., 2018. Marking the Perception – Cognition Boundary: The Criterion of Stimulus-Dependence. Australasian Journal of Philosophy 96, 319–334. https://doi.org/10.1080/00048402.2017.1329329 Beck, J., 2019. Perception Is Analog: The Argument from weber’s Law. Journal
of Philosophy 116, 319–349. https://doi.org/10.5840/jphil2019116621
Beeghly, e., 2020. Bias and knowledge: Two Metaphors, in: Beeghly, e., Madva,
A. (eds.), An Introduction to Implicit Bias: Knowledge, Justice, and the Social Mind. New York: Routledge. Belke, e., Humphreys, G.w., watson, D.G., Meyer, A.S., Telling, A.L., 2008. TopDown effects of Semantic knowledge in Visual Search are Modulated by Cognitive but not Perceptual Load. Perception & Psychophysics 70, 1444– 1458. https://doi.org/10.3758/PP.70.8.1444 Bennett, J., 1990. why Is Belief Involuntary? Analysis 50(2), 87–107. Bergqvist, A., Cowan, R. (eds.), 2018. Evaluative Perception. Oxford: Oxford university Press. Berkeley, G., 1709. An Essay Towards a New Theory of Vision. Dublin: Aaron Rhames. Block, N., 2010. Attention and Mental Paint. Philosophical Issues 20, 23–63. https://doi.org/10.1111/j.1533-6077.2010.00177.x BonJour, L., 1985. The Structure of Empirical Knowledge. Cambridge, MA: Har vard university Press. Brewer, B., 1999. Perception and Reason. Oxford: Oxford university Press. Brewer, B., 2011. Perception and Its Objects. Oxford: Oxford university Press. Brewer, w.F., Lambert, B.L., 2001. The Theory-Ladenness of Observation and the Theory-Ladenness of the Rest of the Scientific Process. Philosophy of Science 3, 176–186. https://doi.org/10.1086/392907 Broadbent, D.e., 1958. Perception and Communication. Oxford: Pergamon Press. Brown, T.G., Sherrington, C.S., 1912. On the Instability of a Cortical Point. Pro ceedings of the Royal Society of London. Series B, Containing Papers of a Biological Character 85, 250–277. https://doi.org/10.1098/rspb.1912.0050 Brownstein, M., 2019. Implicit Bias, in: Zalta, e.N. (ed.), The Stanford Encyclo pedia of Philosophy. Metaphysics Research Lab. Stanford: Stanford uni versity Press. Brownstein, M., Saul, J., (eds.), 2016a. Implicit Bias and Philosophy, Volume 1: Metaphysics and Epistemology. Oxford: Oxford university Press. Brownstein, M., Saul, J., (eds.), 2016b. Implicit Bias and Philosophy, Volume 2: Moral Responsibility, Structural Injustice, and Ethics. Oxford: Oxford uni versity Press.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Bruner, J.S., 1957. On Perceptual Readiness. Psychological Review 64, 123–152. https://doi.org/10.1037/h0043805 Bruner, J.S., Goodman, C.C., 1947. Value and Need as Organizing Factors in Perception. The Journal of Abnormal and Social Psychology 42, 33–44. https://doi.org/10.1037/h0058484 Bukach, C.M., Gauthier, I., Tarr, M.J., 2006. Beyond Faces and Modularity: The Power of an expertise Framework. Trends in Cognitive Sciences 10, 159– 166. https://doi.org/10.1016/j.tics.2006.02.004 Burge, T., 2003. Perceptual entitlement. Philosophy and Phenomenological Research 67, 503–548. https://doi.org/10.1111/j.1933-1592.2003.tb00307.x Burge, T., 2005. Disjunctivism and Perceptual Psychology. Philosophical Topics 33, 1–78. https://doi.org/10.5840/philtopics20053311 Burge, T., 2010. Origins of Objectivity. Oxford: Oxford university Press. Burge, T., 2011. Disjunctivism Again. Philosophical Explorations 14, 43–80. https://doi.org/10.1080/13869795.2011.544400 Burnston, D., 2017. Cognitive Penetration and the Cognition – Perception Inter face. Synthese 194, 3645–3668. https://doi.org/10.1007/s11229-016-1116-y Busey, T.A., Vanderkolk, J.R., 2005. Behavioral and electrophysiological evi dence for Configural Processing in Fingerprint experts. Vision Research 45, 431–448. https://doi.org/10.1016/j.visres.2004.08.021 Butterfill, S.A., 2015. Perceiving expressions of emotion: what evidence Could Bear on Questions About Perceptual experience of Mental States? Consciousness and Cognition 36, 438–451. https://doi.org/10.1016/j.con cog.2015.03.008 Campbell, J., 2002. Reference and Consciousness. Oxford: Oxford university Press. Carey, S., 2009. The Origin of Concepts. Oxford: Oxford university Press. Carrasco, M., Ling, S., Read, S., 2004. Attention Alters Appearance. Nature Neuroscience 7, 308–313. https://doi.org/10.1038/nn1194 Carruthers, P., 2006. The Architecture of the Mind: Massive Modularity and the Flexibility of Thought. Oxford: Clarendon Press. Cartwright, N., 1989. Nature’s Capacities and Their Measurement. Oxford: Oxford university Press. Causer, J., Smeeton, N.J., williams, A.M., 2017. expertise Differences in Antic ipatory Judgements During a Temporally and Spatially Occluded Task. PLoS One 12. https://doi.org/10.1371/journal.pone.0171330 Cecchi, A.S., 2014. Cognitive Penetration, Perceptual Learning and Neural Plas ticity. Dialectica 68, 63–95. https://doi.org/10.1111/1746-8361.12051
273
Copyright © 2021. Taylor & Francis Group. All rights reserved.
274
BIBLIOGRAPHY
Cecchi, A.S., 2018. Cognitive Penetration of early Vision in Face Perception. Consciousness and Cognition 63, 254–266. https://doi.org/10.1016/j. concog.2018.06.005 Chisholm, R.M., 1977. Theory of Knowledge, 2nd edition. englewood Cliffs: Prentice-Hall. Chisholm, R.M., 1988. The Indispensability of Internal Justification. Synthese 74, 285–296. https://doi.org/10.1007/BF00869631 Churchland, P.M., 1979. Scientific Realism and the Plasticity of Mind. Cambridge: Cambridge university Press. Churchland, P.M., 1988. Perceptual Plasticity and Theoretical Neutrality: A Reply to Jerry Fodor. Philosophy of Science 55, 167–187. https://doi. org/10.1086/289425 Churchland, P.S., 1980. Language, Thought, and Information Processing. Noûs 14, 147–170. https://doi.org/10.2307/2214858 Clark, A., 1997. Being There: Putting Brain, Body, and World Together Again. Cam bridge, MA: MIT Press. Clark, A., 2013. whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science. Behavioral and Brain Sciences 36, 181–204. https://doi.org/10.1017/s0140525x12000477 Clark, A., 2016. Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford: Oxford university Press. Clarke, S., forthcoming. Beyond the Icon: Core Cognition and the Bounds of Perception. Mind and Language. https://doi.org/10.1111/mila.12315 Clerkin, e.M., Cody, M.w., Stefanucci, J.k., Proffitt, D.R., Teachman, B.A., 2009. Imagery and Fear Influence Height Perception. Journal of Anxiety Disor ders 23, 381–386. Cohen, J., 2015. Perceptual Constancy, in: Matthen, M. (ed.), The Oxford Handbook of Philosophy of Perception. Oxford: Oxford university Press, pp. 621–639. Coltheart, M., 2001. Assumptions and Methods in Cognitive Neuropsychology, in: The Handbook of Cognitive Neuropsychology: What Deficits Reveal About the Human Mind. New York: Psychology Press, pp. 3–21. Connolly, k., 2017. Perceptual Learning, in: Zalta, e.N. (ed.), The Stanford Ency clopedia of Philosophy. Metaphysics Research Lab. Stanford: Stanford university Press. Connolly, k., 2019. Perceptual Learning: The Flexibility of the Senses, Philosophy of Mind. Oxford and New York: Oxford university Press.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Copenhaver, R., 2013. Review of Origins of Objectivity, by Tyler Burge. Mind 122, 1065–1068. Correll, J., Park, B., Judd, C.M., wittenbrink, B. 2002. The Police Officer’s Dilemma: using ethnicity to Disambiguate Potentially Threatening Indi viduals. Journal of Personality and Social Psychology 83(6), 1314–1329. Cosmides, L., Tooby, J., 1992. Cognitive Adaptations for Social exchange, in: The Adapted Mind: Evolutionary Psychology and the Generation of Culture. New York: Oxford university Press, pp. 163–228. Crane, T., 1992. The Nonconceptual Content of experience, in: Crane, T. (ed.), The Contents of Experience. Cambridge: Cambridge university Press, pp. 136–157. Cross, J.F., Cross, J., Daly, J., 1971. Sex, Race, Age, and Beauty as Factors in Recognition of Faces. Perception & Psychophysics 10, 393–396. https://doi. org/10.3758/BF03210319 Cummins, R., 1975. Functional Analysis. Journal of Philosophy 72, 741–764. https://doi.org/10.2307/2024640 Curby, k.M., Gauthier, I., 2010. To the Trained eye: Perceptual expertise Alters Visual Processing. Topics in Cognitive Science 2, 189–201. https://doi. org/10.1111/j.1756-8765.2009.01058.x Curby, k.M., Glazek, k., Gauthier, I., 2009. A Visual Short-Term Mem ory Advantage for Objects of expertise. Journal of Experimental Psy chology: Human Perception and Performance 35, 94–107. https://doi. org/10.1037/0096-1523.35.1.94 Cutter, B., n.d. Phenomenal Contrast and High-Level Perception. unpublished manuscript. d’Almeida, O.C., Mateus, C., Reis, A., Grazina, M.M., Castelo-Branco, M., 2013. Long Term Cortical Plasticity in Visual Retinotopic Areas in Humans with Silent Retinal Ganglion Cell Loss. Neuroimage 81, 222–230. https://doi. org/10.1016/j.neuroimage.2013.05.032 Delk, J.L., Fillenbaum, S., 1965. Differences in Perceived Color as a Function of Characteristic Color. The American Journal of Psychology 78, 290–293. https://doi.org/10.2307/1420503 Descartes, R., 1641/1984. Meditations on First Philosophy in which are Demonstrated the existence of God and the Distinction Between the Human Soul and Body, in: Cottingham, J., Stoothoff, R., Murdoch, D. (eds.). Cambridge: Cambridge university Press.
275
Copyright © 2021. Taylor & Francis Group. All rights reserved.
276
BIBLIOGRAPHY
Desimone, R., Duncan, J., 1995. Neural Mechanisms of Selective Visual Atten tion. Annual Review of Neuroscience 18, 193–222. https://doi.org/10.1146/ annurev.ne.18.030195.001205 Diamond, R., Carey, S., 1986. why Faces are and are not Special: An effect of expertise. Journal of Experimental Psychology: General 115, 107–117. https:// doi.org/10.1037//0096-3445.115.2.107 Downes, S.M., Matthews, L., 2019. Heritability, in: Zalta, e.N. (ed.), The Stan ford Encyclopedia of Philosophy. Metaphysics Research Lab., Stanford: Stanford university Press. Draganski, B., May, A., 2008. Training-Induced Structural Changes in the Adult Human Brain. Behavioural Brain Research, Biobehavioural Plasticity 192, 137–142. https://doi.org/10.1016/j.bbr.2008.02.015 Dretske, F., 1969. Seeing and Knowing. Chicago: university of Chicago Press. Dretske, F., 1979. Simple Seeing, in: Gustafson, D.F., Tapscott, B.L. (eds.), Body, Mind, and Method. Netherlands: kluwer Academic Publishers, pp. 1–15. Dretske, F., 1981. Knowledge and the Flow of Information. Cambridge: Cam bridge university Press. Dretske, F., 1986. Misrepresentation, in: Bogdan, R. (ed.), Belief: Form, Con tent, and Function. Oxford: Oxford university Press, pp. 17–36. Dretske, F., 1988. Explaining Behavior: Reasons in a World of Causes. Cambridge, MA: MIT Press. Dretske, F., 1993. Conscious experience. Mind 102, 263–283. https://doi. org/10.1093/mind/102.406.263 Dretske, F., 1995. Naturalizing the Mind. Cambridge, MA: MIT Press. Dretske, F., 2003. experience as Representation. Philosophical Issues 13, 67–82. https://doi.org/10.1111/1533-6077.00005 Drew, T., evans, k., Võ, M.L.-H., Jacobson, F.L., wolfe, J.M., 2013. Informat ics in Radiology: what Can You See in a Single Glance and How Might This Guide Visual Search in Medical Images? Radiographics 33, 263–274. https://doi.org/10.1148/rg.331125023 Dreyfus, H.L., 1992. What Computers Still Can’t Do: A Critique of Artificial Rea son. Cambridge, MA: MIT Press. Duhem, P.M.M., 1914. The Aim and Structure of Physical Theory. Princeton, NJ: Princeton university Press. Dunning, D., Balcetis, e., 2013. wishful Seeing: How Preferences Shape Visual Perception. Current Directions in Psychological Science. https://doi. org/10.1177/0963721412463693
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
evans, G., 1982. The Varieties of Reference. Oxford and New York: Oxford uni versity Press. evans, k.k., Georgian-Smith, D., Tambouret, R., Birdwell, R.L., wolfe, J.M., 2013. The Gist of the Abnormal: Above-Chance Medical Decision Mak ing in the Blink of an eye. Psychonomic Bulletin & Review 20, 1170–1175. https://doi.org/10.3758/s13423-013-0459-3 evans, k.k., Haygood, T.M., Cooper, J., Culpan, A.-M., wolfe, J.M., 2016. A Half-Second Glimpse Often Lets Radiologists Identify Breast Cancer Cases even when Viewing the Mammogram of the Opposite Breast. Pro ceedings of the National Academy of Sciences of the United States of Amer ica 113, 10292–10297. https://doi.org/10.1073/pnas.1606187113 Fahle, M., Morgan, M., 1996. No Transfer of Perceptual Learning Between Similar Stimuli in the Same Retinal Position. Current Biology 6, 292–297. https://doi.org/10.1016/s0960-9822(02)00479-7 Fairweather, A., Zagzebski, L. (eds.), 2001. Virtue Epistemology: Essays in Epis temic Virtue and Responsibility, 1st edition. Oxford and New York: Oxford university Press. Fazekas, P., Nanay, B., forthcoming. Attention is Amplification, Not Selection. British Journal for the Philosophy of Science. https://doi.org/10.1093/bjps/ axy065 Feyerabend, P.k., 1958. An Attempt at a Realistic Interpretation of experience. Proceedings of the Aristotelian Society 58, 143–170. Fiorentini, A., Berardi, N., 1980. Perceptual Learning Specific for Orientation and Spatial Frequency. Nature 287, 43–44. https://doi.org/10.1038/287043a0 Firestone, C., Scholl, B.J., 2014. Can You experience ‘Top-Down’ effects on Perception?: The Case of Race Categories and Perceived Lightness. Psychonomic Bulletin & Review 22, 694–700. https://doi.org/10.3758/ s13423-014-0711-5 Firestone, C., Scholl, B.J., 2016. Cognition Does Not Affect Perception: evalu ating the evidence for “Top-Down” effects. Behavioral and Brain Sciences 39, 1–72. https://doi.org/10.1017/S0140525X15000965 Fodor, J.A., 1983. The Modularity of Mind: An Essay on Faculty Psychology. Cam bridge, MA: MIT Press. Fodor, J.A., 1984. Observation Reconsidered. Philosophy of Science 51, 23–43. https://doi.org/10.1086/289162 Fodor, J.A., 1985. Precis of the Modularity of Mind. Behavioral and Brain Sciences 8, 1–42. https://doi.org/10.1017/s0140525x0001921x
277
Copyright © 2021. Taylor & Francis Group. All rights reserved.
278
BIBLIOGRAPHY
Fodor, J.A., 1988. A Reply to Churchland’s ‘Perceptual Plasticity and The oretical Neutrality’. Philosophy of Science 55, 188–198. https://doi. org/10.1086/289426 Fodor, J.A., 2001. The Mind Doesn’t Work that Way: The Scope and Limits of Com putational Psychology. Cambridge, MA: MIT Press. Folstein, J.R., Palmeri, T.J., Gauthier, I., 2013. Category Learning Increases Dis criminability of Relevant Object Dimensions in Visual Cortex. Cerebral Cortex 23, 814–823. https://doi.org/10.1093/cercor/bhs067 Forber, P., 2020. Contemporary Teleology, in: McDonough, J. (ed.) Teleology: A History. Oxford: Oxford university Press, pp. 255–278. Franklin, A., 1986. The Neglect of Experiment. Cambridge: Cambridge university Press. Freire, A., Lee, k., Symons, L.A., 2000. The Face-Inversion effect as a Deficit in the encoding of Configural Information: Direct evidence. Perception 29, 159–170. https://doi.org/10.1068/p3012 Fridland, e., Pavese, C., 2020. Routledge Handbook of Skill and Expertise. Lon don: Routledge. Friston, k., 2005. A Theory of Cortical Responses. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 360, 815–836. https://doi.org/10.1098/rstb.2005.1622 Fuller, S., Carrasco, M., 2006. exogenous Attention and Color Perception: Per formance and Appearance of Saturation and Hue. Vision Research 46, 4032–4047. https://doi.org/10.1016/j.visres.2006.07.014 Gall, F.J., 1835. On the Functions of the Brain and of Each of Its Parts: With Obser vations on the Possibility of Determining the Instincts, Propensities, and Tal ents, or the Moral and Intellectual Dispositions of Men and Animals, by the Configuration of the Brain and Head. Boston: Marsh, Capen & Lyon. Ganeri, J., 2007. The Concealed Art of the Soul: Theories of Self and Practices of Truth in Indian Ethics and Epistemology. Oxford: Clarendon Press. Gauthier, I., Curran, T., Curby, k.M., Collins, D., 2003. Perceptual Interference Supports a Non-Modular Account of Face Processing. Nature Neurosci ence 6, 428–432. https://doi.org/10.1038/nn1029 Gauthier, I., Logothetis, N.k., 2000. Is Face Recognition not So unique After All? Cognitive Neuropsychology 17, 125–142. https://doi.org/10.1080/ 026432900380535 Gauthier, I., Skudlarski, P., Gore, J.C., Anderson, A.w., 2000. expertise for Cars and Birds Recruits Brain Areas Involved in Face Recognition. Nature Neu roscience 3, 191–197. https://doi.org/10.1038/72140
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Gauthier, I., Tarr, M.J., 1997. Becoming a “Greeble” expert: exploring Mecha nisms for Face Recognition. Vision Research 37, 1673–1682. https://doi. org/10.1016/S0042-6989(96)00286-6 Gauthier, I., Tarr, M.J., 2002. unraveling Mechanisms for expert Object Rec ognition: Bridging Brain Activity and Behavior. Journal of Experimental Psychology: Human Perception and Performance 28, 431–446. https://doi. org/10.1037//0096-1523.28.2.431 Gauthier, I., Tarr, M.J., Anderson, A.w., Skudlarski, P., Gore, J.C., 1999. Acti vation of the Middle Fusiform “Face Area” Increases with expertise in Recognizing Novel Objects. Nature Neuroscience 2, 568–573. https://doi. org/10.1038/9224 Gauthier, I., williams, P., Tarr, M., Tanaka, J., 1998. Training “Greeble” experts: A Framework for Studying expert Object Recognition Processes. Vision Research 38, 2401–2428. https://doi.org/10.1016/S0042-6989(97) 00442-2 Gawronski, B., Bodenhausen, G.V., 2006. Associative and Propositional Processes in evaluation: An Integrative Review of Implicit and explicit Attitude Change. Psychological Bulletin 132, 692–731. https://doi.org/ 10.1037/0033-2909.132.5.692 Gazzaley, A., Nobre, A.C., 2012. Top-Down Modulation: Bridging Selective Attention and working Memory. Trends in Cognitive Sciences 16, 129–135. https://doi.org/10.1016/j.tics.2011.11.014 Gendler, T.S., 2011. On the epistemic Costs of Implicit Bias. Philosophical Stud ies 156, 33–63. Gibson, e.J., 1963. Perceptual Learning. Annual Review of Psychology 14, 29–56. https://doi.org/10.1146/annurev.ps.14.020163.000333 Gibson, J.J., 1966. The Senses Considered as Perceptual Systems, revised edition. westport, CT: Praeger. Gibson, J.J., 1979/2014. The Ecological Approach to Visual Perception, classic edi tion. Hove: Psychology Press. Ginet, C., 1975. Knowledge, Perception and Memory, Philosophical Studies Series.
Netherlands: Springer. https://doi.org/10.1007/978-94-010-9451-1
Glüer, k., 2009. In Defence of a Doxastic Account of experience. Mind and Lan guage 24, 297–327. https://doi.org/10.1111/j.1468-0017.2009.01364
Godfrey-Smith, P., 1991. Signal, Detection, Action. Journal of Philosophy 88, 709–722. https://doi.org/jphil199188122 Godfrey-Smith, P., 1992. Indication and Adaptation. Synthese 92, 283–312. https://doi.org/10.1007/BF00414302
279
Copyright © 2021. Taylor & Francis Group. All rights reserved.
280
BIBLIOGRAPHY
Goldman, A., 1979. what Is Justified Belief, in: Pappas, G. (ed.), Justification and Knowledge. Boston: D. Reidel, pp. 1–25. Goldstone, R.L., 1998. Perceptual Learning. Annual Review of Psychology 49, 585–612. Goldstone, R.L., 2003. Learning to Perceive while Perceiving to Learn, in: Per ceptual Organization in Vision: Behavioral and Neural Perspectives. Mah wah, NJ: Lawrence erlbaum Associates Publishers, pp. 233–280. Graham, P.J., 2012. epistemic entitlement. Noûs 46, 449–482. https://doi. org/10.1111/j.1468-0068.2010.00815.x Graham, P.J., 2014. The Function of Perception, in: Fairweather, A. (ed.), Virtue Scientia: Bridges Between Virtue Epistemology and Philosophy of Science. Dordrecht, Netherlands: Synthese Library, pp. 13–31. Greco, J., 1993. Virtues and Vices of Virtue epistemology. Canadian Journal of Philosophy 23, 413–432. https://doi.org/10.1080/00455091.1993.10717329 Greco, J., 2002. Virtue epistemology, in: Moser, P. (ed.), Oxford Handbook of Epistemology. New York: Oxford university Press, pp. 287–315. Greco, J., 2009a. Epistemic Value. Oxford: Oxford university Press. Greco, J., 2009b. knowledge and Success from Ability. Philosophical Studies: An International Journal for Philosophy in the Analytic Tradition 142, 17–26. Greco, J., 2010. Achieving Knowledge: A Virtue-Theoretic Account of Epistemic Normativity. Cambridge: Cambridge university Press. Green, e.J., 2020. The Perception-Cognition Border: A Case for Archi tectural Division. Philosophical Review 129, 323–393. https://doi. org/10.1215/00318108-8311221 Greenwald, A.G., Banaji, M.R., 1995. Implicit Social Cognition: Attitudes, Self-esteem, and Stereotypes. Psychological Review 102, 4–27. https://doi. org/10.1037/0033-295X.102.1.4 Gregory, R.L., 1970. The Intelligent Eye. New York: McGraw-Hill. Griffiths, P.e., 1993. Functional Analysis and Proper Functions. British Jour nal for the Philosophy of Science 44, 409–422. https://doi.org/10.1093/ bjps/44.3.409 Grill-Spector, k., knouf, N., kanwisher, N., 2004. The Fusiform Face Area Subserves Face Perception, not Generic within-Category Identification. Nature Neuroscience 7, 555–562. https://doi.org/10.1038/nn1224 Gross, S., 2017. Cognitive Penetration and Attention. Frontiers in Psychology 8, 1–12. https://doi.org/10.3389/fpsyg.2017.00221 Gross, S., Chaisilprungraung, T., kaplan, e., Menendez, J., Flombaum, J., 2014. Problems for the Purported Cognitive Penetration of Perceptual
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Color experience and Macpherson’s Proposed Mechanism. Baltic Inter national Yearbook of Cognition, Logic and Communication 9. https://doi. org/10.4148/1944-3676.1085 Hacking, I., 1983. Representing and Intervening: Introductory Topics in the Philoso phy of Natural Science. Cambridge: Cambridge university Press. Haddock, A., Macpherson, F., 2008. Disjunctivism: Perception, Action, Knowl edge. Oxford: Oxford university Press. Hansen, T., Olkkonen, M., walter, S., Gegenfurtner, k.R., 2006. Memory Mod ulates Color Appearance. Nature Neuroscience 9, 1367–1368. https://doi. org/10.1038/nn1794 Hanson, N.R., 1958. Patterns of Discovery: An Inquiry Into the Conceptual Foun dations of Science. Cambridge: Cambridge university Press. Hanson, N.R., 1969. Perception and Discovery: An Introduction to Scientific Inquiry. New York: Springer. Harman, G., 1990. The Intrinsic Quality of experience. Philosophical Perspec tives 4, 31–52. Hawley, k., Macpherson, F., 2011. The Admissible Contents of Experience. Malden, MA: wiley Blackwell. Heck, R., 2000. Nonconceptual Content and the “Space of Reasons.” Philo sophical Review 109, 483–523. https://doi.org/10.1215/00318108-109-4-483 Helmholtz, H. von, 1867. Handbuch der physiologischen Optik. Voss. Helton, G., 2016. Recent Issues in High-Level Perception. Philosophy Compass 11, 851–862. https://doi.org/10.1111/phc3.12383 Hinton, J.M., 1967. Visual experiences. Mind 76, 217–227. https://doi. org/10.1093/mind/LXXVI.302.217 Hohwy, J., 2013. The Predictive Mind. Oxford: Oxford university Press. Hopf, J.M., Luck, S.J., Girelli, M., Hagner, T., Mangun, G.R., Scheich, H., Heinze, H.J., 2000. Neural Sources of Focused Attention in Visual Search. Cere bral Cortex 10, 1233–1241. https://doi.org/10.1093/cercor/10.12.1233 Howe, C.Q., Purves, D., 2002. Range Image Statistics Can explain the Anomalous Perception of Length. PNAS 99, 13184–13188. https://doi. org/10.1073/pnas.162474299 Howe, C.Q., Purves, D., 2004. Size Contrast and Assimilation explained by the Statistics of Natural Scene Geometry. Journal of Cognitive Neuroscience 16, 90–102. https://doi.org/10.1162/089892904322755584 Howe, C.Q., Purves, D., 2005. The Müller-Lyer Illusion explained by the Statis tics of Image – Source Relationships. PNAS 102, 1234–1239. https://doi. org/10.1073/pnas.0409314102
281
Copyright © 2021. Taylor & Francis Group. All rights reserved.
282
BIBLIOGRAPHY
Howe, C.Q., Yang, Z., Purves, D., 2005. The Poggendorff Illusion explained by Natural Scene Geometry. PNAS 102, 7707–7712. https://doi.org/10.1073/ pnas.0502893102 Hugenberg, k., Corneille, O., 2009. Holistic Processing is Tuned for In-Group Faces. Cognitive Science 33, 1173–1181. https://doi. org/10.1111/j.1551-6709.2009.01048.x Hume, D., 1739/2000. A Treatise of Human Nature, in: Norton, D.F., Norton M.J. (eds.). Oxford: Oxford university Press. Hume, D., 1757. The Standard of Taste. early Modern Texts 15. Hurley, S.L., 1998. Consciousness in Action. Cambridge, MA: Harvard university Press. Jahoda, G., 1971. Retinal Pigmentation, Illusion Susceptibility and Space Perception. International Journal of Psychology 6, 199–207. https://doi. org/10.1080/00207597108246683 Jenkin, Z., Siegel, S. (eds.), 2015. Issue on Cognitive Penetrability: Modularity, epistemology, and ethics. Review of Philosophy and Psychology 6(4). Johnson, k.e., Mervis, C.B., 1997. effects of Varying Levels of expertise on the Basic Level of Categorization. Journal of Experimental Psychology: General 126, 248–277. https://doi.org/10.1037/0096-3445.126.3.248 Johnston, M., 2004. The Obscure Object of Hallucination. Philosophical Studies 120, 113–183. https://doi.org/10.1023/B:PHIL.0000033753.64202.21 kalderon, M.e., 2011. Color Illusion. Noûs 45, 751–775. https://doi. org/10.1111/j.1468-0068.2010.00781.x kanai, R., Rees, G., 2011. The Structural Basis of Inter-Individual Differences in Human Behaviour and Cognition. Nature Reviews Neuroscience 12, 231– 242. https://doi.org/10.1038/nrn3000 kant, I., 1790/1987. Critique of Judgment, in: Pluhar, w.S., Gregor, M.J., (eds.). Indianapolis, IN: Hackett Publishing. kanwisher, N., McDermott, J., Chun, M.M., 1997. The Fusiform Face Area: A Module in Human extrastriate Cortex Specialized for Face Percep tion. The Journal of Neuroscience 17, 4302–4311. https://doi.org/10.1523/ JNeuROSCI.17-11-04302.1997 kanwisher, N., Yovel, G., 2006. The Fusiform Face Area: A Cortical Region Spe cialized for the Perception of Faces. Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 361, 2109–2128. https://doi. org/10.1098/rstb.2006.1934 karmarkar, u.R., Dan, Y., 2006. experience-Dependent Plasticity in Adult Visual Cortex. Neuron 52, 577–585. https://doi.org/10.1016/j.neuron.2006.11.001
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
kiss, M., Goolsby, B.A., Raymond, J.e., Shapiro, k.L., Silvert, L., Nobre, A.C., Fragopanagos, N., Taylor, J.G., eimer, M., 2007. efficient Attentional Selection Predicts Distractor Devaluation: eRP evidence for a Direct Link Between Attention and emotion. Journal of Cognitive Neuroscience 19, 1316–1322. https://doi.org/10.1162/jocn.2007.19.8.1316 kolb, B., Gibb, R., 2011. Brain Plasticity and Behaviour in the Developing Brain. Journal of the Canadian Academy of Child and Adolescent Psychiatry 20, 265–276. kottenhoff, H., 1957. Situational and Personal Influences on Space Percep tion with experimental Spectacles: I. Prolonged experiments with Inverting Glasses. Acta Psychologica 13, 79–97. https://doi.org/10.1016/ 0001-6918(57)90012-4 kravitz, D.J., Behrmann, M., 2011. Space-, Object-, and Feature-Based Atten tion Interact to Organize Visual Scenes. Attention, Perception, & Psycho physics 73, 2434–2447. https://doi.org/10.3758/s13414-011-0201-z kriegel, u., 2013. Phenomenal Intentionality. Oxford: Oxford university Press. kriegel, u., 2019a. The Perception/Cognition Divide: One More Time, with Feeling, in: Limbeck-Lilienau, C., Stadler, F. (eds.), The Philosophy of Per ception. Berlin and Boston: De Gruyter, pp. 149–170. kriegel, u., 2019b. Phenomenal Intentionality and the Perception/Cognition Divide, in: Sullivan, A. (ed.), Sensations, Thoughts, Language: Essays in Honor of Brian Loar. New York: Routledge, pp. 167–183. kuhn, T.S., 1962. The Structure of Scientific Revolutions. Chicago: university of Chicago Press. kuhn, T.S., 1977. The Essential Tension: Selected Studies in Scientific Tradition and Change. Chicago: university of Chicago Press. kundel, H.L., La Follette, P.S., 1972. Visual Search Patterns and experi ence with Radiological Images. Radiology 103, 523–528. https://doi. org/10.1148/103.3.523 kundel, H.L., Nodine, C.F., 1975. Interpreting Chest Radiographs with out Visual Search. Radiology 116, 527–532. https://doi.org/10.1148/ 116.3.527 kundel, H.L., Nodine, C.F., Carmody, D., 1978. Visual Scanning, Pattern Recog nition and Decision-making in Pulmonary Nodule Detection. Investiga tive Radiology 13, 175–181. kundel, H.L., Nodine, C.F., Conant, e.F., weinstein, S.P., 2007. Holistic Compo nent of Image Perception in Mammogram Interpretation: Gaze-Tracking Study. Radiology 242, 396–402. https://doi.org/10.1148/radiol.2422051997
283
Copyright © 2021. Taylor & Francis Group. All rights reserved.
284
BIBLIOGRAPHY
kvanvig, J.L., 2003. The Value of Knowledge and the Pursuit of Understanding. Cambridge: Cambridge university Press. Laudan, L., 1996. Beyond Positivism and Relativism: Theory, Method, and Evi dence, 1st edition. Boulder, CO: Routledge. Leder, H., Candrian, G., Huber, O., Bruce, V., 2001. Configural Features in the Context of upright and Inverted Faces. Perception 30, 73–83. https://doi. org/10.1068/p2911 Lee, T.S., Mumford, D., 2003. Hierarchical Bayesian Inference in the Visual Cortex. Journal of the Optical Society of America A 20, 1434. https://doi. org/10.1364/JOSAA.20.001434 Levin, D.T., Banaji, M.R., 2006. Distortions in the Perceived Lightness of Faces: The Role of Race Categories. Journal of Experimental Psychology: General 135, 501–512. https://doi.org/10.1037/0096-3445.135.4.501 Livingston, k.R., Andrews, J.k., 1995. On the Interaction of Prior knowl edge and Stimulus Structure in Category Learning. The Quarterly Journal of Experimental Psychology Section A 48, 208–236. https://doi. org/10.1080/14640749508401385 Logan, G.D., 1988. Toward an Instance Theory of Automatization. Psychological Review 95, 492–527. https://doi.org/10.1037/0033-295X.95.4.492 Longino, H., 1990. Science as Social Knowledge. Princeton: Princeton university Press. Longino, H., 2002. essential Tensions-Phase Two: Feminist, Philosophical, and Social Studies of Science, in: Antony, L., witt, C. (eds.), A Mind of One’s Own: Feminist Essays on Reason and Objectivity. Boulder, CO: westview Press, pp. 93–109. Lupyan, G., 2015. Cognitive Penetrability of Perception in the Age of Prediction: Predictive Systems are Penetrable Systems. Review of Philosophy and Psy chology 6, 547–569. https://doi.org/10.1007/s13164-015-0253-4 Lupyan, G., Clark, A., 2015. words and the world: Predictive Coding and the Language-Perception-Cognition Interface. Current Directions in Psycholog ical Science 24, 279–284. https://doi.org/10.1177/0963721415570732 Lupyan, G., Rahman, R.A., Boroditsky, L., Clark, A., 2020. effects of Language on Visual Perception. Trends in Cognitive Sciences 24, 930–944. https:// doi.org/10.1016/j.tics.2020.08.005 Lyons, J.C., 2011. Circularity, Reliability, and the Cognitive Penetrability of Per ception. Philosophical Issues 21, 289–311. https://doi.org/10.1111/j.1533 6077.2011.00205.x
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Lyons, J.C., 2016. Inferentialism and Cognitive Penetration of Perception. Epis teme 13, 1–28. https://doi.org/10.1017/epi.2015.60 Macevoy, S.P., Paradiso, M.A., 2001. Lightness Constancy in Primary Visual Cortex. Proceedings of the National Academy of Sciences of the United States of America 98, 8827–8831. https://doi.org/10.1073/pnas.161280398 Mack, A., Rock, I., 1998. The New School for Social, in: Inattentional Blindness. Cambridge, MA: MIT Press. MacPherson, F., 2012. Cognitive Penetration of Colour experience: Rethinking the Issue in Light of an Indirect Mechanism. Philosophy and Phenome nological Research 84, 24–62. https://doi.org/10.1111/j.1933-1592.2010.00 481.x Macpherson, F., 2015. Cognitive Penetration and Nonconceptual Content, in: Zeimbekis, J., Raftopoulos, A. (eds.), The Cognitive Penetrability of Per ception: New Philosophical Perspectives. Oxford: Oxford university Press. Macpherson, F., Batty, C., 2016. Redefining Illusion and Hallucination in Light of New Cases. Philosophical Issues 26, 263–296. https://doi.org/10.1111/ phis.12086 Macpherson, F., Platchias, D., 2013. Hallucination. Cambridge, MA: MIT Press. Mandelbaum, e., 2018. Seeing and Conceptualizing: Modularity and the Shal low Contents of Perception. Philosophy and Phenomenological Research 97, 267–283. https://doi.org/10.1111/phpr.12368 Mann, D.T.Y., williams, A.M., ward, P., Janelle, C.M., 2007. Perceptual-Cognitive expertise in Sport: A Meta-Analysis. Journal of Sport and Exercise Psychol ogy 29, 457–478. https://doi.org/10.1123/jsep.29.4.457 Martin, M.G.F., 1992. Perception, Concepts, and Memory. The Philosophical Review 101, 745–763. https://doi.org/10.2307/2185923 Martin, M.G.F., 2002. The Transparency of experience. Mind and Language 17, 376–425. https://doi.org/10.1111/1468-0017.00205 Martin, M.G.F., 2004. The Limits of Self-Awareness. Philosophical Studies 120, 37–89. https://doi.org/10.1023/B:PHIL.0000033751.66949.97 Matthen, M., 1988. Biological Functions and Perceptual Content [www Docu ment]. The Journal of Philosophy. https://doi.org/10.2307/2026898 Matthen, M., 2005. Seeing, Doing, and Knowing: A Philosophical Theory of Sense Perception. Oxford: Oxford university Press. May, A., 2011. experience-Dependent Structural Plasticity in the Adult Human Brain. Trends in Cognitive Sciences 15, 475–482. https://doi.org/10.1016/j. tics.2011.08.002
285
Copyright © 2021. Taylor & Francis Group. All rights reserved.
286
BIBLIOGRAPHY
Maya-Vetencourt, J.F., Origlia, N., 2012. Visual Cortex Plasticity: A Complex Interplay of Genetic and environmental Influences. Neural Plasticity 2012, 631965. https://doi.org/10.1155/2012/631965 Mayo, D.G., 1994. The New experimentalism, Topical Hypotheses, and Learn ing from error. PSA: Proceedings of the Biennial Meeting of the Philoso phy of Science Association 1994, 270–279. https://doi.org/10.1086/ psaprocbienmeetp.1994.1.193032 McCauley, R.N., Henrich, J., 2006. Susceptibility to the Muller-Lyer Illusion, Theory-Neutral Observation, and the Diachronic Penetrability of the Visual Input System. Philosophical Psychology 19, 79–101. https://doi. org/10.1080/09515080500462347 McDowell, J., 1996. Mind and World. Cambridge, MA: Harvard university Press. McGugin, R.w., Mckeeff, T.J., Tong, F., Gauthier, I., 2011. Irrelevant Objects of expertise Compete with Faces During Visual Search. Attention, Perception, & Psychophysics 73, 309–317. https://doi.org/10.3758/ s13414-010-0006-5 Mckeeff, T.J., McGugin, R.w., Tong, F., Gauthier, I., 2010. expertise Increases the Functional Overlap Between Face and Object Perception. Cognition 117, 355–360. https://doi.org/10.1016/j.cognition.2010.09.002 Mckone, e., kanwisher, N., Duchaine, B.C., 2007. Can Generic expertise explain Special Processing for Faces? Trends in Cognitive Sciences 11, 8–15, Regular edition. https://doi.org/10.1016/j.tics.2006.11.002 Meissner, C.A., Brigham, J.C., 2001. Thirty Years of Investigating the Own-Race Bias in Memory for Faces: A Meta-Analytic Review. Psychology, Public Pol icy, and Law 7, 3–35. https://doi.org/10.1037/1076-8971.7.1.3 Meyer, A.S., Belke, e., Telling, A.L., Humphreys, G.w., 2007. early Activation of Object Names in Visual Search. Psychonomic Bulletin & Review 14, 710– 716. https://doi.org/10.3758/BF03196826 Millikan, R.G., 1984. Language, Thought, and Other Biological Categories. Cam bridge, MA: MIT Press. Millikan, R.G., 1989. Biosemantics. Journal of Philosophy 86, 281–297. https:// doi.org/jphil198986652 Mole, C., 2011. Attention Is Cognitive Unison: An Essay in Philosophical Psychol ogy. Oxford: Oxford university Press. Mole, C., 2015. Attention and Cognitive Penetration, in: Zeimbekis, J., Raftopou los, A. (eds.), The Cognitive Penetrability of Perception: New Philosophical Perspectives. Oxford: Oxford university Press, pp. 218–238.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Mole, C., 2017. Attention, in: Zalta, e.N. (ed.), The Stanford Encyclopedia of Phi losophy. Metaphysics Research Lab. Stanford: Stanford university Press. Mole, C., Smithies, D., wu, w., 2011. Attention: Philosophical and Psychological Essays. Oxford: Oxford university Press. Montmarquet, J.A., 1993. Epistemic Virtue and Doxastic Responsibility. Boston: Rowman & Littlefield. Moore, T., Zirnsak, M., 2017. Neural Mechanisms of Selective Visual Atten tion. Annual Review of Psychology 68, 47–72. https://doi.org/10.1146/ annurev-psych-122414-033400 Moores, e., Laiti, L., Chelazzi, L., 2003. Associative knowledge Controls Deployment of Visual Selective Attention. Nature Neuroscience 6, 182– 189. https://doi.org/10.1038/nn996 Moran, J., Desimone, R., 1985. Selective Attention Gates Visual Processing in the extrastriate Cortex. Science 229, 782–784. https://doi.org/10.1126/ science.4023713 Morrison, J., 2016. Perceptual Confidence. Analytic Philosophy 57, 15–48. https:// doi.org/10.1111/phib.12077 Nagel, T., 1974. what is It Like to Be a Bat? Philosophical Review 83, 435–450. https://doi.org/10.2307/2183914 Nagel, T., 1986. The View From Nowhere. Oxford: Oxford university Press. Nanay, B., 2010. Attention and Perceptual Content. Analysis 70, 263–270. https://doi.org/10.1093/analys/anp165 Nanay, B., 2013. Between Perception and Action. Oxford: Oxford university Press. Nanay, B., 2016. Aesthetics as Philosophy of Perception. Oxford: Oxford univer sity Press. Neander, k., 1995. Misrepresenting and Malfunctioning. Philosophical Studies 79, 109–141. https://doi.org/10.1007/BF00989706 Neander, k., 2018. Teleological Theories of Mental Content, in: Zalta, e.N. (ed.), The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab. Stanford: Stanford university Press. Neisser, u., Becklen, R., 1975. Selective Looking: Attending to Visually Specified events. Cognitive Psychology 7, 480–494. https://doi. org/10.1016/0010-0285(75)90019-5 Newman, L., 1994. Descartes on unknown Faculties and Our knowledge of the external world. Philosophical Review 103, 489–531. https://doi. org/10.2307/2185790
287
Copyright © 2021. Taylor & Francis Group. All rights reserved.
288
BIBLIOGRAPHY
Nishimura, M., Maurer, D., 2008. The effect of Categorisation on Sensitivity to Second-Order Relations in Novel Objects. Perception 37, 584–601. https://doi.org/10.1068/p5740 Nodine, C.F., krupinski, e.A., 1998. Perceptual Skill, Radiology expertise, and Visual Test Performance with NINA and wALDO. Academic Radiology 5, 603–612. https://doi.org/10.1016/s1076-6332(98)80295-x Nodine, C.F., kundel, H.L., Mello-Thoms, C., weinstein, S.P., Orel, S.G., Sul livan, D.C., Conant, e.F., 1999. How experience and Training Influence Mammography expertise. Academic Radiology 6, 575–585. https://doi. org/10.1016/s1076-6332(99)80252-9 Noë, A., 2004. Action in Perception. Cambridge, MA: MIT Press. Nosofsky, R.M., 1986. Attention, Similarity, and the Identification – Categoriza tion Relationship. Journal of Experimental Psychology: General 115, 39–57. https://doi.org/10.1037/0096-3445.115.1.39 Olkkonen, M., Hansen, T., Gegenfurtner, k.R., 2008. Color Appearance of Familiar Objects: effects of Object Shape, Texture, and Illumination Changes. Journal of Vision 8, 1–16. https://doi.org/10.1167/8.5.13 Olkkonen, M., Hansen, T., Gegenfurtner, k.R., 2012. High-Level Perceptual Influences on Color Appearance. Visual Experience: Sensation, Cognition, and Constancy 177–198. O’Regan, J.k., Noë, A., 2001. A Sensorimotor Account of Vision and Visual Consciousness. Behavioral and Brain Sciences 24, 883–917. https://doi. org/10.1017/s0140525x01000115 Orlandi, N., 2014. The Innocent Eye: Why Vision Is not a Cognitive Process. Oxford: Oxford university Press. Palmer, S.e., 1999. Vision Science: Photons to Phenomenology. Cambridge, MA: Bradford Book. Palmeri, T.J., 1997. exemplar Similarity and the Development of Automaticity. Journal of Experimental Psychology: Learning, Memory, and Cognition 23, 324–354. https://doi.org/10.1037//0278-7393.23.2.324 Papineau, D., 1984. Representation and explanation. Philosophy of Science 51, 550–572. https://doi.org/10.1086/289205 Pascual-Leone, A., Amedi, A., Fregni, F., Merabet, L.B., 2005. The Plastic Human Brain Cortex. Annual Review of Neuroscience 28, 377–401. https:// doi.org/10.1146/annurev.neuro.27.070203.144216 Passmore, J., 1967. Logical Positivism, in: edwards, P. (ed.), The Encyclopedia of Philosophy. London: Macmillan.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Pautz, A., 2009. what are the Contents of experiences. Philosophical Quarterly 59(236), 483–507. Payne, B.k., 2001. Prejudice and Perception: The Role of Automatic and Con trolled Processes in Misperceiving a weapon. Journal of Personality and Social Psychology 81, 181–192. https://doi.org/10.1037/0022-3514.81.2.181 Payne, B.k., 2006. weapon Bias: Split-Second Decisions and unintended Ste reotyping. Current Directions in Psychological Science 15(6), 287–291. Peterson, M.A., Gibson, B.S., 1994. Must Figure-Ground Organization Precede Object Recognition? An Assumption in Peril. Psychological Science 5, 253–259. https:// doi.org/10.1111/j.1467-9280.1994.tb00622.x Phillips, B., 2019. The Shifting Border Between Perception and Cognition. Noûs 53, 316–346. https://doi.org/10.1111/nous.12218 Pinker, S., 1997. How the Mind Works. New York: w. w. Norton & Company. Pitcher, G., 1971. Theory of Perception. Princeton: Princeton university Press. Poggio, T., Fahle, M., edelman, S., 1992. Fast Perceptual Learning in Visual Hyperacuity. Science 256, 1018–1021. https://doi.org/10.1126/science.1589770 Popper, k.R., 1934. The Logic of Scientific Discovery. London: Routledge. Prinz, J.J., 2006. Is the Mind Really Modular?, in: Stainton, R.J. (ed.), Contem porary Debates in Cognitive Science. Malden, MA: Blackwell, pp. 22–36. Prinz, J.J., 2011. Is Attention Necessary and Sufficient for Consciousness?, in: Mole, C., Smithies, D., wu, w. (eds.), Attention: Philosophical and Psycho logical Essays. Oxford: Oxford university Press, pp. 174–204. Proffitt, D.R., 2006. embodied Perception and the economy of Action. Perspec tives on Psychological Science 1, 110–122. Proffitt, D.R., Baer, D., 2020. Perception: How Our Bodies Shape Our Minds. New York: St. Martin’s Publishing Group. Proffitt, D.R., Stefanucci, J., Banton, T., epstein, w., 2003. The Role of effort in Perceiving Distance. Psychological Science 14, 106–112. https://doi. org/10.1111/1467-9280.t01-1-01427 Pylyshyn, Z.w., 1980. Computation and Cognition: Issues in the Foundations of Cognitive Science. https://doi.org/10.1017/S0140525X00002053 Pylyshyn, Z.w., 1984. Computation and Cognition: Toward a Foundation for Cog nitive Science. Cambridge, MA: MIT Press. Pylyshyn, Z.w., 1999. Is Vision Continuous with Cognition?: The Case for Cog nitive Impenetrability of Visual Perception. Behavioral and Brain Sciences 22, 341–365. https://doi.org/10.1017/s0140525x99002022
289
Copyright © 2021. Taylor & Francis Group. All rights reserved.
290
BIBLIOGRAPHY
Quilty-Dunn, J., 2016. Iconicity and the Format of Perception. Journal of Con sciousness Studies 23, 255–263. Quilty-Dunn, J., 2020. Attention and encapsulation. Mind and Language 35(3), 335–349. https://doi.org/10.1111/mila.12242 Quine, w.V.O., 1951. Two Dogmas of empiricism. Philosophical Review 60, 20–43. https://doi.org/10.2307/2266637 Raftopoulos, A., 2001a. Is Perception Informationally encapsulated? The Issue of the Theory-Ladenness of Perception. Cognitive Science 25, 423–451. https://doi.org/10.1207/s15516709cog2503_4 Raftopoulos, A., 2001b. Reentrant Neural Pathways and the Theory-Ladenness of Perception. Philosophy of Science 3, 187–199. https://doi.org/ 10.1086/392908 Raftopoulos, A., 2006. Defending Realism on the Proper Ground. Philosophical Psychology 19, 47–77. https://doi.org/10.1080/09515080500462370 Ransom, M., Fazelpour, S., Mole, C., 2017. Attention in the Predictive Mind. Consciousness and Cognition 47, 99–112. https://doi.org/10.1016/j.con cog.2016.06.011 Reid, T., 1785. Essays on the Intellectual Power of Man: A Critical Edition, 1st edi tion. university Park, PA: Penn State university Press. Richler, J.J., Bukach, C.M., Gauthier, I., 2009. Context Influences Holistic Pro cessing of Nonface Objects in the Composite Task. Attention, Percep tion, & Psychophysics 71, 530–540. https://doi.org/10.3758/APP.71.3.530 Richler, J., wong, Y., Gauthier, I., 2011. Perceptual expertise as a Shift from Stra tegic Interference to Automatic Holistic Processing. Current Directions in Psychological Science 20, 129–134. Riggs, w.D., 2003. understanding “Virtue” and the Virtue of understanding, in: DePaul, M., Zagzebski, L. (eds.), Intellectual Virtue: Perspectives From Ethics and Epistemology. Oxford: Oxford university Press, pp. 203–226. Robbins, P., 2013. Modularity and Mental Architecture. Wiley Interdisciplinary Reviews: Cognitive Science 4, 641–649. https://doi.org/10.1002/wcs.1255 Robbins, P., 2017. Modularity of Mind, in: Zalta, e.N. (ed.), The Stanford Ency clopedia of Philosophy. Metaphysics Research Lab. Stanford: Stanford university Press. Rock, I., 1983. The Logic of Perception. Cambridge, MA: MIT Press. Rosch, e., Mervis, C.B., Gray, w.D., Johnson, D.M., Boyes-Braem, P., 1976. Basic Objects in Natural Categories. Cognitive Psychology 8, 382–439. https://doi.org/10.1016/0010-0285(76)90013-X
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Rossion, B., Collins, D., Goffaux, V., Curran, T., 2007. Long-Term expertise with Artificial Objects Increases Visual Competition with early Face Categori zation Processes. Journal of Cognitive Neuroscience 19, 543–555. https:// doi.org/10.1162/jocn.2007.19.3.543 Rossion, B., Curran, T., 2010. Visual expertise with Pictures of Cars Correlates with RT Magnitude of the Car Inversion effect. Perception 39, 173–183. https://doi.org/10.1068/p6270 Rossion, B., Gauthier, I., Goffaux, V., Tarr, M.J., Crommelinck, M., 2002. exper tise Training with Novel Objects Leads to Left-Lateralized Facelike elec trophysiological Responses. Psychological Science 13, 250–257. https:// doi.org/10.1111/1467-9280.00446 Rossion, B., kung, C.-C., Tarr, M.J., 2004. Visual expertise with Nonface Objects Leads to Competition with the early Perceptual Processing of Faces in the Human Occipitotemporal Cortex. Proceedings of the National Acad emy of Sciences of the United States of America 101, 14521–14526. https:// doi.org/10.1073/pnas.0405613101 Rule, N.O., Ambady, N., Adams, R.B., Macrae, C.N., 2007. us and Them: Mem ory Advantages in Perceptually Ambiguous Groups. Psychonomic Bulle tin & Review 14, 687–692. https://doi.org/10.3758/BF03196822 Rule, N.O., Garrett, J.V., Ambady, N., 2010. Places and Faces: Geographic envi ronment Influences the Ingroup Memory Advantage. Journal of Personal ity and Social Psychology 98, 343–355. https://doi.org/10.1037/a0018589 Sartre, J.-P., 1943. Being and Nothingness: An Essay on Phenomenological Ontol ogy. London: Routledge. Scheck, B., Neufeld, P., Dwyer, J., 2003. Actual Innocence. New York: New Amer ican Library. Schendan, H.e., Ganis, G., 2015. Top-Down Modulation of Visual Processing and knowledge after 250 ms Supports Object Constancy of Category Deci sions. Frontiers in Psychology 6. https://doi.org/10.3389/fpsyg.2015.01289 Schendan, H.e., Stern, C.e., 2008. where Vision Meets Memory: Prefrontal – Posterior Networks for Visual Object Constancy During Categorization and Recognition. Cerebral Cortex 18, 1695–1711. https://doi.org/10.1093/ cercor/bhm197 Scholl, B.J., 2001. Objects and Attention: The State of the Art. Cognition 80, 1–46. https://doi.org/10.1016/s0010-0277(00)00152-9 Schwartz, S., 2007. Functional MRI evidence for Neural Plasticity at early Stages of Visual Processing in Humans, in: Osaka, N., Rentschler, I.,
291
Copyright © 2021. Taylor & Francis Group. All rights reserved.
292
BIBLIOGRAPHY
Biederman, I. (eds.), Object Recognition, Attention, and Action. Japan, Tokyo: Springer, pp. 27–40. https://doi.org/10.1007/978-4-431-73019-4_3 Schwartz, S., Maquet, P., Frith, C., 2002. Neural Correlates of Perceptual Learn ing: A Functional MRI Study of Visual Texture Discrimination. Proceedings of the National Academy of Sciences of the United States of America 99, 17137–17142. https://doi.org/10.1073/pnas.242414599 Scolari, M., ester, e.F., Serences, J.T., 2014. Feature- and Object-Based Atten tional Modulation in the Human Visual System. The Oxford Handbook of Attention. https://doi.org/10.1093/oxfordhb/9780199675111.013.009 Scott, L.S., 2011. Face Perception and Perceptual expertise in Adult and Devel opmental Populations, in: Oxford Handbook of Face Perception. https:// doi.org/10.1093/oxfordhb/9780199559053.013.0011 Scott, L.S., Tanaka, J.w., Sheinberg, D.L., Curran, T., 2006. A Reevaluation of the electrophysiological Correlates of expert Object Processing. Jour nal of Cognitive Neuroscience 18, 1453–1465. https://doi.org/10.1162/ jocn.2006.18.9.1453 Scott, L.S., Tanaka, J.w., Sheinberg, D.L., Curran, T., 2008. The Role of Cate gory Learning in the Acquisition and Retention of Perceptual expertise: A Behavioral and Neurophysiological Study. Brain Research 1210, 204– 215. https://doi.org/10.1016/j.brainres.2008.02.054 Searle, J.R., 1983. Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge university Press. Seeley, w.P., 2020. Attentional Engines: A Perceptual Theory of the Arts. New York: Oxford university Press. Segall, M.H., Campbell, D.T., Herskovits, M.J., 1966. The Influence of Culture on Visual Perception. Oxford: Bobbs-Merrill. Sekuler, A.B., 1994. Local and Global Minima in Visual Completion: effects of Symmetry and Orientation. Perception 23, 529–545. https://doi. org/10.1068/p230529 Sellars, w., 1956. Empiricism and the Philosophy of Mind. Cambridge, MA: Har vard university Press. Sergent, J., Ohta, S., MacDonald, B., 1992. Functional Neuroanatomy of Face and Object Processing. A Positron emission Tomography Study. Brain 115 Pt 1, 15–36. https://doi.org/10.1093/brain/115.1.15 Shea, N., 2015. Distinguishing Top-Down From Bottom-up effects, in: Stokes, D., Matthen, M., Biggs, S. (eds.), Perception and Its Modalities. Oxford: Oxford university Press, pp. 73–91.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Shiffrin, R., Schneider, w., 1977. Controlled and Automatic Human Informa tion Processing: II. Perceptual Learning, Automatic Attending and a Gen eral Theory. Psychological Review 84, 127–190. https://doi.org/10.1037/ 0033-295X.84.2.127 Shriver, e.R., Hugenberg, k., 2010. Power, individuation, and the cross-race recognition deficit. Journal of Experimental Social Psychology 46, 767–774. https://doi.org/10.1016/j.jesp.2010.03.014 Sibley, F., 1959. Aesthetic Concepts. Philosophical Review 68, 421–450. https:// doi.org/10.2307/2182490 Siegel, S., 2005. which Properties Are Represented in Perception, in: Gendler, T.S., Hawthorne, J. (eds.), Perceptual Experience. Oxford: Oxford univer sity Press, pp. 481–503. Siegel, S., 2009. The Visual experience of Causation. Philosophical Quarterly 59, 519–540. https://doi.org/10.1111/j.1467-9213.2008.607.x Siegel, S., 2011. The Contents of Visual Experience. Oxford: Oxford university Press. Siegel, S., 2012. Cognitive Penetrability and Perceptual Justification. Noûs 46. Siegel, S., 2013. The epistemic Impact of the etiology of experience. Philosophi cal Studies 162, 697–722. https://doi.org/10.1007/s11098-012-0059-5 Siegel, S., 2016. The Contents of Perception, in: Zalta, e.N. (ed.), The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab. Stanford: Stanford university Press. Siegel, S., 2017. The Rationality of Perception. Oxford: Oxford university Press. Siegel, S., 2018. The Rationality of Perception: Reply to Begby, Ghijsen, and Samoilova. Analysis 78, 523–536. Siewert, C., 1998. The Significance of Consciousness. Princeton, NJ: Princeton university Press. Silins, N., 2016. Cognitive Penetration and the epistemology of Perception. Philosophy Compass 11, 24–42. https://doi.org/10.1111/phc3.12292 Simons, D.J., Chabris, C.F., 1999. Gorillas in Our Midst: Sustained Inatten tional Blindness for Dynamic events. Perception 28, 1059–1074. Sloan Report, 1978. http://csjarchive.cogsci.rpi.edu/misc/CognitiveScience 1978_OCR.pdf Smeeton, N.J., williams, A.M., Hodges, N.J., ward, P., 2005. The Relative effectiveness of explicit Instruction, Guided-Discovery and Discovery Learning Techniques in enhancing Perceptual Skill in Sport. Journal of Experimental Psychology: Applied, 11, 98–110.
293
Copyright © 2021. Taylor & Francis Group. All rights reserved.
294
BIBLIOGRAPHY
Smith, A.D., 2001. Perception and Belief. Philosophy and Phenomenological Research 62, 283–309. https://doi.org/10.1111/j.1933-1592.2001.tb00057.x Smith, A.D., 2002. The Problem of Perception. Cambridge, MA: Harvard uni versity Press. Snowdon, P.F., 1980. Perception, Vision, and Causation. Proceedings of the Aris totelian Society 81, 175–192. Sosa, e., 1980. The Raft and the Pyramid: Coherence versus Foundations in the Theory of knowledge. Midwest Studies in Philosophy 5, 3–26. https://doi. org/10.1111/j.1475-4975.1980.tb00394.x Sosa, e., 2007. A Virtue Epistemology: Apt Belief and Reflective Knowledge, Vol ume I. Oxford: Oxford university Press. Sosa, e., 2009. Reflective Knowledge: Apt Belief and Reflective Knowledge, Vol ume II. Oxford: Oxford university Press. Sosa, e., 2015a. Judgment and Agency. Oxford: Oxford university Press. Sosa, e., 2015b. Getting It Right. Opinionator-The New York Times. Sowden, P.T., 1999. expert Perceivers and Perceptual Learning. Behavioral and Brain Sciences 22, 396–397. https://doi.org/10.1017/S0140525X995 52023 Sperandio, I., Chouinard, P.A., Goodale, M.A., 2012. Retinotopic Activity in V1 Reflects the Perceived and not the Retinal Size of an Afterimage. Nature Neuroscience 15, 540–542. https://doi.org/10.1038/nn.3069 Sperber, D., 1994. The Modularity of Thought and the epidemiology of Rep resentations, in: Mapping the Mind: Domain Specificity in Cognition and Culture. New York, NY, uSA: Cambridge university Press, pp. 39–67. https://doi.org/10.1017/CBO9780511752902.003 Sperber, D., 2001. In Defense of Massive Modularity, in: Language, Brain, and Cognitive Development: Essays in Honor of Jacques Mehler. Cambridge, MA: The MIT Press, pp. 47–57. Stampe, D.w., 1977. Toward a Causal Theory of Linguistic Representation. Mid west Studies in Philosophy 2, 42–63. https://doi.org/10.1111/j.1475-4975.1977. tb00027.x Stefanucci, J.k., Proffitt, D.R., 2009. The Roles of Altitude and Fear in the Per ception of Height. Journal of Experimental Psychology: Human Perception and Performance 35, 424. Stefanucci, J.k., Proffitt, D.R., Clore, G.L., Parekh, N., 2008. Skating Down a Steeper Slope: Fear Influences the Perception of Geographical Slant. Per ception 37, 321–323. https://doi.org/10.1068/p5796
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Stefanucci, J.k., Storbeck, J., 2009. Don’t Look Down: emotional Arousal elevates Height Perception. Journal of Experimental Psychology: General 138, 131. Stephens, C.L., 2001. when is It Selectively Advantageous to Have True Beliefs? Sandwiching the Better Safe Than Sorry Argument. Philosophical Studies 105, 161–189. https://doi.org/10.1023/A:1010358100423 Sterelny, k., 2003. Thought in a Hostile World: The Evolution of Human Cogni tion, 1st edition. Malden, MA: wB. Steup, M., 2001. Knowledge, Truth, and Duty: Essays on Epistemic Justification, Responsibility, and Virtue. Oxford: Oxford university Press. Stocker, M., 1982. Responsibility especially for Beliefs. Mind 91, 398–417. Stokes, D., 2012. Perceiving and Desiring: A New Look at the Cognitive Pen etrability of experience. Philosophical Studies 158, 479–492. https://doi. org/10.1007/s11098-010-9688-8 Stokes, D., 2013. Cognitive Penetrability of Perception: Cognitive Penetrability of Perception. Philosophy Compass 8, 646–663. https://doi.org/10.1111/ phc3.12043 Stokes, D., 2014. Cognitive Penetration and the Perception of Art (win ner of 2012 Dialectica essay Prize). Dialectica 68, 1–34. https://doi. org/10.1111/1746-8361.12049 Stokes, D., 2015. Towards a Consequentialist understanding of Cognitive Penetra tion, in: Zeimbekis, J., Raftopoulos, A. (eds.), The Cognitive Penetrability of Perception: New Philosophical Perspectives. Oxford: Oxford university Press. Stokes, D., 2018. Attention and the Cognitive Penetrability of Perception. Aus tralasian Journal of Philosophy 96, 303–318. https://doi.org/10.1080/0004 8402.2017.1332080 Stokes, D., 2020a. Cognitive Penetration and the Perception of Colour, in: Brown, D., Macpherson, F. (eds.), The Routledge Handbook of Philosophy of Colour. London: Routledge. Stokes, D., 2020b. On Perceptual expertise. Mind & Language. https://doi. org/10.1111/mila.12270 Stokes, D., Bergeron, V., 2015. Modular Architectures and Informational encap sulation: A Dilemma. European Journal for Philosophy of Science 5, 315–338. https://doi.org/10.1007/s13194-015-0107-z Stokes, D., Matthen, M., Biggs, S. (eds.), 2014. Perception and its Modalities. New York: Oxford university Press. Stokes, D., Nanay, B., 2020. Perceptual Skills, in: Fridland, e., Pavese, C. (eds.), Routledge Handbook on Skill and Expertise. London: Routledge.
295
Copyright © 2021. Taylor & Francis Group. All rights reserved.
296
BIBLIOGRAPHY
Stratton, G.M., 1897. Vision without Inversion of the Retinal Image. Psycholog ical Review 4, 463–481. https://doi.org/10.1037/h0071173 Tanaka, J.w., Curran, T., 2001. A Neural Basis for expert Object Recognition. Psychological Science 12, 43–47. https://doi.org/10.1111/1467-9280.00308 Tanaka, J.w., Curran, T., Sheinberg, D.L., 2005. The Training and Transfer of Real-world Perceptual expertise. Psychological Science 16, 145–151. https:// doi.org/10.1111/j.0956-7976.2005.00795.x Tanaka, S., Fujita, I., 2015. Computation of Object Size in Visual Cortical Area V4 as a Neural Basis for Size Constancy. The Journal of Neuroscience 35, 12033–12046. https://doi.org/10.1523/JNeuROSCI.2665-14.2015 Tarr, M.J., Gauthier, I., 2000. FFA: A Flexible Fusiform Area for SubordinateLevel Visual Processing Automatized by expertise. Nature Neuroscience 3, 764–769. https://doi.org/10.1038/77666 Taylor, J.G., 1962. The Behavioral Basis of Perception. New Have, CT: Yale uni versity Press. Telling, A.L., kumar, S., Meyer, A.S., Humphreys, G.w., 2010. electrophysiolog ical evidence of Semantic Interference in Visual Search. Journal of Cogni tive Neuroscience 22, 2212–2225. https://doi.org/10.1162/jocn.2009.21348 Teufel, C., Dakin, S.C., Fletcher, P.C., 2018. Prior Object-knowledge Sharpens Properties of early Visual Feature-Detectors. Scientific Reports 8, 1–12. https://doi.org/10.1038/s41598-018-28845-5 Thompson, e., 2015. Waking, Dreaming, Being: Self and Consciousness in Neuro science, Meditation, and Philosophy. New York: Columbia university Press. Thompson, P., 1980. Margaret Thatcher: A New Illusion. Perception 9(4), 483– 484. https://doi.org/10.1068/p090483 Treisman, A.M., Gelade, G., 1980. A Feature-Integration Theory of Attention. Cognitive Psychology 12, 97–136. https://doi.org/10.1016/0010-0285(80)90 005-5 Treue, S., Martinez-Trujillo, J.C., 2007. Attending to Features Inside and Outside the Spotlight of Attention. Neuron 55, 174–176. https://doi.org/10.1016/j. neuron.2007.07.005 Tucker, C., 2014. If Dogmatists Have a Problem with Cognitive Penetration, You Do Too. Dialectica 68, 35–62. https://doi.org/10.1111/1746-8361.12050 Turri, M., John, Alfano, Greco, J., 2019. Virtue epistemology, in: Zalta, e.N. (ed.), The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab. Stanford: Stanford university Press. uttal, w.R., 2001. The New Phrenology: The Limits of Localizing Cognitive Pro cesses in the Brain. Cambridge, MA: The MIT Press.
Copyright © 2021. Taylor & Francis Group. All rights reserved.
BIBLIOGRAPHY
Valenti, J.J., Firestone, C., 2019. Finding the “Odd One Out”: Memory Color effects and the Logic of Appearance. Cognition 191, 103934. https://doi. org/10.1016/j.cognition.2019.04.003 Vance, J., 2014. emotion and the New epistemic Challenge From Cognitive Penetrability. Philosophical Studies 169, 257–283. https://doi.org/10.1007/ s11098-013-0181-z Vance, J., 2015. Cognitive Penetration and the Tribunal of experience. Review of Philosophy and Psychology 6, 641–663. https://doi.org/10.1007/s13164 014-0197-0 Vance, J. unpublished ms. Predictive processing and persistent illusions. Vance, J., Stokes, D., 2017. Noise, uncertainty, and Interest: Predictive Cod ing and Cognitive Penetration. Consciousness and Cognition 47, 86–98. https://doi.org/10.1016/j.concog.2016.06.007 van Fraassen, B.C., 2002. The Empirical Stance. New Haven: Yale university Press. van Gulick, R., 1994. Deficit Studies and the Function of Phenomenal Con sciousness, in: Graham, G., Stephens, G.L. (eds.), Philosophical Psycho pathology. Cambridge, MA: MIT Press. van ulzen, N., Semin, G., Oudejans, R., Beek, P., 2008. Affective Stimulus Prop erties Influence Size Perception and the ebbinghaus Illusion. Psychologi cal Research 72, 304–310. https://doi.org/10.1007/s00426-007-0114-6 Varela, F.J., Thompson, e., Rosch, e., 1991. The Embodied Mind: Cognitive Sci ence and Human Experience. Cambridge, MA: MIT Press. Vogt, S., Magnussen, S., 2007. expertise in Pictorial Perception: eye-Movement Patterns and Visual Memory in Artists and Laymen. Perception 36, 91–100. https://doi.org/10.1068/p5262 von Bubnoff, A., 2014. Face-Blind People Can Learn to Tell Similar Shapes Apart. Nature News, 14916. https://doi.org/10.1038/nature ward, P., williams, A.M., 2003. Perceptual and Cognitive Skill Development in Soccer: The Multidimensional Nature of expert Performance. Journal of Sport & Exercise Psychology 25, 93–111. watzl, S., 2011. Attention as Structuring of the Stream of Consciousness, in: Mole, C., Smithies, D., wu, w. (eds.), Attention: Philosophical and Psycho logical Essays. Oxford: Oxford university Press, p. 145. watzl, S., 2017. Structuring Mind: The Nature of Attention and How It Shapes Consciousness. Oxford: Oxford university Press. white, A.L., Carrasco, M., 2011. Feature-Based Attention Involuntarily and Simultaneously Improves Visual Performance Across Locations. Journal of Vision 11. https://doi.org/10.1167/11.6.15
297
Copyright © 2021. Taylor & Francis Group. All rights reserved.
298
BIBLIOGRAPHY
williams, B., 1970. Deciding to Believe, in: williams, B. (ed.), Problems of the Self. Cambridge: Cambridge university Press. pp. 136–151. williams, A.M., Ford, P.R., eccles, D.w., ward, P., 2011. Perceptual-Cognitive expertise in Sport and Its Acquisition: Implications for Applied Cogni tive Psychology. Applied Cognitive Psychology 25, 432–442. https://doi. org/10.1002/acp.1710 witt, J.k., Proffitt, D.R., epstein, w., 2004. Perceiving Distance: A Role of effort
and Intent. Perception 33, 577–590. https://doi.org/10.1068/p5090
witzel, C., 2016. An easy way to Show Memory Color effects. i-Perception 7,
2041669516663751. https://doi.org/10.1177/2041669516663751 witzel, C., Valkova, H., Hansen, T., Gegenfurtner, k.R., 2011. Object knowledge Modulates Colour Appearance. i-Perception. https://doi.org/10.1068/ i0396 wong, A.C.-N., Palmeri, T.J., Gauthier, I., 2009. Conditions for Facelike expertise with Objects: Becoming a Ziggerin expert – but which Type? Psychological Science 20, 1108–1117. https://doi.org/10.1111/j.1467-9280.2009.02430.x wright, L., 1973. Functions. Philosophical Review 82, 139–168. https://doi.org/ 10.2307/2183766 wu, w., 2011. what is Conscious Attention? Philosophy and Phenomenological Research 82, 93–120. https://doi.org/10.1111/j.1933-1592.2010.00457.x wu, w., 2013. Visual Spatial Constancy and Modularity: Does Intention Pene trate Vision? Philosophical Studies 165, 647–669. https://doi.org/10.1007/ s11098-012-9971-y wu, w., 2014. Attention. London: Routledge. wu, w., 2017. Shaking up the Mind’s Ground Floor: The Cognitive Penetra tion of Visual Attention. Journal of Philosophy 114, 5–32. https://doi. org/10.5840/jphil201711411 Young, S.G., Hugenberg, k., Bernstein, M.J., Sacco, D.F., 2011. Perception and Motivation in Face Recognition: A Critical Review of Theories of the Cross-Race effect. Personality and Social Psychology Review 16, 116–142. https://doi.org/10.1177/1088868311418987 Zagzebski, L.T., 1996. Virtues of the Mind: An Inquiry Into the Nature of Virtue and the Ethical Foundations of Knowledge. Cambridge: Cambridge univer sity Press. Zagzebski, L.T., 2001. Recovering understanding, in: Steup, M. (ed.), Knowl edge, Truth, and Duty: Essays on Epistemic Justification, Responsibility, and Virtue. Oxford: Oxford university Press.
BIBLIOGRAPHY
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Zangwill, N., 2019. Aesthetic Judgment, in: Zalta, e.N. (ed.), The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab., Stanford: Stan ford university Press. Zawidski, T., Bechtel, w.P., 2004. Gall’s Legacy Revisited: Decomposition and Localization in Cognitive Neuroscience, in: erneling, C.e., Johnson, D.M. (eds.), Mind as a Scientific Object. Oxford: Oxford university Press. Zeimbekis, J., 2013. Color and Cognitive Penetrability. Philosophical Studies 165, 167–175. Zeimbekis, J., Raftopoulos, A. (eds.), 2015. The Cognitive Penetrability of Percep tion: New Philosophical Perspectives, 1st edition. Oxford: Oxford university Press.
299
INDEX
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Note: Page numbers in italics indicate a figure on the corresponding page. accentuation 75, 140n6 accuracy conditions 31, 187 – 188, 199 – 200, 223 – 231, 242 – 243 action 15, 80 – 81 action potential 16, 102 – 103, 230 aesthetic judgment 236n19 aesthetic properties 38, 231 – 233 aesthetic sensitivity 231 – 234 affordances 102 Ahissar, M. 172n18 Ahluwalia, A. 72n14 Alfano, J. 203 Allard, F. 171n17, 193 Allport, D.A. 139 Alston, W. 72n10 Anderson, M. 40, 71, 71n3, 230 Antony, L. 203n1 architectural claims 149, 162 – 163, 165 – 166, 175, 210, 241 – 242
argument from consequences 135 – 138 argument from directness 131 – 135 arguments from reliability 45, 61 – 70, 174 arguments from stability 55–61, 64, 70 Aristotle 41n11, 47, 173, 180, 196, 203, 210, 250 Armstrong, D. 19 artificial intelligence 14 Ashby, W.R. 41n9 assertive content 32 – 33 attention: biased competition model 123 – 124, 135 – 136, 140n3; bottleneck theories 120; cognitive influence on 134 – 135; competition model of 122 – 124; feature-based 120 – 121, 124 – 125, 127 – 128, 131, 136, 140n3, 170 – 171n10, 198;
INDex
Copyright © 2021. Taylor & Francis Group. All rights reserved.
integrated competition model of attention 132; object-based 120 – 121, 131, 136, 140n3, 198; as part of visual system 132 – 133; role in cognitive effects on perception 117 – 119; selective attention 110, 120 – 130; see also spatial attention attentional weighting 190 attention as act assumption 116, 118, 131, 135, 167, 198 attention as gatekeeper assumption 117, 119, 131–134, 135, 139, 167–168, 199 attention as spotlight assumption 110, 131, 133, 135, 167, 198 attention-shift explanation 76 – 77, 85n2, 97, 117 – 119, 125, 127, 131, 139n1, 140n3, 148 – 149, 152, 161, 197 – 198, 238 – 239 audition 16, 36 – 37, 39 automaticity 152, 160 – 161, 176, 195 – 196 Bach, K. 113n1 Balcetis, e. 77, 85n3 Banaji, M.R. 80, 85n5, 218, 235 Barton, J.J.S. 171n12 basic-level categories 147 Battaly, H. 181, 203 Batty, C. 203n1 Bayne, T. 42n18, 43n22, 43n23 Bechtel, W.P. 71n1 Beck, J. 26, 34, 40, 43n21 Becklen, R. 120 Beeghly, e. 235 behaviourism 240 belief 14 – 15, 16, 18 – 21, 30 – 34, 39, 47, 56, 57, 59, 72n10, 101, 113n1
Belke, e. 126 Bennett, J. 72n10 Berardi, N. 172n18 Bergeron, V. 40n1, 63, 85n5 Bergqvist, A. 43n26 Berkeley, G. 41n12 biased competition model of attention 123 – 124, 135 – 136, 140n3, 198 – 199 binocular rivalry 77 – 78, 78 biological systems 184 – 189, 200 – 201 Block, N. 139 Bodenhausen, G.V. 218, 235 BonJour, L. 32 bottleneck theories of attention 120 Brewer, B. 42n19 Brewer, W.F. 112, 114n6, 236n8 Brigham, J.C. 235 Broadbent, D.e. 120, 139 Brownstein, M. 235, 236n10 Brown, T.G. 202 Bruner, J. 6, 21 – 23, 41n6, 41n7, 41n9, 42n13, 48 – 49, 63, 75 – 77, 91 – 93, 108, 140n6, 244n3 Burge, T. 13, 40, 42n19, 72n15, 186 – 189, 191, 199, 203n4, 225 – 226 Burnston, D. 40 Busey, T.A. 147, 155, 171n12 Butterfill, S.A. 43n24 Campbell, J. 139 Candrakirti 254 Carey, S. 33, 155 “carpentered” environments 57, 72n14 Carrasco, M. 121, 139 Carruthers, P. 70, 72n12 Cartwright, N. 236n8
301
Copyright © 2021. Taylor & Francis Group. All rights reserved.
302
INDex
categorization tasks 146 – 147 causal properties 38 Causer, J. 190 Cecchi, A.S. 236n12 central processes 59 – 61 Chabris, C.F. 41n13 Chisholm, R.M. 203n2 Chomsky, N. 48, 49, 240 Churchland, P.M. 57, 71n5, 76, 97, 112, 114n6, 116, 117, 205n12, 216, 227, 235 Churchland, P.S. 71n5 Clark, A. 22 – 23, 58, 73n20, 85n8, 102 – 103 Clarke, S. 34, 40 Clerkin, e.M. 85n3 cognition: characteristics 13 – 16; format of 34; modes for distinguishing perception from 23 – 29, 40 cognitive impenetrability 54, 64 – 65, 73n18, 101, 151, 221 cognitive neuropsychology 145 cognitive penetration of perception: consequences of 99 – 105; consequentialism 105 – 112, 168 – 169; definitions 89 – 90, 113 – 114n2; definitions applied to cases of diachronic changes 94 – 99; definitions applied to cases of value-influenced perception 91 – 94; discussion of 85n4; effects on perception 131; epistemic threat of 150; evidence of 85; general epistemology of 112 cognitive science 4, 14, 18, 49, 104, 239 – 240
Cohen, J. 60 colour perception 79 – 80, 85n4 Coltheart, M. 145 competition model of attention 122 – 124, 135 – 136, 140n3 composite task 152 – 155, 153, 195, 220 computationalism 51 – 52, 111 computational theories of vision 71n5 computational theory of mind 51, 71n5 concepts: perceptual content and 34 – 36; requirements 26 – 28 conceptual content 34, 150, 162, 165 conjunctive consequentialism 106 Connolly, K. 189 – 190 consequentialism 105 – 112, 168 – 169 consequentialist constraint 105 – 106, 112 constancy mechanisms 24, 59 – 61, 73n16, 81, 170n4 contact 1 – 4, 243 contact hypothesis 219 content: admissible content of perceptual experience 36 – 39, 231 – 234; cognitive content 34 – 35; concepts and 34 – 36; conceptualism 34, 36; digital vs. analog 34; format of 33 – 34, 40; imagistic 47 – 48; importance and variety 32 – 34; mental states 30 – 32; non-conceptual perceptual content 35 – 36; perceptual content 34 – 37; richness of 3, 28, 35, 40 content universalism 229 contextual empiricism 213 – 214 Copenhaver, R. 13 Corneille, O. 220 corollary discharge signal 81
INDex
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Correll, J. 218 Cosmides, L. 70 covert attention 132 Cowan, R. 43n26 Crane, T. 27 Critique of Judgment (Kant) 231 Cross, J.F. 221 cross-race effect (CRe) 219 – 221, 234, 235, 242 cross-race faces (CR) 219 – 220 cross-talk, problem of 84, 87, 88 – 89, 94, 98, 112, 236n9 Cummins, R.183 – 184, 203 Curby, K.M. 155, 157 – 158, 169 Curran, T. 155, 159 Cutter, B. 237n20 d’Almeida, O.C. 205n11 Dan, Y. 205n11 decision-making 16, 39, 59, 146 default position assumption 49 – 50, 70, 114n3, 240, 248 Delk, J.L. 79 – 80 Descartes, R. 15, 23, 41n11, 41n12, 47, 174, 203n2, 250, 254 Desimone, R. 120, 122 – 124, 124, 127, 199 desire 16, 39 diachronic cognitive effect 57, 94 – 99, 109 – 111 Diamond, R. 155 disjunctive consequentialism 106 – 107, 109 – 110 dispositional states 15 distributed attention 139n2 domain-specific expertise 95, 137, 146, 152, 164, 176, 182, 192 – 193,
195, 197, 199, 207 – 209, 233 – 234, 252 – 253 domain-specific faculties 49 domain specificity: in cognitive learning 162, 164–165, 168, 175, 177, 181–182, 192, 194, 202, 208, 233–234, 241; concept in Sloan report 72n9; in functions of perception 200, 206; of modules 49–51, 53; in VSTM performance 158 Downes, S.M. 203 doxastic involuntarism 72n10 doxastic states 15, 20 – 21, 26, 78 – 79, 92, 137, 179, 211, 228 Draganski, B. 202 Dretske, F. 13, 20, 27, 34, 40, 42n15, 42n17, 43n22, 203n4 Drew, T. 137 Dreyfus, H.L. 42n13 duck–rabbit image 76, 115, 117, 118 Duhem, P.M.M. 235n6 Duncan, J. 120, 122 – 124, 199 Dunning, D. 77, 85n3 electroencephalogram (eeG) 136, 148, 161, 167 elite athletes 171 – 172n17, 190, 193, 196, 201, 252 – 253 emotional properties 38 empiricism 42n14 epistemic/epistemological claim 149 – 150, 174 – 175 epistemic externalism, 113n1 epistemic internalism 113n1, 176 – 178 epistemic justification 176 – 177 epistemic normativity 192 epistemic seeing 27
303
304
INDex
epistemic virtue 143 – 144, 175, 176, 179 – 183, 192 – 197, 199, 207, 217, 224, 243 etiological analysis: 185, 203 E E
Copyright © 2021. Taylor & Francis Group. All rights reserved.
expertise-expertise interference 156 eye tracking 82, 126 – 127, 130, 135 – 137, 148, 159, 161, 168, 193, 196 – 198, 202, 208, 241 face perception 146, 155, 158 – 159, 161, 163, 219 – 221, 234, 235, 242, 252 face recognition 146 faculty psychology 47 – 51 faculty virtue 180 – 182 Fahle, M. 190 Fairweather, A. 203 Fazekas, P. 139 feature-based attention 140n3 feature-based attention (FBA) 120 – 121, 124 – 125, 127 – 128, 131, 136, 140n3, 170 – 171n10, 198 Feyerabend, P. 211, 236n9 Fillenbaum, S. 79 – 80 Fiorentini, A. 172n18 Firestone, C. 85, 85n4, 116, 117, 139, 167 fixed neural architectures 50 focused attention 139n2 Fodor, J. 71n6, 71n8, 76, 108 – 109, 114n7, 115, 136, 244n4; arguments from reliability of perception 61 – 70; arguments from stability of perception 55 – 61, 70, 142; attention as spotlight assumption 133; attention-shift explanation
117 – 119, 131; conditions for disproving theory 71n4; definition of cognitive penetrability of perception 112; on function of perception 100 – 101; inaccessibility claim 59 – 60; informational encapsulation 50, 52 – 54, 70, 72n13, 104; inverting-lens adaptation 109; mental modules 50 – 51; Modularity of Mind 48 – 49, 51; modus ponens 224; non-interference claim 59 – 61; non-mandatory mental systems, 52; on perceptual adaptation 97; on perceptual expertise 148, 165 – 167, 170n4; persistent illusions 58 – 59, 144; theory-ladenness 114n6; use of theory from Gall 49; wishful seeing 222 Forber, P. 203 Franklin, A. 236n8 Fregean propositions 34 Fregean Thoughts 34 Freire, A. 171n12 Fridland, e. 169 Friston, K. 41n10, 73n20 Fujita, I. 73n16 Fuller, S. 139 functional analysis, 183 – 188 functional magnetic resonance imaging (fMRI) 148, 159, 161, 167, 202 functional-teleological analysis 175, 183 fusiform face area (FFA) 159, 202 Gall, J. 47 – 51, 53 Ganis, G. 60 Gauthier, I. 146, 154, 155, 157–159, 164, 169, 170n2, 170n3, 171n12, 235n1 Gawronski, B. 218, 235
INDex
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Gazzaley, A. 126 Gelade, G. 120, 139 Gendler, T.S. 236n11 generality constraint 35 gestalt properties 231 – 234, 243 Gibb, R. 202 Gibson, B.S. 190 Gibson, e. 189 Gibson, J.J. 42n13, 80, 102, 114n7, 189, 229 Ginet, C. 203n2 Glüer, K. 41n5 Godfrey-Smith, P. 204n5 Goldman, A. 178 Goldstone, R.L. 190 Goodman, C.C. 75–77, 91–93, 108, 140n6 Graham, P. 185 – 189, 191, 199, 204n5, 204n6 Greco, J. 180, 182, 203n3 Greebles 147, 147, 152, 153, 163, 164, 231 Green, e.J. 38 – 39, 40 Greenwald, A.G. 218, 235 Gregory, R.L. 41n6 Grill-Spector, K. 170n2 Gross, S. 85n4, 139 Guernica (Picasso) 232 Hacking, I. 236n8 Haddock, A. 42n19 Hansen, T. 79 Hanson, N.R. 42n17, 99, 112, 114n5, 210 – 215 Harman, G. 13 Hawley, K. 40 Heck, R. 34 – 35 Helmholtz, H. von 41n9 Henrich, J. 72n13 high-level properties 37, 39, 95
Hinton, J.M. 42n19 Hobbes, T. 41n12 Hohwy, J. 41n10, 73n20 holistic processing 154 – 156, 158 Hopf, J.M. 82, 126 Howe, C.Q. 58 Hugenberg, K. 220 Hume, D. 41n12, 47 – 48, 231, 236n19 Hurley, S.L. 42n13 iconic representations 33 identification tasks 146 – 147 imagining 47 – 48 implicit bias 218, 221 – 222, 225, 235 imprinting 190 inaccessibility claim 59 – 60 inattentional blindness 24 – 25, 41n13, 121 informational encapsulation: arguments from reliability of perception 62 – 63, 69 – 70; arguments from stability of perception 56 – 57, 59; attention as gatekeeper assumption 134; biological-teleological argument for 72n13; as categorical or absolute 62 – 63, 73n18; degrees of 63; as essence of modularity 50, 52 – 54, 70, 104, 174, 239 – 240; formalized “definition” of 72n11; mental modules 52 – 53, 74 – 75, 167 – 168; non-doxastic influences 222; perceptual accuracy and 224; perceptual expertise 148; violation of 136, 148 integrated competition model of attention 132 – 133 intellectual virtue 180, 193
305
Copyright © 2021. Taylor & Francis Group. All rights reserved.
306
INDex
intentions 15, 24, 28, 31, 39, 46, 80 – 81, 110 interference effects 152 – 155, 157, 159 inter-subjective objectivity 213 – 214, 226, 242 intra-perceptual explanation 109 inversion effects 155 – 156, 160 inverting-lens adaptation 109
Kolb, B. 202 Kottenhoff, H. 97 Kriegel, U. 40, 42n18 Krupinski, e.A. 165 Kuhn, T. 112, 211 – 213, 222, 235, 236n8, 236n9 Kundel, H.L. 159, 193 Kvanvig, J. 250, 251
Jahoda, G. 72n14 Jenkin, Z. 85 Johnson, K.e. 147 Johnston, M. 203n1 judgment: aesthetic judgment 236n19; as cognitive state 5, 10, 15 – 16, 29, 37, 74, 95, 150, 247 – 248, 248; forming 60, 151, 160 – 161, 176 – 177; implicit bias 218; inaccurate 31; influence of theory on observational judgment 211; perceptual expertise and 7; perceptual judgment 137; reasoning and 48, 195; recency judgment 82; spatial judgment 80 judgment interpretation 76 – 77, 79, 85n2, 92 – 93, 107 – 108 justified belief 113n1 justified believer 113n1
laboratory-trained expertise 147 – 148 La Follette, P.S. 159 Lambert, B.L. 112, 114n6, 236n8 lateral geniculate nucleus (LGN) 61 Laudan, L. 235, 236n8 Lee, T.S. 41n10 Levin, D.T. 80, 85n5 linguistic knowledge 82, 137 – 138 localization theory 71n3 Locke, J. 26, 41n12, 42n15 logical positivism 211, 214 Logothetis, N.K. 170n3 Longino, H. 213 – 214, 235, 236n8 Lupyan, G. 58 – 59, 73n20, 85n8 Lyons, Jack 101 Lyons, J.C. 90, 101, 112, 137
Kalderon, M.e. 203n1 Kanai, R. 202 Kant, I. 231 Kanwisher, N. 146, 169, 170n2 Karmarkar, U.R. 205n11 Kepler, J. 210, 211 Kiss, M. 82, 127 knowledge 32, 36, 39, 82, 104, 137 – 138, 144, 249 – 250
Macevoy, S.P. 73n16 Mack, A. 41n13 Macpherson, F. 36, 40, 40n1, 42n19, 43, 79, 85n2, 85n4, 85n5, 89, 94, 96 – 98, 112, 117, 131, 203n1 Magnussen, S. 159 Malleability of mind 9, 83 – 84, 139, 169, 172n19, 201 – 202, 217, 221, 223 – 224, 234, 238, 241 – 242, 246, 253 – 256 mandatoriness 51 – 52 Mandelbaum, e. 40, 71n2
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INDex
Mann, D.T.Y. 192, 204n7 Martin, M.G.F. 42n19, 43n22 “massive modularity” theories 70, 72n12 Matthen, M. 41n11, 42n13, 204, 229 – 230 Matthews, L. 203 Maurer, D. 148 May, A. 202 Maya-Vetencourt, J.F. 205n11 Mayo, D.G. 236n8 McCauley, R.N. 72n13 McDowell, J. 34, 36, 40, 73n21 McGugin, R.W. 157 McKeeff, T.J. 156 McKone, e. 170n2 Meditations (Descartes) 254 Meissner, C.A. 235 memory 4, 16, 31, 54, 59, 59 – 60, 76, 79 – 80, 82, 90, 93, 107, 124 – 130, 133, 136, 140, 142, 155, 157, 160, 168, 175, 180 – 181, 194 – 195 memory colour effects 79, 85n4 memory explanation 76 memory interpretation 85n2 mental architecture: arguments from reliability of perception 61–70; arguments from stability of perception 55–61, 70; faculty psychology 47–51; Fodor’s modularity theory 49–70, 87–88, 104; informational encapsulation 52–53; mental modules 49–54; mind/body dualism 47; model 18; perceptual expertise in theories 166–169 mental faculties 47 – 48 mental modules: domain specificity of 49 – 51, 53; features of 50; mandatoriness and speed of 51 – 52
mental representations 30 – 31, 33 – 35, 60, 71n5, 160, 249 mental states 15 – 16, 30 – 32 Mervis, C.B. 147 Meyer, A.S. 82, 126 – 127, 127, 140n5 Millikan, R.G. 203n4, 204n5 mind/body dualism 47 Modularity of Mind (Fodor) 48 – 49, 51 modularity of mind: arguments from reliability of perception 61 – 70, 239; arguments from stability of perception 55 – 61, 70, 239; characteristics 49 – 54; domain specificity of modules 51; informational encapsulation 50, 52 – 54, 62 – 63, 70, 104, 134, 148 – 149, 174, 239; mandatoriness and speed of modules 51 – 52; perceptual systems as modular 87 – 88, 104, 135, 173, 240 Mole, C. 122, 132, 134, 139, 140n9, 140n11, 199 Montague, M. 42n18 Montmarquet, J.A. 203n3 “Mooney” image 1 68 “Mooney” image 2 69 Moores, e. 82, 125, 127, 137, 140n5 moral properties 38 Moran, J. 124, 127 Morgan, M. 190 Morrison, J 228 Müller Lyer illusion 5, 19, 55 – 58, 56, 144, 210 Mumford, D. 41n10 Nagel, T. 28, 213 Nanay, B. 42n13, 43n26, 139, 203, 237n21
307
Copyright © 2021. Taylor & Francis Group. All rights reserved.
308
INDex
nativism 48, 49 natural kinds 38, 183, 231, 233, 236 – 237n20 natural norms 189, 191, 194, 201 Neander, K. 203 Necker Cube 115 Neisser, U. 120 neural reuse theory 71 neuroplasticity 202, 205n12 New Look Psychology: accentuation 140n6; criticism of 22; early study of 91, 93, 107; faculty psychology and 48 – 49; interpretations of 84, 240, 244n3; later proponents of 41n6; on distinguishing perception and cognition 21 – 23, 41n7, 63; predictive coding and 41n9; topdown effects on perception 75 – 77 Newman, L. 41n12 Nicomachean Ethics (Aristotle) 203 Nishimura, M. 148 Nobre, A.C. 126 Nodine, C.F. 159, 165 Noë, A. 42n13 non-conceptual perceptual content 35 – 36 non-doxastic states 15, 26, 28, 54, 68, 69, 108, 111, 150, 217, 221 – 222, 225 non-epistemic seeing 27, 34 non-inferential psychology 162 non-interference claim 60 – 61 non-spatial selective attention 82 non-transfer 165 normative ethics 145 – 146 object-based attention (OBA) 120 – 121, 131, 136, 140n3, 198 objectification 225
Objectivity* 213 – 214, 216 – 217, 222, 224 – 227, 234 objectivity 101, 144, 173, 207, 212 – 216, 223 – 231, 236n19 occipital face area (OFA) 146, 159, 161, 202 olfaction 16, 37, 39 Olkkonen, M. 60, 79 online matching task 76, 79 O’Regan, J.K. 42n13 Origlia, N. 205n11 Orlandi, N. 71n5 overt attention 132 Palmer, S.e. 60 Paradiso, M.A. 73n16 part–whole task 155, 155 Pascual-Leone, A. 202 Passmore, J. 212, 236n7 pattern recognition 171n17, 175, 191 – 194, 196 – 198, 200, 206, 229 – 234, 242 – 243, 250 – 255 Pautz, A. 40 Pavese, C. 169 Payne, B.K. 80, 218 perception: action and 80 – 81; “active” theories of 41 – 42n13, 114n7; activity of 23 – 29, 114n7; admissible content for perceptual experience 36 – 39; behavioural role of 17 – 18, 99, 101 – 103, 105 – 106, 108, 229 – 230; as categorical 21 – 22; characteristics 40n3; cognitive influence on 16 – 18, 33, 44 – 45, 87 – 88, 134 – 135; colour perception 79 – 80, 85n4; conceptual requirements 26 – 27, 34 – 36; diachronic cognitive effect
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INDex
on 57, 94 – 99, 109 – 111; as directed at objects 20; as directed at world 20; direct realist theories of 42n19; doxastic influences on 78 – 79; effect of cognitive penetration 131 – 135; empiricist view 42n14; epistemic role of 16, 32, 57, 87, 99 – 100, 105 – 106, 108, 137; format of 34; as inferential 21 – 22; intentions and 80 – 81; linguistic knowledge effects 137 – 138; modes for distinguishing cognition from 23 – 29, 40; as modular 69; naïve/direct realist theories 42n19; naïve theories of 42n19; as objective sensory representation 12 – 13; perception-is-belief view 18 – 21; phenomenology 28 – 29; presence of appropriate stimuli 25 – 26, 40; properties 21, 37 – 39; Rationality of Perception thesis 177 – 178; reductive and revisionary theories 18 – 23; reliability of 61 – 70; scientific-theoretical role of 99 – 100, 104, 210 – 216; sensory organ activity 25, 40; social categories and 79 – 80; stability of 55 – 61; stereotypes and 79 – 80; subject activity 23 – 25, 29, 40; success conditions 31, 225 – 231; synchronic cognitive effect on 57, 58; thought as indistinguishable from 21; traditional view 23 – 24, 173; value-influenced 75 – 78, 91 – 94 perceptual adaptation 97 perceptual confidence 228 perceptual constancies see constancy mechanisms
perceptual content 26 – 27, 32 – 36, 39, 40, 92, 102, 109, 151, 166, 223 – 231, 233 – 234, 242 – 243, 246 – 253 perceptual convergence 37 perceptual error 145 perceptual experience: accuracy of 224 – 226, 228 – 230; admissible contents of 36 – 39, 225, 231; effect of selective attention mechanisms on 121, 128, 130 – 133, 138 – 139, 198, 239; attention-shift explanation 119; behavioural roles of 135; belief and 18, 20, 36, 88 – 90, 227; cognitive learning and 144; cognitive penetration of 94 – 97, 150, 163; concepts for 34 – 36; as conscious sensory representation 13, 31; demarcating from other mental phenomena: 84; in empiricism 42n14; epistemic role of 32 – 33, 100; higher-level effects on 104; as inferential 21 – 22; notion of self and 255; perceptual expertise and 151; phenomenal 28, 83, 93, 96; and possible cases of cognitive influence on 5 – 7, 74 – 86; prediction error minimization process as 22 – 23; racial biases resident in 218 – 219; as static snapshots 249; theoryladen observation 211; top-down feedback and 68; understanding and 253 perceptual expertise: admissible contents of experience and 231 – 234; architectural claims 149, 162 – 163, 165 – 166, 175, 210, 241 – 242; characteristics 149 – 150;
309
Copyright © 2021. Taylor & Francis Group. All rights reserved.
310
INDex
domain-specific expertise and 95, 137, 146, 152, 164, 176, 182, 192 – 193, 195, 197, 199, 207 – 209, 233 – 234, 252 – 253; epistemic virtue and 179 – 182; epistemology of 175 – 179; generality of 207 – 210; as genuine cognitively sensitive expertise 162 – 165, 241; as genuinely perceptual phenomenon 151 – 162, 241, 243; objections 194 – 202; qualifications 194 – 202; situating in theories of the architecture of mind 166 – 169; top-down or background cognitive states in 67 perceptual experts 148 – 151, 209, 231 perceptual improvement 189 – 194, 197, 199 – 200, 208, 243, 253 perceptual learning 97 – 98, 109 – 110, 150, 162, 165, 189 – 192, 196, 219 perceptual misrepresentation 145 perceptual modules 49 – 54, 55 – 70, 104, 108, 142, 167, 168 perceptual processing: as cognitively impenetrable 53 – 54, 101, 111, 163; cognitive penetration of 113, 136, 168; inaccessibility claim 59 – 60; influence of beliefs 56; influence of intentions 81; influence of mental phenomena 63 – 65; influence of non-perceptual processes 24, 83; influence of perceptual experience 104, 132 – 133; as informationally encapsulated 135 – 136, 166; New Look Psychology 22 – 23, 42n13; perceptual expertise 148 perceptual psychology 138, 145, 150 perceptual recognition 70, 146
perceptual representation 174, 203 – 204n4, 223 – 225 perceptual understanding 252 – 254 perceptual skill 114n4, 165, 169, 181 – 182, 203, 206, 209, 216 – 217, 223 – 224, 229, 234, 242 – 243, 253 perceptual success 87, 223, 228 – 229, 230, 242 pernicious cognitive effects assumption 143, 144, 174, 224 persistent illusions 19, 56 – 59, 63, 70, 72n13, 144 Peterson, M.A. 190 phenomenology 13 – 14, 15, 20 – 21, 28 – 29, 38, 42n18, 129 – 130, 161 – 162 Phillips, B. 40 Pinker, S. 72n12 Pitcher, G. 19 Platchias, D. 203n1 Plato 250 pluralistic realism 229 Poggio, T. 172n18 Pollock, J. 5 pop-out selection 170n10 Popper, K. 211, 213 post-perceptual explanation 150 – 151 prediction error 22 – 23, 41n9 prediction machine 22 – 23 predictive coding/processing 22 – 23, 41n9, 56 – 58, 73n20 pre-perceptual explanation 151 – 152 Prinz, J. 57, 60, 61, 132 – 133 process reliabilism 178 – 179 Proffitt, D. 42n13, 80 – 81, 85n3, 114n7 propositional knowledge 229, 249, 251 proprioception 16, 39 Ptolemy 210
Copyright © 2021. Taylor & Francis Group. All rights reserved.
INDex
Purves, D. 58 Pylyshyn, Z. 41n7, 70, 71n8, 101, 244n3; attention as gatekeeper assumption 167 – 168; attention as spotlight assumption 133; attention-shift explanation 115 – 116, 117, 119, 131, 133, 148; cases of diachronic changes in perception 94 – 99; cases of value-influenced perception 92 – 94; on “cognitive impenetrability” of perception 64, 204n8, 221; on cognitive penetration 89 – 90, 111; definition of cognitive penetrability of perception 112; perception and thought as indistinguishable 21; on perceptual expertise 149, 167 – 168, 170n5, 170n9; on reliability of perception 68
reflex 53, 55 Reid, T. 254, 254 – 255 reliabilism 178 – 181 representational error 186 representational function 183, 186 – 189, 191, 193 – 194, 199 – 200, 206, 242 Richler, J. 154 rich perceptual content 36 – 39, 41n8, 109, 233, 243 Riggs, W. 250, 251 Robbins, P. 60, 71n6 Rock, I. 41n13 Rock, Irving 41n6 Rosch, e. 146 – 147 Rossion, B. 155, 159 Rothko, M. 5 Rubin goblet 128, 129 Rule, N.O. 221
Quilty-Dunn, J. 33, 40, 134, 139 Quine, W.V.O 211, 235n6
saccadic eye movement patterns 82, 126 – 128, 131, 137, 140n4, 159, 167, 198, 202 same-race faces (SR) 219 – 220 Sartre, J-P. 129 – 130, 254 satisfaction conditions 31 Saul, J. 236n10 scepticism 48 Schendan, H.e. 60 Schneider, W. 190 Scholl, B.J. 85, 85n5, 116, 117, 121, 139, 167 Schwartz, S. 205n11 scientific investigation 17 – 18, 54, 99 – 105, 210 – 216, 226, 235 Scott, L.S. 147, 164, 235n1 Seeley, W.P. 237n21 Segall, M.H. 72n14
racial bias 218 – 219 radiologist 67 – 68, 95, 100, 137, 146, 152, 159, 164 – 165, 176, 182, 193, 195, 197, 199, 208, 233 – 234, 242 – 243, 251 – 252 Raftopoulos, A. 70, 85, 90, 114n6, 131, 140n10 Ransom, M. 41n10 rapid serial visual presentation (RSVP) 156 – 157, 160 Rationality of Perception thesis 177 – 178 real-world expertise 147 – 148 reasoning 15, 16, 28, 39, 47–48, 59, 146 Rees, G. 202
311
Copyright © 2021. Taylor & Francis Group. All rights reserved.
312
Index
Sekuler, A.B. 190 selective attention 81, 82, 110, 120 – 131, 134 – 136, 166, 168, 190, 198, 239 self-consciousness 254 – 255 self-understanding 254 Sellarsian dilemma 33, 36, 39 Sellars, W. 32 – 33, 40 sensation 47 – 48, 102 – 103 sensory organ activity 25, 29, 40 sensory perception 4, 12, 15, 18, 23 – 24, 26, 37, 150, 251 Sergent, J. 146 “Shannon number” 17 Shea, n. 40 Sherrington, C.S. 202 Shiffrin, R. 190 Shriver, e.R. 220 Sibley, F. 231 – 234 Siegel, S. 38, 40, 43n23, 43n25, 85, 90, 100, 112, 137, 177 – 178, 235, 236n12 Siewert, C. 43n23, 237n20 Silins, n. 84, 112 Simons, d.J. 41n13 Skinner, B.F. 240 Sloan Institute 50 – 51 Smeeton, n.J. 172n17 Smith, A.d. 20 – 21, 203n1 Snowdon, P.F. 42n19 social categories and 79 – 80 social psychology 80, 218 Sosa, e. 179 – 180, 192, 203, 203n3 Sowden, P. 204n8 sparse perceptual content 37 – 39, 95, 151 spatial attention: attentionshift explanation 76 – 77, 88,
97 – 98, 117 – 119, 125, 127, 131, 139n1, 148 – 149, 152, 161, 167, 197 – 198, 238 – 239; linguistic knowledge effects 82; selection mechanisms as nuanced forms 133 – 135; spotlight of attention assumption 110, 116, 119 – 121 spatial change vs. feature change task 156 spatial constancy mechanisms 81 spatial occlusion task 191 Sperandio, I. 73n16 Sperber, d. 70, 72n12 spotlight of attention assumption 116, 118, 120 – 121, 128 Stampe d. 204 Stefanucci, J.K. 81, 85n3 Stephens, C.L. 204n5 Sterelny, K., 204n5 stereotypes 79 – 80, 218, 234, 235 Stern, C.e. 60 Steup, M. 203 stimuli 25 – 26, 29, 40, 65 – 67, 70, 77 – 78, 122 Stocker, M. 72n10 Stokes, d. 40, 40n1, 41n10, 43n26, 63, 69, 84, 85n2, 85n4, 85n5, 89, 90, 112, 117, 131, 137, 139, 169, 203, 233, 237n20 Storbeck, J. 85n3 Stratton, G.M. 97 subject activity 23 – 25, 29, 40 synchronic cognitive effect 57 – 58 Tanaka, S. 73n16, 159, 163, 235n1 TaP thesis 139, 144, 169, 238 – 239
INDex
Copyright © 2021. Taylor & Francis Group. All rights reserved.
Tarr, M.J. 146, 154, 171n12 taste 16, 39, 234, 236n19 Taylor, J.G. 97 Telling, A.L. 126 Thatcher illusion 155 theory-ladenness 90, 106, 100, 112, 114n6, 114n9, 204n9, 205n12, 210 – 222, 235n5 theory-laden observation 211 Thompson, e. 254 Thompson, P. 155 thought see cognition TiP thesis 144, 169, 221, 223, 231, 239, 241 – 242, 253, 255 Tooby, J. 70 top-down influence on perception 5 – 7, 55, 64 – 67, 74 – 85, 101, 124 touch 16, 39 Treatise of Human Nature, A (Hume) 47 Treisman, A.M. 120, 139 Tucker, C. 112 Turri, M. 203 Tycho Brahe 210, 211 underdetermination 54, 211, 217, 245, 253 understanding 11, 17, 42n14, 67, 100, 176, 215, 246, 250 – 256 Uttal, W.R. 71n3 Valenti, J.J. 85n4 value-influenced perception 75 – 78 Vance, J. 41n10, 56 – 57, 112, 137 Vanderkolk, J.R. 147, 155, 171n12 van Fraassen, B.C. 236n8 van Gulick, R. 43n23
van Ulzen, N. 77, 91 Varela, F.J. 42n13 veridicality 58, 62, 93, 108, 144, 186, 188 – 189, 194, 203 – 204n4 vertical faculties 49 virtue reliabilism 180 – 181 virtue responsibilism 180 – 181 virtue theory 179 – 182, 192 – 193, 195 – 197, 203, 217 vision: admissible content of experience and 36 – 39; informational contact and 4; as input system 53 – 54; linguistic knowledge effects 81 – 82; in linguistic practice 12; as modular 51, 58 – 59, 70; as perceptual process 15 – 16, 24, 29, 39; reliability of 61 – 70; sensory organ activity 25; stability of 55 – 61; top-down effects 81 – 82 visual attention 67, 81 – 84, 122 – 123, 136, 139, 163, 170n10 visual experience 32, 35, 42n19 visual illusions 55 – 61 visual perception 5, 15, 17, 25, 26, 75, 96, 117, 161, 163, 196, 233 visual search task 122 visual short-term memory (VSTM) 157 – 158, 158, 160 visual systems 185 – 186 Vogt, S. 159 Ward, P. 204n7 Watzl, S. 128, 139 Where’s Waldo puzzle, 95 – 98, 121 Williams, A.M. 171n17, 193, 204n7 Williams, B. 72
313
314
INDex
Copyright © 2021. Taylor & Francis Group. All rights reserved.
wishful seeing 222 Witt, J.K. 81 Witzel, C. 79 Wong, A.C.-N. 154 Wright, L. 185, 186, 203 Wu, W. 63, 72n11, 81, 94, 112, 128, 134, 136, 139
Young, S.G. 235 Yovel, G. 170n2 Zagzebski, L. 180, 182, 203, 203n3, 250–251 Zangwill, N. 236n18 Zawidski, T. 71n1 Zeimbekis, J. 70, 85, 85n4, 131