EDITED BY MATTEO COLOMBO ELIZABETH IRVINE and MOG STAPLETON
Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries.
Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above.
You must not circulate this work in any other form and you must impose this same condition on any acquirer.
CIP data is on file at the Library of Congress
ISBN 978–0–19–066281–3
9 8 7 6 5 4 3 2 1
Printed by Sheridan Books, Inc., United States of America
CONTENTS
Foreword ix
Daniel C. Dennett
Acknowledgements xi
List of Contributors xiii
Introduction 1
Matteo Colombo, Liz Irvine, and Mog Stapleton
PART 1
EXTENSIONS AND ALTERATIONS
1. Extended Cognition and Extended Consciousness 9
David J. Chalmers
2. The Elusive Extended Mind: Extended Information Processing Doesn’t Equal Extended Mind 21
Fred Adams
3. Clark on Language, Cognition, and Extended Cognition 32
Kenneth Aizawa
4. Extended Mental Features 44
Katalin Farkas
5. Extended Affectivity, Reconsidered 56
Michelle Maiese
6. Matters of the Flesh: The Role(s) of Body in Cognition 69
Lawrence A. Shapiro
7. Breaking the Waves: Beyond Parity and Complementarity in the Arguments for Extended Cognition 81
Michael Wheeler
PART 2 ON BEING A CYBORG
8. Supercharged Apes versus Super- Sized Minds: Embracing Continuity While Accepting Difference 99
Louise Barrett
9. Building Inner Tools 113
Robert L. Goldstone
10. When Is a Mind Extended? 128
David Kirsh
11. The Archaeology of the Extended Mind 143
Kim Sterelny
12. The World Well Gained: On the Epistemic Implications of Ecological Information 161
Michael Anderson and Anthony Chemero
13. Beyond the Desert Landscape 174
Karl J. Friston
14. Quick’n’Lean or Slow and Rich? Andy Clark on Predictive Processing and Embodied Cognition 191
Jakob Hohwy
15. How Radical Is Predictive Processing? 206
Nico Orlandi and Geoff Lee
16. Ways of Mindmaking 222
Jesse Prinz
17. Being a Beast Machine: The Origins of Selfhood in Control-Oriented Interoceptive Inference 238
Anil K.
Seth
18. The Minds of Insects
Barbara Webb
254
Replies to Critics: In Search of the Embodied, Extended, Enactive, Predictive (EEE-P) Mind 266
Andy Clark
Index 303
FOREWORD
The ideal of “all things considered” is a perpetual motion machine for epistemology. You can’t think effectively about any hard topic without relying on unexamined constraints, self-imposed barricades on your imagination that permit you to dismiss without a hearing most of the myriad candidate solutions to whatever problem you are concentrating on. The need for brusque heuristic pruning of the search tree is a fact of life, not just a fact of chess. My colleague Marcel Kinsbourne has proposed that what makes any problem hard is always the fact that something attractive, and false, stands in the way, securing allegiance that then poisons the investigation. Philosophers, at their best, specialize in “opening our eyes to new perspectives,” helping us overcome our subliminal aversions and loosen our grip on “home truths” that are so familiar we never stop to consider them. This delicate task calls for an artful mingling of arresting observations, vivid language, and a deep understanding of the theoretical work that has created the arena in which current disputes play themselves out.
Andy Clark is a peerless perspective-shifter, and the fruits of his labors are magnificently on display in this volume. There is a palpable sense of intellectual community and progress on the tough issues, a sense that if we put our heads together, we can discern the contributions of a wide variety of apparently warring positions, both philosophical and scientific, and weave them into a cable of mutual agreement, defeasible of course, but a new path of common ground on which to pursue further research. The central cord of Andy’s cable, still only reluctantly endorsed or even resisted by some of the contributors, is (in my hardly disinterested opinion) the working assumption that there is nothing like élan vital, or wonder tissue, or intrinsic intentionality that distinguishes the mind from the rest of nature. One manifestation of this assumption is the Parity Principle (and the Reverse Parity Principle proposed by Goldstone), and another is the fruitful pursuit by all participants of continuities between all evolved organisms and indeed all designed tools for thinking, from writing and prostheses to Scrabble tiles and computers.
We totter, as usual, on the shoulders of giants: Descartes, Bayes, Skinner, Gibson, Fodor, and others, wielding the multitude of isms provoked or inspired by them, and one thing that strikes me about the discussions here is how often they expose a pendulum swinging between overstatement and oversimplification. You pay a price for the vividness without which you cannot hold the attention of your students (or colleagues): the Heartbreak of Premature Definition. What is your definition of functionalism, cognitivism, enactivism, embodied cognition, representation, affordance? There are good reasons to postpone definition until after you’ve elaborated some of the treasures, prospects, and risks of your position, but if you decline to define your terms, your critics will be obliged to define them for you, and the result can degenerate into a food fight of counterexamples and reductios. Is functionalism really just behaviorism unleashed by Skinner’s 1964 observation that “the skin is not that important as a boundary”? What does cognitivism add to behaviorism, if not a dread homunculus? Once you’ve laundered all the intellectualist connotations out of the concept of representation, dismantling the inner user, how do representations differ from resonant loops? Concentrating myopically on devising variant definitions, chisholming away in a fugue state of defensive strategizing, is a well-known philosophical foible, but it is not much in evidence here! Andy Clark sets a fine example, followed by his critics, of really trying to educe the best insights from the opposition, and both the essays and Andy’s gracious response to them provide models of philosophical behavior that should inspire and instruct all who enter these arenas.
This is not your grandfather’s philosophy of mind, and the contributors are the all-stars of the twenty-first century. Among the many volumes of essays exploring the ideas of one philosopher or another, this one has a rare virtue: as you read the essays you keep learning new things, not just novel arguments, novel objections, novel critiques, but facts about the world outside philosophy that philosophers ought to know. Who would have thought that the ingenious devices of insect mating competition, or the ethnography of ritual memory boards among the Luma, or the architecture of Differential Neural Computers, or the function of postural sway, or the effects of left-handedness would illuminate any philosophical controversy? As one who gets impatient with philosophical writing that does not inform me about anything beyond the cleverness of the author, I find rewards aplenty in these pages, and I cannot think of a book that better exposes the limitations of traditional, factually impoverished philosophical combat. And like Clark’s own work, it is all eminently readable by nonspecialists. Too many of our colleagues in philosophy have been subliminally taught that philosophy that isn’t hard to read is not worth reading. This cheerful, and cheer-inducing, book is a fine counterexample.
Daniel C. Dennett
ACKNOWLEDGEMENTS
We would like to thank a number of people for helping us make this volume possible: Abigail Johnson, Judith Hoover, and Lincy Priya for copy-editing and helping us to prepare the volume for final submission; Peter Ohlin for providing us with encouragement and guidance during the project; and, most importantly of all, the authors for contributing thoughtful, bold, and valuable chapters. There are many other researchers from disparate disciplines doing excellent work on Clarkian topics, but far too many to fit in one volume!
We, of course, must acknowledge and thank Andy himself, not only for contributing to, and supporting us in creating this volume, but for contributing to and supporting our academic development. Each of us has taken a different academic path through philosophy of cognitive science, but each has been graciously supported by Andy, and inspired by his spirit of curious openness to new philosophical and scientific developments.
This volume has been in the making since late 2015. Over this time we have had the good fortune to be academically supported by a number of different institutions and funding bodies, which made it possible for us to produce the final book.
Matteo gratefully acknowledges financial support from the Deutsche Forschungsgemeinschaft (DFG) within the priority program “New Frameworks of Rationality” ([SPP 1516]), and from the Alexander von Humboldt Foundation.
Liz has been working as a lecturer at Cardiff University throughout the preparation of the book.
Mog was supported over this period by postdoctoral positions in Macau on Nevia Dolcini’s project, Tübingen in Hong Yu Wong’s research group, and Edinburgh on Duncan Pritchard’s project. She is very grateful for the financial and academic support over this period.
Matteo, Liz, and Mog November 2018
CONTRIBUTORS
Fred Adams, Department of Linguistics & Cognitive Science, University of Delaware
Kenneth Aizawa, Department of Philosophy, Rutgers University, Newark
Michael Anderson, Department of Philosophy, Western University, Ontario
Louise Barrett, Department of Psychology, University of Lethbridge
David J. Chalmers, Department of Philosophy, New York University
Anthony Chemero, Departments of Philosophy and Psychology, University of Cincinnati
Matteo Colombo received his PhD from the University of Edinburgh in 2012, supervised by Andy Clark and Peggy Seriès. He is now an associate professor at the Tilburg Center for Logic, Ethics, and Philosophy of Science, and at the Department of Philosophy, Tilburg University. He’s interested in questions about evidence and explanation in the computational, cognitive, and brain sciences, and more generally in how the scientific and manifest images of mind relate to one another.
Katalin Farkas, Department of Philosophy, Central European University
Karl J. Friston, Institute of Neurology, University College London
Robert L. Goldstone, Department of Psychological and Brain Science, Indiana University
Jakob Hohwy, Department of Philosophy, Monash University
Elizabeth Irvine received her PhD from the University of Edinburgh in 2011, supervised by Andy Clark, and held postdoctoral research positions at the University of Tübingen and the Australian National University before becoming a lecturer at
Cardiff University. Her research interests are primarily in philosophy of cognitive science and psychology and philosophy of science.
David Kirsh, Department of Cognitive Science, University of California, San Diego
Geoff Lee, Department of Philosophy, University of California, Berkeley
Michelle Maiese, Department of Philosophy, Emmanuel College
Nico Orlandi, Department of Philosophy, University of California, Santa Cruz
Jesse Prinz, The Graduate Center, City University of New York
Anil K. Seth, Sackler Centre for Consciousness Science, University of Sussex
Lawrence A. Shapiro, Department of Philosophy, University of Wisconsin, Madison
Mog Stapleton received her PhD from the University of Edinburgh in 2012, supervised by Andy Clark. She held postdoctoral positions at the University of Stuttgart, the University of Macau, and the University of Tübingen before returning to the University of Edinburgh in 2017. Her research interests are primarily in embodied, affective, and enactive approaches to cognition.
Kim Sterelny, School of Philosophy, Australian National University
Barbara Webb, School of Informatics, University of Edinburgh
Michael Wheeler, Division of Law and Philosophy, University of Stirling
Andy Clark and His Critics
Introduction
Matteo Colombo, Liz Irvine, and Mog Stapleton
Where is your mind? According to traditional wisdom in philosophy of mind and cognitive science, the machinery of your mind is just your brain. In Jerry Fodor’s words: “If the mind happens in space at all, it happens somewhere north of the neck” (1999, 69).
While you are reading, your brain is buzzing with neural activity. Some patterns of neural activity support your eyes rapidly moving and then shortly fixating on the symbols in front of you. Other patterns of activity enable you to perceive, decode, and understand what you’re reading. Knock down your frontal eye fields in your frontal cortex, and your eye movements will be impaired. Likewise, knock down your visual cortex, and your visual perception of what’s in front of you will be seriously damaged. All this might suggest that your mind is brain-bound.
According to traditional wisdom, there is another, more fundamental reason why the physical basis of human minds cannot outrun the bounds of skin and skull. If we ask you, “What are you doing now?”, you may reply along the following lines: “I want to read the introductory chapter of the volume on Andy Clark’s philosophy, and I have reason to believe that moving my eyes and body in certain ways, and perceiving and deriving meaning from the symbols in front of me, is one way to get the job done.” You would thus offer us an explanation of your behaviour by citing the beliefs and desires you entertain.
Importantly, the mental states cited in this explanation have a unique property. They are about things: they possess intentionality and have “content” (making them “semantically evaluable”). The sentences on this page have content too, but their content derives from our mental states. Sentences, and indeed any other piece of the external world, are often said to have “derived intentionality.” They would not have any meaning unless it was conferred on them by people who use those sentences to express their thoughts and communicate with others. So, if external items do not mean anything on their own, and the mark of the mental is intentionality, then the mental cannot overflow the boundaries of skin and skull. And citing bits of the environment in explaining intelligent behaviour will not play the unique role played by intrinsically meaningful, content-having, causally efficacious mental states.
In sum, according to traditional wisdom, the physical substrate of the mind cannot reside outside of our heads, and explanations of intelligent behaviour must always look for content-having, causally efficacious states within the boundaries of skin and skull.
Over the course of his career, Andy Clark has systematically challenged both of these tenets in the philosophy of mind and cognitive science. Clark’s challenge has primarily come on two fronts. On the metaphysical front, he has tried to show how a form of functionalism opens the door to the possibility that the vehicles of thought can overflow the causal transactions that take place within our skulls. This form of functionalism can be captured in a “Parity Principle,” which Clark and Chalmers (1998, 8) formulated as follows: “If, as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is . . . part of the cognitive process.”
Here is the basic idea: if what makes something a certain type of cognitive state or process depends not on its internal constitution or on its physical instantiation, but rather on the coarse functional role it plays in a system, then it is a prejudice to maintain that the mind must be instantiated within the skull. When external devices, systems, and other structures in the world function sufficiently like things or processes people would normally regard as cognitive, were they to occur inside the skull, then they too could figure as proper parts of our cognitive system (Clark and Chalmers 1998; Clark 1997, 2008).
This proposal has generated a productive controversy in the metaphysics of mind, under the banner of the “Extended Mind” debate, and in the philosophy of mind and cognitive science more generally, under the banner of “Extended Cognition.” In the first section of this volume David Chalmers, Fred Adams, Katalin Farkas, and Mike Wheeler articulate cutting-edge criticisms of the arguments presented in these debates, and make original proposals for the future direction of this research. Both Chalmers and Farkas focus on how exactly to specify the extended mind thesis, including how to interpret some of its early examples, and press on the possibility of consciousness extending. Adams argues that Clark’s failure to specify what “cognition” is undermines his argument that cognition extends; and Wheeler pushes this style of analysis further in terms of the debates between the first and second “waves” of the extended mind debate.
Clark’s arguments have also extended into a wider program investigating “Embodied, Embedded, Enactive and Affective” cognition, in which he has argued that details of the bodies of agents, as well as worldly resources, make essential contributions to explanations of a great many cases of intelligent behaviour (Clark 1997, 2004, 2008). The remaining chapters in the first section of this volume expand on this theme. Ken Aizawa pursues Clark’s claims about the role of language in metacognition and related implications of extended cognition. Larry Shapiro defends the view (against Clark’s “larger mechanism” framework) that the body
makes a “special contribution” to cognition. Michelle Maiese argues that affectivity is embodied and embedded, but not extended.
The hypotheses of extended, embodied, and embedded cognition have been taken up further in the informatics disciplines and in AI, in terms of Clark’s proposal that humans are “natural-born cyborgs.” This suggests that we are beings who routinely incorporate parts of the world into our cognition, using these not only to extend our cognitive capacities, but also to realise many novel cognitive capacities. The chapters in part 2 explore this theme.
Louise Barrett uses the theme of evolutionary (dis)continuity between humans and other animals to challenge the role of representationalism in Clark’s work, and to question whether reliance on artefacts increases or decreases cognitive load. Rob Goldstone turns the Parity Principle around, and proposes that we can “hack” our own perceptual systems to provide new mental tools. David Kirsh develops a novel way to distinguish cognitive extension from cognitive embedding, and suggests that real cognitive extension is only ever brief and temporary. Kim Sterelny charts a methodologically cautious history and evolution of human reliance on materially scaffolded cognition. This concludes discussion of the first front of Clark’s arguments against the received view about cognition and the nature of the mind.
On the second front, Clark has argued against classicist ideas about cognition. The classicist conception of cognition is one in which the mind functions like a digital computer, manipulating symbols according to a set of rules, where the symbols consist of concepts similar to the lexical items found in natural languages. However, Clark has argued that the functional profile of a system need not feature symbolic items that track natural language concepts, and so need not be “semantically transparent.” This means that the representational items that feature in a system’s processes need not relate in a systematic way to features of the world that can be picked out propositionally, with the expressive resources of public language (Clark 1989, 1993). Even systems without this semantic transparency can be justifiably taken to possess genuinely intentional states that play an important, though not exclusive, explanatory role for cognitive activities (Clark 1997, 2013).
Explanations of intelligent behaviour that rely on connectionist neural networks, dynamical system modelling, Bayesian modelling and predictive coding illuminate this explanatory shift. In these alternative approaches, contentful mental states like thoughts, beliefs and desires can be characterised as distributed sets of biases and structures in a system that is poised to pick up statistical structure in the data it encounters when it interacts with the environment. In this case, mental states need not consist of contentful symbols, but typically consist of fluid, distributed, probabilistic, and superpositional structures acquired at different time scales, which are far from semantically transparent (i.e., it’s not easy to read off their meaning and how they are related to each other). And yet this opacity should not suggest that the system does not possess genuinely intentional states or internal representations, or that it cannot support the productivity and systematicity of thought.
All this may naturally suggest that we should be looking at a diverse and disunified array of explanatory resources in order to understand how minds emerge from the complex cooperation of brain processes with bodily form, action, and the canny use of environmental structures. While this conclusion seems to follow naturally from Clark’s work, Clark (2013, 2016) has in fact resisted it in his most recent project. In this, he argues that the embodied mind’s rag bag of tricks and stratagems may be unified through a few core principles grounded in the view that the brain is a multi-layered probabilistic prediction machine. Instead of being passive, feedforward accumulators of environmental features, brains are active predictors of environmental signals. Hierarchically organized networks of neurons encode hierarchically organized statistical models of the environment, which they employ to make predictions about their next sensory state. As a function of the way in which observed sensory input proves these predictions to be wrong, the neurally encoded statistical models get updated and redeployed to make fresh predictions that organisms use to aptly navigate their environment. In this continuous cycle of prediction-error-based updating guided by action-perception loops, embodied brains become better and better at predicting environmental structures that matter for their own survival and flourishing.
In this exciting picture of brain function, action, perception, attention, and consciousness are painted as continuously co-constructed around the same fundamental computational routine: prediction-error minimization. Thus, our embodied, embedded, extended, and spatially and temporally distributed cognition is to be grounded in an orchestrated attempt to individually and collectively minimize the error in our predictions about specific sensorimotor trajectories in our local environments.
However, it is early days for predicting whether the predictive processing account will deliver a genuinely unified science of the embodied mind. This issue is taken up in part 3 of the volume. A number of authors in this section question the fit between predictive processing and Clark’s earlier work on extended, enactive, and embodied cognition. Mike Anderson and Tony Chemero offer a different reading of how the predictive processing framework might fit with radical embodiment, which highlights the role of ecological information and downplays the concepts of prediction, models, and representation. Somewhat differently again, Karl Friston recommends an enactivist and embodied version of predictive processing in which agents need not always have (representational) models of their environments but simply are models of their environments in virtue of existing. Jakob Hohwy argues that the representational demands of predictive processing are not consistent with the kind of mind-world relationship that Clark is committed to. Nico Orlandi and Geoff Lee examine Clark’s interpretation of the predictive processing framework and argue that it preserves traditional (and non-Clarkian) distinctions between perception and action. Jesse Prinz offers a critical overview of Clark’s work and raises the question of whether it is possible for any single account of cognition to be explanatorily adequate. Anil Seth uses
the resources of predictive processing applied to interoception to explain the phenomenology of embodiment, selfhood, and subjectivity. Barbara Webb uses test cases of insect cognition to analyse whether predictive processing does in fact capture how simpler cognitive organizations such as those of insects function.
In the final section of the volume, Andy Clark offers responses to all the critics (and friends) who have contributed to this volume. His hope is to achieve, as he puts it, “a single, not wildly inconsistent, narrative.”
This volume, like the work it engages with, is incredibly broad in scope and will serve to showcase and encapsulate Clark’s imaginative explorations of mind and its place in nature. Andy has been a sparkly inspiration to us. Smart and curious, he has taught us how to think big and take chances, explore new intellectual territories, and face up to questions that have the potential to impact human life, and all with grace and humility. We hope that this volume will likewise inspire future generations of researchers in the sciences of mind.
I feel released
Bad times deceased
My confidence has increased Reality is here
The game has been disbanded My mind has been expanded Richard O’Brien, “Rose Tint My World,” The Rocky Horror Picture Show (1975)
References
Clark, A. 1989. Microcognition: Philosophy, Cognitive Science and Parallel Distributed Processing MIT Press.
Clark, A. 1993. Associative Engines: Connectionism, Concepts and Representational Change. MIT Press.
Clark, A. 1997. Being There: Putting Brain, Body and World Together Again. MIT Press.
Clark, A. 2004. Natural-Born Cyborgs: Minds, Technologies, and the Future of Human Intelligence. Oxford University Press.
Clark, A. 2008. Supersizing the Mind: Embodiment, Action, and Cognitive Extension. Oxford University Press.
Clark, A. 2013. “Whatever Next? Predictive Brains, Situated Agents, and the Future of Cognitive Science.” Behavioural and Brain Science 36:181–204.
Clark, A. 2014. Mindware: An Introduction to the Philosophy of Cognitive Science. 2nd ed. Oxford University Press.
Clark, A. 2016. Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press.
Clark, A., and D. Chalmers. 1998. “The Extended Mind.” Analysis 58:7–19.
PART 1 EXTENSIONS AND ALTERATIONS
Extended Cognition and Extended Consciousness
David J. Chalmers
It’s a pleasure to contribute to this volume honoring Andy Clark.1 Andy has been an influential figure for me ever since I read his first book, Microcognition, while I was in graduate school. A few years later I took up my first job as a postdoctoral fellow working with him in the newly initiated Philosophy-Neuroscience-Psychology Program at Washington University in St. Louis. We had a lot of adventures, but the one that seems to have left the most traces is our joint paper, “The Extended Mind,” which argued that cognitive processes can extend outside the brain and the body to include objects that we interact with in our environment.
I recently came across hard copies of two early drafts of “The Extended Mind” in my files. The first draft was by Andy on his own and is covered in handwritten comments by me. The second draft is a coauthored version with revisions by me and is covered in handwritten comments by Andy. (I have put these two online at consc. net/e-drafts.html.) There were many further iterations after that. The whole thing is a wonderful record of distributed and extended cognition, involving a complex cognitive process spread between Andy, me, various notes on paper, and computer files. I’m proud to have played a role in constituting this process. That said, I have to acknowledge that, as with many cases of distributed cognition, there was an individual at the core who set the whole process into motion and who bears primary responsibility, and that individual was Andy.
I am still very much sympathetic with the extended mind thesis, though I also have some doubts about its truth that I articulated in my foreword to Andy’s 2008 book, Supersizing the Mind. I will return to those doubts shortly, but here I will focus especially on two other issues. First, what exactly is the extended mind thesis? Second, are there cases of extended consciousness, and if not, why not? In addressing
1 Thanks to the editors and to Andy Clark, Kati Farkas, Jackson Kernion, Tobias Schlicht, and Rob Wilson for comments.
the first issue, I will be criticizing and revising the statement of the thesis in our original article. In addressing the second issue, I will be criticizing and revising a thesis that both of us have endorsed separately in more recent work. So these remarks will to some extent criticize Andy, as the title of this volume requires. But really I am in the situation of a split-brain hemisphere criticizing a previous whole brain of which it was a part, as well as criticizing both itself and the other hemisphere on their own.
1. What Is the Extended Mind Thesis?
What exactly is the extended mind thesis? In “The Extended Mind” we never quite state a single official thesis. We say that we advocate an active externalism based on the active role of the environment in driving cognitive processes. We say that parts of the world can be parts of cognitive processes and that cognitive processes are not all in the head. We argue that some mental states, including especially beliefs, can be constituted partly by features of the environment, when those features play the right sort of role in driving cognitive processes. If there is an official thesis in the article, it is something like this:
A subject’s cognitive processes and mental states can be partly constituted by entities that are external to the subject, when those entities play the right role in driving cognitive processes.
To unpack this a little: the thesis applies to both cognitive processes and mental states, which we separate in the article since it is not obvious that the extension of one implies the extension of the other. “Partly constituted” is naturally understood as what is often now called “partial grounding” (metaphysical dependence on these entities along with others), though one might allow weaker relations in order to accommodate property dualist versions of the extended mind thesis in which mental states depend nomologically on features outside the head. “Entities” can be (perhaps inter alia) objects, instantiated properties, or states of affairs whereby objects instantiate properties. As for “external,” at various points we appeal to the boundaries of brain, body, head, skin, and skull. Each boundary leads to an interesting thesis, but perhaps to distinguish the extended mind thesis from various embodied mind theses, it makes sense to use the body as the key boundary.2 Finally, “the right role in driving cognitive processes” is required to distinguish our externalism from that of Putnam, Burge, and others. The article gives only limited guidance about just what
2 I am on record (www.youtube.com/watch?v=FhybIsN5IZc) as once endorsing the thesis that consciousness is in the hair, which is presumably outside the boundaries of brain, skin, and skull, but still part of the body, and perhaps a borderline case of being outside the head?
roles count as the right ones, but minimally what is needed is an active and ongoing role, as opposed to a passive and distal role.
Something like this is a fairly standard formulation of the extended mind thesis. But there is good reason to doubt that it is the right formulation. This reason has been touched on by various theorists over the years in formulating versions of the thesis, but it has been pressed most forcefully as an objection by Katalin Farkas (2012).
The objection is roughly this: as formulated, the thesis is too weak to be interesting. In particular, it is near-obviously true in extended circuit cases, cases in which a brain circuit is replaced by an external silicon circuit that is connected directly (e.g., by wiring or radio transmitters) to the rest of the brain so that it plays the same role as the original brain circuit.
In his reply to Jerry Fodor’s review of Supersizing the Mind, Andy himself spells out such a case, the case of Diva:
But now imagine a case in which a person (call her Diva) suffers minor brain damage and loses the ability to perform a simple task of arithmetic division using only her neural resources. An external silicon circuit is added that restores the previous functionality. Diva can now divide just as before, only some small part of the work is distributed across the brain and the silicon circuit: a genuinely mental process (division) is supported by a hybrid bio-technological system. That alone, if you accept it, establishes the key principle of Supersizing the Mind. It is that non-biological resources, if hooked appropriately into processes running in the human brain, can form parts of larger circuits that count as genuinely cognitive in their own right.
(Clark 2009a)
Andy says here, in effect, that if one accepts that an extended circuit like Diva’s can partially constitute a mental process (call this the extended circuit thesis), one has thereby accepted the key thesis of his book. Furthermore, if one accepts the extended circuit thesis, one has more or less accepted the official thesis of “The Extended Mind” as described earlier: mental processes will be partly constituted by something external because of the active role it plays in cognition. Now, some later remarks (which I will discuss shortly) suggest that Andy may think of his book’s thesis as something distinct from the original extended mind thesis. Nevertheless all this tends toward a view on which the extended circuit thesis suffices for extended cognition.
Farkas (2012) says, and I agree, that this cannot be quite right. The extended circuit thesis is too weak for it to support the interesting and controversial version of the extended mind thesis. One way to bring this out is to observe that many who take themselves to be opponents of the extended mind thesis explicitly accept the extended circuit thesis. Adams and Aizawa (2008) and Rupert (2009), who have
mounted perhaps the most sustained opposition to the extended mind thesis, are happy to accept the extended circuit thesis: they agree that in science-fiction cases such as Diva’s, there can be extended cognition. This suggests that the official formulation of the extended mind thesis does not really capture what is centrally at issue in the debate.
The point can also be brought out by observing that the Diva case is quite different from the key cases in “The Extended Mind,” such as the Otto case and the Tetris case. Most obviously, the Otto case and the Tetris case use only existing technology, where the Diva case uses hypothetical future technology. The first two cases involve ordinary perception and action, whereas the Diva case requires sciencefiction-style extended circuitry. Furthermore, compared to the other two cases, the Diva case is much more clearly functionally isomorphic to non-extended cases of cognition. Partly as a result, the claim of extended cognition in the Otto case and the Tetris case is far more controversial than the corresponding claim about the Diva case.
Now Andy and I could stand our ground and stick with our stipulated definition of the extended mind thesis, so that Adams, Aizawa, Farkas, and Rupert all count as supporters of the thesis. That would be a little akin to the US declaring victory in Vietnam and going home. I think it makes more sense to find a stronger formulation of the extended mind thesis that captures what is really at issue in the debate. This stronger thesis should be one that is not supported by a verdict of extended cognition in the Diva case, but that is supported by such a verdict in the Otto and Tetris cases.
Various stronger theses have been suggested in the literature, but no thesis that I have seen is ideal. Adams and Aizawa (2008) distinguish a modal version of the thesis (weak), holding that there could be cases of extended cognition, from a nonmodal version (strong), saying that there are cases of extended cognition. But this makes the strong thesis uncomfortably contingent. (If no one used notebooks and the like, would the thesis be false?) One will also need to index the thesis to times, so that if certain opponents are right, the 2017 version of the thesis is false, but the 2117 version of the thesis will be true. (My colleague Ned Block likes to say that the thesis was false in 1995, when we wrote the article, but it has since become true with the advent of smartphones and the like.) But I do not think that a thesis about the years 1995 or 2017 really captures what is of deepest philosophical interest in the article.
A related move distinguishes science-fiction from non-science-fiction cases, so that the strong thesis says there can be externally constituted cognition without science-fiction resources, while the weak thesis allows science-fiction cases such as extended circuits to count. Both the strong and the weak theses here are modal theses, but the modality in the strong thesis invokes only worlds fairly close to the actual world. I think something like this was what Andy and I had in mind in writing the paper. In recent work, Andy appeals to a version of this criterion: “In fairly easily
imaginable circumstances—ones that involved no giant leaps of technology or technique—we would be justified in holding that certain mental and cognitive states extended” (Clark 2010, 82). But the science fiction of today is the non–science fiction of tomorrow, and a giant leap from today is standard tomorrow, so as before, this formulation ties the extended cognition thesis uncomfortably to specific times. Perhaps there is something philosophically important about the difference between present technology and future technology, but if so a good formulation of the thesis ought to capture the philosophically important distinction directly.
Farkas (2012) suggests articulating a stronger thesis specifically in terms of standing states such as beliefs. Her suggested thesis says that the typical role of standing states can be extended to include states that produce conscious manifestations in a somewhat different way than normal beliefs and desires do. I think that while this thesis follows from our analysis of the Otto case and is interesting in its own right, it is not a good candidate to be the core extended mind thesis that is at issue. For a start, the thesis does not say anything directly about extension. Furthermore, the core thesis should certainly cover mental rotation in the Tetris case, but mental rotation is not a standing state so is not covered by Farkas’s thesis. Many wish to apply the extended mind thesis to other non-standing states such as perception, emotion, and the like; perhaps they are wrong to do so, but I do not think their view should be ruled out by definition. So we need to look elsewhere for a core extended mind thesis.
I suggest the following. What is distinctive about the extended mind thesis is that cognition can be extended in a certain way: via perception and action. In the Otto case, his belief is extended in virtue of his interactions with the notebook by perception and action. He sees the notebook, and he writes in it. In the Tetris case, mental rotation is extended by the subject’s seeing the tiles and pressing the button. In the Diva case, by contrast, Diva’s belief is extended by wires or radio transmitters that do not involve perception or action—instead the extension is a subpersonal process. This is why Otto and Tetris provide interesting cases of extended cognition while Diva does not.
This leads to the following formulation of the extended mind thesis:
A subject’s cognitive processes and mental states can be partly constituted by entities that are external to the subject, in virtue of the subject’s interacting with these entities via perception and action.
The “interacting via perception and action” clause plays the key role of counting the Otto and Tetris cases as relevant while counting the Diva case as irrelevant. It captures the plausible core of the nonmodal and non-science-fiction construals of the thesis: what is important about existing and non-science-fiction modes of extension is that they all involve extension through the mundane means of perception and action rather than extension through novel means such as radio transmitters
