Neuroethics
Neuroscience has dramatically increased understanding of how
mental states and processes are realized by the brain, thereby
opening doors for treating the multitude of ways in which minds
become dysfunctional. This book explores questions such as: When
is it permissible to alter a person’s memories, influence personality
traits or read minds? What can neuroscience tell us about free will,
self-control, self-deception and the foundations of morality?
The view of neuroethics offered here argues that many of our new
powers are continuous with much older abilities to alter minds. They
have, however, expanded to include almost all our social, political
and ethical decisions. Written primarily for graduate students,
this book will appeal to anyone with an interest in the more
philosophical and ethical aspects of the neurosciences.
neil levy is a Senior Research Fellow at the Centre for Applied
Philosophy and Public Ethics, University of Melbourne, Australia,
and a James Martin Research Fellow at the Program on Ethics of the
New Biosciences, Oxford. He has published more than fifty articles
in refereed journals, as well as four books previous to this one.

Neuroethics
neil levy

CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
First published in print format
ISBN-13 978-0-521-68726-3
ISBN-13 978-0-511-34272-1
© N. Levy 2007
2007
Information on this title: www.cambridge.org/9780521687263
This publication is in copyright. Subject to statutory exception and to the provision of
relevant collective licensing agreements, no reproduction of any part may take place
without the written permission of Cambridge University Press.
ISBN-10 0-511-34272-1
ISBN-10 0-521-68726-8
Cambridge University Press has no responsibility for the persistence or accuracy of urls
for external or third-party internet websites referred to in this publication, and does not
guarantee that any content on such websites is, or will remain, accurate or appropriate.
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
paperback
eBook (NetLibrary)
eBook (NetLibrary)
paperback

Contents
Preface page ix
Acknowledgements xiv
1 Introduction 1
What is neuroethics? 1
Neuroethics: some case studies 3
The mind and the brain 8
Peering into the mind 17
The extended mind 29
The debate over the extended mind 44
2 Changing our minds 69
Authenticity 73
Self-knowledge and personal growth 76
Mechanization of the self 78
Treating symptoms and not causes 81
3 The presumption against direct manipulation 88
The treatment/enhancement distinction 88
Enhancements as cheating 89
Inequality 92
Probing the distinction 94
Assessing the criticisms 103
Conclusion 129
4 Reading minds/controlling minds 133
Mind reading and mind controlling 133
Mind control 145

Mind reading, mind controlling and the parity principle 147
Conclusion 154
5 The neuroethics of memory 157
Total recall 159
Memory manipulation 171
Moderating traumatic memories 182
Moral judgment and the somatic marker hypothesis 187
Conclusion 195
6 The ‘‘self’’ of self-control 197
The development of self-control 203
Ego-depletion and self-control 206
Successful resistance 215
Addiction and responsibility 219
7 The neuroscience of free will 222
Consciousness and freedom 225
Who decides when I decide? 226
Consciousness and moral responsibility 231
Moral responsibility without the decision constraint 239
Lessons from neuroscience 243
Neuroscience and the cognitive test 246
Neuroscience and the volitional test 250
8 Self-deception: the normal and the pathological 258
Theories of self-deception 259
Anosognosia and self-deception 263
Anosognosia as self-deception 276
Conclusion: illuminating the mind 278
9 The neuroscience of ethics 281
Ethics and intuitions 282
The neuroscientific challenge to morality 288
contentsvi

Responding to the deflationary challenge 293
Moral constructivism 300
Moral dumbfounding and distributed cognition 307
Distributed cognition: extending the moral mind 308
References 317
Index 337
contents vii

objective, a philosophical approach to neuroethics is necessary. I do
not claim that it is the only approach that is necessary: obviously
neuroscientists must contribute to neuroethics, but so must
specialists in other fields. Neuroethics is, by its very nature,
interdisciplinary. But the kind of approach that only philosophy can
provide is indispensable, and, I believe, fascinating. Moreover, I shall
claim, the broader philosophical perspective offered here will help
illuminate the ethical issues, more narrowly construed. Only when
we understand, philosophically, what the mind is and how it can
be altered can we begin properly to engage in the ethics of
neuroethics. Indeed, I shall claim that understanding the mind
properly plays a significant role in motivating an important
alteration in the way ethics is understood, and in what we come to
see as the bearers of moral values. What might be called an
externalist ethics gradually emerges from the pages that follow, an
ethics in which the boundaries between agents, and between agents
and their context, is taken to be much less significant than is
traditionally thought.
Despite this insistence on the necessity for philosophy, I shall
not assume any philosophical background. Since I believe that
philosophical reflection will illuminate the ethical issues, and that
these ethical issues are the concern of all reflective people, I shall
attempt to provide necessary background, and to explain terminology
and debates, as it becomes relevant. I do not aim here to produce a
work of popular philosophy, which too often means philosophy
over-simplified. Instead, I aim to produce genuine philosophy that is
also accessible to non-philosophers. Since I am constructing a case
for a novel view of neuroethics, I expect that professional
philosophers will find a great deal of interest in what follows.
In this brief preface, I have added, in a small way, to the hype
surrounding neuroscience and neuroethics. I have claimed that the
sciences of the mind have the potential to help us understand the
nature of the self, and of humanity, our very identity. These claims
are, I believe, true. Yet this book defends a somewhat deflationary
preface xi

thesis, so far as the ethics of neuroethics is concerned. I shall argue
for what I call the parity thesis: our new ways of altering the mind
are not, for all that, entirely unprecedented, and ought not to be
regarded, as a class, as qualitatively different in kind from the old.
They are, instead, on a par with older and more familiar ways of
altering the mind. New technologies are often treated with suspicion
simply because they are new; sometimes they are celebrated for
precisely the same reason. Neuroscientific technologies ought not be
celebrated or reviled for being new: in fact, they – typically – raise
much the same kinds of puzzles and problems as older, sometimes
far older, technologies. That is not to say that they do not present
us with genuine ethical dilemmas and with serious challenges; they
do. But, for the most part, these dilemmas and challenges are new
versions of old problems.
If the new sciences of the mind often pose serious challenges,
they also present us with opportunities: since the challenges they
pose are often new versions of old challenges, they present us with
the opportunity to revisit these challenges, and the older
technologies that provoke them, with fresh eyes. Sometimes we
accept older practices simply because they are well established, or
because we have ceased to see their problems; reflecting on the new
neurosciences gives us the opportunity to reassess older ways of
altering minds. I hasten to add, too, that the parity thesis defended
here concerns the new technologies of the mind as a class. Some
particular applications of these technologies do raise new, and
genuinely unprecedented, challenges for us. We must assess each on
its own merits, for the powers and perils it actually possesses and
promises.
I will defend the parity thesis, in large part, by way of reflection
on what it means to be human. Thus while the thesis is deflationary
in one sense – deflating the pretensions of the technologies of the
mind to offer entirely novel and unprecedented possibilities for
altering human beings – it is also exciting in another: it offers us a
perspective upon ourselves, as individuals and as a species that is,
prefacexii

Acknowledgements
I have incurred many intellectual debts in the course of writing this
book. Many of the ideas here owe their genesis to discussions and
collaborations with Tim Bayne. Richard Ashcroft, Jill Craigie, Walter
Glannon, Gert-Jan Lokhorst and Saskia Nagel read the entire
manuscript and offered many useful comments. Many more people
read parts of the manuscript, in various stages of composition, or
listened to versions presented at conferences. Their comments saved
me from many embarrassing errors. They include Piers Benn, David
Chalmers, Randy Clark, George Graham Jeanette Kennett, Morten
Kringelbach Al Mele, Dick Passingham Derk Pereboom, Julian
Savulescu and Daniel Weiskopf. Finally, I want to thank Jo Tyszka
for her efficient copy-editing.
The ideas here have developed over many years; earlier
versions of some of them have previously been published. Chapter 6
contains ideas that previously appeared in ‘‘Addiction, autonomy and
ego-depletion,’’ Bioethics, 20 (2006): 16–20 and ‘‘Autonomy and
addiction,’’ Canadian Journal of Philosophy, 36 (2006): 427–47.
Chapter 7 builds on ‘‘Libet’s impossible demand,’’ Journal of
Consciousness Studies, 12 (2005): 67–76. Chapter 8 is a modified
version of ‘‘Self-deception without thought experiments,’’ to appear
in J. Fernandez and T. Bayne, eds., Delusions, Self-Deception and
Affective Influences on Belief-formation, New York: Psychology
Press, and Chapter 9 builds on ‘‘The wisdom of the Pack,’’
Philosophical Explorations, 9 (2006): 99–103 and ‘‘Cognitive
Scientific Challenges to Morality’’ Philosophical Psychology,
forthcoming. I thank the editors of all these journals and books for
permission to reprint relevant sections.

1 Introduction
what is neuroethics?
Neuroethics is a new field. The term itself is commonly, though
erroneously, believed to have been coined by William Safire (2002),
writing in The New York Times. In fact, as Safire himself acknowl-
edges, the term predates his usage.1 The very fact that it is so widely
believed that the term dates from 2002 is itself significant: it indi-
cates the recency not of the term itself, but of widespread concern
with the kinds of issues it embraces. Before 2002 most people saw no
need for any such field, but so rapid have been the advances in the
sciences of mind since, and so pressing have the ethical issues sur-
rounding them become, that we cannot any longer dispense with the
term or the field it names.
Neuroethics has two main branches; the ethics of neu-
roscience and the neuroscience of ethics (Roskies 2002). The ethics
of neuroscience refers to the branch of neuroethics that seeks
to develop an ethical framework for regulating the conduct of
neuroscientific enquiry and the application of neuroscientific know-
ledge to human beings; the neuroscience of ethics refers to the
impact of neuroscientific knowledge upon our understanding of
ethics itself.
One branch of the ‘‘ethics of neuroscience’’ concerns the con-
duct of neuroscience itself; research protocols for neuroscientists, the
ethics of withholding incidental findings, and so on. In this book
I shall have little to say about this set of questions, at least directly
(though much of what I shall say about other issues has implications
for the conduct of neuroscience). Instead, I shall focus on questions to
do with the application of our growing knowledge about the mind
and the brain to people. Neuroscience and allied fields give us an

Whichever interpretation of the aetiology of the disorder is
correct, however, BIID falls within the purview of neuroethics. BIID
is a neuroethical issue because it raises ethical questions, and
because answering those questions requires us to engage with the
sciences of the mind. The major ethical issue raised by BIID focuses
on the question of the permissibility of amputation as a means of
treating the disorder. Now, while this question cannot be answered
by the sciences of the mind alone, we cannot hope to assess it ade-
quately unless we understand the disorder, and understanding it
properly requires us to engage in the relevant sciences. Neuroscience,
psychiatry and psychology all have their part to play in helping us to
assess the ethical question. It might be, for instance, that BIID can
be illuminated by neuroscientific work on phantom limbs. The
experience of a phantom limb appears to be a near mirror image of
BIID; whereas in the latter, subjects experience a desire for removal
of a limb that is functioning normally, the experience of a phantom
limb is the experience of the continued presence of a limb that has
been amputated (or, occasionally, that is congenitally absent).
The experience of the phantom limb suggests that the experi-
ence of our bodies is mediated by a neural representation of a body
schema, a schema that is modifiable by experience, but which resists
modification (Ramachandran and Hirstein 1998). Phantom limbs are
sometimes experienced as the site of excruciating pain; unfortu-
nately, this pain is often resistant to all treatments. If BIID is
explained by a similar mismatch between an unconscious body
schema and the objective body, then there is every chance that it too
will prove very resistant to treatment. If that’s the case, then the
prima facie case for the permissibility of surgery is quite strong: if
BIID sufferers experience significant distress, and if the only way to
relieve that distress is by way of surgery, the surgery is permissible
(Bayne and Levy 2005).
On the other hand, if BIID has an origin that is very dissimilar
to the origin of the phantom limb phenomenon, treatments less
radical than surgery might be preferable. Surgery is a drastic course of
introduction4

action: it is irreversible, and it leaves the patient disabled. If BIID can
be effectively treated by psychological means – psychotherapy,
medication or a combination of the two – then surgery is imper-
missible. If BIID arises from a mismatch between cortical repre-
sentations of the body and the objective body, then – at least given
the present state of neuroscientific knowledge – there is little hope
that psychological treatments will be successful. But if BIID has its
origin in something we can address psychologically – a fall in certain
types of neurotransmitters, in anxiety or in depression, for instance –
then we can hope to treat it with means much less dramatic than
surgery. BIID is therefore at once a question for the sciences of the
mind and for ethics; it is a neuroethical question.
Automatism
Sometimes agents perform a complex series of actions in a state
closely resembling unconsciousness. They sleepwalk, for instance:
arising from sleep without, apparently, fully awaking, they may dress
and leave the house. Or they may enter a closely analogous state, not
by first falling asleep, but by way of an epileptic fit, a blow on the
head, or (very rarely) psychosis. Usually, the kinds of actions that
agents perform in this state are routine or stereotyped. Someone who
enters the state of automatism while playing the piano may continue
playing if they know the piece well; similarly, someone who enters
into it while driving home may continue following the familiar
route, safely driving into their own drive and then simply sitting in
the car until they come to themselves (Searle 1994).
Occasionally, however, an agent will engage in morally sig-
nificant actions while in this state. Consider the case of Ken Parks
(Broughton, et al. 1994). In 1987, Parks drove the twenty-three kilo-
metres to the home of his parents-in-law, where he stabbed them
both. He then drove to the police station, where he told police that he
thought he had killed someone. Only then, apparently, did he notice
that his hands had been badly injured. Parks was taken to hospital
where the severed tendons in both his arms were repaired. He was
neuroethics: some case studies 5

consideration: data about the range of normal cases, and about
various pathologies of agency. Investigating the mind of the acting
subject teaches us important lessons. We learn, first, that our con-
scious access to our reasons for actions can be patchy and unreliable
(Wegner 2002): ordinary subjects sometimes fail to recognize their
own reasons for action, or even that they are acting. We learn how
little conscious control we have over many, probably the majority, of
our actions (Bargh and Chartrand 1999). But we also learn how these
actions can nevertheless be intelligent and rational responses to
our environment, responses that reflect our values (Dijksterhuis
et al. 2006). The mere lack of conscious deliberation, we learn,
cannot differentiate responsible actions from non-responsible ones,
because it does not mark the division between the voluntary and the
non-voluntary.
On the other hand, the sciences of themind also provide us with
good evidence that some kinds of automatic actions fail to reflect our
values. Some brain-damaged subjects can no longer inhibit their
automatic responses to stimuli. They compulsively engage in utili-
zation behavior, in which they respond automatically to objects in
the environment around them (Lhermitte et al. 1986). Under some
conditions, entirely normal subjects find themselves prey to stereo-
typed responses that fail to reflect their consciously endorsed values.
Fervent feminists may find themselves behaving in ways that appar-
ently reflect a higher valuation of men than of women, for instance
(Dasgupta 2004). Lack of opportunity to bring one’s behavior under
the control of one’s values can excuse. Outlining the precise cir-
cumstances under which this is the case is a problem for neuroethics:
for philosophical reflection informed by the sciences of the mind.
Parks was eventually acquitted by the Supreme Court of
Canada. I shall not attempt, here, to assess whether the court was
right in its finding (we shall return to related questions in Chapter 7).
My purpose, in outlining his case, and the case of the sufferer from
BIID, is instead to give the reader some sense of how fascinating, and
how strange, the neuroethical landscape is, and how significant its
neuroethics: some case studies 7

findings can be. Doing neuroethics seriously is difficult: it requires a
serious engagement in the sciences of the mind and in several
branches of philosophy (philosophy of mind, applied ethics, moral
psychology and meta-ethics). But the rewards for the hard work are
considerable. We can only understand ourselves, the endlessly fas-
cinating, endlessly strange, world of the human being, by under-
standing the ways in which our minds function and how they
become dysfunctional.
the mind and the brain
This is a book about the mind, and about the implications for our
ethical thought of the increasing number of practical applications
stemming from our growing knowledge of how it works. To begin
our exploration of these ethical questions, it is important to have
some basic grasp of what the mind is and how it is realized by the
brain. If we are to evaluate interventions into the mind, if we are to
understand how our brains make us the kinds of creatures we are,
with our values and our goals, then we need to understand what
exactly we are talking about when we talk about the mind and the
brain. Fortunately, for our purposes, we do not need a very detailed
understanding of the way in which the brain works. We shall not be
exploring the world of neurons, with their dendrites and axons, nor
the neuroanatomy of the brain, with its division into hemispheres
and cortices (except in passing, as and when it becomes relevant).
All of this is fascinating, and much of it is of philosophical, and
sometimes even ethical, relevance. But it is more important, for our
purposes, to get a grip on how minds are constituted at much a higher
level of abstraction, in order to shake ourselves free of an ancient and
persistent view of the mind, the view with which almost all of us
begin when we think about the mind, and from which few of us ever
manage entirely to free ourselves: dualism. Shaking ourselves free of
the grip of dualism will allow us to begin to frame a more realistic
image of the mind and the brain; moreover, this more realistic image,
of the mind as composed of mechanisms, will itself prove to be
introduction8

see how mere matter could really think; be rational and intelligent,
and not merely flexibly responsive. Equally, it remains difficult to see
howmatter could be conscious. How could amachine, nomatter how
complex or cleverly designed, be capable of experiencing the subtle
taste of wine, the scent of roses or of garbage; how could there be
something that it is like to be a creature built entirely out of matter?
Dualism, with its postulation of a substance that is categorically
different from mere matter, seems to hold out the hope of an answer.
Descartes thought that matter could never be conscious or
rational, and it is easy to sympathize with him. Indeed, it is easy to
agree with him (even today some philosophers embrace property
dualism because, though they accept that matter could be intelligent,
they argue that it could never be conscious). Matter is unconscious
and irrational – or, better, arational – and there is no way to make it
conscious or rational simply by arranging it in increasingly complex
ways (or so it seems). It is therefore very tempting to think that since
we are manifestly rational and conscious, we cannot be built out of
matter alone. The part of us that thinks and experiences, Descartes
thought, must be built from a different substance. Animals and
plants, like rocks and water, are built entirely out of matter, but we
humans each have a thinking part as well. It follows from this view
that animals are incapable not only of thought, but also of experi-
ence; notoriously, this doctrine was sometimes invoked to justify
vivisection of animals. If they cannot feel, then their cries of pain
must be merely mechanical responses to damage, rather than
expressions of genuine suffering (Singer 1990).
It’s easy to share Descartes’ puzzlement as to how mere matter
can think and experience. But the centuries since Descartes have
witnessed a series of scientific advances that have made dualism
increasingly incredible. First, the idea that there is a categorical
distinction to be made between human beings and other animals no
longer seems very plausible in light of the overwhelming evidence
that we have all evolved from a common ancestor. Human beings
have not always been around on planet Earth – indeed, we are a
introduction10

supposed to explain how rationality and consciousness were possible,
given that, supposedly, no arrangement of mere matter could ever
realize these features. The evidence from brain damage suggests that
soul-stuff does not in fact have these alleged advantages, if indeed it
exists: it is itself too closely tied to the material to possess them.
Unexpectedly – for the dualist – mind degrades when matter is
damaged; the greater the damage, the greater the degradation. Given
that cognition degrades when, and to the extent that, matter is
damaged, it seems likely that any mind that could survive the
wholesale decay of matter that occurs after death would be, at best,
sadly truncated, incapable of genuine thought or memory, and
entirely incapable of preserving the identity of the unique individual
whose mind it is. Moreover, the fact that rationality degrades and
consciousness fades or disappears when the underlying neural
structures are damaged suggests that, contra the dualist, it is these
neural structures that support and help to realize thought and con-
sciousness, not immaterial mind – else the coincidental degradation
of mind looks miraculous. Immaterial minds shouldn’t fragment or
degrade when matter is damaged, but our minds do.
Perhaps these points will seem more convincing if we have
some actual cases of brain lesions and corresponding mind mal-
function before us. Consider some of the agnosias: disorders of
recognition. There are many different kinds of agnosias, giving rise to
difficulty in recognizing different types of object. Sometimes the
deficit is very specific, involving, for instance, an inability to identify
animals, or varieties of fruit. One relatively common form is proso-
pagnosia, the inability to recognize faces, including the faces of
people close to the sufferer. What’s going on in these agnosias? The
response that best fits with our common sense, dualist, view of
the mind preserves dualism by relegating some apparently mental
functions to a physical medium that can degrade. For instance, we
might propose that sufferers have lost access to the store of infor-
mation that represents the people or objects they fail to recognize.
Perhaps the brain contains something like the hard drive of a
the mind and the brain 13

computer on which memories and facts are stored, and perhaps the
storage is divided up so that memories of different kinds of things are
each stored separately. When the person perceives a face, she sear-
ches her ‘‘memory of faces’’ store, and comes up with the right
answer. But in prosopagnosia, the store is corrupted, or access to it is
disturbed. If something like this was right, then we might be able to
preserve the view that mind is a spiritual substance, with the kinds
of properties that such an indivisible substance is supposed to pos-
sess (such as an inability to fragment). We rescue mind by delegating
some of its functions to non-mind: memories are stored in a physical
medium, which can fragment, but mind soars above matter.
Unfortunately, it is clear that the hypothesis just sketched is
false. The agnosias are far stranger than that. Sufferers have not simply
lost the ability to recognize objects or people; they have lost a sense-
specific ability: to recognize a certain class of objects visually (or
tactilely, or aurally, and so on). The prosopagnosic who fails to recog-
nize his wife when he looks at her knows immediately who she is
when she speaks. Well, the dualist might reply, perhaps it is not his
store of information that is damaged, but his visual system; there’s
nothingwrongwithhismind at all, butmerelywithhis eyes. But that’s
not right either: the sufferer from visual agnosia sees perfectly well.
Indeed, he may be able to describe what he sees as well as you or I.
Consider Dr. P., the eponymous ‘‘manwhomistook his wife for a hat’’
of Oliver Sack’s well-known book, and his attempts to identify an
object handed to him by Sacks (here and elsewhere I quote at length,
in order to convey the strangeness of many dysfunctions of the mind):
‘About six inches in length,’ he commented. ‘A convoluted red
form with a linear green attachment.’
‘Yes,’ I said encouragingly, and what do you think it is, Dr. P.?’
‘Not easy to say.’ He seemed perplexed. ‘It lacks the simply
symmetry of the Platonic solids, although it may have a higher
symmetry of its own . . . I think this could be an infloresence or
flower.’
introduction14

On Sack’s suggestion, Dr P. smelt the object.
Now, suddenly, he came to life. ‘Beautiful!’ he exclaimed. ‘An
early rose!’
(Sacks 1985: 12–13)
Dr. P. is obviously a highly intelligent man, whose intellect is intact
despite his brain disorder. His visual system functions perfectly well,
allowing him to perceive and describe in detail the object with which
he is presented. But he is forced to try to infer, haltingly, what the
object is – even though he knows full well what a rose is and what it
looks like. Presented with a glove, which he described as a container
of some sort with ‘‘five outpouchings,’’ Dr. P. did even worse at the
object-recognition task.
It appears that something very strange has happened to Dr. P.’s
mind. Its fabric has unravelled, in some way and at one corner, in a
manner that no spiritual substance could conceivably do. It is diffi-
cult to see how to reconcile what he experiences with our common
sense idea of what the mind is like. Perhaps a way might be found to
accommodate Dr. P.’s disorder within the dualist picture, but the
range of phenomena that needs to be explained is wide, and its
strangeness overwhelming; accommodating them all will, I suggest,
prove impossible without straining the limits of our credibility.
One more example: another agnosia. In mirror agnosia,
patients suffering from neglect mistake the reflections of objects for
the objects themselves, even though they know (in some sense) that
they are looking at a mirror image, and only when the mirror is
positioned in certain ways. First, a brief introduction to neglect,
which is itself a neurological disorder of great interest. Someone
suffering from neglect is profoundly indifferent to a portion of their
visual field, even though their visual system is undamaged. Usually,
it is the left side of the field that is affected: a neglect sufferer might
put makeup on or shave only the right side of their face; when asked
to draw a clock, they typically draw a complete circle, but then
stuff all the numbers from one to twelve on the right hand half.
the mind and the brain 15

chapters). All I want to do, right now, is to draw your attention to
how strange malfunctions of the mind can be – far stranger than we
might have predicted from our armchairs – and also how (merely)
physical dysfunction can disrupt the mind. The mind may not be
a thing; it may not be best understood as a physical object that
can be located in space. But it is entirely dependent, not just for
its existence, but also for the details of its functioning, on mere
things: neurons and the connections between them. Perhaps it is
possible to reconcile these facts with the view that the mind is a
spiritual substance, but it would seem an act of great desperation
even to try.
peering into the mind
I introduced some of the disorders of the mind in order to show that
substance dualism is false. Now I want to explore them a little fur-
ther, in order to accomplish several things. First, and most simply,
I want to demonstrate how strange and apparently paradoxical the
mind can be, both when it breaks down and when it is functioning
normally. This kind of exploration is fascinating in its own right, and
raises a host of puzzles, some of which we shall explore further in
this book. I also have a more directly philosophical purpose, however.
I want to show to what extent, contra what the dualist would have us
expect, unconscious processes guide intelligent behaviour: to a very
large extent, we owe our abilities and our achievements to sub-
personal mechanisms. Showing the ways in which mind is built, as it
were, out of machines will lay the ground for the development of a
rival view of the mind which I will urge we adopt. This rival view
will guide us in our exploration of the neuroethical questions we
shall confront in later chapters.
Let’s begin this exploration of mind with a brief consideration
of one method commonly utilized by cognitive scientists, as they
seek to identify the functions of different parts of the brain. Typi-
cally, they infer function by seeking evidence of a double dissocia-
tion between abilities and neural structures; that is, they seek
peering into the mind 17

how our brains and minds make us who, and what, we are, we need
to understand the very large extent to which information processing
takes place automatically, below the level of conscious awareness.
This is exactly what one would predict, on the basis of our
evolutionary past. Evolution tends to preserve adaptations unless
two conditions are met: keeping them becomes costly, and the costs
of discarding them and redesigning are low. These conditions are very
rarely met, for the simple reason that it would take too many steps to
move from an organism that is relatively well-adapted to an envir-
onment, to another which is as well or better adapted, but which is
quite different from the first. Since evolution proceeds in tiny steps,
it cannot jump these distances; large-scale changes must occur via
a series of very small alterations each of which is itself adaptive.
Evolution therefore tends to preserve basic design features, and tin-
ker with add-ons (thus, for instance, human beings share a basic
body plan with all multicellular animals). Now, we know that most
organisms in the history of life on this planet, indeed, most organ-
isms alive today, got along fine without consciousness. They needed
only a set of responses to stimuli that attracted and repelled them
according to their adaptive significance. Unsurprisingly, we have
inherited from our primitive ancestors a very large body of sub-
personal mechanisms which can get along fine without our con-
scious interference.
Another double dissociation illustrates the extent to which our
behavior can be guided and driven by subpersonal mechanisms.
Vision in primates (including humans) is subserved by two distinct
systems: a dorsal system which is concerned with the guidance of
action, and a ventral system which is devoted to an internal repre-
sentation of the world (Milner and Goodale 1995). These systems are
functionally and anatomically distinct; probably the movement-
guidance system is the more primitive, with the ventral system being
a much later add-on (since guidance of action is something that
all organisms capable of locomotion require, whereas the ability to
form complex representations of the environment is only useful to
introduction20

creatures with fairly sophisticated cognitive abilities). Populations of
neurons in the ventral stream are devoted to the task of object dis-
crimination, with subsets dedicated to particular classes of objects.
Studies of the abilities of primates with lesioned brains – experi-
mental monkeys, whose lesions were deliberately produced, and of
human beings who have suffered brain injury – have shown the
extent to which these systems can dissociate. Monkeys who have
lost the ability to discriminate visual patterns nevertheless retain the
ability to catch gnats or track and catch an erratically moving peanut
(Milner and Goodale 1998). Human beings exhibit the same kinds of
dissociations: there are patients who are unable to grasp objects
successfully but are nevertheless able to give accurate descriptions of
them; conversely, there are patients who are unable to identify even
simple geometric shapes but who are able to reach for and grasp them
efficiently. Such patients are able to guide their movements using
visual information of which they are entirely unconscious (Goodale
and Milner 2004).
What’s it like to guide one’s behavior using information of
which one is unconscious? Well, it’s like everyday life: we’re all
doing it all the time. We all have dorsal systems which compute
shape, size and trajectory for us, and which send the appropriate
signals to our limbs. Sometimes we make the appropriate move-
ments without even thinking about it; for instance, when we catch a
ball unexpectedly thrown at us; sometimes we might remain una-
ware that we have moved at all (for instance when we brush away a
fly while thinking about something else). Action guidance without
consciousness is a normal feature of life. We can easily demonstrate
unconscious action-guidance in normal subjects, using the right kind
of experimental apparatus. Consider the Titchener illusion, produced
by surrounding identical sized circles with others of different sizes. A
circle surrounded by larger circles appears smaller than a circle sur-
rounded by small circles. Aglioti and colleagues wondered whether
the illusion fooled both dorsal and ventral visual systems. To test
this, they replaced the circles with physical objects; by surrounding
peering into the mind 21

clever, consciousness.4 Consciousness is reliant on these mechan-
isms, though it can also act upon and shape them. Many of our
actions, too, including some of our most important, are products of
unconscious mechanisms. The striker’s shot at goal happens too fast
to be initiated by consciousness, similarly, the improvising musician
plays without consciously deciding how the piece will unfold. Think,
finally, of the magic of ordinary speech: we speak, and we make
sense, but we learn precisely what we are going to say only when we
say it (as E.M. Forster put it, ‘‘How can I tell what I think till I see
what I say?’’). Our cleverest arguments and wittiest remarks are not
first vetted by consciousness; they come to consciousness at pre-
cisely the same time they are heard by others. (Sometimes we
wonder whether a joke or a pun was intentional or inadvertent.
Clearly, there are cases which fit both descriptions: when someone
makes a remark that is interpreted by others as especially witty, but
he is himself bewildered by their response, we are probably dealing
with inadvertent humor, while the person who stores up a witty
riposte for the right occasion is engaging in intentional action. Often,
though, there may be no fact of the matter whether the pun I make
and notice as I make it counts as intentional or inadvertent.)
Identifying the self with consciousness therefore seems to be
hopeless; it would shrink the self down to a practically extensionless,
and probably helpless, point. Few sophisticated thinkers would be
tempted by this mistake. But an analogous mistake tempts even very
clear thinkers, a last legacy of the Cartesian picture. This mistake is
the postulation of a control centre, a CPU in the brain, where
everything comes together and where the orders are issued.
One reason for thinking that this is a mistake is that the idea of
a control centre in the brain seems to run into what philosophers of
mind call the homunculus fallacy: the fallacy of explaining the
capacities of the mind by postulating a little person (a homunculus)
inside the head. The classic example of the homunculus fallacy
involves vision. How do we come to have visual experience; how,
that is, are the incoming wavelengths of light translated into the rich
introduction24

visual world we enjoy? Well, perhaps it works something like a
camera obscura: the lenses of the eyes project an image onto the
retina inside the head, and there, seated comfortably and perhaps
eating popcorn, is a homunculus who views the image. The reason
that the homunculus fallacy is a fallacy is that it fails to explain
anything. We wanted to know how visual experience is possible, but
we answered the question by postulating a little person who looks at
the image in the head, using a visual system that is presumably
much like ours. How is the homunculus’ own visual experience to be
explained? Postulating the homunculus merely delays answering the
question; it does not answer it at all.
The moral of the homunculus fallacy is this: we explain the
capacities of our mind only by postulating mechanisms that have
powers that are simpler and dumber than the powers they are
invoked to explain. We cannot explain intelligence by postulating
intelligent mechanisms, because then we will need to explain their
intelligence; similarly, we cannot explain consciousness by postu-
lating conscious mechanisms. Now, one possible objection to the
postulation of a control centre in the brain is that the suggestion
necessarily commits the homunculus fallacy: perhaps it ‘‘explains’’
control by postulating a controller. It is not obvious, to me at any
rate, that postulating a controller must commit the homunculus
fallacy. However, recognition of the fallacy takes away much of the
incentive for postulating a control centre. We do not succeed in
explaining how we become capable of rational and flexible behavior
by postulating a rational and flexible CPU, since we are still required
to explain how the CPU came to have these qualities. Sooner or later
we have to explain how we come to have our most prized qualities by
reference to simpler and much less impressive mechanisms; once we
recognize that this is so, the temptation to think there is a controller
at all is much smaller. We needn’t fear that giving up on a central
controller requires us to give up on agency, rationality or morality.
We rightly want our actions and thoughts to be controlled by an
agent, by ourselves, and we want ourselves to have the qualities we
peering into the mind 25

striking feature of everyday life. Consider weakness of the will, the
everyday phenomenon in which we find ourselves doing things of
which we do not rationally approve. We have a second drink when
we know we should stop at one; we resolve to skip dessert but our
resolve crumbles when we glance at the menu, we continue smoking
in spite of our New Year’s resolution to stop. As Plato noticed 2500
years ago, these kinds of incidents seem to indicate the presence
within the single agent of different centres of volition and desire:
parts of the self with their own preferences, each of which battles for
control of the agent so as to satisfy its desires.
This is not to say that our everyday view of ourselves as unified
agents, as a single person with a character and with (relatively)
consistent goals, is false. Rather, the unified agent is an achievement:
we unify ourselves as we mature. If we do not manage to impose a
relatively high degree of unity on ourselves, we shall always be at
odds with ourselves, and our ability to pursue any goal which
requires planning will be severely curtailed. Unification is a neces-
sary condition of planning, for without a relatively high degree of
unity we shall always be undermining our own goals. Lack of unity is
observed in young children, and undermines their ability to achieve
the goals they themselves regard as desirable. Longitudinal studies
show that children who do not acquire the skills to delay gratifica-
tion generally do worse on a range of indicators throughout their
lives, but delaying gratification requires the imposition of unity. In
order to become rational agents, capable of long-term planning and
carrying out our plans, we need to turn diversity into unity. We
achieve this not by eliminating diversity, but by forging coalitions
between the disparate elements of ourselves. These coalitions remain
forever vulnerable to disruption, short and long term. One way to
understand drug addiction, for instance, is as a result of a disruption
of the imposed unity of the agent. The drug-addicted agent might
genuinely desire to give up his drug, but because he cannot extend his
will across time and across all the relevant subagents which con-
stitute him, he is subject to regular preference reversals. When he
introduction28

But the behavior of ventromedial patients was quite different.
There is nothing wrong with these patients’ autonomic system: they
experience SCRs in response to punishment, in just the same way as
do normal subjects (Damasio 1994). But they do not generate
anticipatory SCRs. They do not get the ‘‘warning signals’’ which,
unconsciously, bias normal subjects against certain actions (because
the ventromedial prefrontal cortex stores dispositional knowledge
which activates the relevant parts of the brain which in turn cause
autonomic system response, Bechara et al. (1997) speculate). Ven-
tromedial patients, because they lack this vital source of informa-
tion, find it much more difficult to work out how the decks are
arranged. They are, we might say, thrown back on pure, brain-based,
rationality. But pure rationality, all on its own, is a relatively meagre
resource. Indeed, even when it is sufficient for the production of
knowledge, it may be insufficient to guide rational behavior: though
some ventromedial patients did eventually work out how the decks
were arranged, their behavior continued to differ from that of normal
subjects. These patients remained more likely to risk the large
punishment in order to secure the bigger rewards of decks A and B.
Not only is the ability to generate anticipatory SCRs beneficial to
cognition; it also proves to be an indispensable guide to prudent
action. Pure – that is, brain-based – rationality cannot compensate for
its absence.
We shall return to Damasio’s theory, the so-called somatic-
marker hypothesis (SMH), according to which bodily responses are
an indispensable guide in beneficial decision-making and action, in
later chapters. For the moment, what matters is the way in which the
SMH expands our view of the resources needed for rational decision-
making. Without bodily responses, cognition is impaired. It is, in
important part, by referring to our bodily feelings, when we con-
template different courses of action, that we make good decisions. If
the mind should be understood to consist of all the resources we use
in assessing different courses of action, then the mind includes (parts
of) the body, at least for some purposes and in some contexts. While
introduction32

information store from being mental. Even in terms of the parity
thesis, however, this is a mistake: these conditions are merely suf-
ficient, and not necessary. That is, something might count as part of
memory even if it failed to meet one or more of the conditions out-
lined. Constant availability cannot be a necessary condition for
counting as part of the mind, since the contents of our short-term
memories are, by definition, only available for a brief period (no more
than thirty seconds), and our longer-term memory is (notoriously)
not always available for instant recall. Moreover, people suffering
from dementia might have rather unreliable access to their mem-
ories; so long as they are able to recall memories sometimes, they
seem not (yet) to have lost those elements of their minds. Because the
constantly availability criterion is not necessary, I think there is a
good case (from parity) to be made for regarding the visual scene
before us as the place where our representation of the immediate
environment is stored. Nor can easy accessibility be a necessary
condition of the mental, as the example of the demented individual,
or indeed even ordinary cases of failure of memory retrieval (such as
the tip of the tongue phenomenon) show. Memories are not always
easily accessible, yet they count as part of the mind; so do all the
unconscious mechanisms that figure in information processing.
Given the similarities, in functional terms, of the role that
external representations can play in cognition to the role played by
internal, it seems mere prejudice to insist that they are not mental
just because they are external. Moreover, it is not just external
representations that should be counted as part of the mind, on the
basis that they are an integral part of cognition. As well as external
representations, we use external tools to enhance our thought. These
tools for thinking are almost as old as human beings; indeed, perhaps
we only became truly human when we developed them (Sterelny
2005). The most basic and obvious of these tools is speech. Speech
does not merely allow us to articulate thoughts that we would have
had in any case. Instead, it allows us to externalize our thoughts
and thereby to treat them as objects for contemplation and for
introduction38

manipulation. Externalized thoughts can be worked over, criticized
and improved. They can also be shared, and the mental resources
of two, or many, people brought to bear on them. Once we take
the further step of developing the capacity to write down our thoughts,
the extent to which they can be manipulated, publicly criticized
and thereby improved, increases exponentially. All kinds of ways of
thinking become accessible for the first time with the invention of
ways of keeping track of our thoughts by representing them externally;
paradigmatically (but not only) by writing them down.
Think of mathematics, for instance. Even arithmetic, beyond a
certain level of complexity, requires pen and paper – or a computer
screen, or clay tablets, or what have you – if it is to be performed at
all. Multiply 23 789 by 54 553. Without pen and paper, the calcula-
tion is beyond me. Perhaps you can perform it; there are short cuts
that can be learned (though I suspect that these short cuts themselves
rely on ways of extending cognition beyond the boundaries of skin
and skull). For almost all of us, however, the calculation is impos-
sible without some kind of tool, and as the complexity of the sum
grows, the proportion of those unable to perform it without tools
reaches 100 percent (divide the result of the above multiplication by
0.0034). Without some kind of means of making external repre-
sentations, even arithmetic quickly exceeds our abilities. In fact,
multiplication is multiply extended, across time and across space,
across physical elements and across agents. How do you go about
doing a three-digit multiplication (say 365· 412), in the absence of a
calculator? You probably use the method most of us learned at
school: converting the sum into a series of single-digit multiplication
tasks, remembering to carry the one, writing the results down
sequentially, then moving to the next line, and repeating the process,
until you have a series of numbers which can simply be added to get
the right result. In other words, you use a simple algorithm you
learned long ago. You did not develop the algorithm for yourself; you
probably just accepted, on faith, that it worked. If that’s right, your
cognitive process is reliant not only on pencil and paper, but also on
the extended mind 39

yet it may be the one feature that, more than any other, makes
human thought so powerful. Homo sapiens is (as the name suggests)
a pretty clever beast. But in order to be able to engage in systematic
science, and to develop technology, from cooking to writing, from
steam engines to computers, from water treatment plants to fMRI
machines – we must be able to accumulate knowledge, and hand it
on from generation to generation. If each generation had to start
afresh, we would not still be in the Dark Ages; we could never have
got anywhere near as advanced as the Dark Ages. Moreover, progress
in knowledge requires a distributed approach: many people must
tackle a problem, from many different directions at once, and the
results of their work must be available to others for criticism. This,
too, requires that results and techniques be made available in a
publicly accessible, and long-lived, medium. For these reasons, both
because we need a way of accumulating knowledge, and because
great knowledge can be produced only by a community of enquirers,
we need a means of expressing our ideas, our results, our findings
and our failings, a means that is in principle relatively easy to
access. We need external representations: a system of writing at very
minimum. Obviously writing has its limitations, which makes
it unsuitable for certain kinds of knowledge: for these we need
specialized systems of representations, such as mathematical or
chemical symbols.
Knowledge acquisition must be a public endeavor, if it is to
be successful. The lone scientist, for all the mythology of the
Dr. Frankenstein alone in his lab, doesn’t get very far (even the sci-
entists of earlier ages, who sometimes worked alone, generally
required access to a library). Moreover, even for the individual cog-
nizer, the advancement of knowledge requires that the cognitive
environment is structured in a way that takes the load off their
brains: they need, for instance, to be able to write down their cal-
culations so as to ensure that they free up mental resources that
would otherwise be drained by the effort of memory. One way we
take the load off our brains is by developing tools that embody our
the extended mind 43

knowledge; they preserve previous results for us, and suggest ways in
which we could tackle new problems. We extend our cognitive
capacities by building external tools and representations, and we are
able to do that only because we already have minds that make use of
external representations – for instance, by relying on the stability of
the environment to spare us the effort of representing it – thus freeing
cognitive resources for the building of new tools.
the debate over the extended mind
The extended mind hypothesis is very controversial. Many cognitive
scientists and philosophers have rushed to defend the view that the
mind, and with it all properly cognitive processes, is entirely con-
tained within the skull. It’s worth pausing for a moment over these
debates. As I’ve already pointed out, these debates have generally
focused on the parity thesis; this is, I think (as I’ve also already
mentioned), a mistake. Nevertheless, I have also said that the parity
thesis is a useful heuristic for expanding minds; for this reason it’s
worth seeing just how far it can take us. Seeing just how difficult it is
to defend the claim that the mind is inside the head, even when we
accept the parity thesis, will help to shake us free from this last
legacy of Cartesianism.
The most influential criticisms of the extended mind hypoth-
esis have been those advanced by Adams and Aizawa (2001) and
Rupert (2004). Adams and Aizawa argue that in order to assess the
claim that something counts as properly part of the mind, we need an
account of what cognition consists in: we need to identify ‘‘the mark
of the mental.’’ They do not attempt to advance anything like a full
theory of what cognition is; instead they argue for two necessary
conditions which any process or state must satisfy in order to count
as cognitive. First, it must involve intrinsic content. Second, cogni-
tive processes must be causally individuated.
The notion of intrinsic content is best understood by contrast
to its opposite, derived content. Roughly, a representation has
derived content if it refers in virtue of a convention. The letters CAT
introduction44

refer to the animal ‘‘cat’’ in virtue of a set of nested conventions: the
convention, on the part of speakers of English, of using the sounds
represented by those letters to refer to cats, and the convention of
using the letters CAT to refer to those sounds. Our world is full of
representations that have derived content: words of natural lan-
guages, mathematical symbols, traffic signs, and so on. But if a sign
refers in virtue of a convention, its referential power is derivative
from the referential power of the human mind. Now, how do the
paradigm mental states of human beings come to refer? If we are to
avoid an infinite regress, Adams and Aizawa argue, we must recog-
nize that human minds are capable of states that are intrinsically
referential. When I read CAT, I have a mental state that refers to cats,
but my mental state itself refers to cats in virtue of its nature, not in
virtue of any convention. If this were false, if my mental state
referred only in virtue of a convention, we would need to ask how
that convention, in turn, came to be meaningful: eventually, we
must stop at a representation that is intrinsically meaningful, or
representation could never get off the ground in the first place.
Now for the second of Adams’ and Aizawa’s conditions, a
condition also emphasized by Rupert. Why think that genuine cog-
nitive processes must be causally individuated; that is, distinguished
from each other and from anything else in terms of their physical
causes? The motivation for this claim seems to be this: if a science of
cognition is to be possible, then there had better be a discoverable set
of causal regularities or laws for that science to capture. And the
available evidence suggests that there are indeed such laws. The
processes that occur in human minds are somewhat diverse, but they
are nevertheless relatively unified, inasmuch as they can be descri-
bed by a relatively small set of laws. Within more narrowly cir-
cumscribed domains, the degree of unification is even higher: thus,
there is a set of rough, ceteris paribus, laws that describe a wide range
of human memory systems, short and long term (Rupert 2004);
hence, the domain of the mental should be expected to be causally
individuated.7
the debate over the extended mind 45

content, and ask whether such states must consist only of such
content? Adams and Aizawa are unsure (2001: 50), and as Clark sees,
this opens the way for admitting derived content within the purview
of the mental. Clark asks us to consider the mental manipulation of
Venn diagrams. Suppose that Otto sees that some Xs are also Ys by
picturing Venn diagrams to himself. Is he not thereby engaged in
cognition? Yet clearly Venn diagrams get their content from a con-
vention (Clark 2005). Similarly, some ordinary human thought seems
to be conducted in a natural language. We have no reason to exclude
such thought from the domain of the cognitive.8
Clark also asks us to consider a thought experiment. Imagine
alien beings capable of storing bit-mapped images of text. These
aliens would have representations of text in their minds, not
representations of the meaning of the text. When they wanted to
know what the text meant, they would picture it and read it in the
privacy of their heads. Clark asks us to think about the stretch of
time prior to their reading the text. Wouldn’t we count the content
of the text as (dispositionally) part of their mind at that time? If so,
then it is simply the prejudice that mind is within the skull that
prevents us from counting Otto’s similarly poised notebook as part
of his mind.
In reply, Adams and Aizawa (forthcoming) insist that they have
no inclination to count alien bit-mapped texts as part of alien minds.
Such texts have no intrinsic content, prior to recall, and therefore fail
to be cognitive. I wonder, though, whether they might not feel some
pull toward Clark’s view of the case if they compared it to actual
cases of mental recall here on Earth. Some people – mainly children,
but some adults as well – have what is known as eidetic recall. In
certain circumstances, and with varying degrees of reliability, they
are able to ‘‘picture,’’ in their ‘‘mind’s eye,’’ a scene that is no longer
before them. Of course, we are all capable of mental imagery, but
eidetikers seem capable of something beyond most of us, in most
circumstances: they are able to extract information from their
mental image of which they were not previously aware.9 They can,
the debate over the extended mind 47

just too large and too disparate to ever form the object of a science.
Whereas even now cognitive scientists are discovering the causal
regularities governing the mind, traditionally conceived, there is no
chance that they will discover an interesting set of regularities which
circumscribes the domain of brains-plus-notebooks, brains-plus-cal-
culators, brains-plus-marks-in-the-sand, and so on.
In response, Clark stresses that we ought not to try to second-
guess the progress of science (Clark 2006). We cannot infer, from our
current inability to unify extended cognition under a set of laws,
that future science will not succeed where we fail. I think, however,
that we ought to concede to Clark’s critics that the probability of
future unification is low. The set of processes and objects already
supposed to extend our minds into the world – marks in clay, tattoos
on the body, digitally encoded information, and many more besides –
is already very disparate, and we have every reason to think that the
current technological explosion will deliver us many more, and
different, ways of manipulating and storing information. Of course
we shall discover – are already in the process of discovering – laws
that apply to all these resources, as well as to the brain. But that’s
not the question at issue; the question is, will we be able to
delineate laws which uniquely circumscribe the domain of thought,
where ‘‘thinking’’ is an activity distributed across brains and tools?
The fact that the laws of physics (for instance) will apply to all these
elements is neither here nor there: the laws of physics apply
everywhere, and therefore cannot be used to delineate the subject
matter of a special science. I think we ought to concede, then, that
Adams, Aizawa and Rupert are on solid ground when they argue that
we are unlikely to be able to develop a science of the extended
mind, at least a science which is as well unified as the science of the
mind/brain.
Why should this matter? Adams, Aizawa and Rupert are
somewhat unclear on this point. Here is one reason to think that the
contours of the science of mind ought to be taken seriously in these
debates. Science, one may believe, is such a powerful means of
the debate over the extended mind 49

gaining knowledge in part because each science ‘‘carves nature at its
joints’’ (in Plato’s phrase). In other words, scientific theories might be
so powerful because they describe or pick out parts of the universe
that are genuinely unified. If that’s the case, then science may be a
good guide to ontology, to the nature of reality: if a relatively discrete
part of the universe can be captured by a set of scientific general-
izations, then we have good reason to think that that part of the
universe constitutes a natural kind. If this line of argument is right,
then the fact (if it is a fact) that there is a set of causal regularities
which applies to the brain/mind, and no such set that applies to the
brain/mind plus its various add-ons gives us reason to think that the
brain/mind constitutes a discrete entity, and the extended mind does
not. If that’s right, then the mind is identical to an entity circum-
scribed at the relatively low level of causal regularities, and not at the
functional level.
Critics of the extended mind could invoke this argument in an
effort to confine mind to the skull. It is not, they can claim, mere
prejudice that prevents us from extending the mind, but arguments,
and a respect for the power of science. However, the critics adopt this
line at their own peril. It will not vindicate the traditional conception
of mind; instead it will undermine it.
The mind/brain, as we saw in the first chapter, contains a
number of quite different mechanisms, which are sufficiently diverse
to cast doubt on the claim that they form a single domain. Consider,
for instance, the distinction often made between controlled and
automatic processes (Bargh and Chartrand 1999; Wegner 2005). Both
processes issue in complex behavior in human beings. But there are
significant differences between them. Controlled processes are gen-
erally conscious; they are also very demanding of cognitive resources,
such that performance at them degrades significantly under cognitive
load (we place subjects under cognitive load by requiring them to do
two tasks at once: the controlled task we want to study and another,
loading, task, such as counting backwards from 1000 in threes).
Automatic processes are not consciously initiated, and do not
introduction50

related beliefs are automatically updated. But Otto’s relevantly
related beliefs are not automatically updated. Suppose he has written
in his diary ‘‘Karl Rove is political advisor to the president.’’ The act
of crossing out ‘‘George W. Bush’’ and replacing it with ‘‘Hilary
Clinton’’ does not automatically alter what is contained in his diary
about Karl Rove. His beliefs are not informationally integrated, in the
manner in which Inga’s are.
As Weiskopf points out, informational integration is not a
marginal feature of ordinary belief. The concept of belief plays an
essential role in explaining and in enabling the prediction of human
behavior, but only relatively integrated beliefs can guide behavior.
The agent who believes p and that p entails q, but who believes not-
q, is an irrational agent and it is difficult to know how they will act.
It is only because normal agents who believe p and that if p then q
also automatically go on to believe q that we can reliably predict and
explain human behavior.12
It might be pointed out that the degree of informational inte-
gration of beliefs is variable, and that some perfectly ordinary beliefs
are isolated, so that updating them has little or no effect on the
texture of our web of belief. But this observation is of little help to
the proponent of the extended mind hypothesis, since beliefs with a
low degree of integration are also, at least typically, trivial beliefs.
Their insignificance both explains and is explained by their isolation:
they are isolated, in part, because we do not connect them to the
ongoing narrative of our lives, and we do not connect them to this
narrative because we perceive them as isolated (I have in mind such
prima facie trivial beliefs as: I saw a man wearing a green coat on the
train yesterday; three people living in my street own red cars, and so
on). I take it that proponents of the extended mind hypothesis do not
want to hang their hat on the claim that only trivial beliefs can leak
out of skulls and into the world; the hypothesis loses all interest if
that’s what it amounts to.
I think we ought to concede that paradigm beliefs are normally
informationally integrated, and that this is an important fact about
introduction54

them. The information stored in the environmental resources and
props upon which we lean so heavily is not informationally inte-
grated to anything like the same extent, or with anything like the
same degree of automaticity (at least, not yet: perhaps databases
constructed so that changes in one element ramify throughout the
system in the appropriate manner will one day exist), and this is a
significant difference between internal and external information
stores. However, is this difference significant enough to require us
to reserve the term ‘‘belief’’ only for internal representations? The
answer to this question is not clear.
Informational integration is a feature of paradigm beliefs, and it
is not a feature of extended information stores, not, at least, to any-
thing like the same extent. So extended information differs from
paradigm beliefs. But it does not differ in this manner from some
internal mental states that many people are disposed to call beliefs.
Consider the representational content of some delusions. Deluded
patients sometimes present with delusions that are not merely
impossible but contradictory. Sufferers from Cotard’s delusion, for
instance, claim to be dead, which is a claim that seems to be prag-
matically self-defeating. Breen et al. (2000) report the case of a
woman who believed (correctly) that her husband had died four years
earlier, and had been cremated, and also that he was currently a
patient in the same hospital she found herself in. Clearly, in both the
case of the Cotard’s sufferer, and in the case of the delusional woman,
there is an extreme failure of informational integration: if one
believes that one is dead, one also ought to give up the belief that one
(say) needs to, or indeed is able to, eat, but Cotard’s sufferers usually
(though not always) continue to eat. If one believes that one’s hus-
band is in the next ward, one ought to give up the belief that he is
dead and has been cremated. Ought we to conclude that delusional
patients, or at least some of them, do not actually believe the content
of their delusion?
Philosophers and neuroscientists are divided on this ques-
tion. Delusions are bizarre, relatively unconnected to evidence,
the debate over the extended mind 55

weak version, corresponding to the hypotheses of extended and
embedded cognition respectively:
EPP (strong): Since themind extends into the external environment,
alterations of external props used for thinking are (ceteris paribus)
ethically on a par with alterations of the brain.
EPP (weak): Alterations of external props are (ceteris paribus)
ethically on a par with alterations of the brain, to the precise
extent to which our reasons for finding alterations of the brain
problematic are transferable to alterations of the environment in
which it is embedded.
EPP (strong) claims that our ethical responses to interventions into
the cognitive environment ought to be consistent with our ethical
responses to interventions into the brain. If we worry, say, that
enhancing the brain pharmacologically is (for whatever reason)
wrong, or that transforming it using, say, magnetic stimulation or
surgery risks inauthenticity, then we should worry equally about
analogous interventions into the extended mind. If Otto were to
replace his notebook with a PDA, we might want to claim that he is
cheating, just as some people claim that agents who take Ritalin to
improve their cognitive perfomance are cheating. More plausibly
perhaps, if interventions into the extended mind are not ethically
problematic – if Otto’s replacing his notebook with a PDA raises no
ethical questions at all – then we ought not to regard analogous
interventions into the brain as ethically problematic (other things
equal; that is, not wrong simply because they are interventions into
the brain). What counts as analogous is, however, a difficult question,
and it is to this and related issues that the ceteris paribus (other
things being equal) qualification directs us. There are often differ-
ences between brain-based and external resources, and some of these
differences are ethically relevant. Most obviously, at our present
level of technology, external resources have a much higher degree of
fungibility than internal. Otto’s notebook is far more easily replaced
than his hippocampus (which, like his notebook, plays an important
the debate over the extended mind 61

nothing so much as prejudice, this is a prejudice that it is difficult to
overcome.Given the level of resistance I expect to encounter, it would
be foolish to require EPP (strong) to do much work in what follows.
Accordingly, I shall avoid invoking it. Instead, my arguments will
turn only on EPP (weak), a principle that does not antecedently
commit those who appeal to it to any particular view of the mind. In
assessing interventions into the mind, I shall persistently return to
the question whether there are actual or possible external interven-
tions that raise precisely the same problems as the new neu-
roscientific techniques. Of course, asking this question is simply the
first step in assessing new techniques. That there are, for instance,
existing techniques that are ethically analogous to new technologies
does not show that the new technologies are permissible. It might
show that neither is permissible: to that extent assessing the new
technologies affords us the opportunity to reassess the old. In any case,
in the absence of relevant moral differences between the cases, we
ought to treat them alike. The weak parity principle is important not
because it provides us with the answers to our ethical questions, but
because it allows us to set aside one bad, but tempting, response: that a
neuroscientific technology is especially bad (or especially good) just
because it is an intervention into the mind, narrowly construed. If, as
I shall claim, the parallels between new, internal, interventions and
older external techniques run deep, approaching neuroethics with the
weak parity principle in mind serves to deflate the hype surrounding
the new technologies, and encourages sober reflection on the ethical
issues that are genuinely significant.16
I will not often have cause to invoke the extended, or, indeed,
the embedded, mind hypothesis explicitly in what follows. Instead,
I shall go directly to consideration of relevant differences between
kinds of interventions. I take the very fact that EPP, in its weak form,
can do so much ethical work in guiding our deliberations about
permissibility without our invoking the extended or the embedded
mind hypothesis as itself indirect evidence for the thesis: were it not
true, it would not be able to guide us so reliably.
the debate over the extended mind 63

It is now time to turn directly to the neuroethical issues. New
neurotechnologies offer us unprecedented powers to alter brains, and
thereby thoughts; these technologies promise much, but for many
people they are also profoundly troubling. Are there good reasons to
find these technologies troubling, and if so, should we find some
environmental interventions equally troubling? Or are the horrified
responses to neurotechnologies a reflection of the prejudice that
would confine mind to skull?
End notes
1. Safire acknowledged that the term predates his use in discussion with two
leading neuroethicists, the neuroscientist Michael Gazzaniga and the
legal scholar Henry Greely, at the Library of Congress in May 2005.
A transcript of that discussion is available at http://www.nyas.org/snc/
readersReport.asp?articleID=34&page=2. Safire suggests, however,
that the earliest usage predated his by only a few months. In fact,
‘‘neuroethics,’’ and cognates like ‘‘neuroethical’’ and ‘‘neuroethicist’’
date back more than a decade earlier. Illes (2003) records uses, from the
scientific literature, from 1989 and 1991. I thank Morten Kringelbach for
correcting my misapprehension that the term originated with Safire.
2. Readers who have a passing familiarity with Steven Jay Gould’s theory of
punctuated equilibrium might dispute the claim that evolution is a story
of continuous change. Gould argues that evolution is characterized by
long periods of stasis interrupted by sudden change. But Gould’s claim
does not in fact conflict with the view I’m putting forward here: speciation
is sudden, in Gould’s view, only on the geological timescale upon which
evolution is normally measured.
3. On primate proto-morality, see de Waal (1996), and my own (2004).
4. In a later chapter, I shall argue that we cannot even take a great deal of
comfort from the fact that consciousness is clever. This cleverness is itself
a product of unconscious mechanisms: consciously processed problems
get better solutions because consciousness allows more and better
informed unconscious mechanisms to work on these problems.
5. More recent work suggests that Bechara et al. may have underestimated
subjects’ conscious knowledge of the payoff structure in the Iowa
Gambling Task. Maia and McClelland (2004) repeated the experiment on
introduction64

13. In correspondence, Weiskopf has argued that we ought to refuse the
status of beliefs to the extended states of an agent like Otto, because
there is no explanatory payoff in doing so. I find this reply mystifying.
Recall that the basic argument for the extended mind thesis is
functionalist: we ought to regard extended processes and stores as mental
because they play a similar role in guiding cognitive processes (that is,
roughly, processes that allow us to solve problems in navigating the
external world and in satisfying our needs and desires) as do internal
structures and processes. It is explanatory payoff that motivates
the entire project. Unless the opponent of the extended mind has the
resources to dispute the very many cases cited in support of the
functionalist claim, it seems foolhardy for them to contest the claim on
these grounds. Moreover, even with regard to Otto’s extended states it
seems that the extended mind has an explanatory payoff. We shall best
explain the inconsistencies in Otto’s behavior that might arise from a
lack of informational integration by attributing to him a quasi-belief
state, one that guides his behavior when it is on-line, but which goes
off-line whenever Otto is not accessing it.
14. Let me mention just one more objection sometimes voiced to the
extended mind thesis. Some philosophers take the conditions advanced
by Clark and Chalmers (1998) to be necessary for a resource to count
as mental, and have objected that we do not automatically endorse
external representations. Thus, they claim, there is a difference between
internal and external resources, sufficient to ensure that the only first
count as mental. Once again, I do not believe that the demonstration that
internal and external resources differ amounts to a refutation of the
thesis; however (once again) it is worth pointing out that it is more
difficult than the critics believe to demonstrate such differences. In fact,
there is ample evidence that endorsement of propositions, external as
well as internal, is the automatic default response: only subsequently – if
we have the time, the attentional resources and the motivation – do we
come to doubt them (Gilbert 1991; 1993).
15. Menary (2006) is similarly critical of the parity thesis.
16. David Chalmers has suggested that something like the weak parity
principle is all I need, and all I am entitled to. He argues that the
neuroethical issues can all be understood parsimoniously in terms of
differing causal routes whereby we can alter the mind, where ‘‘mind’’ is
the debate over the extended mind 67

conceived of traditionally. Pointing out, as I will do at length, that causal
routes that are largely external to the mind are not, in any ethically
relevant way, different from causal routes that are largely internal does
not require us to subscribe to the extended mind hypothesis (Gert-Jan
Lokhorst has offered a very similar criticism independently). I find this a
surprising line of argument. Its similarity to defences of internalism
against Chalmers himself should make us suspicious of it. We can, if we
like, reserve the term ‘‘mind’’ for processes, states and mechanisms that
are internal to the skull, but there is no explanatory gain in doing so;
insofar as we want to understand cognition, we shall be forced to take
into consideration the environmental scaffolding of cognition, as well
as internal resources. Similarly, if we want to understand what is
permissible and what impermissible, there is no explanatory gain in
understanding the bearer of value in the traditional manner. Here’s an
illustration: some philosophers argue that to the extent to which an
agent is reliant upon others, especially, but not only, cognitively, they are
non-autonomous (Weithman 2005). But if we take the embedding of the
agent in the social world seriously (and once we grasp the extent to which
the agent shrinks to an extensionless, and powerless, point if we take the
internalist view to its logical conclusion), this view loses its force. A
responsible doxastic agent trusts others, rather than relying upon their
own cognitive resources; similarly, a self-controlled agent allows the
environment, including the social environment, to carry much of the
load in controlling their behavior in the manner she wants (we shall
explore the issue of autonomy more fully in a later chapter). Just as
importantly, the extended mind hypothesis plays a significant heuristic
role, enabling us to recognize the degree to which the intuition that
external means of altering minds – means which impact on the mind by
way of impacting on the environment – are less problematic than
internal manipulations is the product of the prejudice that confines mind
to skull. Finally, though this is not the place to develop the view, the
extended mind hypothesis might play an important role in motivating an
externalist meta-ethics. As Rowlands (2003) has pointed out, if we
take externalism seriously, the traditional options for the location of
value – either entirely in the world, or entirely in the minds of agents – no
longer seem exhaustive. If mind extends into the world, then value can
depend upon mind without being located only in the skulls of agents.
introduction68

departures from the pure form of the traditional way of changing
minds – presenting reasons and arguments – are permissible. But
the degree to which psychoanalysis departs from tradition pales
into insignificance compared to the second way in which we
might go about changing someone’s mind (putting concerns about
extended cognition to one side for the moment, and assuming that
the mind is wholly bounded by the skull): by direct manipulation of
the brain.
Of course, there is a sense in which presenting evidence is a
kind of (indirect) manipulation of the brain – it alters connections
between neurons, and might contribute, in a very small way, to
changing the morphology of the brain (enough evidence, presented
over a long enough period of time, can produce alterations which are
large enough to be visible to the naked eye: a study of London taxi
drivers demonstrated that the posterior hippocampus, which stores
spatial information, was larger in drivers than in controls [Maguire
et al. 2000]). But direct manipulation of the brain differs from indirect
in an extremely significant way: whereas the presentation of evi-
dence and argument manipulates the brain via the rational capacities
of the mind, direct manipulation bypasses the agent’s rational
capacities altogether. It works directly on the neurons or on the larger
structures of the brain. There are many different kinds of direct
manipulation in use today: electroconvulsive therapy (ECT), in
which a current is passed through the brain, inducing seizures; psy-
chosurgery, which may involve the severing of connections in the
brain surgically, or may involve the creation of lesions using radio
frequency; transcranial magnetic stimulation of superficial struc-
tures of the brain; and deep brain stimulation.1 But the most widely
used kind of direct manipulation is, of course, pharmacological.
Many millions of people have taken one or another drug designed to
alter brain function: antipsychotics, lithium for manic-depression,
Ritalin for attention-deficit/hyperactivity disorder (ADHD) and,
most commonly, antidepressants. Here we shall focus, mainly, on
psychopharmacological treatment of depression; the issues raised by
changing our minds70

(Angell 2004). These are very serious objections to the use of
psychopharmaceuticals, but I set them aside here in order to focus
better on the philosophical issues, relatively narrowly construed.
The question I want to examine is not whether, given the current
state of the art, use of these drugs is advisable; rather I want to
concentrate upon in principle objections to direct manipulations. In
principle objections are objections that remain sound no matter how
much the technologies improve, and no matter what the political and
social context in which they are developed, prescribed and taken.
I therefore set aside these objections, and assume that the products
concerned are safe, or at least safe enough (inasmuch as their
expected benefits to patients outweigh their risks), and ignore con-
cerns about the industries that create and distribute them.
It might be objected that in setting these matters aside, we set
aside the ethics of neuroethics: the very heart and soul of the ques-
tions. There is some justice in this accusation: of course, it will be
necessary to factor these concerns back into the equation in coming
to an all-things-considered judgment of the advisability of using or
promoting these drugs in actual circumstances. But clarity demands
that we treat the issues raised by direct manipulation one by one, and
that requires isolating them from one another, not conflating them.
Moreover, in a book defending a conception of the mind as extended
and knowledge acquisition as distributed, it is not special pleading to
note that others – policy specialists, lawyers, sociologists and many
kinds of medical professionals – are better placed than philosophers
to analyze the issues set aside here. By focusing on the questions
where I can best contribute, I hope thereby to advance the entire
neuroethical agenda all the more effectively.
What remains, once we set these concerns aside? No doubt
some of the opposition to psychopharmacology that remains is
simply irrational. In part, it seems to be an expression of a deep-
seated prejudice against technologies regarded as ‘‘unnatural.’’ When
we investigate the roots of this kind of objection, we frequently find
that by ‘‘unnatural’’ people tend to mean no more than ‘‘unfamiliar’’
changing our minds72

society makes available. Instead, they look for and actively create
their own way, by reference to who they, truly and deeply, are.
Authenticity is a modern ideal. It could not exist in premodern
societies, in which social roles were relatively few, and people had
little freedom to move between them. A medieval peasant had few
options available to him or her, and little leeway to choose between
them. Authenticity requires the growth of cities, and the consequent
decrease in the social surveillance and mutual policing characteristic
of village life. In the anonymity of the city, people were free to
remake themselves. They could, if they wished, break free (at least
somewhat) from the expectations of their family, their church, their
friends and even of social conventions, and remake themselves in
their own image.2
Authenticity, the search for a way of life that is truly one’s
own, has gradually gone from a mere possibility to an ideal. Today,
most of us feel stung by the charge of inauthenticity. Conformism,
going along with the crowd, the herd mentality – all of these are, if
not quite vices, at least imperfections to be avoided. Of course, we
may authentically choose to do what most everyone else is doing, but
to choose it authentically is to choose it because it is right for us, and
not because it is what everyone else does.
Authenticity, as Charles Taylor (1991) has argued, exists in an
unstable tension with other ideals and standards of a good life. It can
easily tip over into selfishness and a shallow form of individualism. It
can lead us to overlook the fact that values are sustained socially, and
that each of us must forge our own way of life in an ongoing dialogue
with others: with those close to us, and with the ideals and standards
of our culture. Nevertheless, though it is an ideal that becomes self-
defeating if it is taken too far, it is unrepudiable by us moderns.
Authenticity is so deeply woven into our cultures and our values that
few, if any, of us can simply turn our backs upon it. It is true that
some of us choose to embrace, or to remain in, ways of life that are in
some ways antithetical to the ideal of authenticity – we join mon-
asteries, or adhere to religions that regulate every aspect of our lives,
changing our minds74

even to the point of deciding who we shall marry and what careers (if
any) we shall have. But even when we embrace ways of life that
require us to cede control of our significant choices to others, we
often justify our decision in ways that invoke authenticity: we find
this way of life personally fulfilling; it is, after all, our way of being
ourselves.
It is easy to see why the use of direct manipulations of the
mind might be thought antithetical to the ideal of authenticity. To
be authentic is to find one’s way of life and one’s values within; it is
to make one’s entire life an expression of who one truly is. But
antidepressants, psychosurgery and the other technologies of direct
manipulation introduce an alien element into the equation: after
treatment with these technologies, I am no longer the person I was.
Either I have changed, as a result of this intrusion from outside me,
or (less radically) who I really am has been covered over by the foreign
element. This cheerfulness, this sunny disposition, this is not really
me; it is the antidepressants. As Carl Elliott, the bioethicist who has
insisted most forcefully and persuasively on the problem of inau-
thenticity puts it:
It would be worrying if Prozac altered my personality, even if
it gave me a better personality, simply because it isn’t my
personality. This kind of personality change seems to defy an
ethics of authenticity.
(Elliott 1998: 182)
To this extent, psychotherapy is preferable to direct manip-
ulation. Psychotherapy explores my self, my inner depths. It seeks
coherence and equilibrium between my inner states, and between
my inner states and the world. But direct manipulation simply
imposes itself over my self.
One might illuminatingly compare the effects of direct
manipulation in treating a mental illness to the effects of more
familiar direct manipulations of the mind: drugs consumed for
recreational purposes. When someone behaves aggressively, or breaks
authenticity 75

unable to feel pain are at much greater risk of serious injury or death
because they lack these generally reliable signals of somatic pro-
blems. For this reason, too much analgesic, or too effective an
analgesic, is risky, leaving those who take it similarly exposed. But
mightn’t direct manipulations such as antidepressants constitute
just such an analgesic for emotional pain? Emotional pain, like
somatic pain, can be a signal that something is wrong: that our
choices have not been wise, and that we ought not to repeat them,
that our relationships need work, that the way we are acting toward
others is inappropriate. Blocking these signals prevents us from
learning from our failures, and therefore removes one of our most
important opportunities for personal growth. Direct manipulation, if
used to medicalize normal suffering, turns opportunities for growth
into symptoms to be eliminated (Manninen 2006).
mechanization of the self
A further reason to favor traditional means of changing minds over
direct manipulations is closely connected to the forgoing. Bypassing
our rational capacities in order to change our minds might carry a
cost that is potentially far greater than merely passing up the
opportunity to gain self-knowledge: it risks the very existence of our
self, as it has traditionally been understood. The self is, or has as an
essential component, the capacity to respond to reasons. We are not
machines, but living creatures, with rich inner lives constituted by
our emotional and cognitive responses to our environment and to the
people within it. When we treat ourselves as if we were machines, by
modifying our brains, our emotions and our cognitive processes using
direct means, we risk everything that makes us more than mere
machines. If we treat ourselves as if we did not live in a space
of reasons, we risk making it true, and that is to risk, literally,
everything.
One way to understand the claim being made here is by analogy
with the kind of disrespect we show others when we treat their
emotional and cognitive responses as mere symptoms. Think, for
changing our minds78

instance, of one all-too-real response to feminists, especially in the
early days of the movement. Critics rejected feminist arguments, but
not on the grounds that they were false. Rather than engage with
them as arguments, they were wont to reject them as hysterical; as
expressions of psychological maladjustment. If I reject your argu-
ments because I think they are false, and especially if I give reasons
why I reject them, I manifest respect for you as a rational being. But if
I reject your arguments without even engaging with them, because
I see them as expressions of your psychology and not as rational
responses to the world as you perceive it, I treat you with profound
disrespect. I treat you as an object to be managed, not as a person to
engage with. More generally, when we treat mental states, of our-
selves or of others, as objects to be manipulated, we treat ourselves as
things (Freedman 1998).
What’s wrong with treating ourselves as things? Well, it may be
that being a self requires being treated, and treating ourself, as a self.
Human identity is relatively fragile. It can be disrupted or degraded
by inappropriate treatment. We are self-interpreting animals, Freed-
man argues (following Charles Taylor), and the terms of our inter-
pretations profoundly affect the kinds of creature we are. How we
understand ourselves, the kinds of stories we tell ourselves about
ourselves, actually affect the kinds of beings we are. So we must be
careful to avoid the kinds of mechanistic explanations of our actions
and thoughts that are appropriate to mere machines. We can appro-
priately explain why a computer responded in a certain way by
referring to its internal states in mechanical terms, but treating each
other as selves involves treating our responses as reflecting assess-
ments of situations, not as mere mechanical processes. The com-
puter responded like that because there is a bug in its program (or a
malfunction in its coprocessor, or a power failure), but the person
responded as they did because they thought he was rude (or whatever
the case might be).4
The problem with Prozac, then (where ‘‘Prozac’’ refers to any
chemical means of treating mental problems) is that it treats people
mechanization of the self 79

responses are adaptive (Watson and Andrews 2002): withdrawing
allows one time to recuperate from the sudden decline in status, and
from any wounds that may have been incurred in fights over hier-
archy. Decreased serotonin is also associated with rumination and
with hypervigilance, which are likely also adaptive responses:
rumination ensures that all one’s intellectual resources are con-
centrated on the problem of how to ensure that the decline in status
is reversed or at least halted, and hypervigilance ensures that one is
able successfully to navigate the dangerous environment of the low-
status animal. In short, a fall in serotonin levels, and subsequent
sadness, is not a disease. It is a rational response to external events.
Today serotonin levels are more likely to fall in response to
various kinds of life events: losing one’s job, or getting divorced.
Nevertheless, it remains plausible to think of the response as adap-
tive and even as rational. It is normal to feel depressed in response to
such events. Indeed, one might go further: failure to feel depressed
in response to these life events would itself be pathological. These
events are depressing, and if one simply brushes them aside and
carries on cheerfully, one fails to appreciate their proper significance.
The person who shrugs off the failure of a marriage, for instance, fails
to appreciate the importance that intimate relationships have in a
properly human life, which is a reliable indication that they fail to
live a properly human life. The proper response to intimate rela-
tionship breakdown, the only response that takes the measure of its
true significance, is deep sadness; a sadness which, moreover, per-
sists for some extended period of time (think, for example, of how we
look askance at someone who remarries very soon after divorce or
the death of a spouse). Similarly, the appropriate response to the loss
of a job might well be depression, as the unemployed person con-
templates the struggle to pay the mortgage or the rent, to feed their
family or to forge a new place for themselves within the social world.
None of this is to say that depression may not sometimes be
a pathological response to life events. Depression is pathological
when it is deeper and far more prolonged than life events warrant.
changing our minds82

own way, and unable to achieve the full range of goods constitutive
of a flourishing life.
Carl Elliott (1998) has developed this theme in a way that links
it closely with the problem of authenticity. Authenticity can be
contrasted with alienation; a state in which we feel estranged from
our social world and cut off from sources of meaning. In seeking
authenticity, we normally seek a place in the world that is fulfilling
for us, and which satisfies our deepest needs as human beings. But
(and here I go beyond what Elliott himself says), authenticity-as-
fulfilment understood in this way has social preconditions. Only in
social environments which are conducive to meaningful lives can
authenticity-as-fulfilment be achieved. In less favourable environ-
ments, authenticity will be a less happy – literally and figuratively –
affair. In some environments, authenticity consists not in fulfilment,
but in the clear-headed acknowledgement that we confront serious
and perhaps insoluble problems.
Authenticity must not be confused with happiness. The
authentic individual is not necessarily happy, and the happy indivi-
dual is not necessarily authentic. Elliott asks us to consider the
example of the stereotypical conformist, an accountant living in the
American suburbs. Suppose that such a person presents to a physician
suffering from depression. Should the physician treat him with anti-
depressants (or, indeed, with any other direct manipulations)? The
case for treatment is simple: the man is suffering, and the physician
has it within their power to relieve that suffering. But it is far from
clear that in treating the patient’s suffering, the physician actually
makes them better off. Elliott contrasts two possible (inauthentic)
suburban accountants (the qualification is necessary, since it is at
least possible that someone could be an authentic suburban accoun-
tant). Both live a life of dull conformity to social norms which they
have accepted unthinkingly, neither lives according to a conception of
the good life they have reflectively endorsed. Both suffer a growing,
gnawing, malaise, as their opportunities for a life endorsed from the
inside slip away. But suppose that one of them now seeks professional
changing our minds84

treatment for their depression, and is prescribed Prozac. They respond
well to the treatment, and their mood lifts. They are more energetic at
work, and happier at home. But are they really better off than their
counterpart across the road, whose despondency is untreated?
The point here is subtly different to the objection that in
covering over our pain, Prozac obstructs a path to useful self-
knowledge – knowledge that can enable us to achieve a better life and
more satisfying relationships. The claim Elliott makes is that inde-
pendently of its instrumental value, a clear-eyed appreciation of
inauthenticity or other malaises is valuable. Even if there is no
practical means available to the suburban accountant to achieve
authenticity, even if seeing the truth comes at the cost of permanent
and irredeemable psychic pain for him, he is better off for suffering
that pain than his medicated counterpart.
There is more, far more, to a properly human life than mere
happiness, as a little reflection on some possible ways in which
humans can achieve happiness demonstrates. Most people fear the
prospect of dementia, sometimes as much as or even more than
death. They dread the prospect of having their mind, their memory
and their very selfhood slipping away. But if happiness were all that
mattered, this fear would be irrational. Dementia is not incompatible
with happiness; though some demented individuals experience a
great deal of frustration and unhappiness, some seem quite content.
Our fear of dementia therefore seems to be due to our recognition
that well-being consists in more than just feelings of contentment.
Similarly, few of us believe that the life of the drug addict would be
a good life, even if we could ensure a steady supply of high-quality
drugs and suffered no adverse health effects (both of which are rea-
sonable assumptions, for addicts who are wealthy). We think that a
good life is a life not just of happy experience, but of contact with
reality (Nozick 1974). Happiness, if it comes, is certainly to be wel-
comed, and perhaps it ought to be sought. But it is not to be sought at
the expense of contact with reality. The happiness we seek is the
happiness which comes through the achievement of intrinsically
treating symptoms and not causes 85