D.M. Armstrong - Nature of Mind-Harvester Press (1981)

T~e
N~lure oF
Mind
Te
Ma ureof
Mind
TLe Harve1~er Prell

First published in Great Britain in 1981 by
THE HARVESTER PRESS LIMITED
Publishers: john Spiers and Margaret A. Boden

16 Ship Street, Brighton, Sussex
© University of Queensland Press, 1980
British Library Cataloguing in Publication Data
Armstrong, D.M.
The nature of mind
-(Harvester studies in philosophy).
1. Intellect
2. Knowledge, Theory of
I. Title
128'.2'08 BF161
ISBN 0-7108-0027-4
Manufactured in Hong Kong by Silex Enterprise & Printing Co.
All rights reserved

To the staff, secretaries
and students of the
Department of Traditional
and Modern Philosophy,
University of Sydney.
Introduction ix
Acknowledgements xi
1 The Nature of Mind 1

2 The Causal Theory of the Mind 16
3 Epistemological Foundations for a Materialist Theory of the
Mind 32
4 What is Consciousness? 55
5 Acting and Trying 68
6 The Nature of Tradition 89
7 Colour Realism and the Argument from Microscopes 104
8 Immediate Perception 119
9 Perception, Sense-data and Causality 132
10 Naturalism, Materialism and First Philosophy 149
References 166
Bibliography: Writings of D.M. Armstrong up to and including
1979 169
Index 173
ln~roduc~ion
These papers deal with the philosophy of mind, the philosophy of

action and the philosophy of perception. They explain, develop
and correct some of the things said in my book A Materialist
Theory of the Mind (1968), but they may be read independently.
The first two papers ("The Nature of Mind" and "The Causal
Theory of the Mind") are general introductions to my philosophy
of mind, but published eleven years apart. The third and somewhat
more technical paper, "Epistemological Foundations", tries to
give reasons why we should be inclined to accept a Materialist or
Physicalist theory of the mind. The second section of this paper
argues that, when philosophers try to articulate broad general
views of the nature of the world, they should look for guidance to
the general findings of science, together with (bed-rock) common
sense, rather than to the procedures of philosophy itself.
The notion of consciousness is a notoriously puzzling one. It has
often been thought to present a special difficulty for Materialism.
In the final section of the fourth paper, "What is Consciousness?",
I try to show why consciousness seems to us to be so special a
phenomenon.
The next two papers are concerned with the theory of action. In
"Acting and Trying", I try to improve upon the theory of the will
advanced in A Materialist Theory. In particular, I try to show that
all intentional action involves trying so to act. My view has been
criticized by Donald Davidson (1973), and in the course of
developing my argument, I defend myself against his criticisms.
David Lewis has recently shown in his book Convention (1969)
X introduction
that certain social concepts can be analysed in terms drawn from

the theory of action. Notions such as tradition are less complex
than the very difficult notions, such as that of convention,
apparently successfully analysed by Lewis, but might be thought
to be too vague and sloppy to be given precise treatment. In "The
Nature of Tradition", I try to show that this is not so.
Any satisfactory philosophy of mind must involve a satisfactory
theory of perception, but the latter is notoriously difficult to
achieve. Three papers in this collection supplement the Direct
Realist, "information-flow", theory advanced in A Materialist
Theory. Colours provide notorious problems. In "Colour Realism
and the Argument from Microscopes", it is argued that the
changes in colour perception that occur when we look at a col-
oured surface under a microscope do not, as they are often sup-
posed to do, force us to deny that colours are "out there". The
paper "Immediate Perception" is concerned with the ever-puzzling
notion of immediate or direct perception. I try to bring out a new
casual dimension in the notion. The great rival to Direct Realist
theories of perception are Representative theories. Most of the tra-
ditional criticisms of the latter (including criticism that I have pre-
viously endorsed) seem to be unsound. In "Perception, Sense-data
and Causality", I try to develop a new objection to the Representa-
tive theory. The causal relations between perceiver and object per-
ceived are once again the object of scrutiny.
I have placed at the end a paper, "Naturalism, Materialism and
First Philosophy", that is not primarily concerned with philosophi-
cal psychology. However, it takes up some of the themes in the
other papers and points the way ahead to the metaphysical or
ontological concerns that are the present focus of my philosophical
investigations.
"The Nature of Mind" was originally printed in Arts, the Proceed-
ings of the Sydney University Arts Association (1966). "The
Causal Theory of the Mind" appeared in Neue Hefte fur
Philosophie, Heft 11, (Gottingen: Vandenhoeck & Ruprecht,
1977) and is reprinted by permission of the editors and publisher.
"Epistemological Foundations for a Materialist Theory of the
Mind" appeared in Philosophy of Science (1973) and is reprinted
by permission of the Philosophy of Science Association. A version
of "What is Consciousness?" appeared in the Proceedings of the
Russel/ian Society (Sydney, 1978). "Acting and Trying" appeared
in Philosophical Papers (1973). A new section has been added to
this paper, based upon "Beliefs and Desires as Causes of Action",
which appeared in Philosophical Papers (1975). The two papers
are reprinted by permission of the editor. "The Nature of Tradi-
tion" appeared in Liberty and Politics, edited by 0. Harries
(Sydney: Workers' Educational Association of N.S.W. and
Pergamon Press (Australia), 197 6) and is reprinted by permission
of the publishers. "Colour Realism and the Argument from
Microscopes" appeared in Contemporary Philosophy in Australia,
edited by R. Brown and C.D. Rollins, (London: Allen & Unwin,
1969). It is reprinted by permission of the publisher. "Immediate
Perception" appeared in Essays in Memory of Imre Lakatos,
Boston Studies in the Philosophy of Science, edited by R.S.
Cohen, P.K. Feyerabend and W.M. Wartofsky (Dordrecht-Holland:
D. Reidel, 1976) and is reprinted by permission of the publisher.
"Perception, Sense-data and Causality" appeared in Perception
xii acknowledgements
and Identity: Essays presented to A.J. Ayer with his replies to

them, edited by G.F. Macdonald (London: Macmillan, 1979) and
is reprinted by permission of the publisher. "Naturalism, Material-
ism and First Philosophy" appeared in 1st Systematische
Phi/osophie Mog/ich ?, Proceedings of the Stuttgarter Hegel-
Kongress, 1975 (Bonn: Herbert Grundmann, 1977) and also in
Philosophia, (1978) and is reprinted by permission of the editors
and publishers.
In all cases, minor revisions have been made to the papers.
1
T~e Nolure of Mind
Men have minds, that is to say, they perceive, they have sensa-
tions, emotions, beliefs, thoughts, purposes and desires. What is it
to have a mind? What is it to perceive, to feel emotion, to hold a
belief or to have a purpose? Many contemporary philosophers
think that the best clue we have to the nature of mind is furnished
by the discoveries and hypotheses of modern science concerning
the nature of man.
What does modern science have to say about the nature of
man? There are, of course, all sorts of disagreements and
divergencies in the views of individual scientists. But I think it is
true to say that one view is steadily gaining ground, so that it bids
fair to become established scientific doctrine. This is the view that
we can give a complete account of man in purely physico-chemical
terms. This view has received a tremendous impetus in recent
decades from the new subject of molecular biology, a subject that
promises to unravel the physical and chemical mechanisms that lie
at the basis of life. Before that time, it received great encourage-
ment from pioneering work in neurophysiology pointing to the
likelihood of a purely electro-chemical account of the working of
the brain. I think it is fair to say that those scientists who still reject
the physico-chemical account of man do so primarily for
philosophical, or moral or religious reasons, and only secondarily,
and half-heartedly, for reasons of scientific detail. This is not to say
that in the future new evidence and new problems may not come
to light that will force science to reconsider the physico-chemical
view of man. But at present the drift of scientific thought is clearly
2 The Nature of Mind
set towards the physico-chemical hypothesis. And we have

nothing better to go on than the present.
For me, then, and for many philosophers who think like me, the
moral is clear. We must try to work out an account of the nature of
mind which is compatible with the view that man is nothing but a
physcio-chemical mechanism.
And in this paper, I shall be concerned to do just this: to sketch
(in barest outline) what may be called a Materialist or Physicalist
account of the mind.
THE AUTHORITY OF SCIENCE
But before doing this, I should like to go back and consider a cri-
ticism of my position that must inevitably occur to some. What
reason have I, it may be asked, for taking my stand on science?
Even granting that I am right about what is the currently dominant
scientific view of man, why should we concede science a special
authority to decide questions about the nature of man? What of
the authority of philosophy, of religion, of morality, or even of
literature and art? Why do I set the authority of science above all
these? Why this "scientism"?
It seems to me that the answer to this question is very simple. If
we consider the search for truth, in all its fields, we find that it is
only in science that men versed in their subject can, after
investigation that is more or less prolonged, and which may in
some cases extend beyond a single human lifetime, reach substan-
tial agreement about what is the case. It is only as a result of scien-
tific investigation that we ever seem to reach an intellectual con-
sensus about controversial matters.
In the Epistle Dedicatory to De Corpore, Hobbes wrote of
William Harvey, the discoverer of the circulation of the blood, that
he was: "the only man I know, that conquering envy, hath estab-
lished a new doctrine in his life-time."
Before Copernicus, Galileo and Harvey, Hobbes remarks:
"there was nothing certain in natural philosophy." And we might
add, with the exception of mathematics, there was nothing certain
in any other learned discipline.
These remarks of Hobbes are incredibly revealing. They show
The Nature of Mind 3
us what a watershed in the intellectual history of the human race

the seventeenth century was. Before that time, enquiry proceeded,
as it were, in the dark. Men could not hope to see their doctrine es-
tablished, that is to say, accepted by the vast majority of those
properly versed in the subject under discussion. There was no
intellectual consensus. Since that time, it has become a com-
monplace to see new doctrines, sometimes of the most far-reach-
ing kind, established to the satisfaction of the learned, often within
the lifetime of their first proponents. Science has provided us with
a method of deciding disputed questions. This is not to say, of
course, that the consensus of those who are learned and competent
in a subject cannot be mistaken. Of course such a consensus can be
mistaken. Sometimes it has been mistaken. But, granting
fallibility, what better authority have we than such a consensus?
Now this is of the utmost importance. For in philosophy, in
religion, in such disciplines as literary criticism, in moral questions
in so far as they are thought to be matters of truth and falsity, there
has been a notable failure to achieve an intellectual consensus
about disputed questions among the learned. Must we not then
attach a peculiar authority to the discipline that can achieve a con-
sensus? And if it presents us with a certain vision of the nature of
man, is this not a powerful reason for accepting that vision?
I will not take up here the deeper question why it is that the
methods of science have enabled us to achieve an intellectual con-
sensus about so many disputed matters. That question, I think,
could receive no brief or uncontroversial answer. I am resting my
argument on the simple fact that, as a result of scientific investiga-
tion, such a consensus has been achieved.
It may be replied - it often is replied - that while science is all
very well in its own sphere - the sphere of the physical, perhaps
- there are matters of fact on which it is not competent to pro-
nounce. And among such matters, it may be claimed, is the ques-
tion: what is the whole nature of man? But I cannot see that this
reply has much force. Science has provided us with an island of
truths, or, perhaps one should say, a raft of truths, to bear us up on
the sea of our disputatious ignorance. There may have to be revi-
sions and refinements, new results may set old findings in a new
perspective, but what science has given us will not be altogether
superseded. Must we not therefore appeal to these relative certain-

ties for guidance when we come to consider uncertainties
elsewhere? Perhaps science cannot help us to decide whether or
not there is a God, whether or not human beings have immortal
souls, or whether or not the will is free. But if science cannot assist
us, what can? I conclude that it is the scientific vision of man, and
not the philosophical or religious or artistic or moral vision of
man, that is the best clue we have to the nature of man. And it is
rational to argue from the best evidence we have. 1
DEFINING THE MENTAL

Having in this way attempted to justify my procedure, I turn back
to my subject: the attempt to work out an account of mind, or, if
you prefer, of mental process, within the framework of the
physico-chemical, or, as we may call it, the Materialist view of
man.
Now there is one account of mental process that is at once
attractive to any philosopher sympathetic to a Materialist view of
man: this is Behaviourism. Formulated originally by a psy-
chologist, J.B. Watson, it attracted widespread interest and con-
siderable support from scientifically oriented philosophers. Tradi-
tional philosophy had tended to think of the mind as a rather
mysterious inward arena that lay behind, and was responsible for,
the outward or physical behaviour of our bodies. Descartes
thought of this inner arena as a spiritual substance, and it was this
conception of the mind as spiritual object that Gilbert Ryle
attacked, apparently in the interest of Behaviourism, in his impor-
tant book The Concept of Mind (1949). He ridiculed the Cartesian
view as the dogma of "the ghost in the machine". The mind was
not something behind the behaviour of the body, it was simply
part of that physical behaviour. My anger with you is not some
modification of a spiritual substance that somehow brings about
aggressive behaviour; rather it is the aggressive behaviour itself;
1. The view of science presented here has been challenged in recent years by
new Irrationalist philosophies of science. See, in particular, Thomas Kuhn
(1962) and Paul Feyerabend (1975). A complete treatment of the problem
would involve answering their contentions.
my addressing strong words to you, striking you, turning my back

on you, and so on. Thought is not an inner process that lies
behind, and brings about, the words I speak and write: it is my
speaking and writing. The mind is not an inner arena, it is outward
act.
It is clear that such a view of mind fits in very well with a com-
pletely Materialistic or Physicalist view of man. If there is no need
to draw a distinction between mental processes and their expres-
sion in physical behaviour, but if instead the mental processes are
identified with their so-called "expressions", then the existence of
mind stands in no conflict with the view that man is nothing but a
physico-chemical mechanism.
However, the version of Behaviourism that I have just sketched
is a very crude version, and its crudity lays it open to obvious
objections. One obvious difficulty is that it is our common
experience that there can be mental processes going on although
there is no behaviour occurring that could possibly be treated as
expressions of those processes. A man may be angry, but give no
bodily sign; he may think, but say or do nothing at all.
In my view, the most plausible attempt to refine Behaviourism
with a view to meeting this objection was made by introducing the
notion of a disposition to behave. (Dispositions to behave play a
particularly important part in Ryle's account of the mind.) Let us
consider the general notion of disposition first. Brittleness is a dis-
position, a disposition possessed by materials like glass. Brittle
materials are those that, when subjected to relatively small forces,
break or shatter easily. But breaking and shattering easily is not
brittleness, rather it is the manifestation of brittleness. Brittleness
itself is the tendency or liability of the material to break or shatter
easily. A piece of glass may never shatter or break throughout its
whole history, but it is still the case that it is brittle: it is liable to
shatter or break if dropped quite a small way or hit quite lightly.
Now a disposition to behave is simply a tendency or liability of a
person to behave in a certain way under certain circumstances.
The brittleness of glass is a disposition that the glass retains
throughout its history, but clearly there also could be dispositions
that come and go. The dispositions to behave that are of interest to
the Behaviourist are, for the most part, of this temporary

character.
Now how did Ryle and others use the notion of a disposition to
behave to meet the obvious objection to Behaviourism that there
can be mental process going on although the subject is engaging in
no relevant behaviour? Their strategy was to argue that in such
cases, although the subject was not behaving in any relevant way,
he or she was disposed to behave in some relevant way. The glass
does not shatter, but it is still brittle. The man does not behave,
but he does have a disposition to behave. We can say he thinks
although he does not speak or act because at that time he was dis-
posed to speak or act in a certain way. If he had been asked,
perhaps, he would have spoken or acted. We can say he is angry
although he does not behave angrily, because he is disposed so to
behave. If only one more word had been addressed to him, he
would have burst out. And so on. In this way it was hoped that
Behaviourism could be squared with the obvious facts.
It is very important to see just how these thinkers conceived of
dispositions. I quote from Ryle:
To possess a dispositional property is not to be in a particular
state, or to undergo a particular change; it is to be bound or lia-
ble to be in a particular state, or to undergo a particular change,
when a particular condition is realized. 2
So to explain the breaking of a lightly struck glass on a particular
occasion by saying it was brittle is, on this view of dispositions,
simply to say that the glass broke because it is the sort of thing that
regularly breaks when quite lightly struck. The breaking was the
normal behaviour, or not abnormal behaviour, of such a thing. The
brittleness is not to be conceived of as a cause for the breakage, or
even, more vaguely, a /actor in bringing about the breaking. Brit-
tleness is just the fact that things of that sort break easily.
But although in this way the Behaviourists did something to
deal with the objection that mental processes can occur in the
absence of behaviour, it seems clear, now that the shouting and the
dust have died, that they did not do enough. When I think, but my
thoughts do not issue in any action, it seems as obvious as any-
2. Ryle, 1949: 43; emphasis added.

thing is obvious that there is something actually going on in me

that constitutes my thought. It is not simply that I would speak or
act if some conditions that are unfulfilled were to be fulfilled.
Something is currently going on, in the strongest and most literal
sense of "going on", and this something is my thought. Rylean
Behaviourism denies this, and so it is unsatisfactory as a theory of
mind. Yet I know of no version of Behaviourism that is more
satisfactory. The moral for those of us who wish to take a purely
physicalistic view of man is that we must look for some other
account of the nature of mind and of mental processes.
But perhaps we need not grieve too deeply about the failure of
Behaviourism to produce a satisfactory theory of mind. Behaviour-
ism is a profoundly unnatural account of mental processes. If
somebody speaks and acts in certain ways, it is natural to speak of
this speech and action as the expression of his thought. It is not at
all natural to speak of his speech and action as identical with his
thought. We naturally think of the thought as something quite dis-
tinct from the speech and action that, under suitable circums-
tances, brings the speech and action about. Thoughts are not to be
identified with behaviour, we think; they lie behind behaviour. A
man's behaviour constitutes the reason we have for attributing
certain mental processes to him, but the behaviour cannot be iden-
tified with the mental processes.
This suggests a very interesting line of thought about the mind.
Behaviourism is certainly wrong, but perhaps it is not altogether
wrong. Perhaps the Behaviourists are wrong in identifying the
mind and mental occurrences with behaviour, but perhaps they are
right in thinking that our notion of a mind and of individual mental
states is logically tied to behaviour. For perhaps what we mean by
a mental state is some state of the person that, under suitable cir-
cumstances, brings about a certain range of behaviour. Perhaps
mind can be defined not as behaviour, but rather as the inner
cause of certain behaviour. Thought is not speech under suitable
circumstances, rather it is something within the person that, in
suitable circumstances, brings about speech. And, in fact, I believe
that this is the true account, or, at any rate, a true first account, of
what we mean by a mental state.
How does this line of thought link up with a purely Physicalist
view of man? The position is that while it does not make such a
Physicalist view inevitable, it does make it possible. It does not
entail, but it is compatible with, a purely Physicalist view of man.
For if our notion of the mind and of mental states is nothing but
that of a cause within the person of certain ranges of behaviour,
then it becomes a scientific question, and not a question of logical
analysis, what in fact the intrinsic nature of that cause is. The
cause might be, as Descartes thought it was, a spiritual substance
working through the pineal gland to produce the complex bodily
behaviour of which men ar~. capable. It might be breath, or
specially smooth and mobile atoms dispersed throughout the
body; it might be many other things. But in fact the verdict of
modern science.seems to be that the sole cause of mind-betoken-
ing behaviour in man aryd th~ higher animals is the physico-
chemical workings of the central nervous system. And so, assum-
ing we have correctly characterized our concept of a mental state
as nothing but the cause of certain sorts of behaviour, then we can
identify these mental states with purely physical states of the
central nervous system.
At this point we may stop and go back to the Behaviourist's dis-
positions. We saw that, according to him, the brittleness of glass
or, to take another example, the elasticity of rubber, is not a state
of the glass or the rubber, but is simply the fact that things of that
sort behave in the way they do. But now Jet us consider how a
scientist would think about brittleness or elasticity. Faced with the
phenomenon of breakage under relatively small impacts, or the
phenomenon of stretching when a force is applied followed by
contr9-ction when the force is removed, he will assume that there is
some current state of the glass or the rubber that is responsible for
the characteristic behaviour of samples of these two materials. At
the beginning, he will not know what this state is, but he will
endeavour to find out, and he may succeed in finding out. And
when he has found out, he will very likely make remarks of. this
sort: "We have discovered that the brittleness of glass is in fact a
certain sort of pattern in the molecules of the glass." That is to say,
he will identify brittleness with the state of the glass that is respon-
sible for the liability of the glass to break. For him, a disposition of
an object is a state of the object. What makes the state a state of
brittleness is the fact that it gives rise to the characteristic

manifestations of brittleness. But the disposition itself is distinct
from its manifestations: it is the state of the glass that gives rise to
these manifestations in suitable circumstances.
This way of looking at dispositions is very different from that of
Ryle and the Behaviourists. The great difference is this: If we treat
dispositions as actual states, as I have suggested that scientists do,
even if states the intrinsic nature of which may yet have to be dis-
covered, then we can say that dispositions are actual causes, or
causal factors, which, in suitable circumstances, actually bring
about those happenings that are the manifestations of the disposi-
tion. A certain molecular constitution of glass that constitutes its
brittleness is actually responsible for the fact that, when the glass
is struck, it breaks.
Now I cannot argue the matter here, because the detail of the
argument is technical and difficult, but I believe that the view of
dispositions as states, which is the view that is natural to science, is
the correct one. 3 I believe it can be shown quite strictly that, to the
extent that we admit the notion of dispositions at all, we are com-
mitted to the view that they are actual states of the object that has
the disposition. I may add that I think that the same holds for the
closely connected notions of capacities and powers. Here I will
simply have to assume this step in my argument.
But perhaps it will be seen that the rejection of the idea that
mind is simply a certain range of man's behaviour in favour of the
view that mind is rather the inner cause of that range of man's
behaviour, is bound up with the rejection of the Rylean view of
dispositions in favour of one that treats dispositions as states of
objects and so as having actual causal power. The Behaviourists
were wrong to identify the mind with behaviour. They were not so
far off the mark when they tried to deal with cases where mental
happenings occur in the absence of behaviour by saying that these
are dispositions to behave. But in order to reach a correct view, I
am suggesting, they would have to conceive of these dispositions
as actual states of the person who has the disposition, states that
have actual causal power to bring about behaviour in suitable cir-
3. I develop the argument in Belief, Truth and Knowledge (1973), ch. 2, sect. 2.
cumstances. But to do this is to abandon the central inspiration of

Behaviourism: that in talking about the mind we do not have to go
behind outward behaviour to inner states.
And so two separate but interlocking lines of thought have
pushed me in the same direction. The first line of thought is that it
goes profoundly against the grain to think of the mind as
behaviour. The mind is, rather, that which stands behind and
brings about our complex behaviour. The second line of thought is
that the Behaviourist's dispositions, properly conceived, are really
states that underlie behaviour and, under suitable circumstances,
bring about behaviour. Putting these two together, we reach the
conception of a mental state as a state of the person apt for pro-
ducing certain ranges of behaviour. This formula: a mental state is
a state of the person apt for producing certain ranges of behaviour,
I believe to be a very illuminating way of looking at the concept of
a mental state. I have found it fruitful in the search for detailed
logical analyses of the individual mental concepts.
I do not think that Hegel's Dialectic has much to tell us about
the nature of reality. But I think that human thought often moves
in a dialectical way, from thesis to antithesis and then to the syn-
thesis. Perhaps thought about the mind is a case in point. I have
already said that classical philosophy has tended to think of the
mind as an inner arena of some sort. This we may call the Thesis.
Behaviourism moves to the opposite extreme: the mind is seen as
outward behaviour. This is the Antithesis. My proposed Synthesis
is that the mind is properly conceived as an inner principle, but a
principle that is identified in terms of the outward behaviour it is
apt for bringing about. This way of looking at the mind and mental
states does not itself entail a Materialist or Physicalist view of man,
for nothing is said in this anlaysis about the intrinsic nature of
these mental states. But if we have, as I have argued that we do
have, general scientific grounds for thinking that man is nothing
but a physical mechanism, we can go on to argue that the mental
states are in fact nothing but physical states of the central nervous
system.
THE PROBLEM OF CONSCIOUSNESS

Along these lines, then, I would look for an account of the mind
that is compatible with a purely Materialist theory of man. There
are, as may be imagined, all sorts of powerful objections that can
be made to my view. But in the rest of this paper, I propose to do
only one thing: I will develop one very important objection to my
view of the mind -an objection felt by many'philosophers - and
then try to show how the objection should be met.
The view that our notion of mind is nothing but that of an inner
principle apt for bringing about certain sorts of behaviour may be
thought to share a certain weakness with Behaviourism. Modern
philosophers have put the point about Behaviourism by saying
that, although Behaviourism may be a satisfactory account of the
mind from an other-person point of view, it will not do as a first-
person account. To explain. In my encounters with other people,
all I ever observe is their behaviour: their actions, their speech,
and so on. And so, if we simply consider other people, Behaviour-
ism might seem to do full justice to the facts. But the trouble about
Behaviourism is that it seems so unsatisfactory as applied to our
own case. In our own case, we seem to be aware of so much more
than mere behaviour.
Suppose that now we conceive of the mind as an inner principle
apt for bringing about certain sorts of behaviour. This again fits the
other-person cases very well. Bodily behaviour of a very sophisti-
cated sort is observed, quite different from the behaviour that
ordinary physical objects display. It is inferred that this behaviour
must spring from a very special sort of inner cause in the object
that exhibits this behaviour. This inner cause is christened "the
mind", and those who take a Physicalist view of man argue that it
is simply the central nervous system of the body observed. Com-
pare this with the case of glass. Certain characteristic behaviour is
observed: the breaking and shattering of the material when acted
upon by relatively small forces. A special inner state of the glass is
postulated to explain this behaviour. Those who take a purely
Physicalist view of glass then argue that this state is a material
state of the glass. It is, perhaps, an arrangement of its molecules
and not, say, the peculiarly malevolent disposition of the demons
that dwell in glass.
But when we turn to our own case, the position may seem less
plausible. We are conscious, we have experiences. Now can we say
that to be conscious, to have experiences, is simply for something
to go on within us apt for the causing of certain sorts of
behaviour? Such an account does not seem to do any justice to the
phenomena. And so it seems that our account of the mind, like
Behaviourism, will fail to do justice to the first-person case.
In order to understand the objection better, it may be helpful to
consider a particular case. If you have driven for a very long dis-
tance without a break, you may have had experience of a curious
state of automatism, which can occur in these conditions. One can
suddenly "come to" and realize that one has driven for long dis-
tances without being aware of what one was doing, or, indeed,
without being aware of anything. One has kept the car on the road,
used the brake and the clutch perhaps, yet all without any aware-
ness of what one was doing.
Now if we consider this case, it is obvious that in some sense
mental processes are still going on when one is in such an auto-
matic state. Unless one's will was still operating in some way, and
unless one was still perceiving in some way, the car would not still
be on the road. Yet, of course, something mental is lacking. Now, I
think, when it is alleged that an account of mind as an inner princi-
ple apt for the production of certain sorts of behaviour leaves out
consciousness or experience, what is alleged to have been left out
is just whatever is missing in the automatic driving case. It is con-
ceded that an account of mental processes as states of the person
apt for the production of certain sorts of behaviour very possibly
may be adequate to deal with such cases as that of automatic driv-
ing. It may be adequate to deal with most of the mental processes
of animals, which perhaps spend most of their lives in this state of
automatism. But, it is contended, it cannot deal with the con-
sciousness that we normally enjoy.
I will now try to sketch an answer to this important and power-
ful objection. Let us begin in an apparently unlikely place and con-
sider the way that an account of mental processes of the sort I am
giving would deal with sense-perception.
Now psychologists, in particular, have long realized that there is
a very close logical tie between sense-perception and selective
behaviour. Suppose we want to decide whether an animal can per-

ceive the difference between red and green. We might give the
animal a choice between two pathways, over one of which a red
light shines and over the other of which a green light shines. If the
animal happens by chance to choose the green pathway, we
reward it; if it happens to choose the other pathway, we do not
reward it. If, after some trials, the animal systematically takes the
green-lighted pathway, and if we become assured that the only
relevant differences in the two pathways are the differences in the
colour of the lights, we are entitled to say that the animal can see
this colour difference. Using its eyes, it selects between red-lighted
and green-lighted pathways. So we say it can see the difference
between red and green.
Now a Behaviourist would be tempted to say that the animal's
regular selection of the green-lighted pathway was its perception
of the colour difference. But this is unsatisfactory, because we all
want to say that perception is something that goes on within the
person or animal - within its mind - although, of course, this
mental event is normally caused by the operation of the environ-
ment upon the organism. Suppose, however, that we speak instead
of capacities for selective behaviour towards the current environ-
ment, and suppose we think of these capacities, like dispositions,
as actual inner states of the organism. We can then think of the
animal's perception as a state within the animal apt, if the animal
is so impelled, for selective behaviour between the red- and green-
lighted pathways.
In general, we can think of perceptions as inner states or events
apt for the production of certain sorts of selective behaviour
towards our environment. To perceive is like acquiring a key to a
door. You do not have to use the key: you can put it in your
pocket and never bother about the door. But if you do want to
open the door, the key may be essential. The blind man is a man
who does not acquire certain keys and, as a result, is not able to
operate in his environment in the way that somebody who has his
sight can operate. It seems, then, a very promising view to take of
perceptions that they are inner states defined by the sorts of selec-
tive behaviour that they enable the perceiver to exhibit, if so
impelled.
Now how is this discussion of perception related to the question

of consciousness or experience, the sort of thing that the driver
who is in a state of automatism has not got, but which we normally
do have? Simply this. My proposal is that consciousness, in this
sense of the word, is nothing but perception or awareness of the
state of our own mind. The driver in a state of automatism per-
ceives, or is aware of, the road. If he did not, the car would be in a
ditch. But he is not currently aware of his awareness of the road.
He perceives the road, but he does not perceive his perceiving, or
anything else that is going on in his mind. He is not, as we nor-
mally are, conscious of what is going on in his mind.
And so I conceive of consciousness or experience, in this sense
of the words, in the way that Locke and Kant conceived it, as like
perception. Kant, in a striking phrase, spoke of "inner sense". We
cannot directly observe the minds of others, but each of us has the
power to observe directly our own minds, and "perceive" what is
going on there. The driver in the automatic state is one whose
"inner eye" is shut: who is not currently aware of what is going on
in his own mind.
Now if this account is along the right lines, why should we not
give an account of this inner observation along the same lines as
we have already given of perception? Why should we not conceive
of it as an inner state, a state in this case directed towards other
inner states and not to the environment, which enables us, if we
are so impelled, to behave in a selective way towards our own
states of mind ? One who is aware, or conscious, of his thoughts
or his emotions is one who has the capacity to make discrimina-
tions between his different mental states. His capacity might be
exhibited in words. He might say that he was in an angry state of
mind, when, and only when, he was in an angry state of mind. But
such verbal behaviour would be the mere expression or result of
the awareness. The awareness itself would be an inner state: the
sort of inner state that gave the man a capacity for such
behavioural expressions.
So I have argued that consciousness of our own mental state
may be assimilated to perception of our own mental state, and
that, like other perceptions, it may then be conceived of as an
inner state or event giving a capacity for selective behaviour, in
this case selective behaviour towards our own mental state. All this
is meant to be simply a logical analysis of consciousness, and none
of it entails, although it does not rule out, a purely Physicalist
account of what these inner states are. But if we are convinced, on
general scientific grounds, that a purely physical account of man is
likely to be the true one, then there seems to be no bar to our iden-
tifying these inner states with purely physical states of the central
nervous system. And so consciousness of our own mental state
becomes simply the scanning of one part of our central nervous
system by another. Consciousness is a self-scanning mechanism in
the central nervous system.
As I have emphasized before, I have done no more than sketch a

programme for a philosophy of mind. There are all sorts of expan-
sions and eludications to be made, and all sorts of doubts and
difficulties to be stated and overcome. But I hope I have done
enough to show that a purely Physicalist theory of the mind is an
exciting and plausible intellectual option.
2
The Cou1ol T~eory
of ~he Min~-
IS PHILOSOPHY JUST CONCEPTUAL ANALYSIS?
What can philosophy contribute to solving the problem of the rela-

tion to mind to body? Twenty years ago, many English-speaking
philosophers would have answered: "Nothing beyond an analysis
of the various mental concepts. " If we seek knowledge of things,
they thought, it is to science that we must turn. Philosophy can
only cast light upon our concepts of those things.
This retreat from things to concepts was not undertaken lightly.
Ever since the seventeenth century, the great intellectual fact of
our culture has been the incredible expansion of knowledge both
in the natural and in the rational sciences (mathematics, logic).
Everyday life presents us with certain simple verities. But, it
seems, through science and only through science can we build
upon these verities, and with astonishing results.
The success of science created a crisis in philosophy. What was
there for philosophy to do? Hume had already perceived the prob-
lem in some degree, and so surely did Kant, but it was not until the
twentieth century, with the Vienna Circle and with Wittgenstein,
that the difficulty began to weigh heavily. Wittgenstein took the
view that philosophy could do no more than strive to undo the
intellectual knots it itself had tied, so achieving intellectual release,
and even a certain illumination, but no knowledge. A little later,
and more optimistically, Ryle saw a positive, if reduced, role for
philosophy in mapping the "logical geography" of our concepts:
how they stood to each other and how they were to be analysed.
On the whole, Ryle's view proved more popular than Wittgens-
The Causal Theory of the Mind 17
tein's. After all, it retained a special, if much reduced, realm for

philosophy where she might still be queen. There was better hope
of continued employment for members of the profession!
Since that time, however, philosophers in the "analytic" tradi-
tion have swung back from Wittgensteinian and even Rylean
pessimism to a more traditional conception of the proper role and
tasks of philosophy. Many analytic philosophers now would accept
the view that the central task of philosophy is to give an account,
or at least play a part in giving an account, of the most general
nature of things and of man. (I would include myself among that
many.)
Why has this swing back occurred? Has the old urge of the
philosopher to determine the nature of things by a priori reason-
ing proved too strong? To use Freudian terms, are we simply wit-
nessing a return of what philosophers had repressed? I think not.
One consideration that has had great influence was the realization
that those who thought that they were abandoning ontological and
other substantive questions for a mere investigation of concepts
were in fact smuggling in views on the substantive questions. They
did not acknowledge that they held these views, but the views
were there; and far worse from their standpoint, the views imposed
a form upon their answers to the conceptual questions.
For instance, in The Concept of Mind (1949), Gilbert Ryle,
although he denied that he was a Behaviourist, seemed to be
upholding an account of man and his mind that was extremely
close to Behaviourism. Furthermore, it seemed in many cases that
it was this view of the mind-body problem that led him to his par-
ticular analyses of particular mental concepts, rather than the
other way around. Faced with examples like this, it began to
appear that, since philosophers could not help holding views on
substantive matters, and the views could not help affecting their
analyses of concepts, the views had better be held and discussed
explicitly instead of appearing in a distorted, because
unacknowledged, form.
The swing back by analytic philosophers to first-order questions
was also due to the growth of a more sophisticated understanding
of the nature of scientific investigation. For a philosophical tradi-
tion that is oriented towards science, as, on the whole, Western
philosophy is, the consideration of the methods of science must be

an important topic. It was gradually realized that in the past scien-
tific investigation had regularly been conceived in far too
positivistic, sensationalistic and observationalistic a spirit. (The
influence of Karl Popper has been of the greatest importance in
this realization.) As the central role of speculation, theory and
reasoning in scientific investigation began to be appreciated by
more and more philosophers, the border-line between science and
philosophy began to seem at least more fluid, and the hope arose
again that philosophy might have something to contribute to first-
order questions.
The philosopher has certain special skills. These include the
stating and assessing of the worth of arguments, including the
bringing to light and making explicit suppressed premisses of argu-
ments, the detection of ambiguities and inconsistencies, and,
perhaps especially, the analysis of concepts. But, I contend, these
special skills do not entail that the objective of philosophy is to do
these things. They are rather the special means by which philoso-
phy attempts to achieve further objectives. Ryle was wrong in tak-
ing the analysis of concepts to be the end of philosophy. Rather,
the analysis of concepts is a means by which the philosopher
makes his contribution to great general questions, not about con-
cepts, but about things.
In the particular case of the mind-body problem, the proposi-
tions the philosopher arrives at need not be of a special nature.
They perhaps might have been arrived at by the psychologist, the
neuro-physiologist, the biochemist or others, and, indeed, may be
suggested to the philosopher by the results achieved or pro-
grammes proposed by those disciplines. But the way that the argu-
ment is marshalled by a philosopher will be a special way. Whether
this special way has or has not any particular value in the search
for truth is a matter to be decided in particular cases. There is no a
priori reason for thinking that the special methods of philosophy
will be able to make a contribution to the mind-body problem.
But neither is there an a priori reason for assuming that the
philosopher's contribution will be valueless.
THE CONCEPT OF A MENTAL STATE

The philosophy of philosophy is perhaps a somewhat joyless and
unrewarding subject for reflection. Let us now turn to the
mind-body problem itself, hoping that wha,J: is to be said about
this particular topic will confirm the general remarks about
philosophy that have just been made.
If we consider the mind- body problem today, then it seems
that we ought to take account of the following consideration. The
present state of scientific knowledge makes it probable that we can
give a purely physico-chemical account of man's body. It seems
increasingly likely that the body and the brain of man are con-
stituted and work according to exactly the same principles as those
physical principles that govern other, non-organic, matter. The
differences between a stone and a human body appear to lie solely
in the extremely complex material set-up that is to be found in the
living body and which is absent in the stone. Furthermore, there is
rather strong evidence that it is the state of our brain that com-
pletely determines the state of our consciousness and our mental
state generally.
All this is not beyond the realm of controversy, and it is easy to
imagine evidence that would upset the picture. In particular, I
think that it is just possible that evidence from psychical research
might be forthcoming that a physico-chemical view of man's brain
could not accommodate. But suppose that the physico-chemical
view of the working of the brain is correct, as I take it to be. It will
be very natural to conclude that mental states are not simply
determined by corresponding states of the brain, but that they are
actually identical with these brain-states, brain-states that involve
nothing but physical properties.
The argument just outlined is quite a simple one, and it hardly
demands philosophical skill to develop it or to appreciate its force!
But although many contemporary thinkers would accept its con-
clusion, there are others, including many philosophers, who would
not. To a great many thinkers it has seemed obvious a priori that
mental states could not be physical states of the brain. Nobody
would identify a number with a piece of rock: it is sufficiently
obvious that the two entities fnll under different categories. In the
20 The Nature af Mind
same way, it has been thought, a perception or a feeling of sorrow

must be a different category of thing from an electro-chemical dis-
charge in the central nervous system.
Here, it seems to me, is a question to which philosophers can
expect to make a useful contribution. It is a question about mental
concepts. Is our concept of a mental state such that it is an
intelligible hypothesis that mental states are physical states of the
brain? If the philosopher can show that it is an intelligible proposi-
tion (that is, a non-self-contradictory proposition) that mental
states are physical states of the brain, then the scientific argument
just given above can be taken at its face value as a strong reason
for accepting the truth of the proposition.
My view is that the identification of mental states with physical
states of the brain is a perfectly intelligible one, and that this
becomes clear once we achieve a correct view of the analysis of the
mental concepts. I admit that my analysis of the mental concepts
was itself adopted because it permitted this identification, but such
a procedure is commonplace in the construction of theories, and
perfectly legitimate. In any case, whatever the motive for propos-
ing the analysis, it is there to speak for itself, to be measured
against competitors, and to be assessed as plausible or implausible
independently of the identification it makes possible.
The problem of the identification may be put in a Kantian way:
"How is it possible that mental states should be physical states of
the brain?" The solution will take the form of proposing an inde-
pendently plausible analysis of the concept of a mental state that
will permit this identification. In this way, the philosopher makes
the way smooth for a first-order doctrine, which, true or false, is a
doctrine of the first importance: a purely physicalist view of man.
The analysis proposed may be called the Causal analysis of the
mental concepts. According to this view, the concept of a mental
state essentially involves, and is exhausted by, the concept of a
state that is apt to be the cause of certain effects or apt to be the
effect of certain causes.
An example of a causal concept is the concept of poison. The
concept of poison is the concept of something that when
introduced into an organism causes that organism to sicken and/or
die. I This is but a rough analysis of the concept the structure of

which is in fact somewhat more complex and subtle than this. If A
pours molten lead down B 's throat, then he may cause B to die as
a result, but he can hardly be said to have poisoned him. For a
thing to be called a poison, it is necessary that it act in a certain
sort of way: roughly, in a biological as opposed to a purely physi-
cal way. Again, a poison can be introduced into the system of an
organism and that organism fail to die or even to sicken. This
might occur if an antidote were administered promptly. Yet again,
the poison may be present in insufficient quantities to do any
damage. Other qualifications could be made.
But the essential point about the concept of poison is that it is
the concept of that, whatever it is, which produces certain effects.
This leaves open the possibility of the scientific identification of
poisons, of discovering that a certain sort of substance, such as
cyanide, is a poison, and discovering further what it is about the
substance that makes it poisonous.
Poisons are accounted poisons in virtue of their active powers,
but many sorts of thing are accounted the sorts of thing they are by
virtue of their passive powers. Thus brittle objects are accounted
brittle because of the disposition they have to break and shatter
when sharply struck. This leaves open the possibility of discover-
ing empirically what sorts of thing are brittle and what it is about
them that makes them brittle.
Now if the concepts of the various sorts of mental state are con-
cepts of that which is, in various sorts of way, apt for causing cer-
tain effects and apt for being the effect of certain causes, then it
would be a quite unpuzzling thing if mental states should turn out
to be physical states of the brain.
The concept of a mental state is the concept of something that
is, characteristically, the cause of certain effects and the effect of
certain causes. What sort of effects and what sort of causes? The
effects caused by the mental state will be certain patterns of
behaviour of the person in that state. For instance, the desire for
1. "Any substance which, when introduced into or absorbed by a living organ-

ism, destroys life or injures health." (Shorter Oxford Dictionary, 3rd edn.,
rev., 1978.)
food is a state of a person or animal that characteristically brings

about food-seeking and food-consuming behaviour by that person
or animal. The causes of mental states will be objects and events in
the person's environment. For instance, a sensation of green is the
characteristic effect in a person of the action upon his eyes of a
nearby green surface.
The general pattern of analysis is at its most obvious and plausi-
ble in the case of purposes. If a man's purpose is to go to the
kitchen to get something to eat, it is completely natural to con-
ceive of this purpose as a cause within him that brings about, or
tends to bring about, that particular line of conduct. It is, further-
more, notorious that we are unable to characterize purposes
except in terms of that which they tend to bring about. How can
we distinguish the purpose to go to the kitchen to get something to
eat from another purpose to go to the bedroom to lie down? Only by
the different outcomes that the two purposes tend to bring about.
This fact was an encouragement to Behaviourism. It is still more
plausibly explained by saying that the concept of purpose is a
causal concept. The further hypothesis that the two purposes are,
in their own nature, different physical patterns in, or physical
states of, the central nervous system is then a natural (although, of
course, not logically inevitable) supplement to the causal analysis.
Simple models have great value in trying to grasp complex con-
ceptions, but they are ladders that may need to be kicked away
after we have mounted up by their means. It is vital to realize that
the mental concepts have a far more complex logical structure
than simple causal notions such as the concept of poison. The fact
should occasion no surprise. In the case of poisons, the effect of
which they are the cause is a gross and obvious phenomenon and
the level of causal explanation involved in simply calling a subs-
tance "a poison" is crude and simple. But in the case of mental
states, their effects are all those complexities of behaviour that
mark off men and higher animals from the rest of the objects in the
world. Furthermore, differences in such behaviour are elaborately
correlated with differences in the mental causes operating. So it is
only to be expected that the causal patterns invoked by the mental
concepts should be extremely complex and sophisticated.
In the case of the notion of a purpose, for instance, it is plausible
to assert that it is the notion of a cause within which drives, or

tends to drive, the man or animal through a series of actions to a
certain end-state. But this is not the whole story. A purpose is only
a purpose if it works to bring about behavioural effects in a certain
sort of way. We may sum up this sort of way by saying that pur-
poses are information-sensitive causes. By this is meant that pur-
poses direct behaviour by utilizing perceptions and beliefs, percep-
tions and beliefs about the agent's current situation and the way it
develops, and beliefs about the way the world works. For instance,
it is part of what it is to be a purpose to achieve X that this cause
will cease to operate, will be "switched off", if the agent perceives
or otherwise comes to believe that X has been achieved.
At this point, we observe that an account is being given of that
special species of cause that is a purpose in terms of further mental
items: perceptions and beliefs. This means that if we are to give a
purely causal analysis even of the concept of a purpose we also will
have to give a purely causal analysis of perceptions and beliefs. We
may think of man's behaviour as brought about by the joint opera-
tion of two sets of causes: first, his purposes and, second, his per-
ceptions of and/or his beliefs about the world. But since percep-
tions and beliefs are quite different sorts of thing from purposes, a
Causal analysis must assign quite different causal roles to these
different things in the bringing about of behaviour.
I believe that this can be done by giving an account of percep-
tions and beliefs as mappings of the world. They are structures
within us that model the world beyond the structure. This model is
created in us by the world. Purposes may then be thought of as
driving causes that utilize such mappings.
This is a mere thumb-nail, which requires much further
development as well as qualification. One point that becomes clear
when that development is given is that just as the concept of pur-
pose cannot be elucidated without appealing to the concepts of
perception and belief, so the latter cannot be elucidated without
appealing to the concept of purpose. (This comes out, for instance,
when we raise Hume's problem: what marks off beliefs from the
mere entertaining of the same proposition? It seems that we can
only mark off beliefs as those mappings in the light of which we are
prepared to act, that is, which are potential servants of our pur-
poses.) The logical dependence of purpose on perception and

belief, and of perception and belief upon purpose, is not circularity
in definition. What it shows is that the corresponding concepts
must be introduced together or not at all. In itself, there is nothing
very surprising in this. Correlative or mutually implicated concepts
are common enough: for instance, the concepts of husband and
wife or the concepts of soldier and army. No husbands without
wives or wives without husbands. No soldiers without an army, no
army without soldiers. But if the concepts of purpose, perception
and belief are (i) correlative concepts and (ii) different species of
purely causal concepts, then it is clear that they are far more com-
plex in structure than a simple causal concept like poison. What
falls under the mental concepts will be a complex and interlocking
set of causal factors, which together are responsible for the
"minded" behaviour of men and the higher animals.
The working out of the Causal theory of the mental concepts
thus turns out to be an extremely complex business. Indeed when
it is merely baldly stated, the Causal theory is, to use the phrase of
Imre Lakatos, a research programme in conceptual analysis rather
than a developed theory. I have tried to show that it is a hopeful
programme by attempting, at least in outline, a Causal analysis of
all the main mental concepts in A Materialist Theory of the Mind
(1968); and I have supplemented the rather thin account given
there of the concepts of belief, knowledge and inferring in Belief,
Truth and Knowledge (1973).
Two examples of mental concepts where an especially complex
and sophisticated type of Causal analysis is required are the
notions of introspective awareness (one sense of the word "con-
sciousness") and the having of mental imagery. Introspective
awareness is analysable as a mental state that is a "perception" of
mental states. It is a mapping of the causal factors themselves. The
having of mental imagery is a sort of mental state that cannot be
elucidated in directly causal terms, but only by resemblance to the
corresponding perceptions, which are explicated in terms of their
causal role.
Two advantages of the Causal theory may now be mentioned.
First, it has often been remarked by philosophers and others that
the realm of mind is a shadowy one, and that the nature of mental
states is singularly elusive and hard to grasp. This has given aid
and comfort to Dualist or Cartesian theori~s of mind, according to
which minds are quite different sorts of thing from material
objects. But if the Causal analysis is correct, the facts admit of
another explanation. What Dualist philosophers have grasped in a
confused way is that our direct acquaintance with mind, whiCh
occurs in introspective awareness, is an acquaintance with some-
thing that we are aware of only as something that is causally
linked, directly or indirectly, with behaviour. In the case of our
purposes and desires, for instance, we are often (though not
invariably) introspectively aware of them. What we are aware of is
the presence of factors within us that drive in a certain direction.
We are not aware of the intrinsic nature of the factors. This empti-
ness or gap in our awareness is then interpreted by Dualists as
immateriality. In fact, however, if the Causal analysis is correct,
there is no warrant for this interpretation and, if the Physicalist
identification of the nature of the causes is correct, the interpreta-
tion is actually false.
Second, the Causal analysis yields a still more spectacular
verification. It shows promise of explaining a philosophically
notorious feature of all or almost all mental states: their inten-
tionality. This was the feature of mental states to which Brentano
in particular drew attention, the fact that they may point towards
certain objects or states of affairs, but that these objects and states
of affairs need not exist. When a man strives, his striving has an
objective, but that objective may never be achieved. When he
believes, there is something he believes, but what he believes may
not be the case. This capacity of mental states to "point" to what
does not exist can seem very special. Brentano held that inten-
tionality set the mind completely apart from matter.
Suppose, however, that we consider a concept like the concept
of poison. Does it not provide us with a miniature and unsophisti-
cated model for the intentionality of mental states? Poisons are
substances apt to make organisms sicken and die when the poison
is administered. So it may be said that this is what poisons "point"
to. Nevertheless, poisons may fail of their effect. A poison does not
fail to be a poison because an antidote neutralizes the customary
effect of the poison.
May not the intentionality of mental states, therefore, be in

principle a no more mysterious affair, although indefinitely more
complex, than the death that lurks in the poison? As an intermedi-
ate case between poisons and mental states, consider the mechan-
isms involved in a homing rocket. Given a certain setting of its
mechanism, the rocket may "point" towards a certain target in a
way that is a simulacrum of the way in which purposes point
towards their objectives. The mechanism will only bring the rocket
to the target in "standard" circumstances: many factors can be
conceived that would "defeat" the mechanism. For the mechanism
to operate successfully, some device will be required by which the
developing situation is "mapped" in the mechanism (i.e. what
course the rocket is currently on, etc.). This mapping is an elemen-
tary analogue of perception, and so the course that is "mapped" in
the mechanism may be thought of as a simulacrum of the percep-
tual intentional object. Through one circumstance or another (e.g.
malfunction of the gyroscope) this mapping may be "incorrect".
It is no objection to this analogy that homing rockets are built
by men with purposes, who deliberately stamp a crude model of
their own purposes into the rocket. Homing rockets might have
been natural products, and non-minded objects that operate in a
similar but far more complex way are found in nature. The living
cell is a case in point.
So the Causal analyses of the mental concepts show promise of
explaining both the transparency and the intentionality of mental
states. One problem quite frequently raised in connection with
these analyses, however, is in what sense they can be called
"analyses". The welter of complications in which the so-called
analyses are involved make it sufficiently obvious that they do not
consist of synonymous translations of statements in which mental
terms figure. But, it has been objected, if synonymous translations
of mental statements are unavailable, what precisely can be meant
by speaking of "analyses of concepts"?
I am far from clear what should be said in reply to this objec-
tion. Clearly, however, it does depend upon taking all conceptual
analyses as claims about the synonymy of sentences, and that
seems to be too simple a view. Going back to the case of poison:
it iS surely not an empirical fact, to be learnt by experience, that
poisons kill. It is at the centre of our notion of what poisons are

that they have the power to bring about this effect. If they did not
do that, they would not be properly called "poisons". But
although this seems obvious enough, it is extremely difficult to
give exact translations of sentences containing the word "poison"
into other sentences that do not contain the word or any synonym.
Even in this simple case, it is not at all clear that the task can
actually be accomplished.
For this reason, I think that sentence translation (with syn-
onymy) is too strict a demand to make upon a purported concep-
tual analysis. What more relaxed demand can we make and still
have a conceptual analysis? I do not know. One thing that we
clearly need further light upon here is the concept of a concept,
and how concepts are tied to language. I incline to the view that
the connection between concepts and language is much less close
than many philosophers have assumed. Concepts are linked pri-
marily with belief and thought, and belief and thought, I think,
have a great degree of logical independence of language, however
close the empirical connection may be in many cases. If this is so,
then an analysis of concepts, although of course conducted in
words, may not be an investigation into words. (A compromise
proposal: analysis of concepts might be an investigation into some
sort of "deep structure" - to use the currently hallowed phrase
- which underlies the use of certain words and sentences.) I wish I
were able to take the topic further.
THE PROBLEM OF THE SECONDARY QUALITIES
No discussion of the Causal theory of the mental concepts is com-

plete that does not say something about the secondary qualities. If
we consider such mental states as purposes and intentions, their
"transparency" is a rather conspicuous feature. It is notorious that
introspection cannot differentiate such states except in terms of
their different objects. It is not so immediately obvious, however,
that perception has this transparent character. Perception involves
the experience of colour and of visual extension, touch of the
whole obscure range of tactual properties, including tactual exten-
sion, hearing, taste and smell the experience of sounds, tastes and
smells. These phenomenal qualities, it may be argued, endow

different perceptions with different qualities. The lack of
transparency is even more obvious in the case of bodily sensations.
Pains, itches, tickles and tingles are mental states, even if mental
states of no very high-grade sort, and they each seem to involve
their own peculiar qualities. Again, associated with different emo-
tions it is quite plausible to claim to discern special emotion
qualities. If perception, bodily sensation and emotions involve
qualities, then this seems to falsify a purely Causal analysis of
these mental states. They are not mere "that whiches" known only
by their causal role.
However, it is not at all clear how strong is the line of argument
sketched in the previous paragraph. We distinguish between the
intention and what is intended, and in just the same way we must
distinguish between the perception and what is perceived. The
intention is a mental state and so is the perception, but what is
intended is not in general something mental and nor is what is per-
ceived. What is intended may not come to pass, it is a merely inten-
tional object, and the same may be said of what is perceived. Now
in the case of the phenomenal qualities, it seems plausible to say
that they are qualities not of the perception but rather of what is
perceived. "Visual extension" is the shape, size, etc. that some
object of visual perception is perceived to have (an object that
need not exist). Colour seems to be a quality of that object. And
similarly for the other phenomenal qualities. Even in the case of
the bodily sensations, the qualities associated with the sensations
do not appear to be qualities of mental states but instead to be
qualities of portions of our bodies: more or less fleeting qualities
that qualify the place where the sensation is located. Only in the
case of the emotions does it seem natural to place the quality on
the mental rather than the object side: but then it is not so clear
whether there really are peculiar qualities associated with the emo-
tions. The different patterns of bodily sensations associated with
the different emotions may be sufficient to do phenomenological
justice to the emotions.
For these reasons, it is not certain whether the phenomenal
qualities pose any threat to the Causal analysis of the mental con-
cepts. But what a subset of these qualities quite certainly does pose
a threat to, is the doctrine that the Causal analysis of the mental
concepts is a step towards: Materialism or Physicalism.
The qualities of colour, sound, heat and cold, taste and smell
together with the qualities that appear to be involved in bodily
sensations and those that may be involved in the case of the emo-
tions, are an embarrassment to the modern Materialist. He seeks to
give an account of the world and of man purely in terms of physi-
cal properties, that is to say in terms of the properties that the
physicist appeals to in his explanations of phenomena. The
Materialist is not committed to the current set of properties to
which the physicist appeals, but he is committed to whatever set of
properties the physicist in the end will appeal to. It is clear that
such properties as colour, sound, taste and smell - the so-called
"secondary qualities" - will never be properties to which the
physicist will appeal.
It is, however, a plausible thesis that associated with different
secondary qualities are properties that are respectable from a
physicist's point of view. Physical surfaces appear to have colour.
They not merely appear to, but undoubtedly do, emit light-waves,
and the different mixtures of lengths of wave emitted are linked
with differences in colour. In the same way, different sorts of sound
are linked with different sorts of sound-wave and differences in
heat with differences in the mean kinetic energy of the molecules
composing the hot things. The Materialist's problem therefore
would be very simply solved if the secondary qualities could be
identified with these physically respectable properties. (The
qualities associated with bodily sensations would be identified
with different sorts of stimulation of bodily receptors. If there are
unique qualities associated with the emotions, they would
presumably be identified with some of the physical states of the
brain linked with particular emotions.)
But now the Materialist philosopher faces a problem. Previously
he asked: "How is it possible that mental states could be physical
states of the brain?" This question was answered by the Causal
theory of the mental concepts. Now he must ask: "How is it possi-
ble that secondary qualities could be purely physical properties of
the objects they are qualities of?" A Causal analysis does not seem
to be of any avail. To try to give an analysis of, say, the quality of
being red in Causal terms would lead us to produce such analyses

as "those properties of a physical surface, whatever they are, that
characteristically produce red sensations in us." But this analysis
simply shifts the problem unhelpfully from property of surface to
property of sensation. Either the red sensations involve nothing
but physically respectable properties or they involve something
more. If they involve something more, Materialism fails. But if
they are simply physical states of the brain, having nothing but
physical properties, then the Materialist faces the problem: "How is
it possible that red sensations should be physical states of the
brain?" This question is no easier to answer than the original
question about the redness of physical surfaces. (To give a Causal
analysis of red sensations as the characteristic effects of the action
of red surfaces is, of course, to move round in a circle.)
The great problem presented by the secondary qualities, such as
redness, is that they are unanalysable. They have certain relations
of resemblance and so on to each other, so they cannot be said to
be completely simple. But they are simple in the sense that they
resist any analysis. You cannot give any complete account of the
concept of redness without involving the notion of redness itself.
This has seemed to be, and still seems to many philosophers to be,
an absolute bar to identifying redness with, say, certain patterns of
emission of light-waves.
But I am not so sure. I think it can be maintained that although
the secondary qualities appear to be simple, they are not in fact
simple. Perhaps their simplicity is epistemological only, not
ontological, a matter of our awareness of them rather than the way
they are. The best model I can given for the situation is the sort of
phenomena made familiar to us by the Gestalt psychologists. It is
possible to grasp that certain things or situations have a certain
special property, but be unable to analyse that property. For ins-
tance, it may be possible to perceive that certain people are all
alike in some way without being able to make it clear to oneself
what the likeness is. We are aware that all these people have a cer-
tain likeness to each other, but are unable to define or specify that
likeness. Later psychological research may achieve a specification
of the likeness, a specification that may come as a complete
surprise to us. Perhaps, therefore, the secondary qualities are in
fact complex, and perhaps they are complex characteristics of a

sort demanded by Materialism, but we are unable to grasp their
complexity in perception.
There are two divergences between the model just suggested
and the case of the secondary qualities. First, in the case of grasp-
ing the indefinable likeness of people, we are under no temptation
to think that the likeness is a likeness in some simple quality. The
likeness is indefinable, but we are vaguely aware that it is complex.
Second, once research has determined the concrete nature of the
likeness, our attention can be drawn to, and we can observe
individually, the features that determine the likeness.
But although the model suggested and the case of the secondary
qualities undoubtedly exhibit these differences, I do not think that
they show that the secondary qualities cannot be identified with
respectable physical characteristics of objects. Why should not a
complex property appear to be simple? There would seem to be
no contradiction in adding such a condition to the model. It has
the consequence that perception of the secondary qualities
involves an element of illusion, but the consequence involves no
contradiction. It is true also that in the case of the secondary
qualities the illusion cannot be overcome within perception: it is
impossible to see a coloured surface as a surface emitting certain
light-waves. (Though one sometimes seems to hear a sound as a
vibration of the air.l But while this means that the identification of
colour and light-waves is a purely theoretical one, it still seems to
be a possible one. And if the identification is a possible one, we
have general scientific reasons to think it a plausible one.
The doctrine of mental states and of the secondary qualities

briefly presented in this paper seems to me to show promise of
meeting many of the traditional philosophical objections to a
Materialist or Physicalist account of the world. As I have
emphasized, the philosopher is not professionally competent to
argue the positive case for Materialism. There he must rely upon
the evidence presented by the scientist, particularly the physicist.
But at least he may neutralize the objections to Materialism
advanced by his fellow philosophers.
3
Ef?it~emolo~ical Founda~ion/ for a
Ma~etiali.tl Theory of ~he Mind
In this paper, I shall be arguing for the general plausibility of a

Materialist theory of the mind. The argument I present is, in a
broad sense, epistemological. Hence the title of the paper. I recog-
nize, of course, that the ultimate fate of this view of the mind
depends upon whether some formulation of the theory stands up
well to sustained and detailed examination. And so I claim nothing
more for my argument here than that it establishes a prima facie
case for a Materialist theory of the sort I favour. But in view of the
complexity and difficulty of the detailed considerations for or
against any particular view of any great question of philosophy,
those arguments that seem to weigh heavily for or against some
general standpoint must not be despised.
In the first section, I set out in general terms the view of the
mind which I favour. In the second, key, section, I give a general
argument for thinking that some view of this sort is correct. In the
last three sections, I consider three particular applications of what
is said in the second section.
THE GENERAL NATURE OF A MATERIALIST THEORY OF THE MIND

The theory of the mind that I wish to defend involves two distinct
contentions. The first, but only the first, of these may be presented
in the familiar brief slogan, "Mental processes are physical pro-
cesses in the brain." My first task will be to say something about
this claim.
Epistemological Foundations for a Materialist Theory of the Mind 33
THESIS A
A slogan of this sort is not to be despised. It is like a view of terrain

from a great height. We lose detail and accuracy, but we gain an
overall view. There is no doubt, however, that it requires com-
ment, qualification and expansion.
The word "process" must not be taken too seriously. A
developed theory will, at least, distinguish between mental pro-
cesses, states and events. In the slogan, however, the word "pro-
cess" is no more than a convenient blanket-term. Again, it is
not intended to assert that all processes in the brain are mental
processes. One possible expansion of the slogan would be "mental
processes are identical with (certain) processes in the brain".
Which particular brain processes these are, the Materialist
philosopher does not presume to say. That is left to
neurophysiology and, for the most part, to future neurophysiology.
This expansion of the slogan enables me to bring out a very
important ambiguity. "Mental processes are identical with (cer-
tain) processes in the brain" might be thought to assert no more
than:
(1) (x) (Mx-B 1 x)
where B 1 stands for the particular sort of brain processes that are
identical with mental processes - a specification of the nature of
the brain processes being still to be effected by neurophysiology.
However, the slogan can also be given a stronger interpretation:
(2) M=B 1
(2) is stronger than (1) because it entails (1), but (1) does not
entail (2). Now it seems that the Materialist must assert (2),
because if he asserts no more than (1), what he asserts is compati-
ble with an anti-Materialist view.
For suppose that (1) is true, but that M-;r.B 1 • The two predicates
"M" and "B 1 " will then apply to the individuals that they apply to
in virtue of distinct properties of the individuals. It is then left at
least open that what makes these individuals mental processes is
some further property that no ordinary physical individual has.
This further property (or properties) might be extraordinary
enough to set a wide metaphysical gulf between individuals who
have both physical and mental properties, on the one hand, and
individuals who have physical properties only, on the other.
In other words, the assertion, simply, of (1) does not rule out
what have been called "Double-aspect" theories of mind, but what
I would prefer to call "Attribute", or perhaps even better, "Dual-
attribute" theories. 1 Now, my Materialism is intended to exclude
theories of the latter sort.
Perhaps we can signal the exclusion of Dual-attribute theories
by recasting our formula to read "mental processes are nothing but
a certain sort of physical process in the brain," although it cannot
be claimed that the new form of words is self-explanatory.
It is very important to notice that the distinction between Dual-
attribute and purely Materialist theories cannot be drawn without
presupposing the existence of objective properties of individuals.
As William Kneale notes (1969: 293-94), the distinction cannot
be drawn by the Nominalist, for both sorts of theory assert that the
class of mental processes is identical with a certain class of brain
processes. I accept the existence of objective properties of
individuals, and so I accept that there is a real distinction between
Dual-attribute and Materialist theories of mind. Furthermore, I
reject the former theory.
The asserted identity M=B 1 is, of course, a contingent identity.
I am therefore committed to a contingent identification of proper-
ties. As the view may be put, the two properties, M and B1, are
identical, although the two predicates "M" and "B 1 " are not.
M=B)' but "M" ~ "B 1 ". 2
The doctrine that I defend, then, may be encapsulated in the
slogan, "Mental processes are nothing but a certain sort of physical
process in the brain." But despite even the explanations already
1. The term "Double-aspect" is misleading because even a Central-state

Materialist can admit that there is a sense in which mental processes may pre-
sent different "aspects". Our introspective awareness of a mental process that
is a brain process is very different from the brain surgeon's awareness of that
same process. But only the terms "Attribute" or "Dual-attribute" indicate a
commitment to the view that introspective awareness is an awareness of
extra, non-physical, properties.
2. For further discussion of this topic, see my Universals and Scientific Realism
(1978), vol. 2, ch. 17.
made, unclarity about the nature of the claim made by this slogan
may persist. After all, a contingent identification of properties,
which I have argued to be involved, is a philosophically puzzling
notion. What is required, it seems, is an account of the concept of
a mental process that will make clearer the exact nature of the
identification made, when it is claimed that mental processes are
nothing but a certain sort of brain process. Until we have such an
account, it will not be clear just what scientific evidence and just
what methodological considerations do or do not count as evi-
dence for the identification.
In their pioneer papers, U. T. Place (1956) and J. J. C. Smart
(1959) were concerned only to argue for the contingent identity of
sensations with brain processes. They tried to give an analysis of
the concept of sensation by analysing sensation statements in
"topic-neutral" terms. To have a sensation (sense-impression) is to
have something going on in one like what goes on when physical
objects of a certain sort are acting upon one's sense-organs. The
contingent identification proposed is the identification of this
something with a brain process.
I have tried to extend an analysis of this same general sort to
cover all sorts of mental processes, but have put greater weight
upon responses rather than stimuli in the elucidation of the various
mental concepts. My formula was that a mental state is to be con-
ceived of as a state apt for the production of certain ranges of
behaviour, or, in some cases, a state apt for being produced by a
certain range of stimuli. Now since a state is, presumably, a species
of property, this formula is somewhat obscure as it stands, for how
can a property be a cause or an effect? But things have the powers
they have in virtue of the properties they have. We can therefore
form the notion of that property of a thing, whatever it be, in vir-
tue of which the thing can bring about certain results (certain
behaviour). We can also form the notion of that property of a
thing, whatever it be, that the thing acquires as a result of other
things of certain sorts (certain stimuli) acting upon it. These pro-
perties and things are then identified by a scientific argument with
purely physical properties of portions of the brain.
David Lewis (1966) has suggested a more general, and even
more helpful, formula for capturing the concept of a mental pro-
cess or state.He proposes that all mental concepts are concepts of

that which plays a certain causal role in the physical behaviour of
the organism. If we see the identification of the mental with the
physical as a contingent identification of a property, then the men-
tal concepts will be concepts of that property in virtue of which
things with that property play a certain causal role in the physical
behaviour of the organism.
It is necessary, of course, to follow up these general formulas by
specific logical analyses of the individual mental concepts. I have
attempted this task in A Materialist Theory of the Mind (1968).
Much turns on the plausibility of the specific analyses, but they
cannot be discussed here.
We may call Lewis's and my account of the mental concepts the
"Causal" account of the mental concepts. The Causal account
itself may be in dispute. But it has this great attraction for the
Materialist: if it is correct, the identification of mental with brain
processes is of a quite unpuzzling sort. It is a familiar type of iden-
tification in science. Suppose, for instance, that brittleness is
defined as that feature of the brittle thing, whatever that feature
may be, in virtue of which the brittle thing shatters when struck
sharply. There is clearly nothing logically puzzling about identify-
ing this feature with a certain sort of bonding of the molecules of
the disposed object. Given the truth of the Causal account of the
mental concepts, the identification of the mental with the physical
is no more puzzling than this identification of brittleness with a
certain sort of molecular bonding. It is a further attraction that,
given the truth of the Causal analysis, the identification, although
perhaps not a certain scientific winner, is a promising bet in the
present state of scientific knowledge.
THESISB
So much for the first Materialist thesis that I wish to defend. But a
modern Materialist cannot remain content simply to assert that
mental processes are nothing but certain sorts of physical pro-
cesses in the brain. There remains the problem of the
"phenomenal" qualities - most conspicuously, the qualities
apparently associated with bodily sensations and the perceptual
Epistemological Foundations/or a Materialist Theory of the Mind 37
"secondary qualities". Whether these qualities be treated as

qualities of what is perceived (the Direct Realist view) or of the
perceiving of what is perceived (the Subjectivist view), they are a
problem for anybody who tries to give an account of physical
phenomena purely in terms of the properties attributed to the
phenomena by modern science, and in particular modern physics.
The problems involved have been spelt out in particular by
Smart. 3 (How are occurrences of such qualities to be correlated
plausibly with complexes of "fundamental particles"?)
What the Materialist must assert is that the phenomenal
qualities are in fact ("can be contingently identified with") com-
plex properties of a sort that are respectable from the physicist's
point of view. Mere perceptiou and introspection do not enable us
to grasp this identity, just as mere introspection does not enable us
to grasp the identity of mental processes with brain processes. But
it is vital to realize that in the case of the phenomenal qualities the
identification is not like the identification of mental processes with
brain processes. Both identifications are contingent identifications
of properties. But the identification of the phenomenal qualities is
not the identification of a feature previously specified only in
terms of the causal role of things that have that feature. Identifying
phenomenal with physical properties is instead a matter of iden-
tifying a property, grasped in a totalistic, holistic, unanalyzed way
by sense and/or introspection, with a complex physical property
either of the physical phenomena perceived or of the brain.
An imaginary illustration of such an identification would be the
case of beings who were able to pick identities, resemblances and
differences between a great variety of geometric shapes by means
of some sense, who were able to recognize that these identities and
differences were identities and differences in the intrinsic proper-
ties of the shaped objects, but who were completely unable to
analyse (even at a subverballevel) the nature of the properties they
could discriminate in this way, even to the extent of recognizing
them to be shapes. We then could imagine these beings gaining
knowledge of the property of shape by some other channel of
awareness. They then might proceed to the speculative step of
3. See, in particular, his Philosophy and Scientific Realism (1963).

identifying the properties that they could classify perceptually only

in terms of identity, resemblance and difference to each other, with
certain physical properties: namely various sorts of shape. They
would have made a contingent identification of phen-
omenologically unanalysable properties with complex physical
properties. Along these lines, I believe, the Materialist must
attempt to solve the problem of the phenomenal qualities.
EPISTEMOLOGICAL FOUNDATIONS
Suppose that we want to cast light upon some great metaphysical

theory, such as that of the nature of mind. To what body or bodies
of thought should we look for guidance? Here are five candidates.
(1) We might look to the common sense of mankind, to those
propositions that everybody knows or takes themselves to know.
(2) We might look to the results of particularly careful observa-
tion and scrutiny of the facts, to what might be called "the
phenomenological method", although I imply no particular link
with the philosophical movement called "Phenomenology".
(3) We might look to philosophical argument and philosophical
analysis. (4) We might look to the results of science, including
both the natural and the rational sciences. (5) We might look to
"higher" sources of illumination, to the religions, to the mystics
and to the moralists.
Different philosophies, and different treatments of particular
topics in philosophy, rely on these different bodies of thought iri
different degrees. Are there any reasons for relying on one rather
than another, or is it simply a matter of "ultimate presupposi-
tions"? Can any rational weighting be attached to these bodies of
thought?
The preliminary answer seems uncontroversial enough. We
should rely principally upon those bodies of thought, and upon
those propositions within those bodies, where we have knowledge
or, failing knowledge, where we have a high degree of rational
assurance.
But how do we determine what it is that we know or what
assurance is rational? It seems to me that the answer to this ques-
tion is also fairly uncontroversial, although the answer, if accepted,

yields further results that many might wish to reject. I suggest that
the best mark (it forms no part of the definition) of knowledge or
rational assurance concerning a proposition or body of proposi-
tions is the existence of a rational consensus. And I offer the
following definition of "rational consensus". If a proposition or
body of propositions is accepted as true by all those who are con-
versant with the matter in question, and whose judgement is not
hopelessly impaired, then, and only then, there is a rational con-
sensus about these propositions. As an instance of a person whose
judgement is hopelessly impaired about a particular matter, I sug-
gest a twentieth-century "flat-earther".
As the reader will undoubtedly have remarked, it is not easy to
say what constitutes a rational consensus without seeming to verge
on circularity. Yet I think almost all of us do accept some such
principle, and as soon as it is formulated, we recognize that we
accept some such principle. We give the principle special weight in
practical affairs, where reliability of belief is all important. It
should not be thought that the consensus is infallible, still less that
it is logically guaranteed. But it is the best guide to truth that we
have. If all save the ignorant or the incurably foolish agree, then it
is rational, although it may not always be correct, to assume that
what is agreed upon is true. 4
I proceed to apply the principle of rational consensus to the five
bodies of thought already enumerated.
It seems that we must reject almost entirely the claims of
religion, mysticism and morality to yield knowledge or even a
high degree of rational belief. Whatever claim is made in these
fields, be it positive or negative ("God exists," "God does not
4. The interesting objection has been suggested to me that the acceptance of
this principle of rational consensus makes one the epistemological prisoner of
the rational consensus of one's own time. Such epistemological conservatism,
it is then implied, can only be a bar to intellectual progress.
However, I do not think that this is correct. The rational consensus, like
democratic institutions, contains within itself resources for the revision of the
consensus. And it is in any case inevitable that one should be in a great degree
the intellectual prisoner of one's own time. One whose imprisonment went no
further than the rational consensus would be an intellectually free person
indeed!
exist," "The moral law exists," "There is no moral law") we find

informed, sober and responsible opinion on both sides of the
debate. Almost everything in this field is unproven. And so we
have here no secure epistemological base that we can use in
attempting to determine the trutn of such questions as the nature
of mind.
Indeed, to digress for a moment, I think that the rational con-
sensus in other bodies of opinion, in particular in natural science,
is one of our best guides to the truth in the fields of religion,
mysticism and morality. Scientific results, it seems to me, support
this-worldly conclusions in these fields. I am not, of course, argu-
ing that a this-worldly view of these matters is part of the rational
consensus. What I do maintain is that the results of the rational
consensus in natural science are the premisses of a good
philosophical argument for this-worldly views of religion, mystic-
ism and moral phenomena.
Given the criterion of rational consensus, we must attach great
weight to (1), to the commonsense beliefs of mankind. Common-
sense beliefs can be mistaken. In the part, some have turned out to
be mistaken. But there is a great central core of these beliefs
that almost everybody accepts, where serious doubt is considered
a clear proof of madness. We must take such beliefs to be
knowledge. I draw the moral that philosophers' theories should not
contradict these beliefs. It was the special contribution of G.E.
Moore to emphasize the importance of commonsense beliefs as an
epistemological foundation in philosophy. 5 It will be argued
shortly that the belief that there are mental processes is part of the
rational consensus.
What should we say about (2), about the very careful observa-
tion and scrutiny of facts? It is obvious that in many fields the
results of such observation and scrutiny yield knowledge. But if we
consider only matters that bear on philosophical issues, then the
"phenomenological method" has proved a disappointment. It has
not produced a rational consensus. We are now beginning to
understand more fully the psychological facts that stand behind
this: the incredible extent to which hopes, fears and preconceived
5. Most famously, in his "A Defence of Common Sense" (1925).

theories distort even the most scrupulous attempts to make

unbiased scrutiny of any range of phenomena. These distortions
are peculiarly difficult to avoid in the case of those very general
observations that are relevant to philosophy. Consider, in particu-
lar, the attempt to discover the nature of the mind simply by
inspecting it carefully. There are gross and obvious phenomena in
the field of the mental that can be discovered by this method and
about which there is a rational consensus. But consensus about the
relatively fine detail required to settle the issue between different
philosophies of mind is quite lacking.
Philosophers must naturally be tempted by (3), by the attempt
to seize hold of truth by means of philosophical argument and
analysis. It is their job to engage in these things, and they would
like to make high claims for the results of this activity. But their
endless disagreements show that philosophy yields few secure
intellectual foundations. The wise philosopher will be suspicious of
philosophy. (Including, of course, the argument of this paper.) So,
it seems to me, he should be inclined to seek his clues to the
general nature of, say, the mind outside philosophy. He should
look to common sense, or at least to its central core, and, as will
now be argued, to science.
In the case of the sciences, both natural and rational, we do not
meet with as complete an agreement, and as perfect an assurance
of knowledge, as in the case of the more obvious verities of com-
mon sense. But by fits and starts in the last three millenia, and
steadily in the last four hundred years, we have seen something the
intellectual significance of which cannot be underestimated: a
growth in our knowledge and/or rational assurance concerning
innumerable matters of fact and theory. This new knowledge and
rational assurance is securely attested by the intellectual consensus
that the scientific enterprise generates among those who embrace
or even examine it.
This consensus is a little obscured by the method of the scien-
tific undertaking, involving as it does the proposing of hypotheses,
argument, criticism and controversy, together with the continual
refining, revising and setting in a new light of previously accepted
results. But all this goes on at the frontiers of knowledge. As these
expand, behind them the secure territory of scientific knowledge
also expands, if much more slowly. In the present, at least, the

rational consensus is growing. When the scientist says "We do not
yet know ... " and "We now know ... ", we will be wise to take
his use of the word "know" pretty seriously.
It folio~. I think, that scientific results, and the probabilities that
appear to flow from these results, must be taken with the utmost
seriousness by the philosopher. (I am not talking about the latest
scientific fashion, or about what scientists say, but about firm and
secure scientific results and the perspectives these open up.) If the
scientific results point in a certain direction concerning some great
philosophical question, then that is one of the best lights that the
philosopher has for deciding that question. In contemporary
philosophy, J. J. C. Smart, in particular, has emphasized the
special importance of scientific knowledge in reaching philosophi-
cal conclusions. 6 (It seems to me that Moore and Smart, taken
together, indicate to us the proper epistemological foundations for
philosophy.)
It must be admitted that there are certain philosophers, at the
present time - in particular, some of those influenced by the
thought of Karl Popper - who are inclined to deny that science
does yield knowledge. They cannot be accused of not being con-
versant with the scientific enterprise, and it would be impolite to
urge that their judgement is hopelessly impaired on the matter.
Does their existence show that there is no rational consensus in
science?
If, however, we turn from their explicit pronouncements to what
they implicitly assume in the same page or lecture, I think that
there is reason to assert that, although they say these things about
scientific knowledge, they do not really believe them. Their scien-
tific references make it quite clear that there is a huge body of
scientific fact that they do not seriously doubt. They are like the
lady who wondered why more people were not convinced by the
arguments for solipsism.
In the case of the philosophy of mind, it is the natural rather
than the rational sciences that are important, although it has been
argued, for example, that Godel's theorem refutes a Materialist
6. See, again, Philosophy and Scientific Realism (1963).

account of the mind. 7 Now the progress of the natural sciences

seems to point rather strongly to a purely Materialist account of
the nature of man, and so to a purely Materialist account of the
mind.
It may be objected that just by themselves our accumulated
scientific results do not give any reason to accept a Materialist
account of the mind. To yield such a result, they must be taken in
conjunction with certain methodological principles, in particular
with Occam's razor, which bids us adopt the simplest and most
economical hypothesis. But it is a philosophically controversial
question whether we should accept such principles. So one essen-
tial leg of the support for a Materialist theory rests not upon (relia-
ble) science, but upon (unreliable) philosophy.
I reply that this objection rests upon a very narrow view of the
nature of the rational consensus achieved by science. It is clear that
we have no reason to accept particular scientific results as correct
unless we accept certain principles of method, including the princi-
ple of making the simplest and most economical postulations com-
patible with undeniable observations. It also seems clear that if we
pay attention simply to the scientific facts and likelihoods about
man, the simplest and most economical hypothesis about him is
that he is a purely physical being. My argument simply combines
this fact with the observation that we have no better epistemologi-
cal base, besides the more obvious deliverances of common sense
(which give no particular help here), than scientific fact and likeli-
hood.
In the next three sections, I consider some particular issues that
arise out of and illustrate, the general points made in this section.
COMMON SENSE AND THE DISAPPEARANCE THEORY
Some Materialists have asserted that there are no mental processes

at all. Instead, there are purely physical processes in the brain.
This "disappearance" view has been put forward by Paul Feyera-
bend (1963) and, in a more cautious form, by Richard Rorty
7. For example, Lucas, 1961. Smart replies to Lucas in 1963, ch. 6.

(1965, 1970). What should be said about this radical form of

Materialism?
In Feyerabend's case, this view is based upon the conviction
that the concept of a mental process is such that it entails that
mental processes are immaterial, together with a conviction that a
purely Physicalist account can be given of man. Rorty also accepts
the second assumption and combines it with the view that the con-
cept of a mental process entails that persons have a certain
(empirical) epistemological authority about the nature of their own
mental states. Feyerabend concludes that there are no mental pro-
cesses; Rorty that the advance of science may well break down the
individual's special authority and so show us that there are no
mental processes.
In terms of the argument of the previous section, considerations
drawn from (4) - the results of modern science - together with
considerations drawn from (3) - the results of philosophical
analysis of the mental concepts - are thought to outweigh con-
siderations drawn from (1) - the commonsense view that there
are mental processes. But to have said this is already to have pro-
duced a strong prima facie argument against the "Disappearance"
theory. It is much more obvious that there are pains, perceptions,
emotions and thoughts, than that any particular logical analysis of
the concepts of pain, perception, emotion and thought is correct. It
is just about as obvious that all these sorts of things, together with
some others, all resemble each other in some way and are different
from more ordinary physical processes. In this minimal sense, it is
obvious that there are mental processes and it is more obvious that
there are mental processes than that any particular analysis of the
concept of a mental process is correct. The existence of pain, per-
ception and of mental processes is a deeply entrenched assumption
of the common sense of mankind. The logical analysis of the con-
cept of pain, of perception, of mental process, is a philosophers'
battlefield. If it is further agreed, as the Disappearance theorist
agrees, that the scientific evidence for a Physicalist view of man is
very strong, then it seems methodologically correct to look for an
account of these concepts that permits the scientific identification
of the corresponding processes with physical processes. What we
look for, we may fail to find, and we may have to conclude even-
Epistemological F oundationsfor a Materialist Theory of the Mind 45
tually that it is not there to find. But it seems quite premature at

this stage of the philosophical enterprise to conclude that the pro-
ject of reconciling the existence of pains, of perceptions, and of
mental processes with a purely materialist account of man, has
failed.
Suppose, however, that the argument goes the way I do not
expect it to go. Suppose it is shown that the concept of a mental
process as we have it at present entails, say, that such processes
are immaterial or that each man has a quite special and metaphysi-
cal authority about the nature of his own current mental processes.
Should these entailments, if established, make Physicalists deny,
now or in the future, that there are mental processes? While recog-
nizing ;hat the question may be a relatively trivial matter calling
for little more than a decision how to use words, I suggest that
even these developments would not force the Physicalist to deny
the existence of mental processes.
The first point to be made is this. If it has shown anything,
research in the philosophy of mind in the last few decades has
shown that the logical structures and interrelations of the mental
concepts are exceedingly complex. Now, at least where an object
is complex in nature, it is often the case that changes in its nature,
if they be not too radical, are not thought to destroy the identity of
that object. We continue to treat the changed thing as a thing of
the same sort. The possibility is therefore opened up of a concept
changing its conceptual structure, in some degree at least, and yet
still being accounted the same concept. The notion of a concept
retaining an identity through various changes is familiar to stu-
dents of the history of thought. Might not the Physicalist claim that
the revisions that Materialism might force in our concept of mind
would be mere revisions?
How much revision can a concept stand and yet still be
naturally accounted the same concept? Let us make the question
more precise and ask under what conditions it would be natural to
drop a necessary condition, N, for a thing being a C, from the con-
cept of C. I suggest ihat two rather simple conditions must be
satisfied.
First, if N be compared with the set of the other necessary con-
ditions for C, then N must play a relatively minor role in making

an object an instance falling under C.
I am sorry to say that I can do no more than appeal to an intui-
tive, largely un-analysed notion of "minor role". But I think I can
give examples. In the concept of father, the two necessary condi-
tions of being a male and being a parent would appear to have
equal weight. For this reason, the extension of the concept of
father is naturally represented as the intersection of two circles:
fathers
Contrast this with the concept of murder. Necessary conditons

here are that a murder be a killing, that it be an illegal act and that
it be an act involving malice aforethought. But in the concept of
murder, the necessary condition that it be a killing appears to play
a major role in comparison with the other two necessary condi-
tions. This is shown by the fact that, for instance, we can
immediately understand what is meant when a fox is said to
murder chickens, although the fox does nothing but kill the
chickens. Yet we would not understand calling a theft of money
"murder" (without some further explanation), although the act
was illegal and done with malice aforethought. It may be said that
we understand the statement about the fox metaphorically. But
while this may be true, it only illustrates my point. Why is the
metaphor immediately understood, while it would not be under-
stood without further explanation if we attempted to characterize
the theft as murder? Instead of three intersecting circles we might
represent the extensions of the concepts in the folowing way:
murder
(killing)
illegal
acts
Here the two classes of illegal and malicious acts are both sub-
divided into those that are cases of killing and those that are not.
Acts of killing here appears as a genus of which illegal killings and
killings involving malice aforethought are mere species. Illegality
and malice are necessary conditions for murder, yet they do not
rank with the major necessary condition: that the act be one of
killing.
I think that cases of this sort of hierarchy among necessary con"
ditions are quite common. So one necessary condition for drop-
ping a necessary conditon(s) from a concept, yet still speaking of
the same concept, is that the necessary condition(s) be minor in
this way.
The second necessary condition for dropping a necessary condi-
tion(s) is in one way so trivial that it may seem that there is hardly
need to mention it. It is this. It must be possible to drop the condi-
tion(s) either without affecting the truth-value of the other necess-
ary conditions, or, if they are affected, affecting these others only
in a relatively minor way.
As an example of how easy it is to overlook this simple point,
consider the following. It is generally held by philosophers that a
statement of the form "A sees an object 0" entails that "0
exists". But it is easy to think of this necessary condition for seeing
an object as something that can be negated without any further

damage to the concept of seeing an object. It is then tempting to
introduce a sense of "seeing an object" in which it is not
demanded that 0 exist. The new concept may be different, but is it
not a very close relative of the old concept? In fact, however, it
can be very plausibly argued (and I will here assume it to be true)
that another necessary condition of A's seeing 0 is that 0 be
causally responsible for those visual experiences of A that con-
stitute his seeing of 0. But if 0 does not exist, it cannot cause A's
visual experiences. Hence the condition that 0 exists cannot be
detached by itself. If it goes, at least one more necessary condition,
the causal condition, must go too. The condition that 0 exists is
therefore more deeply embedded in the concept of seeing an
object than we might have realized. And it will then have to be
considered whether "seeing" without objects to see is really so
close to ordinary seeing after all. (Aside: I believe that the condi-
tion that if A knows p, then p is true, is also much more deeply
embedded in the concept of knowledge than has been generally
realized.)
So to drop a necessary condition for a concept, and yet still be
justified in speaking of the concept as the same concept, it is
necessary both that the condition be minor and that its dropping
result in no more than minor further damage to the concept. It
must be like removing a relatively small portion of a house, which,
furthermore, can be removed without bringing down too much of
the rest of the house. Call such a necessary condition that answers
to these two conditions a "detachable" condition.
Now suppose it is true, as is alleged, that the concept of a men-
tal process is the concept of a non-physical process, or is the con-
cept of something that the person undergoing that process has a
metaphysically privileged cognitive access to. It may still be the
case that these conditions are "detachable". Perhaps mental pro-
cesses are necessarily non-physical or necessarily such that their
owners have some privileged cognitive access to them. I doubt it,
but it may be so. But it is hard to accept that these are major
features of the concept of a mental process. However philosophers
may argue, these features do not seem to figure centrally in ordin-
ary talk and thought about the mind. Nor does it seem to me that
removal of these necessary conditions would have any particular

effect upon the remaining necessary conditions. (It would be
interesting if argument could be produced to show that there is
such an effect.) So perhaps these conditions, even if they obtain,
are "detachable". And if, as Physicalists, including Feyerabend
and Rorty, hold, there are no non-physical processes nor any pro-
cesses to which their owners must stand in a special cognitive rela-
tion, then there will be the strongest motive for detaching these
conditions, yet still asserting that there are mental processes. It
would be yet another case of the superiority of piecemeal reform
over revolution.
PHENOMENOLOGY AND MATERIALISM
It is a common complaint that a Materialist theory fails to do

justice to the data of introspection and/or perception. (In particu-
lar, it is alleged that it fails to do justice to our experience of
phenomenal quality.) Theoretical considerations drawn from
science may favour a purely Materialist view of man, but, it is said,
direct observation makes it clear that this cannot be the whole
story.
It does seem to me that there is some justice in this complaint.
Phenomenological results, so far as they are obtainable, do seem
to make against Materialism. There is something quite profoundly
counter-intuitive about the idea that, when we introspect, what we
are in fact introspecting, whether we realize it or not, are purely
physical processes in the brain, or that when we are aware of col-
ours, what we are aware of, whether we realize it or not, are
nothing but packets of light-waves or are again purely physical
processes in the brain. But with intuitions, one swallow does not
make a summer. We may well be justified in discarding an intui-
tion if it conflicts with a systematic body of other propositions that
we are unwilling to discard. Now the scientific evidence for a
purely Materialist theory of man constitutes just such a systematic
body of propositions.
(Compare the situation in linguistics. We may have a strong
intuition that a certain proposed sentence is ungrammatical. But
an intuition may have to give way to intuitions. If a certain theory

explains other data, that is, is itself verified by other intuitions
about other sentences, yet the theory entails that the proposed
sentence is grammatical, we may be justified in concluding after all
that the sentence is grammatical.)
In the case of the mind, the phenomenological method has
inherited prestige from the Cartesian doctrine that first-person
judgements of current mental state are logically indubitable. The
latter doctrine itself derives from something more general, the
"search for certainty", which has led and misled so many
philosophers. (Not the same thing, I hope, as the search for a
reasonably secure epistemological base in philosophy!) I take it,
perhaps over optimistically, that fairly conclusive arguments have
been advanced against such logical indubitability. 8 But although
these arguments have received widespread acceptance, they have
not always worked themselves deep into philosophers' minds.
Perhaps it is the indubitability doctrine, working in an under-
ground way, that has emboldened some philosophers so confi-
dently to back their introspective intuitions about the mind against
the systematic scientific considerations that favor Materialism.
But I do concede that introspection and/or perception appears
to yield some evidence against a Materialist view of man.
Phenomenal qualities, in particular, yield a rather overpowering
impression of (a) relative simplicity and (b) irreducibility. One
thing that would greatly strengthen the Materialist case here would
be the production of an independently plausible explanation of
why Materialism is introspectively implausible.
I think I can suggest at least part of the explanation. In the first
place, it does seem clear that introspection fails to make us aware
of the physical (e.g. spatial) nature of mental processes. In the
same way, introspection, or perhaps more correctly, perception,
fails to make us aware of any great complexity in the phenomenal
qualities, and further fails to make us aware of an identity of these
qualities with complex physical properties. But it is clearly invalid
to argue from lack of awareness of the complex physical nature of
mental processes and phenomenal qualities to the conclusion that
8. See, for instance, my A Materialist Theory of the Mind (1968), ch. 6. sec. 10.
we are aware that these processes and qualities lack this complex
physical nature. The move from "I am not aware that p" to "I am
aware that not-p" is an illegitimate shifting of the negation sign.
But what evidence is there that this shift has actually occurred?
Here I think that evidence can be brought to show that this partic-
ular shift is a fallacy to which the human mind is naturally prone.
A spectacular case in point is the Headless Woman illusion. If an
illusionist presents a woman on a stage, brightly lighted, but
against a perfectly black background and with a black cloth over
her head, then it will look to the audience as if they are seeing a
woman who has no head. Unsophisticated persons might be
deceived. It is clear that the audience lacks any perceptual aware-
ness of the woman's head. Further than that they are not entitled
to judge. But nature, by an uncontrollable necessity, determines
them to judge, or at least determines that it appears to them, as if
the woman actually lacks a head. Out of sight is not simply out of
mind. It seems to be out of existence.
A more prosaic case is that of perceiving what looks to be an
absolutely straight line or edge. The line or edge will undoubtedly
have minor irregularities. Our eyes are only capable of informing
us that the line is straight relative to some limit: namely whatever
is the limit of our capacity for visual discrimination. But our
inability to perceive any irregularity generates in us the additional,
false, impression that the line lacks any irregularity at all.
Once these cases are before us, it becomes clear that such false
reasoning is a natural tendency of the human mind. We see,
indeed, that it is quite a sophisticated intellectual achievement to
believe in the unobservable, a fact that is no doubt one of the
underground sources of the appeal of Operationalism. The way is
then open for a Materialist to explain away our introspective pre-
judice against Materialism along the same lines. Introspection
quite fails to indicate that mental processes are material. It
therefore must generate the illusion that they are immaterial.
Introspection and/or perception fails to indicate that the
phenomenal qualities are both complex, and complexes of proper-
ties of the sort recognized by physics. The illusion therefore must
be generated that these properties are simple and irreducible.
I do not think that this suggestion completely explains away the

phenomenological implausibility of a purely Materialist theory of
man. It certainly does something to solve the problem. But in the
case of the phenomenal qualities, in particular, it seems that more
needs to be said. But I do not know what.
NATURAL SCIENCE AND CONCEPTUAL ANALYSIS
It was pointed out above that a complete defence of the view that
mental processes are nothing but a certain sort of physical process
in the brain demands not only scientific arguments for a purely
Materialist view of man but also an account of the concept of a
mental process. Only if we have the latter can it be seen how the
identification is possible. This was the occasion for proposing the
"Causal" analysjs of t,he mental concepts.
It is to be noticed that the Causal analysis was not arrived at
simply as a result of investigating the mental concepts. Rather
there seemed to be good scientific reasons for accepting a purely
Materialist account of man, which led to a search for an account of
the mental concepts that would make Materialism possible. Place
and Smart were originally attracted to (Rylean) Behaviourism for
just this reason. But the Behaviourist analyses of the mental con-
cepts were not independently plausible enough, the breaking-point
for Place and Smart being first-person reports of such things as
pains and after-images. At this point, a transition was made from a
peripheralist to a Central-state account of the mind; and an
attempt was made to find an account of the mental concepts com-
patible with a Central-state view.
I think that some philosophers are sceptical about the Causal
proposal because it has this origin. They suspect it of being a
philosophical epicycle, introduced to save the doctrine of Material-
ism.
Now it of course would have been a very striking thing if the
Causal account of the mental concepts had been arrived at inde-
pendently, and was already there waiting to join the scientific evi-
dence. But too much weight need not be placed on this point. The
really important point to consider is that we have much better
reason to accept a purely Materialist account of man than we have

to accept ANY philosophical analysis of the mental concepts. The
reason for this is simply the one I have given above. In the present
state of our knowledge, scientific results and probabilities are a
relatively secure epistemological base, philosophical analyses an
extremely insecure one. And current scientific results and pro-
babilities do point to the likely truth of a purely Materialist
account of man. In arriving at the Causal analysis, therefore, we
are doing something of a sort that is commonly done, and is
rationally done. We are arriving at one hypothesis under the gui-
dance of another hypothesis, where we have good reasons to think
the second hypothesis true but where the second hypothesis
requires some such theory as the first hypothesis if the second is to
be sustained.
It is true, of course, that it will have to be shown that the Causal
analysis can be developed fruitfully and can stand up well to inde-
pendent critical examination. But when we take into account the
incredible difficulty of philosophy, including the incredible
difficulty of deciding on the truth or falsity of conceptual analyses,
then I think it is fair to say that this account of the mental concepts
compares quite plausibly with at least any alternative proposed so
far. (The Causal analysis has nothing like the conspicuous
difficulties that attend Behaviourist alternatives.)
In general, it seems that there may be fruitful interaction bet-
ween any two intellectual disciplines, and that therefore there may
be fruitful interaction between scientific results (and speculation)
and conceptual analysis. The mere logical independence of pro-
positions must not blind us to the fact that they may hang together
coherently in logically looser but intellectually significant ways.
The fact that we do not understand explicitly very much about
such hanging together should not deter us. We can recognize it
when we see it. And where a subset of a coherent body of proposi-
tions is epistemologically more secure than the remainder of the
set, as is the case where the subset is made up of scientific results
and probabilities while the remainder of the set are philosophical
propositions, then it often will be plausible to argue from the truth
of the members of the subset to the truth of the members of the

remainder. 9
One source of resistance to the idea that science can and ought
to help to determine philosophical truth may be this. It is often
assumed, and may be true, that the discoveries of science are con-
tingent truths, while conceptual analysis issues in necessary truths.
Now there is an old tradition in philosophy that links necessary
truth with self-evidence, or logical indubitability, while contingent
truth is linked with lack of self-evidence, or the possibility of error.
It is easy to show, and now widely appreciated, that there is in fact
no such link. Mistake about any moderately complex logical
necessity is quite easy. Contrariwise, the class of truths for which
contemporary philosophy most frequently claims indubitability are
first-person reports about current mental states, propositions that
are certainly not necessary truths. Nevertheless, the old linking
persists in philosophers' thoughts, even though it is recognized to
be an error. From the perspective of the search for certainty,
necessity is thought to be an epistemologically desirable charac-
teristic. And so there is resistance to the idea that the contingent
truths of science could be any guide to the conceptual truths of
philosophical analysis.
But whatever may be the epistemological credit of the proposi-
tions put forward by logicians and mathematicians, the conceptual
analyses of philosophers are matters of the greatest dispute among
those most competent to judge (the philosophers themselves), as
well as among everybody else. The rational man therefore will not
give them a high epistemological credit rating. He will recognize
that philosophy needs all the help it can get in deciding upon the
truth or falsity of conceptual claims. He therefore will take the
most careful notice of any relevant relative certainties that science
or common sense can provide.
9. A good illustration of what I am arguing for is provided by Keith Gundersen

0971: 92), when he says: "I want to suggest that by trying to program a
machine with recognition capacities we will perhaps improve our understand-
ing of the concept of recognition."
4
W~a~ i1 ConJciouJne/1?
The notion of consciousness is notoriously obscure. It is difficult to

analyze, and some philosophers and others have thought it
unanalysable. It is not even clear that the word "consciousness"
stands for just one sort of entity, quality, process, or whatever.
There is, however, one thesis about consciousness that I believe
can be confidently rejected: Descartes' doctrine that consciousness
is the essence of mentality. That view assumes that we can explain
mentality in terms of consciousness. I think that the truth is in fact
the other way round. Indeed, in the most interesting sense of the
word "consciousness", consciousness is the cream on the cake of
mentality, a special and sophisticated development of mentality. It
is not the cake itself. In what follows, I develop an anti-Cartesian
account of consciousness.
MINIMAL CONSCIOUSNESS
In thinking about consciousness, it is helpful to begin at the other

end and consider a totally unconscious person. Somebody in a
sound, dreamless sleep may be taken as an example. It has been
disputed whether unconsciousness is really ever total. There is
some empirical evidence that a person in dreamless sleep, or even
under a total anaesthetic, still has some minimal awareness.
Minimal behavioural reactions to sensory stimuli have been
observed under these conditions. But let us take it, if only as a
simplifying and perhaps unrealistic assumption, that we are deal-
ing with total unconsciousness.
Notice first that we are perfectly happy to concede that such a

person, while in this state of total unconsciousness, has a mind.
Furthermore, although by hypothesis this mind is in no way active
- no mental events take place, no mental processes occur within
it - we freely allow that this mind is in various states.
The totally unconscious person does not lack knowledge and
beliefs. Suppose him to be a historian of the mediaeval period. We
will not deny him a great deal of knowledge of and beliefs about
the Middle Ages just because he is sound asleep. He cannot give
current expression to his knowledge and his beliefs, but he does
not lack them. The totally unconscious person also may be cred-
ited with memories. He also can be said to have skills, including
purely mental skills such as an ability for mental arithmetic. The
ability is not lost during sound sleep just because it then cannot be
exercised, any more than an athlete loses his athletic abilities dur-
ing sound sleep, when he cannot exercise them. A totally
unconscious person may be credited with likes and dislikes,
attitudes and emotions, current desires and current aims and pur-
poses. He may be said to have certain traits of character and
termperament. He may be said to be in certain moods: "He has
been depressed all this week."
How are we to conceive of these mental states (it seems natural
to call them "states") we attribute to the unconscious person?
Some decades ago, under the influence of positivistic and pheno-
mentalistic modes of thought, such attributions of mental states to
an unconscious person would not have been taken very seriously,
ontologically. It would have been thought that to say that the cur-
rently unconscious person A believes that p, is simply to refer to
various ways in which A 's mind works, or would work in suitable
circumstances, before and/or after he wakes up. (The same
positivist spirit might try further to reduce the way that A 's mind
works to A's peripheral bodily behaviour or to the behaviour A
would exhibit in suitable circumstances.)
In historical perspective, we can see clearly how unsatisfactory
such a view is. Consider two persons, A and B, unconscious at the
same time, where it is true of A that he believes that p, but false of
B. Must there not be a difference between A and B at that time to
constitute this difference in belief-state? What else in the world
What is Consciousness? 57
could act as a truth-maker (the ground in the world) for the

different conditional statements that are true of A and B ? The
mind of the unconscious person cannot be dissolved into state-
ments about what would be true of the person if the situation were
other than it was; if, in particular, he were not unconscious.
In considering this point, I find very helpful the analogy bet-
ween an unconscious person and a computer that has been pro-
grammed in various ways, that perhaps has partially worked
through certain routines and is ready to continue with them, but is
not currently operating. (I do not think that anything in the
analogy turns on the material, physical nature of the computer.
Even if the mind has to be conceived of in some immaterial way,
the analogy will still hold.) The computer, perhaps, will have a cer-
tain amount of information stored in its memory-banks. This
stored information may be compared to the knowledge, belief and
memories the unconscious person still has during unconscious-
ness. If a Materialist account of the mind is correct, then, of course,
knowledge, belief and memory will be physically encoded in the
brain in some broadly similar way to the way in which information
is stored in the computer. But the Dualist, say, will equally require
the conception of immaterial storage of knowledge, belief and
memory.
What we can say both of the knowledge, beliefs, etc. possessed
by the totally unconscious person, and also of the information
stored in the switched-off computer, is that they are causally
quiescent. Of course, nothing is causally quiescent absolutely:
while a thing exists, it has effects upon its environment. But the
information stored in the switched-off computer is causally quies-
cent with respect to the computing operations of the computer,
and for our purposes this may be called causal quiescence. (The
information may remain causally quiescent even after the com-
puter has been switched on, unless that piece of information is
required for current calculations.) In the same way, knowledge and
beliefs may be said to be causally quiescent while they are not pro-
ducing any mental effect in the person. The mental states of a
totally unconscious person are thus causally quiescent (if they are
not, we may stipulate that the person is not totally unconscious).
Knowledge, beliefs, and so on may remain causally quiescent in
this sense even when the mind is operational, for instance, where
there is no call to use a particular piece of knowledge.
It seems, then, that we attribute mental states of various sorts to
a totally unconscious person. But there are certain mental attribu-
tions we do not make. The totally unconscious person does not
perceive, has no sensations, feelings or pangs of desire. He cannot
think, contemplate or engage in any sort of deliberation. (He can
have purposes, because purposes are capable of causal quiescence,
but he cannot be engaged in carrying them out.) This is because
perception, sensation and thinking are mental activities in a way
that knowledge and beliefs are not. The distinction appears,
roughly at any rate, to be the distinction between events and
occurrences on the one hand, and states on the other. When a
mental state is producing mental effects, the comings-to-be of such
effects are mental events: and so mental activity is involved.
We now have a first sense for the word "consciousness". If there
is mental activity occurring in the mind, if something mental is
actually happening, then that mind is not totally unconscious. It is
therefore conscious. A single faint sensation is not much, but if it
occurs, to that extent there is consciousness. Unconsciousness is
not total. I call consciousness in this sense "minimal" conscious-
ness.
It is alleged that it sometimes occurs that someone wakes up
knowing the solution to, say, a mathematical problem, which they
did not know when they went to sleep. If we rule out magical
explanations, then there must have been mental activity during
sleep. To that extent, there was minimal consciousness. This is
:ompatible with the completest "unconsciousness" in a sense still
to be identified.
PERCEPTUAL CONSCIOUSNESS
Among the mental activities, however, it appears that we make a

special link between consciousness and perception. In perception,
there is consciousness of what is currently going on in one's
environment and in one's body. (Of course, the consciousness may
involve illusion.) There is an important sense in which, if a person
is not perceiving, then he is not conscious, but if he is perceiving,
then he is conscious. Suppose somebody to be dreaming. Since

there is mental activity going on, the person is not totally uncon-
scious. He is minimally conscious. Yet is there not some obvious
sense in which he is unconscious? Now suppose that this person
starts to perceive his environment and bodily state. (I do not want
to say "suppose he wakes up", because perhaps there is more to
waking up than just starting to perceive again.) I think that we
would be inclined to say that the person was now conscious in a
way that he had not been before, while merely dreaming. Let us
say, therefore, that he has regained "perceptual" consciousness.
This is a second sense of the word "consciousness". Perceptual
consciousness entails minimal consciousness, but minimal con-
sciousness does not entail perceptual consciousness.
INTROSPECTIVE CONSCIOUSNESS
Let us suppose, now, that there is mental activity going on in a

person, and that this activity includes perception. If what has been
said so far is accepted, then there are two senses in which such a
person can be said to be conscious. He or she has minima/ con-
sciousness and has perceptual consciousness. There is, neverthe-
less, a third sense, in which such a person may still "lack con-
sciousness". Various cases may be mentioned here. My own
favourite is the case of the long-distance truck-driver. It has the
advantage that many people have experienced the phenomenon.
After driving for long periods of time, particularly at night, it is
possible to "come to" and realize that for some time past one has
been driving without being aware of what one has been doing. The
coming-to is an alarming experience. It is natural to describe what
went on before one came to by saying that during that time one
lacked consciousness. Yet it seems clear that, in the two senses of
the word that we have so far isolated, consciousness was present.
There was mental activity, and as part of that mental activity, there
was perception. That is to say, there was minimal consciousness
and perceptual consciousness. If there is an inclination to doubt
this, then consider the extraordinary sophistication of the activi-
ties successfully undertaken during the period of "unconscious-
ness".
A purpose was successfully advanced during that time: that of

driving a car along a road. This purpose demanded that various
complex sub-routines be carried out, and carried out at appropri-
ate points (for instance, perhaps the brake or the clutch was used).
Were not these acts purposeful? Above all, how is it possible to
drive a car for kilometres along a road if one cannot perceive that
road? One must be able to see where one is going, in order to
adjust appropriately. It would have to be admitted, at the very
least, that in such a case, eyes and brain have to be stimulated in
just the same way as they are in ordinary cases of perception. Why
then deny that perception takes place? So it seems that minimal
consciousness and perceptual consciousness are present. But
something else is lacking: consciousness in the most interesting
sense of the word.
The case of the long-distance truck-driver appears to be a very
special and spectacular one. In fact, however, I think it presents us
with what is a relatively simple, and in evolutionary terms rela-
tively primitive, level of mental functioning. Here we have more or
less skilled purposive action, guided by perception, but apparently
no other mental activity, and in particular no consciousness in
some sense of "consciousness", which differs from minimal and
perceptual consciousness. It is natural to surmise that such rela-
tively simple sorts of mental functioning came early in the course
of evolutionary development. I imagine that many animals, par-
ticularly those whose central nervous system is less developed
than ours, are continually, or at least normally, in the state in
which the long-distance truck-driver is in temporarily. The third
sort of consciousness, I surmise, is a late evolutionary develop-
ment.
What is it that the long-distance truck-driver lacks? I think it is
an additional form of perception, or, a little more cautiously, it is
something that resembles perception. But unlike sense-perception,
it is not directed towards our current environment and/or our
current bodily state. It is perception of the mental. Such "inner"
perception is traditionally called introspection, or introspective
awareness. We may therefore call this third sort of consciousness
"introspective" consciousness. It entails minimal consciousness. If
perceptual consciousness is restricted to sense-perception, then
introspective consciousness does not entail perceptual conscious-

ness.
Introspective consciousness, then, is a perception-like aware-
ness of current states and activities in our own mind. The current
activities will include sense-perception: which latter is the aware-
ness of current states and activities of our environment and our
body. And (an important and interesting complication) since intro-
spection is itself a mental activity, it too may become the object of
introspective awareness.
Sense-perception is not a total awareness of the current states
and activities of our environment and body. In the same way,
introspective consciousness is not a total awareness of the current
states and activities of our mind. At any time there will be states
and activities of our mind of which we are not introspectively
aware. These states and activities may be said to be unconscious
mental states and activities in one good sense of the word
"unconscious". (It is close to the Freudian sense, but there is no
need to maintain that it always involves the mechanism of repres-
sion.) Such unconscious mental states and activities of course may
involve minimal and/or perceptual consciousness, indeed the
activities involve minimal consciousness by definition.
Just as perception is selective - not all-embracing - so it also
may be mistaken. Perceptions may fail to correspond, more or less
radically, to reality. In the same way, introspective consciousness
may fail to correspond, more or less radically, to the mental reality
of which it is a consciousness. (The indubitability of consciousness
is a Cartesian myth, which has been an enemy of progress in
philosophy and psychology.)
Following Locke, Kant spoke of introspection as "inner sense",
and it is essentially Kant's view I am defending here. By "outer
sense", Kant understood sense-perception. There is, however, one
particular form of "outer sense" that bears a particularly close for-
mal resemblance to introspection. This is bodily perception or
proprioception, the perception of our own current bodily states
and activities. If we consider the objects of sight, sound, touch,
taste and smell, then we notice that such objects are intersubjec-
tively available. Each of us is capable of seeing or touching
numerically the very same physical surface, hearing numerically
the very same sound, tasting numerically the same tastes or

smelling numerically the same smell. But the objects of proprio-
ception are not intersubjectively available in this way.
Consider, for instance, kinaesthetic perception, which is one
mode of proprioception. Each person kinaesthetically perceives
(or, in some unusual cases, misperceives) the motion of his own
limbs and those of nobody else. There is no overlap of kinaesthetic
objects. This serves as a good model for, and at the same time it
seems to demystify, the privacy of the objects of introspection.
Each of us perceives current states and activities in our own mind
and that of nobody else. The privacy is simply a little more com-
plete than in the kinaesthetic case. There are other ways to per-
ceive the motion of my limbs besides kinaesthetic perception -
for instance, by seeing and touching. These other ways are inter-
subjective. But, by contrast, nobody else can have the direct
awareness of my mental states and activities that I have. This pri-
vacy, however, is contingent only. We can imagine that somebody
else should have the same direct consciousness of my mental states
and activities that I enjoy. (They would not have those states, but
they would be directly aware of them.)
Perception is a causal affair. If somebody perceives something,
then it is involved in the perception; it is even involved in the con-
cept of perception: that the thing perceived acts upon the per-
ceiver, causing the perception of the object. If introspective
consciousness is to be compared to perception, then it will be
natural to say that the mental objects of introspection act within
our mind so as to produce our introspective awareness of these
states. Indeed, it is not easy to see what other naturalistic account
of the coming-to-be of introspections could be given. If intro-
spection is a causal process, then it will follow, incidentally, from
our earlier definition of causal quiescence that whenever we are
introspectively aware of one of our mental states, then that state is
not at that time causally quiescent.
TYPES OF INTROSPECTIVE CONSCIOUSNESS
Perhaps we still have not drawn enough distinctions. Sometimes

the distinction is drawn between mere "reflex" consciousness,
which is normally always present while we are awake (but which is

lost by the long-distance truck-driver), and consciousness of a
more explicit, self-conscious sort.
This difference appears to be parallel to the difference between
mere "reflex" seeing, which is always going on while we are awake
and our eyes are open, and the careful scrutinizing of the visual
environment that may be undertaken in the interest of some pur-
pose we have. The eyes have a watching brief at all times that we
are awake and have our eyes open; in special circumstances, they
are used in a more attentive manner. Un close scrutiny by human
beings, introspective consciousness is often, although not invaria-
bly, also called into play. We not only give the object more atten-
tion but have a heightened awareness of so doing. But, presuma-
bly, in lower animals such attentive scrutiny does not have this
accompaniment.) Similarly, introspective consciousness normally
has only a watching brief with respect to our mental states. Only
sometimes do we carefully scrutinize our own current state of
mind. We can mark the distinction by speaking of "reflex" intro-
spective awareness and opposing it to "introspection proper". It is
a plausible hypothesis that the latter will normally involve not only
introspective awareness of mental states and activities but also
introspective awareness of that introspective awareness. It is in any
case a peculiarly sophisticated sort of mental process.
WHAT IS SO SPECIAL ABOUT INTROSPECTIVE CONSCIOUSNESS?
There remains the feeling that there is something quite special

about introspective consciousness. The long-distance truck-driver
has minimal and perceptual consciousness. But there is an impor-
tant sense, we are inclined to think, in which he has no exper-
iences, indeed is not really a person, during his period of intro-
spective unconsciousness. Introspective consciousness seems like a
light switched on, which illuminates utter darkness. It has seemed
to many that with consciousness in this sense, a wholly new thing
enters the universe.
I now will attempt to explain why introspective consciousness
seems to have, but does not necessarily actually have, a quite
special status in the world. I proceed by calling attention to two
points, which will then be brought together at the end of the sec-
tion.
First, it appears that introspective consciousness is bound up in
a quite special way with consciousness of self. I do not mean that
the self is one of the particular objects of introspective awareness
alongside our mental states and activities. This view was some-
what tentatively put forward by Russell in The Problems of Phil-
osophy (1912: Ch. 5), but had already been rejected by Hume.
and by Kant. It involves accepting the extraordinary view that
what seems most inward to us, our mental states and activities, are
not really us. What I mean rather is that we take the states and
activities of which we are introspectively aware to be states and
activities of a single continuing thing.
In recent years, we have often been reminded, indeed admon-
ished, that there is a great deal of theory involved even in quite
unsophisticated perceptual judgements. To see that there is a
tomato before our body is already to go well beyond anything that
can be said to be "given", even where we do not make excessive
demands (such as indubitability) upon the notion of the given.
Consider knowingly perceiving a tomato. A tomato, to be a
tomato, must have sides and back, top and bottom, a certain
history, certain casual powers; and these things certainly do not
seem to be given in perception. If we consider the causal situation,
it is only the shape, size and colour of some portions of the surface
of the tomato (the facing portions) that actually determine the
nature of the stimulation that reaches our eyes. This suggests that,
at best, it is only these properties that are in any way "given" to
us. The rest is, in some sense, a matter of theory, although I do not
think that we should take this to mean that the perceptual judge-
ment that there is a tomato before us is a piece of risky specu-
lation.
It is therefore natural to assume that the perceptions of "inner
sense" involve theory, involve going beyond the "given", in the
same general way that the perceptions of "outer sense" do. In par-
ticular, whatever may be the case with other animals, or with small
children, or with those who, like the Wild Boy of Aveyron, have
not been socialized, for ordinary persons, their mental states and
activities are introspected as the states and activities of a single

thing.
Once again, the comparison with proprioception seems to be
instructive. We learn to organize our proprioceptions so that they
yield us perceptions of a single, unitary, physical object, our body,
concerning which our proprioceptions give us certain information:
its current posture, temperature, the movement of its limbs, and so
on. This is clearly a theoretical achievement of some sophisti-
cation.
In the same way, we learn to organize what we introspect as
being states of, and activities in, a single continuing entity: our
self. Mere introspective consciousness, of course, is not at all clear
just what this self is. At a primitive level perhaps, no distinction is
made between the self and the body. Identification of the thing
that is introspected as, say, a spiritual substance, or as the central
nervous system, goes far beyond the level of theorizing involved in
ordinary introspection. But the idea that the states and activities
observed are states and activities of a unitary thing is involved.
Introspective consciousness is consciousness of self.
If it is asked why introspection is theory-laden in this particular
way, then an answer can be suggested. It is always worth asking
the question about any human or animal organ or capacity: "What
is its biological function?" It is therefore worth asking what is the
biological function of introspective consciousness. Once the ques-
tion is asked, then the answer is fairly obvious: it is to sophisticate
our mental process in the interests of more sophisticated action.
Inner perception makes the sophistication of our mental pro-
cesses possible in the following way. If we have a faculty that can
make us aware of current mental states and activities, then it will
be much easier to achieve integration of the states and activities,
to get them working together in the complex and sophisticated
ways necessary to achieve complex and sophisticated ends.
Current computer technology provides an analogy, though I
would stress that it is no more than an analogy. In any complex
computing operation, many different processes must go forward
simultaneously: in parallel. There is need, therefore, for an overall
plan for these activities, so that they are properly co-ordinated.
This cannot be done simply in the manner in which a "command
economy" is supposed to be run: by a series of instructions from

above. The co-ordination can only be achieved if the portion of the
computing space made available for administering the overall
plan is continuously made "aware" of the current mental state of
play with respect to the lower-level operations that are running in
parallel. Only with this feedback is control possible. Equally, intro-
spective consciousness provides the feedback (of a far more
sophisticated sort than anything available in current computer
technology) in the mind that enables "parallel processes" in our
mind to be integrated in a way that they could not be integrated
otherwise. It is no accident that fully alert introspective conscious-
ness characteristically arises in problem situations, situations that
standard routines cannot carry one through.
We now can understand why introspection so naturally gives
rise to the notion of the self. If introspective consciousness is the
instrument of mental integration, then it is natural that what is per-
ceived by that consciousness should be assumed to be something
unitary.
There is nothing necessary about the assumption. It may even
be denied on occasion. Less sophisticated persons than ourselves,
on becoming aware of a murderous impulse springing up, may
attribute it not to a hitherto unacknowledged and even dissociated
part of themselves, but to a devil who has entered them. In
Dickens' Hard Times, the dying Mrs Gradgrind says that there
seems to be a pain in the room, but she is not prepared to say that
it is actually she that has got it. In her weakened condition, she has
lost her grip upon the idea that whatever she introspects is a state
of one unitary thing: herself.
But although the assumption of unity is not necessary, it is one
we have good reason to think true. A Physicalist, in particular, will
take the states and activities introspected to be all physical states
and activities of a continuing physical object: a brain.
That concludes the first step in my argument: to show that, and
in what sense, introspective awareness is introspective awareness
of self. The second step is to call attention to the special connec-
tion between introspective consciousness and event-memory, that
is, memory of individual happenings. When the long-distance
truck-driver recovers introspective consciousness, he has no
memory of what happened while it was lacking. One sort of

memory-processing cannot have failed him. His successful navi-
gation of his vehicle depended upon him being able to recognize
various things for what they were and treat them accordingly. He
must have been able to recognize a certain degree of curve in the
road, a certain degree of pressure on the accelerator, for what they
were. But the things that happened to him during introspective
unconsciousness were not stored in his event-memory. He lived
solely in the present.
It is tempting to suppose, therefore, as a psychological hypo-
thesis, that unless mental activity is monitored by introspective
consciousness, then it is not remembered to have occurred, or at
least it is unlikely that it will be remembered. It is obvious that
introspective consciousness is not sufficient for event-memory.
But perhaps it is necessary, or at least generally necessary. It is
notoriously difficult, for instance, to remember dreams, and it is
clear that, in almost all dreaming, introspective consciousness is
either absent or is at a low ebb.
So it may be that introspective consciousness is essential or
nearly essential for event-memory, that is, memory of the past as
past. A fortiori, it will be essential or nearly essential for memory
of the past of the self.
The two parts of the argument now may be brought together. If
introspective consciousness involves (in reasonably mature human
beings) consciousness of self, and if without introspective con-
sciousness there would be little or no memory of the past history
of the self, the apparent special illumination and power of intro-
spective consciousness is explained. Without introspective con-
sciousness, we would not be aware that we existed - our self
would not be self to itself. Nor would we be aware of what the par-
ticular history of that self had been, even its very recent history.
Now add just one more premiss: the overwhelming interest that
human beings have in themselves. We can then understand why
introspective consCiousness can come to seem a condition of any-
thing mental existing, or even of anything existing at all.
s
Ac~in~ and Tryin~
In this paper, it will be argued first that if a person, A, performs an

intentional action, P, then it is entailed that A tried (attempted) to
do P. This conclusion raises two problems. What further con-
siderations must be added to "A tried to do P " in order to yield
necessary and sufficient conditions for "A did P intentionally"?
The second task of this paper is to sketch an answer to this prob-
lem. A second question raised by the alleged entailment between
doing something intentionally and trying to do that thing is what
we are to understand by trying more or less hard. The third objec-
tive of the paper, which can be achieved briefly, is to give an
analysis of trying hard.
"A DOES P INTENTIONALLY" ENTAILS "A TRIES TO DO P"
It is not the case that if A performs an action, P, then A must have

tried or attempted to do P. (I cannot see any important theoretical
distinction between trying and attempting.) For A may have done
P accidentally or inadvertently. But if A did P intentionally then, I
suggest, the entailment holds.
The main argument that I have for this contention is a single,
very striking case. I will call it "The Case of the Unexpected
Paralysis". An "Argument from Paralysis" is already known in the
literature, with the object of proving that acting intentionally
entails trying so to act. Strangely enough, the argument is given its
most sympathetic treatment by a philosopher who rejects the
entailment: Richard Taylor, in his book Action and Purpose
Acting and Trying 69
(1966). Taylor rejects the argument but admits its power. I think
that this power is still greater, or at any rate is more clearly evi-
dent, if the paralysis is unexpected by the agent.
Suppose, then, that somebody asks A to get something down
from the shelf, and that A is perfectly willing and, as he thinks,
able to comply at once. Suppose, however, that before A begins to
comply, he is struck by an instantaneous and complete paralysis of
the arm that is tC' move. Suppose, further, that this arm does not
feel different to him in any way, so that he does not have any clue
to his strange situation. The situation is a strange one, but it is
surely a possible one. In such a situation, I suggest, A can, and pre-
sumably will, try or attempt to move his arm. Unless he does this,
he is not likely to discover that his arm is paralysed, for to say that
the arm is paralysed is to say that it will not move if he tries to
move it.
We remember the classical problem. When a man raises his
arm, his arm rises. But his arm can rise although he did not raise
his arm. What condition must be added to the rising of the arm to
yield its raising? In the case we have imagined, actual motion of
the arm has been totally removed. What is left? The almost
inevitable answer seems to be: a mental event, the trying or
attempting to raise the arm. This mental event is required to turn
the rising into a raising. (Though necessary, it is not sufficient. As
we shall see, it is requisite also that the mental event stand in the
right relations to the rising of the arm.)
The argument can now be taken further. It is always possible (at
least logically possible) that in the course of doing the easiest and
most routine action the agent may suddenly and unexpectedly be
prevented from doing it. Suppose that this happens. In all such
cases, it would seem true to say that the agent was trying or
attempting to do certain things. In our original case, the trying or
attempting had to be a mental event because there was nothing
else for it to be. If it is a mental event in this case, then it is natural
to suppose that it is a mental event in all cases of sudden and unex-
pected interruption. A third step in the argument follows. If a men-
tal event of trying is present in all cases of sudden and unexpected
interruption, then it is natural to suppose that it is present in all
intentional action.
Obviously, this argument is not an apodeictic one. But I think it

is powerful and persuasive. It should be noticed, however, that not
one, but two conclusions have been drawn from it. It was argued
both that all intentional action involves a mental event, and that
this event is a "trying" or "attempting" to do something. The first
of these conclusions seems to be especially hard to get away from.
What can go on in the agent in the original case except an inner
mental event? (Although no special force should be given to the
word "event" here.) There must be something that marks him off
from an unexpectedly paralysed man who does not try to move the
paralysed limb. By hypothesis, the something that marks him off is
not any motion of the arm. What else can it be but something in
his mind? There is no other plausible candidate. And once this is
admitted, by far the simplest conceptual hypothesis is that this sort
of mental event is an ingredient in all action, the paralysis case
simply revealing clearly what is present in all cases.
But even if this first point is granted, and from here on it will be
assumed, it still may be questioned whether this mental event is
properly called "trying" or "attempting". Perhaps there is no term
for this mental something in our ordinary discourse, and some
term of art, such as "volition", is required.
The obvious preliminary argument for the view that the mental
event is properly called "trying" or "attempting" is that this is the
description we naturally give in cases such as the unexpected
paralysis. The victim will describe himself as having tried to move
his arm, even although he had no inkling that the paralysis had
occurred until he went to move his arm.
It is very important to notice here that this sort of trying is a
different sort of thing from trying more or less hard. The latter sort
of trying necessarily involves the agent in more or less strain and/
or difficulty in his attempt to achieve some objective. Nothing of
that sort need be involved in the unexpected paralysis case. An
account of trying hard, linguistically the more conspicuous notion,
but, I believe, much the less important notion for the analysis of
action, is left for the brief final section of this paper.
The obvious preliminary argument against the view that trying
or attempting is present in all cases of intentional action is that
in ordinary situations involving no difficulty we never use the
terms "trying" or "attempting". If I reach up to the kitchen shelf

without any trouble at all, it sounds absurd to say that I tried or
attempted to do this.
This linguistic point is, however, far from conclusive. As is now
well appreciated, there are propositions that are true but, if given
linguistic expression in certain situations, naturally lead an
audience to form false beliefs. Suppose that I know that Smith is
gone, never to return, that you ask for him and that I tell you that
he is not in at present. I have told you the truth. The trouble is that
I have not told you the whole truth, and that my saying what I said
will naturally lead you to conclude that, as far as I know, Smith
will be back later. To the extent that I can anticipate that you will
come to this conclusion, I may be said to have implied that, as far
as I know, Smith would be back later. I could have "cancelled"
this implication - to use Grice's term - by adding, "Mind you,
he will never be back," although this would raise the question why
I ever made the misleading remark in the first place.
The moral of this is: what is misleading to say can still be true.
And so, given a case where it is misleading to say that A tried or
attempted to raise his arm, it does not follow that it is false that A
tried or attempted to raise is arm. Perhaps the misleadingness of
the utterance can be explained in some other way. In particular,
perhaps, it says too little relative to what the speaker knows, just
as saying that Smith is not in at present says too little relative to
what the speaker knows. If all action involves trying, how pointless
to say so in the case of a routine action routinely performed!
In this way, one who maintains as I do that all intentional action
involves trying or attempting can very plausibly explain away the
objection from "what we naturally say". However, those who
maintain that only some intentional actions involve trying can also
make shift to explain the linguistic situation. The word "trying",
they may maintain, involves a sort of retrospective baptism. Look-
ing back upon the earlier situation of a man who, it later tran-
spires, was headed for disaster, we may refer to him as "the
doomed man". We speak of him as doomed then, although what
constitutes his being doomed is simply and solely the disaster to
come. In the same way, it may be suggested, in the case of the
unexpected paralysis, the mental event that fails to issttP. in mntinn
of the arm is called "trying" only because of the subsequent

failure.
How are we to decide between these competing hypotheses? I
cannot see how to force a conclusive decision, but I think I can
create a fair amount of embarrassment for the second view. For
there are other sorts of case where we speak of "trying" (and do
not mean "trying more or less hard") where it is not the case that
the term is a retrospective baptism sanctioned by the failure of the
thing attempted to occur.
Consider a case where a man opens a door without any diffi-
culty. However, shortly before doing this, he had tried unsuc-
cessfully to open that door. As a result, his effortless success on the
second occasion was a complete surprise to him. We can say truly
and unmisleadingly of such a man that he tried to open the door
on the second occasion. Yet he did not fail (nor did he put forth
any particular effort). The only special factor in the situation was
his belief that he was quite likely to fail.
In the unexpected paralysis case, there is failure, but no belief in
the likelihood of failure. In the new case, there is belief in the like-
lihood of failure, but no failure or even relative difficulty in per-
formance. The cases have no more than a family resemblance. Yet
the term "trying" is naturally used in both cases.
There is at least one other sort of case where we can naturally
use the word "trying", although again it is not a case of trying
more or less hard. Suppose that I know that A is currently pursuing
some objective, but I know very little else. I do not know whether
success is, or is not, a foregone conclusion. I do not know anything
about what A believes about his chances of success. All I know is
what his objective is and that he is pursuing it. In such circum-
stances, will it not be linguistically legitimate for me to say that I
know that A is at least attempting or trying to achieve whatever his
objective is? "I don't know what his chances of success are. For all I
know, the thing is a pushover, and he knows it is a pushover. But I
do know that that is what he is trying to do." There seems to be no
infelicity in using the word "trying" here. We instead could speak of
what A 's objective was, or what he planned to do. But we can also
speak of what he is trying to do. Suppose, further, as may well be the
case, that A never anticipated any difficulty, nor did he find
any. The action was routine. In this case, the agent was properly
spoken of as trying, although the action involved no difficulty and
the agent anticipated no difficulty. The only shadow of failure in
the situation is that the speaker could not definitely rule out the
possibility of the agent's failure and/or the agent's belief in the
possibility of failure.
So the conditions where we normally would speak of trying or
attempting (and do not mean trying more or less hard) are very
heterogeneous. It is hard to discern any unified set of necessary
and sufficient conditions that unites the different sort of case. We
could rest in the conclusion that the cases have a mere "family
resemblance", but this is an hypothesis that should be embraced
only where other hypotheses have failed. It is natural to suppose,
instead, that these cases are the mere visible part of the iceberg,
and that trying is involved in all cases of intentional action.
Contrariwise, it would not be at all surprising that the conditions
under which it is natural to use the word "trying" should be of a
heterogeneous sort.
Two final pieces of evidence. First, consider the third-person
sentences, "He did it intentionally, but he made no attempt to do
it" and "He did it intentionally, but he did not in any way try to do
it." These sentences seem in some way incoherent. Why? A
natural hypothesis to explain the incoherence is that the sentences
are self-contradictory.
Second, we have agreed that the unexpected paralysis case does
make it plausible, at least, that a mental event, distinct from any
action, is involved in making an action an action. Now, if this men-
tal event is not trying or attempting, then ordinary discourse gives
it no name. We have to introduce a philosopher's term of art, such
as "volition". It would be rather surprising if ordinary language
had overlooked the task of providing a name for this vital mental
constituent of all action.
NECESSARY AND SUFFICIENT CONDITIONS FOR "A DOES P

INTENTIONALLY"
Suppose it now be granted that A's doing P intentionally entails

that A tried or attempted to do P. Trying is then a necessary condi-
tion for acting. We can go on to ask the question, "What further

conditions must be added to trying to do P to yield the doing of P
intentionally?" We can compare this question with the question,
"Given that belief is a necessary condition for knowledge, what
further conditions must be added to yield knowledge?"
If a belief is to be a case of knowledge, then it is necessary that
the thing believed be true. Similarly, if A does P, not only must A
try to P, but the thing A tries to do - P - must actually occur.
Is this extra condition sufficient to yield the intentional doing of
P ? It is notorious that true belief is insufficient for knowledge.
The same is true in the case of action. A man goes to raise his arm,
say to vote, but is struck by a sudden paralysis. By a coincidence,
his arm happens to be entangled in a rope, which happens to be
pulled up at that moment. He tried to raise his arm, and his arm
rose. But he did not raise his arm.
What must be added? The obviously unsatisfactory feature of
the case just mentioned is that the man's arm going up had nothing
to do with his trying to put his arm up. The two were found in
accidental conjunction, in just the same way that fantasy and fact
can coincide to produce a true belief that is not knowledge. But
there is an obvious way to eliminate this unsatisfactory feature of
the case. Let us say that not only does P occur, but that P occurs
as a causal result of A 's trying to bring about P. Perhaps a P with
that causal ancestry will be an intentional action of A 's.
There has been much criticism, notably, of course, by Ryle
(1949), of the view that such things as tryings, attemptings, pur-
posings, intendings, etc. are, or can be, causes of the thing
attempted, purposed, intended. But since Donald Davidson's arti-
cle, "Actions, Reasons and Causes" (1963), casual accounts are
respectable again, and there is no need to spend a great deal of
time defending the view that A's trying to do P can bring about P.
The trickiest argument to be met is that such causes can only be
characterized in terms of their effects, thus destroying the alleged
contingency of the casual relation. The arm-raising attempt can
only be characterized as that mental cause that, with luck, brings it
about that the arm goes up, while the head-turning attempt is that
mental cause that, with luck, brings it about that the head is
turned. It sounds suspicious! In fact, however, causes are often
characterized in terms of their effects. Parents bring children into

existence; genes produce hereditary characteristics. It is true that
parents and genes can be characterized independently of their
parental or genetic casual role: as men and women, as DNA. But
who has shown that similar independent characterizations of try-
ings and attemptings are impossible? It is true that we seem to lack
any introspective awareness of this independent nature, a differ-
ence that has helped to obscure the resemblance of the situation to
that of parents and genes. However, as I have argued elsewhere at
length, 1 this merely epistemological difference does not prevent
our conceiving of tryings, attemptings, etc. as causes. Further-
more, I think that there is a plausible hypothesis concerning their
independent nature: they are physical processes in the brain.
Is is true also that tryings must not be conceived as actions in
the sense that A's raising his arm is an action. There are mental
actions, as opposed to mere mental happenings. For instance,
doing a piece of mental arithmetic is an action. This does not just
unroll in your head: you have to keep at it. But tryings cannot be
mental actions, or rather, not all tryings can be mental actions, on
pain of an infinite regress where a trying stands behind every try-
ing. This Ryle really did prove, and it is a point of the utmost
importance. But there is no reason why many tryings should not
be mental happenings "arising in the soul from unknown causes".
Here it is worth calling attention to a case that gave Richard
Taylor a good deal of puzzlement in Action and Purpose (1966:
194-95). A 's arm moves as the causal result of certain electrical
impulses travelling along certain nerve paths. Suppose that, for
scientific purposes, A desires to bring this pattern of impulses into
existence. How is it to be done? The obvious answer is: let A
move his arm. But this is rather baffling. A's moving his arm
appears to be a means to the end of producing the pattern of
impulses. But are not means causes of their ends? Yet A's moving
his arm is no cause of the pattern of impulses. Indeed the situation
is exactly the reverse. The impulses precede and cause the motion
of the arm.
Taylor, who rejects the view that actions are caused by tryings
1. See my Materialist Theory of the Mind (1968), ch. 7.

or attemptings, or indeed any other mental event, cannot handle

this case. He simply says that it is "frightfully puzzling". A causal
account, and I think a causal account alone, dissolves the problem
with ease. What causes the pattern of impulses, and so what is the
means of producing them, is not moving the arm, but rather trying
to move the arm. This mental event, whether it be motion in the
soul or process in the brain, brings about the pattern of impulses,
which in turn brings about the motion of the arm. The motion of
the arm is not a means to the end, but rather an almost inevitable
further effect of the means adopted. 2
Are we at the end of our search now? Can we say that A does P
intentionally if, and only if, P occurs as a causal result of A trying
to do P? Unfortunately not. The conditions are still too wide.
Consider a case similar to that given by Roderick Chisholm
(1964). A sets out across town with the object of killing his uncle.
The excitement generated by his plan causes him to drive fast and
carelessly. As a result, he knocks down and kills a pedestrian
whom he does not see. It turns out to be his uncle. A's attempt to
murder his uncle brought it about that he killed his uncle. But his
killing was not intentional. A's killing of his uncle belongs to that
wide class of happenings that are the causal result of A's trying to
do something. In some wide sense, all these happenings are A's
acts. In some wide sense, he did these things. But we are interested
in, and wish to mark off, a subclass of the acts, a subclass that are
A's acts par excellence, namely those acts A does intentionally.
For although A tried to kill his uncle, his killing of his uncle was
not an intentional act.
How is a case like Chisholm's to be excluded? The reflection
that naturally occurs is that, although A's attempt brought about
the death of his uncle, it did not bring it about in the right way.
2. It perhaps should be noted that if tryings are identical with physical processes
in the brain, then the agent deliberately might bring a certain brain process
into existence by trying to move his arm, but the brain process might not be
the effect of the trying because it was identical with the trying. If means are
always causes of their ends, then this trying would not be a means of pro-
ducing this brain process. I am uncertain whether or not the concept of a
means should be extended to cover this sort of case. It is clear, however, that
means are never the effects of their ends.
The causal pattern in which A's attempt to bring about his uncle's
death brings about that death is not the right sort of pattern. What
would the right sort of pattern be? The special mark of the opera-
tion of the will is that it is a cause that operates by utilizing the
agent's knowledge and beliefs. As I have put it elsewhere, the will
is an "information-sensitive cause". Perhaps, then, in Chisholm's
case the agent's knowledge and beliefs are not playing the causal
role, which they should play in a case of intentional action. Let me
try to spell this out.
But first a preliminary remark. The intentional action of killing
one's uncle is a complex and sophisticated affair. It involves all
sorts of complex adjustments of means to ends. Suppose, instead,
we consider some bedrock or basic action such as moving one's
arm. Could a Chisholm-type counter-example be constructed
here? Suppose that A tries to move his arm and suppose as a result
that A's arm moves, but not as a result of any means that A uses in
order to get this result (A does not make his arm move by pulling
on a rope with his teeth, for instance). It is plausible to suggest that
the motion of A's arm then must be an intentional action. There
seems to be no room for counter-examples like the one proposed
by Chisholm. I will take up the point again at the end of this sec-
tion.
Returning to the main line of the argument, suppose that at a
certain time A embarks upon the attempt to bring about P, and at
a later time succeeds. He does P intentionally, and Pis some medi-
ate or non-basic type of action, such as killing one's uncle. Not all
of A's actions during this time-interval need be relevant to the
bringing about of P, but some will be. We can think of these rele-
vant actions, including the bringing about of P, as resulting from a
certain chain of practical reasoning. The reasoning will be mixed,
involving (a) those things A has as his objectives; (b) things that A
believes. (Taking what A knows as a subclass of the things that A
believes.)
Consider first premisses of the sort (a). These will include A's
objective of bringing about P. But A will have other desires, disin-
clinations, etc., which will complicate the situation. For instance,
it is unlikely that A will want P at any price. This may prevent A
from adopting certain means to the end P, even though he believes
that the means would be effective. Again, of two means judged by

A to be equally effective in bringing about P, one may be intrins-
ically much more attractive to A (driving by the lake rather than
driving by the factory). Again, as the action develops, situations
not originally planned for will regularly arise. As a result, A will
have to form new sub-objectives (under the same constraints of
other desires, disinclinations, etc.l as means to the end P.
The things of the sort (b) will be the things that A believes are
relevant to the bringing about of P. They further divide into (i)
beliefs about particular matters of fact - beliefs the contents of
which involve only spatia-temporally limited situations; and (ii)
beliefs about general connections, where beliefs about purely
mathematical or logical matters are included under the latter head-
ing.
(i) The beliefs about particular matters of fact will involve
various beliefs that the agent holds about himself and his situation
at the beginning of the action. They include, in particular, the true
belief that P does not (yet) obtain. If these beliefs are taken along
with the objective of achieving P (and any other initially present
desires, disinclinations, etc. that are relevant to what is under-
taken), the whole complex of beliefs and objectives may be
thought of as providing the premisses in the reasoning that deter-
mines what A begins by trying to do. As the situation develops,
further beliefs about the nature of the situation will become rele-
vant, and will play their part in determining what A tries to do
next. Finally, there will be the belief that P has been achieved, or
that some situation obtains that will in time engender P, a belief
that will bring the action to a close.
(ii) By contrast, A's relevant beliefs about general connections
generally will not be premisses of his reasoning, but rather will be
principles according to which the reasoning is conducted. 3 The
principles may include purely intellectual principles, but what may
be called "action-recipes" will regularly be involved. Action-
recipes are routines that the agent has in his repertoire, such as the
physical routines involved in riding a bicycle, which he can use in
the attempt to achieve certain results in certain situations. I have
3. See my Belief, Truth and Knowledge (1973), ch. 6.

not spoken of "skills" or "know-how", only because I wish to

include routines that are not tied to success in the way that skills
and know-how are tied. The witch-doctor's rain dance is an action-
recipe.
Now we can imagine A's progress towards P mapped in a series
of pieces of reasoning having these mixed premisses and these
principles of reasoning. The results of the pieces of reasoning, I
think, should be conceived of as "determinations of the will", that
is, as A's trying or attempting to do certain things. And if pur-
poses, beliefs, etc. can be thought of as causes, then we can think
of A acting as he does because, in the casual sense of "because",
he has these objectives, these beliefs about the current and
developing situation, these principles of reasoning and acting. The
pattern of the practical reasoning shadows out a pattern of opera-
tion of causal factors in A's mind.
(Although I cannot go into detail here, I think that A's beliefs
about general connections, including "action-recipes", may be
thought of as dispositions of A. 4 They will be dispositions to move
from certain beliefs to certain further beliefs, or, in the case of
inferences from mixed premisses, from certain beliefs and desires,
to certain determinations of the will, according to certain general
patterns of antecedent and consequent. And like all dispositions,
they should be conceived of as actual states of A, states that are
causal factors in the situation.)
We are now in a position to solve the problem raised by
Chisholm's counter-example. A's doing P intentionally entails that
A's trying to do P caused the occurrence of P. But the latter condi-
tion is not sufficient for A's doing P intentionally, at any rate
where P is not a simple or basic action. I suggest that the missing
factor is this: A's trying to do P must bring about P according to a
causal pattern involving purposes and beliefs, a pattern that can be
shadowed out in a train of practical reasonings of the sort that has
been indicated.
It should be emphasized that this chain of practical reasonings is
not the account of the situation that the agent himself can supply.
4. See again ibid., where, however, the discussion is confined to intellectual as

opposed to practical principles.
The agent's account may be a very thin affair indeed compared to

the pattern of purposes, desires, beliefs, etc. that actually moves
him. We can only give a very partial account of what our purposes
are, what our beliefs are, what principles of inference we actually
reason in accordance with, what "action-recipes" we actually use,
in the course of our activities. The spelling-out of the structure of
the practical reasoning involved would be a major research project
in psychology, not something to be discovered by direct intro-
spection. This is peculiarly so in the case of action-recipes for
physical routines. It is well known that in such cases the agent may
have quite misleading ideas of the way that he actually goes about
things.
But this raises a problem. In what fineness of detail should the
chain of practical reasonings be specified? For instance, does an
action-recipe for a physical routine require that the whole physio-
logical sequence of the contraction of muscles should be specified,
or the whole sequence of electrical impulses from the brain to
muscles? This seems absurd. Yet if the reasoning that the subject
could specify introspectively is not to be the guide to the practical
reasoning involved, where are we to stop?
In A Materialist Theory of the Mind (1968: ch. 7, sect. 5), I
have discussed what is in effect this problem. I there inquire into
the notion of immediate acts of the will, or, as Arthur Danto
(1965) has put it, "basic actions". When I move my arm, the
motion occurs as a result of trying to move my arm, which in turn
causes electrical impulses to travel down my arm, which causes
muscles to contract, which causes my arm to move. Yet the con-
duction of electrical impulses and the contraction of muscles are
not thought to be means to my end of moving my arm. I simply
move my arm. It is an "immediate act of the will", a "basic
action". But the question, however, is what makes us stop here.
Why should an action-recipe, for instance, naturally speak about
motions of the arm, but only most unnaturally make any reference
to the electrical impulses that cause these motions?
I think that the answer here is that the motion of the arm is the
first perceived effect of the operation of the will. The electrical
impulses and the contraction of muscles are not perceived. We
may know of their existence as a result of scientific investigation.
But we can see and, still more important, feel the arm moving.
But what is the special importance of perception here? It is this.
We have noted that the peculiar mark of the operation of the will is
that a vital causal role is played by the agent's beliefs, and in par-
ticular his beliefs about the nature of his current situation. Only so
can the agent adjust means to ends as the situation develops. With-
out a feedback of information about the current situation there
could be almost no possibility of an effective chain of actions
developing. Now it is the perception of the arm moving that pro-
vides the first information about the result of the original attempt,
and so gives the first opportunity to adjust conduct in the light of
the new situation. Only at this point does the characteristic mode
of operation of the will become possible. Hence we think of such
things as the moving of the arm as the basic units of action. No
action-recipe need specify the action more closely than this.
Suppose, however, that we are considering a brief motion of A's
arm, too brief to permit an adjustment of motion as a result of
feedback. Suppose that this motion of the arm has occurred
because (that is, as a causal result of) A's attempt to move his arm.
If the solution I have given to Chisholm's problem is correct, then
there can be no prospect of applying that solution to an~1 case that
may be proposed in connection with this arm motion. Suppose, for
instance, that we contemplate the possibility of a quite eccentric
causal chain leading from the attempt to move the arm to the
actual motion. Provided that that causal chain does not involve
any actions by A, my solution to Chisholm's problem is unavaila-
ble.
But do we need a solution? It seems to me not unreasonable to
say that, for the basic units of action, if the attempt to do the
action brings about the action, without the intervention of any
further action of the agent, then in all cases the action was the
agent's intentional action. For this case, I suggest, it does not mat-
ter what the nature of the intervening causal process is, provided
that it does not include actions of the agent. (It perhaps also may
be necessary to stipulate that the process remain within the agent.)
What may make this solution difficult to accept is the failure to
make the distinction between a mere intentional carrying out of a
basic action and the intentional carrying out of such an action that
is a manifestation of a skill or ability. If A tried to do P, and as a

result he does P, and does P intentionally, it does not follow that A
has the skill or ability to do P. For A could have been lucky. He
might lack the skill to do P, but he might have succeeded in bring-
ing off P by luck or a fluke. Suppose, for instance, that A tries to
move his normally paralysed arm and, because of a once only and
quite eccentric causal mechanism, his arm moves. I suggest that he
moved his arm intentionally. But A lacks the ability to move his
arm, unless we uselessly extend the notion of ability to cover
"once only" situations.
I have not discussed what it is for A's doing P intentionally to
be a manifestation of a skill or ability. That would be a further
topic, and quite an extensive one. Sufficient unto the day ...
DAVIDSON'S OBJECTION
The original version of this paper (1973) drew a criticism from

Donald Davidson (1973), which whether or not it is true, is cer-
tainly well worth answering. It seems necessary to quote him in
full:
David Armstrong, in a recent paper called "Acting and Trying",
comes closest to seeing the nature of the difficulty. He asks the
question what we must add to "A tried to do x" in order to have
necessary and sufficient conditions for "A did x intentionally".
Most of us would say - I certainly would - that trying itself
isn't necessary in many cases, but this point, though at the heart
of Armstrong's interest, is largely irrelevant to the present
theme. According to Armstrong, A's doing x intentionally
entails that A's trying to do x caused the occurrence of x, but
A's trying to do x, even if it causes the occurrence of x, does not
prove that A did x intentionally. The difficulty is, that the
attempt may bring about the desired effect in an unexpected or
undesired way. Here is an example of Daniel Bennett's. A man
may try to kill someone by shooting at him. Suppose the killer
misses his victim by a mile, but the shot stampedes a herd of
wild pigs that trample the intended victim to death. Do we want
to say the man killed his victim intentionally ? The point is that
not just any casual connection between rationalizing attitudes
and a wanted effect suffices to guarantee that producing the

wanted effect was intentional. The causal chain must follow the
right sort of route.
Armstrong tries to fill this gap by saying that the wanted
effect must be produced by a causal chain that answers, at least
roughly, to the pattern of practical reasoning. In Bennett's
example, we must suppose the agent intended to kill the victim
by pulling the trigger because he reasoned that pulling the trig-
ger would cause the gun to fire, which would cause the bullet to
fly, which would cause the bullet to penetrate the body of the
victim, thus causing his death. But the pattern of events por-
trayed by the pattern of practical reasoning was not produced by
the action of pulling the trigger. This throws in doubt the ques-
tion whether the agent intentionally killed his victim.
I am not sure whether or not this difficulty can be overcome,
but there is a related problem that Armstrong does not consider
which seems to be insurmountable. This is the problem, not of
quaint external causal chains, but of non-standard or lunatic
internal causal chains. Armstrong, in trying to mend the trouble
about unwanted external causal relations, was pushed into talk-
ing of the course of practical reasoning, the way in which beliefs
and desires interact to produce action. (Given this, !doubt that
there is any gain to him in the strained thesis that everything we
do intentionally is caused by trying to do it. But never mind
that.) And here we see that Armstrong's analysis ... must cope
with the question how beliefs and desires cause intentional
actions. Beliefs and desires that would rationalize an action if
they caused it in the right way - through a course of practical
reasoning, as we might try saying - may cause it in other ways.
If so, the action was not performed with the intention that we
could have read off from the attitudes that caused it. What I
despair of spelling out is the way in which attitudes must cause
actions if they are to rationalize the action.
Let a single example serve. A climber might want to rid him-
self of the weight and danger of holding another man on a rope,
and he might know that by loosening his hold on the rope he
could rid himself of the weight and danger. This belief and want
might so unnerve him as to cause him to loosen his hold, and
yet it might be the case that he never chose to loosen his hold,
nor did he do it intentionally. It will not help, I think, to add that
the belief and the want must combine to cause him to want to
loosen his hold, for there will remain the two questions how the
belief and the want caused the second want, and how wanting to
loosen his hold caused him to loosen his hold.
It is clear that Davidson is right to contend that it is possible
that a belief and a desire should combine to cause an action, yet
fail, in Davidson's phrase, to "rationalize the action". Yet at the
same time, the belief and desire might be such that they would
naturally serve to rationalize the action. It seems to me, however,
that the possibility of such cases involves no excessive difficulty
for my analysis.
The first thing to see is that there is the possibility of similar
phenomena in the purely physical sphere. Consider, first, a dis-
position such as brittleness. A brittle object is hit sharply, and as a
result, the object shatters. Is this a manifestation of the object's
brittleness? Surprisingly enough: not necessarily. Suppose that
the shattering is brought about by the striking by means of a once-
only, non-standard or lunatic casual chain. (We may suppose, in
addition, that this chain of events is of a sort that, if it occurred in
certain non-brittle objects, might shatter them.) I do not think that
this would be accounted a manifestation of the brittleness. We
might call it a pseudo-manifestation.
We may now come closer to Davidson's case. Consider a com-
puter that accepts input P. If the computer was in good working
order, then it would process this information and yield output Q. It
so happens, however, that this computer is not, for the moment at
least, in good working order. As a result, it does not process the
input P correctly. P is subjected to random manipulation. It hap-
pens further, however, that, by coincidence, the non-standard or
lunatic internal casual chain brings it about that output Q is pro-
duced. Input P has produced output Q, but, we may say, it has not
"rationalized" Q. In other words, the case is parallel to Davidson's
case of the climber.
But now, I suggest, to have the parallel is to be released from
Davidson's case. For it is easy to see why P does not "rationalize"
Q. P does not bring about Q as a result of standard computing
practices. Given the disorganization of the computer's internal

processes, inputs P', P", etc. would not have produced the correct
result Q', Q", etc. (If they would have produced the right result,
then the computer is not after all disorganized, but simply
eccentrically organized.)
It is worth noting, also, that it would be possible to determine,
in principle at least, that the computer had produced the standard
result Q by means of a non-standard or lunatic procedure.
The computer has within itself certain mechanisms with certain
powers. In the same way, we have within ourselves certain
"mechanisms" with certain powers, including certain "mecha-
nisms" capable of practical reasoning under the influence of beliefs
and desires. In Davidson's case, if the climber's belief and desire
work within him in the ordinary way, via the mechanism of prac-
tical reasoning, then his will is determined in the ordinary way. He
deliberately lets go. But in the case that Davidson describes, the
physical event of letting go occurs as a result of belief and desires,
but not as a result of the operation of the "mechanism" of prac-
tical reasoning. The letting-go is not therefore an intentional
action by the climber.
It would seem possible, also, at least in principle, to determine,
if the question arose, which sort of causal chain actually led to a
particular climber letting go on a particular occasion.
I suggest, then, that Davidson's lunatic internal causal chains
pose no more threat to the analysis of intentional action proposed
in this paper than do Chisholm's lunatic external causal chains.
TRYING HARD
But I still have to complete my account of trying by saying what it

is to try hard. All action, it has been argued, involves trying, but it
is clear that not all action involves trying hard. What is peculiar to
the situation of trying hard?
It was pointed out that the trying that is not trying more or less
hard, although involved in all action, can occur in the complete
absence of action (the paralysis case). Can trying more or less hard
occur in the complete absence of action? It can plausibly be
argued that it cannot. A paralysed man may "try hard" to move a
paralysed limb. But will this not involve action, whether physical
or mental? Will he not be tensing such muscles as he can still con-
trol, or effortfully concentrating his thoughts upon the matter in
hand?
Originally, I thought this view was correct, but now I am not
sure. Consider again the man who is as yet unaware that his arm is
paralysed. Suppose that what he tries to do, in the sense in which
trying is involved in all action, is something that is right up to the
limit of his (unparalysed) capacity. Perhaps, on a given signal, he
tries to get his arm up faster than a competitor. Once he knows
that his arm is paralysed, he cannot try hard to defeat his oppo-
nent, for once he knows for certain that his arm is paralysed, he
cannot even try to move his arm: the will is an information-sensi-
tive cause. But if he is unaware of his paralysis, can he not try hard
to defeat his opponent? Tentatively, I suggest that he can.
The following definition is designed to cover this unusual case,
as well as more usual ones:
A tries more or less hard (tries2 ) to do P if, and only if:
(i) A tries 1 (the trying that, I have argued, is involved in all
action) to do P ;
(ii) A tries 1 to put a relatively great effort of some sort into
the achieving of P ;
(iii) A tries 1 to put in a relatively great effort of some sort
because A believes that, by comparison with a lesser
effort of that sort, this effort is necessary to achieve P.
Strictly, condition (iii) makes condition (ii) redundant, and con-
dition (ii) makes condition (i) redundant, but I hope that the mode
of presentation used makes the analysis somewhat clearer. The
"relatively great effort of some sort" that A tries (tries 1 ) to exert,
but which he may fail to exert, is to be explained in the following
way. There are various sorts of activities of which A is capable, or
believes himself to be capable, where the degree of the activity
exhibited can be quantified, either precisely or roughly. For ins-
tance, restricting ourselves to capacities that A actually has, there
are various degrees of certain sorts of muscular force, all degrees of
which A is capable of exerting. Less precisely, there are various
degrees of mental concentration on certain sorts of topic, all
degrees of which A is capable of exhibiting. These activity-scales
may be much more complex than the examples given. All such
scales will have rather imprecise and fluctuating upper limits, the
limits of A's capacities. The limits themselves will be little
different from, say, the limits of various capacities of a motor-car.
It may be noted that in our formula the phrase "a relatively
great effort of some sort" occurs in a referentially transparent
rather than a referentially opaque context. A's belief may simply
be that the achieving of P requires Q, where Q in fact involves this
relatively great deal of effort, but this fact is not something A
believes. The importance of this point is that pretty unsophisti-
cated beings can try hard, and that we want to attribute to them
nothing but pretty unsophisticated beliefs.
If our definition had included only the first two conditions, it
would have admitted certain cases that we wish to exclude. Sup-
pose, for instance, that A wants to reach a certain house, runs with
might and main towards it, but runs in this way only through high
spirits. He knows that an easy stroll would have sufficed. Perhaps
he tried hard to run fast, but I do not think that he tried hard to
reach the house. Our third condition ensures this desirable result.
However, the definition does have a consequence that may
seem strange. Suppose that A runs with might and main towards
the house because he believes that this is what he must do if he is
to reach the house. But in fact his belief is false, and an easy stroll
would have sufficed. The definition yields the consequence that he
tried hard to reach the house. This may be thought a strange thing
to have to say.
We could exclude this case by adding the extra condition that
A's belief must be true. But I am dubious whether we should try to
exclude it. For I think that in such a case it is true that A tried hard
to reach the house, but it is misleading to say so. Saying so carries
the implication to a hearer that A's special exertion was necessary
given his purpose. But I think it is intelligible to "cancel" the
implication by saying, "He tried very hard to reach the house, but
his great efforts were quite unnecessary: he could have got there
without exerting himself."
To conclude: I have argued that trying more or less hard is a
mere species of that basic sort of trying that I previously argued
was involved in all action. Why, then, is trying more or less hard
thought of as trying par excellence, the notion of trying to which

our thoughts naturally turn when we begin philosophical reflection
upon the topic?
One reason may be this. Cases of trying hard are phenomeno-
logically the most conspicuous form of trying. Whenever an agent
acts, it has been argued, he is trying to do something. But it is not
necessary that he should be aware of what he is trying to do, and
although he generally will be so aware, the awareness involved will
normally be of a reflex and automatic sort. It will not involve a
great deal of attention. Now it is possible to try hard to achieve
some objective without being aware of this fact, but the case is
uncommon. When we try hard to achieve something, we are
generally sharply aware that we are trying hard to do this thing. So
when we think about trying, cases of trying hard dominate our
imagination. They ought not to dominate our logical analyses.
6
T~e Na~ure of Tradi~ion
The notion of tradition arouses strong political emotions. "The

progressive in the street" thinks of tradition as standing in the way
of rational social conduct. Traditions cramp and bind energies,
which if allowed to operate freely, could be used effectively for the
greater good of society and the greater happiness of the individual.
"The conservative in the street", on the other hand, is instinctively
drawn towards tradition. He has a sense that what is traditional is
natural and, because it is natural, worthy of being continued. His
attachment to tradition springs largely from unreflective affection.
But if pressed for a justification of this attachment, he might point
to the role of traditions in maintaining social order and stability.
I have my bias, and it will become evident during the paper.
However, before embarking upon a criticism or a defence of the
role of tradition in social life, it might be useful, and is in any case
a task well-suited to a philosopher, to consider what a tradition is.
This is the object of this paper, an object that also might be des-
cribed as an analysis of the concept of tradition. I hope to show
that such an investigation can at least illuminate, though it can
hardly decide, more obviously exciting questions. At various
points I will compare and contrast traditions with customs, habits,
rituals, practices and rules for action.
TYPES OF TRADITION
It may be a tradition in a certain family that the eldest son goes

into the navy. We may call this a simple tradition of action. Such a
tradition enjoins on those who answer to a certain description and/

or are in certain circumstances that they should act in a certain
definite way. It may be a tradition in the same family that their
ancestors came from Cornwall. We may call this a simple intel-
lectual tradition. By contrast with these simple traditions, we have
complex traditions such as the Western academic tradition, the
German military tradition or the "great tradition" of the English
novel.
It seems to be the complex traditions, for the most part, that are
the socially interesting phenomena. If we are concerned with
analysis, however, it is obvious methodological good sense to
begin by considering simple traditions. This paper in fact will be
confined almost entirely to a consideration of simple traditions of
action. (For convenience I will, for the most part, refer to them as
"traditions".) Only at the end of the paper will I say anything
about simple intellectual traditions and the relations that simple
traditions bear to complex traditions.
A TRADITION IS A SERIES OF ACTS
Whatever else a tradition is, it is a series of acts, acts by human

beings or perhaps by some higher animals. Ordinary speech does
not seem to provide a name for these acts. Let us call them the
"enactments" of the tradition.
The word "act" here is already something of a technical term,
although one that is familiar to philosophers, at least. When a
stone breaks a window, it may be said to act. But its breaking the
window is not an act in the sense in which the enactments of a tra-
dition are acts. The latter are intentional acts, things done on pur-
pose, things that spring from the will. Philosophers dispute about
what makes intentional acts intentional. I accept the (traditional)
view that what makes an intentional act intentional is that it has a
certain sort of mental cause. But here we cannot discuss the prob-
lem of the defining conditions of an action any further. 1 It will
simply be assumed that we understand what an intentional act is.
However, one point about intentional actions is quite generally
1. See Essay 5 in this volume, "Acting and Trying".
The Nature of Tradition 91
.accepted by contemporary philosophers and is important for our

purposes. Intentional acts are intentional only under certain
descriptions. A man who raises his arm is likely to displace a cer-
tain amount of air. It is unlikely, however, that this was any part of
his purpose. If his action is described as "raising his arm", "vot-
ing" or "trying to defeat the amendment", then it may well be an
intentional act under all these descriptions. But if it is described as
"displacing air" or "casting a shadow on the wall", it is unlikely to
be intentional under those descriptions.
This point applies to the enactments of a tradition. Many
descriptions will be true of the enactments, but they will only be
enactments of the tradition under some of these descriptions.
The distinction between a tradition and its enactments will au-
tomatically remind philosophers of the distinction between dis-
positions and their manifestations. Brittleness is a typical disposi-
tion. A brittle object manifests its brittleness by shattering when it
is struck sharply. Could the enactments of a tradition be the
manifestations of a disposition possessed by certain people to act
in a certain way in certain circumstances?
The suggestion must be rejected. It is a mark of a disposition
that a particular object that has the disposition need not manifest
it. A brittle object may never be struck sharply and so never shat-
ters. But if a tradition exists in a certain society or social group,
then there must be enactments of that tradition. No enactments,
no tradition. A tradition lives in, is constituted by, its enactments
in a way that the disposition of an object is not constituted by its
manifestations.
In this, traditions resemble customs, habits, practices and
rituals. Unless there are actual acts, there is no custom, habit,
practice or rituaJ.2 They may be contrasted in this respect with
rules for action. In this respect, such rules resemble dispositions.
2. In The Concept of Mind (1949: ch. 2, sect. 7), Ryle misleadingly classified
habits as dispositions. He wrote: "In discussing dispositions, it is initially
helpful to fasten on the simplest models, such as the brittleness of glass or the
smoking habit of a man." But if brittleness is a paradigm of a disposition, we
have reason to deny that a habit is. Brittleness is defined by reference to
manifestations of brittleness, but these manifestations need not actually
occur. Habits, however, logically demand the corresponding habitual actions.
An authority may lay down a certain rule of action, but there may
be no acts performed in obedience to the rule. Nor need this spring
simply from disobedience. The authority may be accepted by ar.
agent, but the agent may never find himself in a situation where
the rule applied. For instance, there may be rules to be followed in
the case of a nuclear attack, which the community accepts and
would obey if need be. Yet it may be lucky enough never to have
to obey such rules.
THE ENACTMENTS OF A TRADITION INVOLVE CERTAIN

REGULARITIES OF CONDUCT
A tradition demands a series of enactments. Suppose that there is

a tradition of performing acts of the sort X in circumstances Y.
This entails that the acts are intentional under that description. X
is a type of intentional action and, following C.S. Peirce, we can
call the individual enactments tokens of that type. A tradition of
doing X in circumstance Y demands a series of tokens of doing
acts of the type X in circumstances of type Y.
It is clear that the same must be said of customs, habits and
practices. (By contrast, a particular ritual might be performed only
once, and then abandoned. It would still have been a ritual.) In the
case of traditions, customs and habits, furthermore, the acts that
go to make up the tradition, custom or habit must stand in certain
relations to each other. (More of this later.) However, in the case of
a practice, no particular relationship between the tokens is
entailed. The notion of a practice is a much looser and more
general notion.
Traditions, customs and habits do not merely entail the exis-
tence of a related series of acts of the same sort. They entail also
that where circumstances Y occur, X is done with some regularity.
In the case of a practice, again, no such regularity is necessary. It is
true that a plurality of acts is necessary before we speak of a prac-
tice, but the practice may still be isolated or sporadic.
ENACTMENTS OF A TRADITION ARE (RELATIVELY) FREE ACTS
Intentional acts may be of various sorts. Some of these sorts are

unsuitable to be enactments of a tradition. In particular, wherever

the agent takes the act to be necessary in some way, then the act is
unsuitable. Such necessity is usually a matter of a means necessary
to achieve some desired end or avoid some undesired conse-
quence.
For ins•ance, suppose that in a certain society there is a
widespread practice of informing upon one's neighbour. But sup-
pose that this practice only exists because of very coercive pres-
sures applied by the authorities of that society. If the coercion was
stopped, then the practice would stop. Now, although what we
have here is a practice, it is wrong to speak of a tradition of
informing on one's neighbour in such a case. Again, if a landslide
blocks a path, forcing everybody to take the only other path, this is
coercion by the world. The enforced taking of the second path is
not the enactment of a tradition. But what is not enforced is (at
least relatively) free, and so the enactments of a tradition are free
acts.
I intend to beg no metaphysical questions here. "Free act" cer-
tainly does not mean "uncaused act" here, but simply "uncoerced
act". An act is uncoerced if there is no special compulsion, from
men or from the world, to act in that way. It, of course, must be
conceded that in all our actions the world, if not our fellows,
coerces us to some degree. A free act is simply one where the
degree of coercion is relatively low.
Is the fact that the enactments of a tradition are free acts simply
a trivial, verbal, point? The enactments of a tradition are not
coerced acts, because, if they were, we should not speak of "a tra-
dition" any more. I think, however, that the point is far from tri-
vial. Its importance can be seen by putting it another way. The
authority of a tradition is to a great extent internal to the tradition
itself. It does not rest upon externally applied sanctions. Now con-
servatives (or at least liberal conservatives) have always seen this
as one of the.virtues of tradition. They argue, first, that a certain
amount of order, a certain stability of expectation about the way in
which men will act in certain situations, is necessary if most social
enterprises, including highly valued ones, are to proceed. Now to
the extent that this order and stability of expectation can be pro-
duced by a relatively uncoercive mechanism, to that extent order
and freeom are reconciled. Tradition is such a mechanism. It is for

this reason that such conservatives have argued that if the
authority of tradition is overthrown, order will depend instead
upon coercive power exerted by authoritarians.
At this point, however, we must qualify the statement that the
enactments of a tradition are free acts. We considered the case
where, owing to state pressure, people unwillingly informed on
their neighbours. This does not constitute a tradition of informing.
Suppose, however, that in the end the state pressure moulds peo-
ple's wills so that they inform willingly, even in the absence of
state pressure. A tradition of informing may then come into exis-
tence. Although the will involved in the enactments of this tradi-
tion is relatively uncoerced now ("internalization" has occurred),
it was produced as a result of coercion. Such action may be
thought to hover ambiguously between coerced and truly free
action.
The ambiguous nature of the freedom involved in some tradi-
tional conduct furnishes the case for the anti-conservative liberal.
He rejects both the notion of conformity with an externally
imposed rule, and the enactment of a tradition, in favour of con-
duct quite freely and uncoercedly chosen by the agent.
At this point the dispute appears to depend upon empirical
facts. How far is it possible to live a tradition-free life? Marx said
in his Eighteenth Brumaire: "The tradition of all the dead genera-
tions weighs like a nightmare on. the brain of the living." That is
one view of the matter. At the opposite end of the spectrum is the
Anarchist vision (also entertained by Marx, but only for the future)
of people making up their lives as they go on. A logical analysis of
the notion of tradition can hardly adjudicate here.
One very important point about our own very complex society
is that, to a greater extent than ever before, there are competing
traditions of conduct. So, it may be argued, even if we cannot help
but live within traditions, we are at least free to choose between
traditions. However, Karl Popper (1969) has replied that even this
freedom depends upon a second-order tradition of choosing bet-
ween competing traditions. I mention these points only to illustr-
ate the complexity of the dispute between conservatives and
liberals at this point, and will not pursue the matter.
At any rate, enactments of tradition must be at least relatively
free. The same may be said in the case of observances of custom

and of habitual actions. To the extent that a thing is done from
necessity, it is not the observance of a custom, nor is it a habitual
action. Consider habitual actions, in particular. A man who drives
a non-automatic car does not have the habit of depressing the
clutch-pedal before attempting to engage a gear. A man who
always deals with a certain merchant, but only because and so long
as he believes that he gets the best deal there, is not dealing with
the merchant from habit. In each case, what the agent does is a
necessary means to his end, and hence it is not an action done out
of habit. (Though he may be said to go through a certain routine.)
It is true that we may develop habits as means, perhaps means that
we judge to be necessary to some further purpose. A man may
form the habit of doing exercises, judging it essential to his health.
But then it is the habit as a whole, not the individual actions, that
is judged to be the necessary means. In the same way, rulers may
set out to foster a tradition of obedience in their subjects, with the
object of simplifying the task of government. But individual enact-
ments of the tradition thus fostered need have no link with the
rulers' aims.
The link between habits and freedom may be queried. May not
a habit have an unbreakable grip upon a person? This can occur,
but what has the grip upon the person is one of his own impulses
to action. The habit springs from his nature, even if it is part of his
nature that the rest of his nature is unable to affect. This is surely a
state of (relative) freedom.
A TRADITION IS A SOCIAL AFFAIR
The enactments of a tradition may be actions of an individual or

the joint, co-operative, action of a number of individuals. (I will
refer to the latter as "collectives".) The same holds of customs and
habits. But in the case of a tradition, the enactments must be the
actions of more than one individual or, in the case of co-operative
enactments, more than one collective. Something must be handed
on.
By contrast with traditions, habits are the habits of individuals
or of individual collectives. A man may have the habit of sleeping
on his right side or two men may have the habit of meeting for a
drink after work. There must be, of course, a causal link between
the series of individual actions that are required for, and make up,
the habit. Suppose that a man has usually slept on his right side.
But suppose that, as is at least logically possible, the side he slept
on was a matter of chance, and all that had occurred was an
extremely improbable distribution of chances. It is clear that in
these circumstances he did not have a habit of sleeping on his right
side. To yield a habit, his regular turning to that side must have
sprung from a continuing bias of the will towards that action.
Without this causal substructure (an actual mechanism in his mind,
I would argue), his actions would not constitute a habit. But the
causally linked series of actions in a habit does not extend beyond
an individual or an individual collective.
However, while habits pertain to individuals or individual col-
lectives, but traditions demand a series of individuals or individual
collectives, customs can resemble either habits or traditions in this
respect. We can speak of an individual's custom: the word is then
nearly synonymous with "habit". We can speak also of a supra-
individual custom: there is then only a relatively fine contrast bet-
ween customs and traditions.
It may be wondered whether tradition can be defined in terms of
habit. 3 Could a tradition be a series of habits of individuals or col-
lectives causally linked together in some way? The suggestion
must be rejected.
In the case of a tradition, the enactments of different individuals
(to take the simplest case) are causally related. The enactors doX in cir-
cumstances Y as a causal result of others having done it before
them - right back to the founder(s) of the tradition. But just
because a habit is widespread in a community, it by no means
follows that the habit has been handed on from person to person in
3. "Such terms as custom, institution, tradition and mores are, however, hardly
capable of a precise scientific definition. All of them are reducible to social
habit or, if one prefers the anthropological to the psychological point of view,
to cultural pattern." (Sapir, 1930; emphasis added).
It seems to me that Sapir's position on the definition of these notions is too
pessimistic. The suggestions made in this paper are tendered as evidence
against him.
this way. It is possible that each person came to the habit for them-
selves, and then the actions involved fail to constitute a tradition.
And even if, as we would expect, there is a common cause that is
responsible for the widespread incidence of the habit, the habitual
acts themselves may play no causal part. Common circumstances,
ideology and character structure might act by themselves to pro-
duce the same habit in different people.
The considerations of the previous paragraph would be com-
patible with defining a tradition as a habit that is handed on. But
there are also traditions that do not involve any habits at all on the
part of persons who enact the tradition. Self-immolation of a
widow on her husband's funeral pyre (suttee) was a tradition or
custom in certain societies. But the lady could not have made a
habit of it.
A TRADITION INVOLVES THE CAUSAL MECHANISM OF IMITATION
In the passing on of a tradition (or a custom) from one individual

to another, the causal mechanism involved is imitation. The imita-
tion here must be of a fairly sophisticated sort. There is what might
be called stimulus-controlled imitation where A does X, B per-
ceives this and immediately does likewise. A cattle stampede or
the movement of a shoal of fish or a flock of birds may be exam-
ples of this phenomenon. But in the imitation required for a tradi-
tion, there is normally a considerable space of time between A 's
original enactment and its effect, B 's act, which constitutes the
carrying on of the tradition. This gap must be filled by a continu-
ing bias of B's will towards doing X (in circumstances Y), a bias
appropriately generated in B by A 's action. By contrast, a habit
need not involve any such imitation. In habitual action, a man
need not be imitating what he did before. All that is necessary is
that there be a continuing underlying bias of his will towards that
action, a bias that may of course be strengthened by the repetition.
Many species have the capacity to imitate conduct (in particu-
lar, the conduct of conspecifics) where the imitation is stimulus-
controlled. In some of these species, the capacity for the sophisti-
cated sort of imitation required for carrying on of a tradition is also
present. Where one, and still more where both, of these capacities
is present and is exercised, a special character is imparted to the

social life of that species. Insects lack these capacities and so,
although some species of insects have a social life, it is of a quite
different character from, say, the social life of mammals.
Imitative capacity of the sophisticated sort, involving a time-
gap, makes possible the passing on to others of individual discov-
eries. One bluetit in England discovers that it is possible to get at
the cream on the top of the milk by pecking through the foil bot-
tle-top. Within a few years all the birds of that species in the coun-
try appear to be aware of the possibility. Conservation of practical
knowledge, or at least of know-how, a conservation going beyond
the lifetime of an individual, thus becomes possible. The species
can have at least a primitive culture.
The case of the bluetits, of course, involves passing on a prac-
tice or a technique. Their actions do not constitute a tradition. The
biological value of imitating practices and techniques found to be
successful is obvious enough. Is there any biological value in the
imitations of conduct involved in tradition and custom? Or is such
imitation simply a biologically valueless by-product of the
mechanism that preserves practical knowledge?
One suggestion about the value of such imitation is this. Con-
scious thought, choice and decision are difficult matters. They
occupy no very extensive part of a human being's life. (And, of
course, still less of the life of other animals.) Whitehead said that
thought was the cavalry charge of the intellect. His point was that
cavalry charges, though vitally important, could form no very
extensive part of battles. Conscious deliberation followed by deci-
sion might be said to be the cavalry charge of the will. Now sup-
pose that one is free to adopt a wide variety of courses of action,
but no particular course is obviously superior. It will be an impor-
tant volitional economy simply to do what one remembers that
some conspecifics did.
A second suggestion is this. Where there is widespread imita-
tion of what was done in the past, not only inherited characteristics
but also behaviour will in some degree "breed true". The enact-
ments of a custom'or tradition are not means to an end, or, where
they are, alternative means to the same end are available.
Nevertheless, in the long run, different customs and different tradi-
tions will have different biological value. In the state of nature, at

least, some will be more conducive to species survival than others.
In nature, the descendants of those with good genes get on, the
decendants of those with bad genes are eliminated. Similarly, the
enactors of a biologically valuable custom or tradition will get on,
and so it will tend to survive, while the enactors of a biologically
harmful custom or tradition will not get on, and so it will tend to
vanish. A form of natural selection thus might act upon customs
and traditions. This may select not only the biologically valuable
customs and traditions but also, in turn, those individuals who are
predisposed to conduct of this sort.
THE FLEXIBILITY OF TRADITIONS
However, we must not think of traditions (and customs) as com-

pletely rigid affairs. The mechanism of transmission in a tradition
is imitation. Now, the imitation of conduct is not necessarily a
transitive relation. If B imitates A 's conduct with tolerable
accuracy, and C imitates B 's imitation with tolerable accuracy,
then C 's conduct is not necessarily a tolerably accurate imitation
of A 's conduct. Imitation, even close imitation, permits a certain
"wander". This links up with a point made on a number of occa-
sions by Michael Oakeshott (1962), that traditions, customs and
practices are not the inflexible things that progressivist stereotypes
present them as. Even when "left to themselves", they are capable
of a good deal of change. Oakeshott, of course, argues that social
change that occurs through the unselfconscious modification of
traditions is generally preferable to social change deliberately
instituted. But whatever stand we take on this issue, it seems that
the flexibility of tradition is built in to the very mechanism by
which it is transmitted.4
4. Burke wrote: "We must all obey the great law of change. It is the most power-
ful law of Nature, and the means perhaps of its conservation. All that we can
do, and that human wisdom can do, is to provide that change shall proceed by
insensible degrees." (Works, vol. 4, p. 301, letter to Sir Hercules Langrishe. I
am indebted to Roger Wilkins for discovering the source of this quotation.)
The mechanism of imitation provides for change by insensible degrees.
DISTINCTIONS BETWEEN TRADITIONS AND CUSTOMS
We have seen that it is possible to speak of the customs of single

inividuals, but not the traditions of single individuals. In this con-
text, however, the word "custom" may be little more than a syn-
onym of "habit". So let us restrict ourselves here to social
customs. What distinctions, if any, should we make between tradi-
tions and such customs? It will be remembered that the traditions
we are concerned with are "simple traditions of action".
It sounds natural to say that there is a tradition in a certain
family of the eldest son going into the navy. Again, it sounds
natural to say that in a certain nation it is the custom to dine at a
particular hour. By contrast, it sounds unnatural to speak of a
custom of the eldest son going into the navy or a tradition of din-
ing at a particular hour. If the latter statements were made, special
explanation would be called for.
One difference seems to be this. In the case of a tradition, more
is involved than simply doing what somebody else did just as a
causal result of the latter action. More than mere imitation is
involved. We speak of customs being "adopted" or "spreading"
and when a custom is adopted or spreads, the causal mechanism
may involve nothing more than imitation. In such a case, the
causal initiative comes from those who imitate. But a tradition
cannot be adopted, nor does it spread. It is handed on. That is to
say, although the result is that the successors in the tradition imi-
tate their predecessors, their predecessors and/or the social group
generally are not simply passive but in some way encourage this
imitation.
It follows that a tradition always involves a normative element.
The predecessors present the traditional conduct to their suc-
cessors as in some way admirable. So may the public generally.
The same holds in the case of many customs, or course, but it does
not seem to be involved in the notion of a custom. Those who
adhere to a certain custom may think of it as nothing more than
what they are accustomed to do.
If traditions involve a normative element, then it seems that
their enactors almost inevitably must have a language, one that
can be used to recommend the course of conduct involved in the
tradition. Here we seem to have come to a feature that makes tra-

dition a peculiarly human thing: a mark of the rational animal.
Some other primate societies appear to have customs, customs
that differ from group to group within the same species. But only
human beings have real traditions.
It may be observed, however, that while it m<lkes sense to speak
of instituting a custom, a tradition cannot be instituted. It can at
best be encouraged or fostered. Customs may be brought into exis-
tence by a decision, but a tradition must grow out of some actual
enactment(s).
Another difference between traditions and (social) customs is
this. Traditions are unitary things in a way that mere customs are
not. Suppose it is given that in two social groups there exists a tra-
dition the enactments of which are tokens of exactly the same
type. It follows that there are then two numerically distinct,
although generically identical, traditions. But if it is given only
that in both communities it is the custom to act in the same sort of
way in the same sort of circumstances, then I do not think that we
would be prepared to say that there are two numerically distinct,
although generically identical, customs. All that can be said is that
the same custom obtains in both communities. To be enactments
of a tradition, the enactments must be set in a certain causal con-
text. To be customary acts, also, the acts must be set in a certain
causal context. But in the case of the enactments of a tradition,
they must also hang together to make up a single unitary thing. No
such unitary object is involved in the case of custom.
These differences between traditions and (social) customs,
though rather fine, show that the concept of tradition is the more
complex and sophisticated notion of the two. If we are given that a
certain act is the enactment of a tradition, then we are given more
information than if we are given simply that it is a customary act.
Perhaps all enactments of simple traditions of action must be
enactments of a custom. But the reverse entailment certainly does
not hold.
OTHER TYPES OF TRADITION
This concludes my analysis of simple traditions of action. Simple

intellectual traditions do not seem to pose any particular new
problems. The tradition in a certain family that the family origi-
nally came from Cornwall is a belief that is handed down from per-
son to person in the same sort of way that a simple tradition of
conduct is handed on. In army manoeuvres, messages used to be
passed across the front from person to person with the additional
order, "Pass it on". This seems to provide a very good model for
the causal mechanisms involved in simple intellectual traditions.
Such a mechanism is by no means foolproof: the message may
become garbled. In terms of communication theory, the ratio of
noise to signal may become very high. Equally, tradition can be a
noisy channel for the passing on of a message.
This leads on to a second point about intellectual traditions. In
the case where knowledge is transmitted from one person to
another over time, we do not speak of a tradition that something is
the case. In the case of traditional conduct, something is done but
something else might have been done. In the case of traditional
beliefs, something is believed but something else might have been
believed. Even if the belief is true, the believers do not have that
degree of justification for their belief that entitles one to say that
what they possess is knowledge. (They may or may not be aware of
this shortcoming.) Of course, this is not to say that such beliefs are
cognitively valueless. They may preserve a good deal of important
information, and in pre-literate cultures they regularly do so.
But what, finally, of complex traditions? I do not have a great
deal to say about them. My hypothesis, however, is that complex
traditions are some sort of fused mass of simple traditions, tradi-
tions both of intellect and action. The set of simple traditions that
make up the complex tradition will have a certain unity. One part
at least of this unity will be causal. The different simple traditions
making up the complex tradition will tend to sustain each other in
existence, in the same way that the parts of an organism do,
though to a lesser extent. Furthermore, there will be something
more elusive: a Certain resemblance or unity of temper among the
various simple traditions. This unity of temper may spring from
higher-order traditions, or traditional attitudes, which encourage

certain first-order traditions and discourage others. 5 The con-
tinuing of, say, academic or military or literary traditions is com-
patible with a great deal of modification and change among the
simple traditions that go to make them up. But at any given
moment in the history of a complex tradition, a large number of
the simple traditions that it involves will be secure, although
others are in the process of being modified, abandoned or created.
In this way, a complex tradition is well suited to meeting demands
both for continuity and change in society. Burke held that a condi-
tion of social health was the maintaining of a partnership between
the past, present and future of a society. Evolving complex tradi-
tions seem well suited to maintain such a partnership.
5. In a review of the collection in which this paper originally appeared, Brian

Beddie {1977) suggested that this approach is "atomistic". It might be better,
he thought, to try to account for simple traditions in terms of complex tradi-
tions. But if we remember that man is a creature whose social life has gra-
dually evolved, becoming more and more complex as it evolved, we shall
have reason to suspect Beddie's suggestion. Surely complex traditions must
have been preceded by something simpler, such as social customs? The
emergence of simple traditions from social customs is then a natural inter-
mediate step between social customs and complex traditions. But if simple
traditions can exist before there are complex traditions, the latter are not
likely to be required for an analysis of what a simple tradition is.
7
Colour Realitm and ~~e Ar~umen~
from Microtcopel
In the philosophical theory of perception, a Realist is one who

holds with Berkeley's Hylas that "to exist is one thing, to be per-
ceived another". A Colour Realist is therefore one who holds that
the colour of the surface of an object, or the colour of an object as
a whole, exists independently of its being perceived.
A Realist may be either a Direct or a Representative Realist. A
Direct Realist holds that, in at least some cases, perception gives
us a direct, that is, non-inferential, awareness of the nature of
physical phenomena. A Representative Realist denies that percep-
tion ever can give us such non-inferential awareness. Clearly, it
would be possible to be either a Direct or a Representative Realist
about colour. In this paper, however, I will only be concerned with
a Direct Realist account of colour.
What I shall contend, in particular, is that an argument that is
given its classical statement in Berkeley's First Dialogue, an argu-
ment that may be baptized the "Argument from Microscopes", and
which Berkeley and many others have thought refutes Direct Real-
ism about colour, in fact does not do so. I have, of course, an
interest in rebutting the Argument from Microscopes. I believe that
Direct Realism about colour is correct. But here I will only be con-
cerned with rebutting the Argument from Microscopes. Other, and
powerful, arguments against a Direct Colour Realism will go
unanswered. Still less will I be concerned to defend Direct Realism
generally, although I also believe the general doctrine to be true.
One final preliminary. A Direct Colour Realism may be held in
two forms, which may be called the Anti-Reductivist and the
Colour Realism and the Argument /rom Microscopes 105
Reductivist form respectively. An Anti-Reductivist holds that the

colour of a surface, for instance, is an irreducible quality perfectly
distinct from, although no doubt correlated with, the light-waves
emitted from that surface. A Reductivist, however, holds that the
colour of the surface is nothing but the light-waves emitted from
that surface, or something of that sort. I myself accept not only a
Direct Realist view of colour, but also a Reductivist account of the
nature of colour. 1 But the rebuttal of the Argument from
Microscopes here attempted is entirely independent of whether an
Anti-Reductivist or a Reductivist view is taken of the nature of col-
our.
The point is important, because in this paper I shall speak of
"correlating" colour of surface with the light-waves emitted from
that surface. Now it is tempting to say that we can only correlate
what are distinct things, and so that to talk of correlation is to be
committed to an Anti-Reductivist view.
But this is a mistake. We can talk of "correlating" while leaving
it open whether the things correlated are distinct or not. The
following example was given by Keith Campbell in discussion. We
might correlate the characteristics of the criminal as deduced from
clues at the scene of the crime with the characteristics known to be
possessed by a certain suspect. Such correlation may leave it an
open question whether the criminal and the suspect are, or are not,
one and the same person. In the same way, talk of "correlating"
light-waves emitted from surfaces with colours of surfaces can
leave it an open question whether light-waves and colour are, or
are not, identical.
THE ARGUMENT FROM MICROSCOPES
It can hardly be denied, even by the Direct Realist about colour,

that some colour appearances are mere appearances. The distant
hills look blue, but their surface is not really blue in colour. The
problem arises for the Realist, therefore, how we can distinguish
real colour of surfaces or objects from merely apparent colour.
1. I have argued for this view of colour, and other secondary qualities, in A
Materialist Theory of the Mind (1968), ch. 12.
In the First Dialogue, Hylas begins by saying that the merely

apparent colours are those that "appearing only at a distance,
vanish upon a nearer approach". 2 A contemporary philosopher
might say instead that real colours are those presented to normal
observers in standard conditions. The observer's power of sight
must be normal, light must be adequate and the observer must be
able to inspect the object quite closely. Under such conditions, the
colours that appear are the real colours. ·
It is at this point that Berkeley brings in the objection from
microscopes. Microscopes are instruments of revelation. They
make us aware of many of the small-scale features of objects that
we would not be aware of otherwise. But in the case of colours, "a
microscope often discovers colours in an object different from
those perceived by the unassisted sight." A drop of blood looks red
all over to the naked eye. But, under the microscope, only a small
amount of the area appears to be red: it is mostly colourless. The
same thing, at the same time, and in the same part, cannot be
differently coloured, or simultaneously be coloured and colourless.
So the colour appearance presented to the naked eye and the col-
our appearance presented under the microscope cannot both be
the real colour.
But now we are faced with the difficulty that there seems to be
no reason to prefer one viewing condition to the other. It is
logically possible that the lenses in our eyes should have had the
magnifying power of the lenses in microscopes. In these circums-
tances, Hylas would have said that the microscopic colours were
the real colours. It therefore seems totally arbitrary to select any
particular magnification as the revealer of reality. So how do we
determine real colour?
I shall first consider two answers to this question that I think are
unsatisfactory, and then propose my own solution of the problem.
THE "PHENOMENALIST" SOLUTION
Some philosophers react to the Argument from Microscopes by

saying that what it proves is that all we mean by the real as
2. Berkeley, 1948-56: vol. 2, p. 184.

Colour Realism and the Argument from Microscopes 107
opposed to the apparent colour of a surface or object is "the colour

appearance presented to a normal observer under standard condi-
tions". A normal observer is an ordinary observer with ordinary
sensory equipment, including ordinary eyes. Standard conditions
involve ordinary light, ordinary distances, ordinary physical condi-
tions generally. If we look at an object under a microscope, then
conditions of observation cannot be said to be standard. So the
colour presented, if different from that presented in the standard
cases, is a mere appearance. Philosophers who take this line freely
admit, indeed insist, that if men had microscopic eyes, then the
colour appearance presented to normal observers in standard con-
ditions would be different, and so the real colour would be
different. But men have not got such eyes, and so a drop of blood
really is red, although it looks a different colour under the
microscope.
This solution of the problem does offer an account of the dis-
tinction between real and apparent colour. But it does so at the
cost of abandoning a Direct Realism about colour, for it gives an
account of real colour as being nothing but a sub-species of
apparent colour: apparent colour presented to normal observers
under standard conditions. If the Argument from Microscopes is
simply presented as an argument against the possibility of dis-
tinguishing real from apparent colour, then it seems to be met by
drawing this distinction between standard and non-standard condi-
tions of observation. But we are considering the Argument from
Microscopes as an argument against a Direct Realist account of
colour. The claim that we are trying to combat is that, in view of
the evidence provided by microscopes, it is impossible to draw the
distinction between real and apparent colour without abandoning
the view that the (real) colours of things exist independently of
their perceivers and can be (sometimes) directly perceived. So the
"Phenomenalist" solution is not available to us. It may be that the
Argument from Microscopes forces us to abandon Direct Realism
about colour. But let us see first whether there are any alternatives.
THE "MINIMUM VISIBILE" SOLUTION
I pass on to consider a second solution. It is sometimes pointed out

that when we look at an object under a microscope, then, in

general, the amount of the surface area perceived is so small that it
is less than the least area that can be discerned by the naked eye.
But where this is the case, do the observations made by the
microscope really conflict with the observations made by the
naked eye? Is there any difficulty in supposing that very tiny areas
may differ in colour from the overall colour of the surface?
Another observation made with microscopes may be adduced as
a parallel. A line may look perfectly straight to the naked eye, but
portions of the line, when viewed under the microscope, may not
look straight at all. Is there any conflict of appearances here? It
would seem not. The naked eye, let us say, is capable of dis-
criminations of straightness to the nearest hundredth of an inch.
Now the line may not merely look, but actually be, straight to that
degree of approximation. The microscope, let us say, is capable of
discriminations of straightness to the nearest ten-thousandth of an
inch. Now the line viewed under the microscope may not merely
look, but actually be, crooked to this stricter degree of approxima-
tion. So these observations with naked eye and microscope are
perfectly compatible. It is true that there is an element of illusion
in the case. The portion viewed under the microscope may look as
big as the whole line viewed with the naked eye. But if this illusion
is allowed for, both perceptions can be accepted as otherwise
veridical. This, it may be suggested, is a model for colours per-
ceived under a microscope.
In fact, however, it seems that the parallel cannot be main-
tained. Imagine that a microscope is deployed bit by bit over the
whole of an area easily visible to the naked eye. A certain colour
picture of the area could be built up - it could be literally pic-
tured on some much larger area - which will, in general, be
incompatible with the colour appearances presented to the naked
eye. Thus a drop of blood looks red all over to the naked eye, but a
colour picture of the same drop obtained by deploying a
microscope over the whole surface of the drop would not contain
very much red.
This incompatibility between the deliverances of the naked eye
and the microscope becomes still clearer when we contemplate the
(logical) possibility of perceiving the whole drop of blood in one
Colour Realism and the Argument /rom Microscopes 109
view, but with eyes that have the resolving power of microscopes.
The colour pattern presented would be incompatible with the col-
our pattern presented to an ordinary eye. But in the case of the line
there would be no conflict between ordinary eye and microscopic
eye. The microscopic eye would simply see more detail. We might
compare the latter to the case of a man who can see that the hen is
speckled, but cannot make out how many speckles it has, but who
then comes closer and can distinguish and number the speckles.
So it seems that, although this solution is not incompatible with
a Direct Realist account of colour, it fails to solve the microscope
problem for independent reasons.
THE "SCIENTIFIC" SOLUTION
I pass on to what I take to be the proper answer of a Direct Realist

about colour to the Argument from Microscopes. The principle of
the answer is this: it is a question to be decided on scientific
grounds what are the real colours of surfaces and objects. No mere
thinking, no mere analysis of the concept of colour, can settle the
matter. There is no question of finding a criterion that logically
will determine the q)..lestion. The best we can hope to do is to find
an answer that is scientifically plausible.
And so, since it is colour that we are discussing, we are brought
to the conclusion that the real colour of a surface is determined by
the nature of the light-waves emitted from that surface. If a surface
looks red, and is emitting light-waves of a sort or sorts charac-
teristic of red surfaces, it is red. If it looks red, but is not emitting
such a sort or sorts of light-waves, it only looks red. There is
nothing logically sacrosanct about this answer, it is simply the best
scientific guess we can make on the basis of available knowledge.
(All this, remember, is abstracting from the further question -
which also, I maintain, is to be decided on grounds of scientific
plausibility - whether the colour is or is not identical with the
light waves.)
But does the appeal to scientifically established correlates really
do much to answer the Argument from Microscopes? It is natural
to counter what has been said so far in the following way. It is clear
that any scientific hypothesis about the correlation of colour and
light-waves must ultimately be tested by observation. We must see

that a certain object has a certain colour, and then establish that
the object does emit light-waves of a sort predicted by the
hypothesis. But this means that we must begin with some idea of
which are the real, as opposed to the merely apparent, colours.
Once we have a plausible theory, it will be possible to turn back
and, in the interests of coherence, eliminate some of the cases that
were originally taken to be veridical colour perceptions. But we
cannot saw away the whole branch that we sit upon. It therefore
must be accepted that, by and large, the cases taken to be veridical
perceptions during the investigation will reappear as veridical in
the final theory. But now, the argument continues, what cases do
we begin by taking to be veridical? Surely it must be the colour
appearances presented to normal observers under standard condi-
tions? But, if so, it is these cases that, by and large, will reappear
in the final theory as cases of veridical perception. In that case,
however, the appearances exhibited under a microscope, being
incompatible in content with our ordinary perceptions, must, by
and large, be illusory.
Once this conclusion is granted, the question can be renewed
why normal vision should be thus honoured. Have not the colours
that appear under the microscope at least as much right to be
accounted the real colours of objects? It seems that the appeal to
light-waves simply has gone a long way round to enforce the view
that the real colours are those that appear to normal observers
under standard conditions. Yet the Argument from Microscopes
seems to show that this account is quite arbitrary.
I shall now try to show, however, that the "scientific" solution
to the problem posed by the Argument from Microscopes need not
simply end up by endorsing ordinary perception. The Direct
Realist about colour may conclude, if detailed scientific considera-
tions seem to warrant it, that the microscope, not the naked eye, is
the best guide to the real colour of objects.
The first point to be made is that the colour of a surface clearly
must be correlated with the light-waves that are actually being
emitted at the surface. For surely the colour of a surface is an
intrinsic, as opposed to a relational, property of the surface? The
real colour of the surface therefore must be a function of what goes
on at the surface. It is only indirectly a function of the stimulation

that actually enters the eye.
Let us now consider the case of the hills that look blue in the
distance although they are not in fact blue. I discussed this illusion
recently with a psychologist who works in the field of perception,
and he denied that it was an illusion at all. I think his denial was
mistaken, because the hills look to have a colour that they in fact
do not have, and failure of correspondence of perception to reality
is a sufficient condition for perceptual illusion. But we can sym-
pathize with his denial, because it is true that the perceiver is in no
way causally responsible for the illusory character of the percep-
tion. The illusion is not due to the way the perceiver's eye or brain
functions. For the illusion is brought about by the presence of air
between the perceiver and the surface of the object. The air pre-
vents all except those patterns of light-waves that are associated
with blue surfaces from reaching the eye. As we might put it, the
eye does its best with the light-waves that it is given. Illusion arises
because what it is given is not a true sample of the light-waves
actually emitted at the surface.
There is a connected point about this case, a point of the
greatest importance. Despite the illusion about the colour of the
hills involved, we can use the case to build up or test hypotheses
about the correlation of light-waves with the (real) colours of sur-
faces. For suppose we now consider, not the light-waves emitted .at
the surface of the hills, but the light-waves that actually enter the
eye. We can properly correlate these survivors with blueness of sur-
face. For if the sort or sorts of light-waves that actually entered the
eye had been a representative sample of the light-waves actually
emitted at the surface of the hill, then we would have been justified
in saying that that surface was really blue. To repeat the point
briefly: The wave-lengths that enter the eye are the "blue" wave-
lengths. So if these wave-lengths only are emitted at a certain sur-
face, we are justified in saying that the surface is blue.
Now to use the discussion of this case to bolster the "scientific"
solution to the problems raised by the Argument from
Microscopes. My suggestion is that the case of the distant hills
stands to normal perception under standard conditions much as
this normal perception stands to perception aided by microscopes.
For what does a microscope do? It brings to the eye a much larger
sample of the light-rays emitted from a particular portion of a sur-
face than reach the naked eye from that portion of the surface.
Relative to that portion of the surface, much more "information"
(in the engineering sense) enters the eye. Now the correlations bet-
ween colour and wave-length that are built up in standard condi-
tions of colour perception can in fact only be correlations between
colour and the waves that actually enter, and so influence, the eye.
But if the waves that actually affect the eye are a bad or misleading
sample of the waves emitted at the surface, then it would be possi-
ble to conclude that the real colour of the surface is not the colour
that appears to normal perceivers in standard conditions. Yet, at
the same time, the correlations between colour and wave-length
built up in standard conditions still could be used as a guide to the
real colours of surfaces, provided the correlations were based on
the wave-lengths that actually managed to arrive at, and so to
affect, the eye.
And so, if detailed considerations of physical theory should
warrant it, it would seem perfectly possible for the Direct Realist to
maintain that it is the microscope, and not the naked eye, that is
our best guide to the colour of a surface. The microscope certainly
brings us a bigger sample of the light-waves from a certain area.
The way is therefore open to argue that it is a less misleading sam-
ple. Whether detailed considerations of physical theory do in fact
enforce this conclusion is hardly for the philosopher to say.
Nor does the contemplation of endlessly more powerful
microscopes create any theoretical difficulty. For there is a
theoretical limit to the process: the perfect instrument that cap-
tured, without distorting, all the light-waves emitted from a certain
surface.
In this way, I suggest, the Direct Realist about colour can-come
to terms with the Argument from Microscopes. The line of argu-
ment involves allowing that there is nothing logically sacrosanct
about the ordinary paradigms of colour. A drop of freshly drawn
blood looks red all over to normal perceivers in standard condi-
tions, and is a very suitable object to teach a child "what redness
is". We can admit these facts, and yet still say it makes sense to
deny the drop is really red all over. I do not think that there is any-
thing counter-intuitive in this, although it does run counter to the

prejudices of some modern philosophers (currently, perhaps, a
decreasing band). Consider the following case. An object looks to
be pink all over, but when examined closely is found to be a
multitude of tiny red and white dots. Is it not logically possible
that every "pink" surface should turn out, on examination, to be a
similar red and white mosaic? And, if so, would we not be justified
in saying that, although many surfaces looked pink in colour, there
was no surface that actually was pink in colour? (For the sake of
convenience, we might still talk about "pink surfaces" as opposed
to "pink-looking surfaces" in ordinary life.) This example should
embolden us to declare that the microscope may show us that our
ordinary paradigms of surfaces of a certain colour do not in fact
have that colour.
I append the following quotations as evidence that scientists, at
least, are ready to speculate about the real colour of objects on the
basis of theoretical considerations, even though such colour can-
not be directly perceived:
At a conference in 1953 Platzman was asked: "Do I unders-
tand that you think that irradiated ammonia ought to turn
blue if it is pure?" He answered with remarkable foresight:
"More than that, I think irradiated water turns blue and we
just don't see it."
Since the initial observation of the absorption spectrum of

the hydrated electron, numerous studies have been made on
sodium radiation-produced solvated electrons in water and
many other solvents. The lifetimes are shorter than the "twin-
kling of an eye", so that only instruments can "see" the col-
our, but the spectrum leaves no doubt that the characteristic
blue colour of solvated electrons is present for a short time
after irradiation.3
There are two further objections to the line of reasoning
adopted in this section that deserve brief consideration.
1. It may be questioned whether, in answering the Argument
3. Dye, 1967: 80.

114 The Nature af Mind
from Microscopes, I have not abandoned Direct Realism about col-

our. For if we can only know the real colour of a surface after we
know what light-waves it emits, we can hardly talk of a non-
inferential awareness of the colour of the surface.
My answer to this objection is that to defend Direct Realism
about colour is not to defend Naive Realism about colour. I have
argued elsewhere4 that it is perfectly legitimate for the Direct
Realist to maintain that sensory illusion is a much more
widespread affair than common sense allows. The scientific evi-
dence suggests that this may be the case for most of our colour
perceptions. But if, by luck, a particular colour perception is veridi-
cal, then, in default of further arguments against Direct Realism,
why should we not say that our awareness is direct or non-inferen-
tial? (Non-inferential awareness is not, of course, incorrigible
awareness, nor is it necessarily knowledge. ) At this point, the
question may be only one of terminology, but it seems to me that
it is legitimate to call the view I am defending a Direct Realist
theory of colour.
2. I have assumed that, by and large, the same surfaces will
appear to be of the same colour to different observers whenever
the light actually entering their eyes has the same composition, for
otherwise there would be no possibility of correlating colour and
light-waves. But this assumption may be questioned. The same
cause may produce different effects upon different things, and may
not the same light-waves produce different colour appearances
when acting upon different eyes and nervous systems?
In fact, of course, such variations in the colour appearances are
not merely possible, they actually occur5 . A Direct Realist about
colour, such as myself, will therefore be obliged to give reasons for
treating some sub-class of the varying appearances as the real col-
ours of the objects perceived. And it may turn out that it is not
possible to find such a non-arbitrary criterion for determining real
colour.
But the point to be made here is that all this has nothing to do
with the Argument from Microscopes. If there is no clear way of
4. In Perception and the Physical World (1961), pt. 5.

5. For a summary of the evidence, see Campbell, 1969: sects. 5 and 6.
Colour Reo/ism and the Argument from Microscopes 115
correlating certain wave-lengths with colours, then there is cer-

tainly no hope for any Realist theory of colours, Direct or Repre-
sentative, Reductivist or Anti-Reductivist. But then the Argument
from Microscopes is completely redundant. Here, however, I am
only attempting to answer the Argument from Microscopes. If we
are to achieve clarity in this difficult field, it seems to me to be
very important to distinguish clearly, and evaluate independently,
the different arguments that are used against the objectivity of col-
our.
TWO WAYS OF TALKING ABOUT COLOUR
There is, however, one troublesome objection to the procedure of

correlating the colour of a surface or an object with the sort or
sorts of light-waves emitted by that surface or object and one
which it does seem proper to consider here. Consider the following
case. A piece of red cloth is put under a mercury-vapour lamp. It
looks chocolate-brown. The light-waves being emitted from the
cloth, or at any rate those that actually reach the eye, are presuma-
bly those associated with the colour of chocolate-brown. But we
say that the cloth is red, not chocolate-brown. This suggests that,
when correlating colour with light-waves, as well as reference to
the light-waves emitted from a surface, reference to the conditions
of the illumination is also required.
I think that this difficulty can be cleared up if the Direct Realist
about colour asserts that there are two distinct senses in which sur-
faces and objects are (objectively) coloured.
The point can be illustrated by distinguishing two senses in
which physical objects such as squash balls may be (objectively)
spherical. In one sense of the word "spherical", squash balls are,
and remain, spherical. Following a terminology suggested by Keith
Campbell, we may call this the "standing" sense of the word.
Manufacturers make squash balls spherical in this sense of the
word, and if the balls ever cease to be spherical they are of no
further use as squash balls. But in another sense of the word
"spherical", the "transient" sense of the word, squash balls are,
from time to time, not spherical in shape at all, for instance at the
moment of being hit by a squash racquet or of rebounding off a
wall. And although this second sense in which a squash ball is

(only sometimes) spherical is not one that finds much employment
in ordinary language, it can be used as a primitive notion to give a
definition of what it is to be spherical in the standing sense. A
thing is spherical in the standing sense if, and only if, it is of such a
nature that it is transiently spherical in normal conditions, that is,
when no especial forces are acting upon it. Being spherical in the
standing sense, and (normally) spherical in the transient sense, are
objective characteristics of squash balls, characteristics that are
perfectly independent of perceivers.
I suggest that there is room to distinguish two parallel senses of
colour words like "red".
In one sense of the word "red", the "standing" sense, red
dresses are, and remain, red until the colour fades or the dress is
dyed another colour. Like the standing sense of "spherical", this is
a relatively permanent characteristic of the object, and is the sense
of "red" usually employed in practical life. But there is another
sense of "red", the "transient" sense, in which red dresses are,
from time to time, not red at all, for instance at the moment that
they are under a mercury-vapour lamp. Red dresses are transiently
red only under normallighting conditions, for instance in ordinary
daylight. Under mercury-vapour lamps, red dresses are transiently
chocolate-brown. This second sense of red, in which what we
ordinarily call a red dress is not always red, finds less employment
than the first sense. It is, however, of interest to those whose
special concern is with visible things, for instance painters. It can
also be used as a primitive notion to give a definition of standing
redness. A thing is red in the standing sense if, and only if, it is of
such a nature that it is transiently red in normal lighting condi-
tions. Being red in the standing and in the transient senses, the
Direct Realist will assert, are both objective characteristics of sur-
faces or objects, characteristics that are perfectly independent of
perceivers. (Of course, if our earlier argument has been correct,
objects that the ordinary perceiver in standard conditions takes to
be red, either in the standing or the transient sense, may in fact not
have those properties, and the evidence of microscopes etc., may
be relevant in showing that ordinary perception is so mistaken.)
It must be confessed here that when a red object is under a mer-
cury-vapour lamp, we are inclined to say that the object looks

brown, rather than say it is (transiently) brown. But I think that
this is simply a mistake embodied in ordinary language. Perhaps
the mistake arises because of our tendency, embodied here in the
use of the word "looks", to assimilate the impermanent to the
unreal.
Now that the distinction has been drawn between two senses in
which a surface or object can be red, it can be suggested that it is
transient redness, the less usual sense, and not standing redness,
that ought to be correlated with the light-waves actually emitted at
the surface. By contrast, objects or surfaces that are red in the
standing sense ~re simply ones that emit these wave-lengths in
ordinary light: in daylight.
This distinction between two senses of colour is not easily
parallelled in the case of the other "secondary" qualities. Sounds,
for instance, would seem to be correlated with sound-waves, and
there is no need for the distinguishing of standing and transient
senses in which sounds attach to bodies. But sounds are not
attributed to bodies in the way that colours are attributed to them
or their surfaces. Heat, cold and taste are attributed to the bodies
themselves, but there still seems no special call to distinguish per-
manent from transient modes of attachment.
I will conclude by noting that the distinction of senses in which
colour may be (objectively) attributed to objects clears up the
problem raised by Locke about the colour of porphyry in the
dark. 6 Locke says that it is plain that porphyry has no colour in the
dark, and draws the conclusion that Direct Realism is false with
respect to colour. For if colour is really in the object, he argues,
the property would not alter in this way with every variation of
light.
Consider a piece of red and white porphyry that is in the dark.
Using our introduced terminology, we can say that the standing
colour of the stone remains red and white, but that its transient
colour is black. This is the sense in which it truly has no colour in
the dark. But none of this is any objection to a Direct Realist view
of colour. In the dark, the porphyry remains the sort of thing that,
6. See Locke, 1975: Bk. 2, ch. 8, sect. 19.

brought into light again, emits the "red" and "white" wave-
lenghts. In the dark, it emits no wave-lengths, and lack of emission
of light-waves from surfaces is correlated with the colour black.
The standing red and white, and the transient blackness are, for
any consideration that Locke has advanced, objective properties of
the porphyry. We simply have a distinction between relatively sta-
ble and relatively evanescent properties of the stone. All cats really
are grey in the dusk: transiently grey.
8
lmmedia~e Percep~ion
THE CONCEPT OF IMMEDIATE PERCEPTION
The philosopher who, above all others, we associate with the

notion of immediate perception is Berkeley. He says in the First
Dialogue: "in truth the senses perceive nothing which they do not
perceive immediately: for they make no inferences." 1 It is clear
that the central notion that Berkeley has in mind is that of the
completely uninferred element in perception. Perceptions may
involve inference, as in the case when, seated in the study, I auto-
matically take it for granted, on the basis of what I hear, that a
coach has gone past in the street. But, in default of infinite regress,
whatever is inferred must be based upon what is not inferred. But
since our knowledge and/or belief about our current environment
begins with perception, there must in every case be perceptions
that involve no inference. These are our "immediate perceptions".
This line of argument seems to be sound. Unfortunately,
however, a number of other much more dubious notions cluster
around this central notion of a perception involving no inference,
both in Berkeley and in later philosophers. Some of these notions
receive little countenance among contemporary philosophers. I
devote the rest of this section to noticing and dismissing these
notions briefly. This leaves me free to attend to the main objective
of this paper: criticism of the notion that there are such things as
immediate objects of perception.
1. Berkeley, 1948-56: Vol. 2, pp. 174-75.

The first correction we need to make is to reject a notion that is

present in the quotation from Berkeley. This is the notion that
immediate perception is the only true perception and that mediate
perception, because it involves inference, is not really perception
at all. If a man in his study takes it for granted that he hears a
coach go by outside, then it seems clear that this should be said to
be a perception. It would be awkward to call it anything else. At
the same time, the perception clearly involves inference, even if
inference of a quite automatic, instantaneous and unselfconscious
sort. Berkeley seems wrong to oppose perception and inference so
sharply.
A second notion, which Berkeley and others have associated
with immediate perception, is that it is an awareness of a certain
special class of objects: "ideas", sense-data or the perceiver's own
sensory states. I reject this notion, first on the ground that there
are no~ch things as "ideas" or sense-data, second on the ground
that, although we do have sensory states, we do not perceive them
and so cannot immediately perceive them. We have, or are in, sen-
sory states and we may, or we may not be, introspectively aware of
them. I excuse myself from arguing for this now fairly orthodox
view.
The third notion, also to be found in Berkeley, is that this com-
pletely uninferred element in perception constitutes logically
indubitable knowledge. Here again I believe that this notion must
be rejected, first on the ground that there is no such knowledge 2 ,
second that there is no reason to think that what is completely
uninferred, in perception or any other field of cognitive activity,
must even be knowledge, let alone indubitable knowledge. It may
simply be belief, and false belief at that. The notion that immediate
perception involves logically indubitable knowledge is, of course,
linked with the idea that it is perception of a special class of mental
or at least private objects. But it has even deeper roots in the
"search for super-certainty" which has bedevilled philosophy from
the time of Descartes almost up to the present day.
But even when we have rejected the notions that immediate
perception is the only true perception, that it is perception of pri-
2. See my A Materialist Theory of the Mind (1968), ch. 6, sect. 10.

Immediate Perception 121
vate objects or states, and that it furnishes us with logically

indubitable knowledge, still the notion of immediate perception
requires further purging before it can be accepted.
IMMEDIATE PERCEPTION IS NOT PERCEPTION OF OBJECTS
Immediate perception has traditionally been thought to be the per-

ception of objects. As usual, the notion is found in Berkeley. Even
those who, like myself, 3 have rejected the idea that immediate per-
ception is of private objects, have simply substituted the notion of
immediate perception of certain public objects. To try to make it
clear that immediate perception is not perception of objects at all,
I will now say something about the vocabulary of perception. I will
distinguish between:
Object-perception statements
Propositional-perception statements
Object-appearance statements
Pure-appearance statements
I believe that most, if not all, ordinary statements of perception
can be brought under these four headings.
1. OBJECT-PERCEPTION STATEMENTS
"Object-perception" statements are statements like, "Jones saw a

(the) fox." Their general form is "A perceives 0. "But the object
perceived need not be an object in any narrow sense of the word.
"The child saw the march-past" counts as an object-perception
statement, although the march-past would be normally classed as
an event or happening, not as an object.
It is an essential feature of object-perception statements that
they entail that the object perceived has existence. If Jones saw a
fox, then, of necessity, there was a fox there to see. "A perceived
0" entails "0 existed". The statements have what may be called
"existence-grammar" (Armstrong 1968: 215).
It must be admitted that it is possible, even if uncommon, to
3. Perception and the Physical World (1961).

122 TheNatureofMind
have sentences of the form "A perceives 0 " where what is stated
by the sentence carries no commitment to 0 's existence. An
oculist may say to a patient, "Tell me what you see", and the
patient may say that he sees something or other without being
committed to asserting the existence of that thing. In the case of
sentences of this sort, I simply stipulate that they do not express
object-perception statements. They express statements of type (3)
or (4).
But although object-perception statements entail that the thing
perceived exists, they do not entail that the perceiver knows, or
truly believes or even has the thought that the object perceived
exists. Nor need the object perceptually appear to him to be the
object that it in fact is. This point was first emphasized, to my
knowledge, in an important paper by Warnock (1954-55). It can
be true that Jones saw a fox, yet also true that Jones neither knew
nor believed that he saw a fox, that the thought of a fox did not
cross Ms mind, and that his perceptual experiences were not fox-
like. Suppose, for instance, there looked to him to be some sort of
object in the far distance the nature of which he was quite unable
to make out, but which was in fact a fox. It might still truly be said
of him that he saw a fox.
The fact is that object-perception statements are hardly cogni-
tive idioms at all. This is connected with the fact that the argu-
ment-form:
(1) A perceives 0
(2) O=P
(3) A perceives P
is a valid form. If A sees a table, and the table is a cloud of funda-
mental particles, then A sees a cloud of fundamental particles,
although A be the most ignorant peasant in the world. As Quine
would put it, in "Jones saw a table", the phrase "a table" occurs in
a referentially transparent context.
But once we see the non-cognitive nature of object-perception
statements, we see further that we cannot draw the immediate/
mediate distinction in respect of such statements. For the distinc-
tion between the uninferred and the inferred is a cognitive distinc-
tion. The uninferred is something known or believed, while the
inferred is something known or believed on the basis of what is

uninferred.
Berkeley gives as an instance of immediate perception hearing a
sound while seated indoors and as an instance of mediate percep-
tion hearing the coach that made that sound. But if we take the
object-perception statement "B heard a coach", then this does not
entail that B made any inference from the occurrence of the sound
to the presence of a coach. A man who was quite ignorant of the
noise coaches make could still truly be said to have heard the
coach, provided only that he heard the noise. Even a man who
knew perfectly well how to recognize the presence of coaches by
their sound might truly be said to hear the sound but fail to realize
that a coach was passing, for he might hear the sound, but because
of some disturbance of auditory perception, it might not sound to
him like the sound of a coach.
2. PROPOSITIONAL-PERCEPTION STATEMENTS
By contrast with object-perception statements, "propositional-

perception" statements, which have the form "A perceived that
p ", are cognitive statements par excellence. If Jones sees that
there is a fox before him, then it is entailed that at that time Jones
knows that there is a fox before him. This entailment is pointed out
by Unger (1972). To see that it holds, consider the unacceptable
sentence, "Jones sees that there is a fox over there, but he does
not know that there is a fox over there.'' There seems no way to
explain the unacceptability of this third-person sentence without
postulating the entailment. With knowledge, referential opacity is
at once introduced. If a peasant sees that there is a table before
him, then it by no means follows from the additional fact that a
table is a collection of fundamental particles that the peasant sees
that there is a collection of fundamental particles before him.
Knowledge may be inferred or it may be uninferred and, on pain
of vicious regress, all inferred knowledge must ultimately rest
upon uninferred knowledge. The distinction between immediate
and mediate perception therefore can be very easily drawn in the
case of propositional-perception statements. "He heard that there
was a coach going past in the street" seems very clearly mediate
perception, involving inferential knowledge. But "He heard that a
124 The Nature a! Mind
certain sort of sound was sounding in his environment" is a good

candidate for a statement of immediate perception, unless one
holds the view, which I am taking to be incorrect, that the presence
of sounds in the environment is only inferred from the awareness
of having sensations of sound. (If the latter view is correct, it
would seem reasonable to speak of the "immediate perception" of
sensations of sound.)
3. OBJECT-APPEARANCE STATEMENTS
"Object-appearance" statements have the form "0 perceptually

appears C/> to A ". "The rose looks red to him" is such a statement.
Such statements entail the truth of a corresponding object-percep-
tion statement. Thus "The rose looks red to him" entails "He sees
the rose". It follows that the word "rose" occurs in a referentially
transparent context, permitting free substitution. From "He sees
the rose", however, all that can be deduced is that "The rose looks
some way to him." Some more concrete object-appearance state-
ment must be true, but it cannot be deduced what one. The word
"red", of course, occurs in an opaque context. The rose need not
look to him to have the colour associated with the red wave-
lengths.
A distinction between immediate and mediate perception can
be drawn in the case of object-appearance statements. But need-
less repetition will be avoided if we pass straight on to pure-
appearance statements where the same distinction appears in a
more disentangled way.
4. PURE-APPEARANCE STATEMENTS
"Pure-appearance" statements have as their form "To A, there

perceptually appears to be an X which is ¢."To Macbeth, there
looks to be a dagger before him. Pure-appearance statements are
characteristically used to describe hallucinations, such as the one
Macbeth suffered from, but their meaning goes beyond such situa-
tions. Indeed, for every perception that a man has, a pure-
appearance statement might be found to render it, lack of vocabul-
ary only excepted. Such statements create purely opaque contexts
(after the verb "appears").
It would seem possible to draw the distinction between immedi-

ate and mediate perception within pure-appearance statements.
Thus "There looks to him to be a dagger before him" would seem
to be a mediate perception, based upon an immediate perception.
"There looks to him to be a thing having a certain shape and col-
our before him" (the visual properties associated with a dagger)
would be a plausible candidate for the statement of the immediate
perception involved.
It is, presumably, an empirical question just what is or is not the
completely uninferred element in a perception. It therefore will be
a matter for psychological theory rather than philosophy to decide
just where the line is to be drawn. Philosophy simply discusses
the nature of the line.
In the case of appearance-statements of both sorts, however, we
must reckon with a difficulty that does not arise in the case of pro-
positional-perception statements. It is possible that, although there
looks to somebody to be a dagger before him, he does not actually
believe that there is a dagger before him. Now, in inferring,
presumably we move from belief to belief, and so it might seem
that there is no room for the immediate/mediate distinction in
belief-free perception.
The point is well taken, but I think that even in belief-free per-
ception, an analogue of the original immediate/mediate distinction
can be found . Suppose that there looks to somebody to be a dag-
ger or a table before him, but he believes that there is no dagger or
table before him (whether truly or falsely is immaterial). We can
surely ask on what basis the world looks to him to be this way. The
natural answer would seem to be that there looks to him to be
things having certain shapes, colours and perhaps other visual pro-
perties before him: things that, if they exist, naturally would be
identified as daggers and tables. These are his immediate percep-
tions. The immediate/mediate distinction is here drawn in terms of
the beliefs that would have been held if the perceptions had not
been "belief-free".
WHY WAS THE MISTAKE MADE?
It seems, then, that the immediate/mediate distinction can be

drawn in the case of propositional perception, and again in the

realm of perceptual appearance, but not in the case of object-per-
ception. It may be wondered why philosophers from Berkeley to
Armstrong persisted in trying to force a cognitive distinction upon
a non-cognitive idiom.
One reason may be the influence of the doctrine that it is sense-
impressions that are the immediate objects of perception. They are
very peculiar objects. They might be described as propositions that
have been congealed into objects, and they retain a propositional
flavour. As a result, it is relatively easy to treat them as what is
immediately perceived.
A second reason may have been the fatal influence of first-per-
son present-tense object-perception statements. If I say that I see a
fox, then I imply that I know or I believe there is a fox there. "I can
see a fox but I do not believe there is one there" is a paradoxical
statement, although not self-contradictory. It thus became easy, by
concentrating on this particular case, to overlook that object-per-
ception statements are non-cognitive.
In this connection, therefore, it is worth noticing that "I can see
a fox, but I cannot see that it is a fox" is not a paradoxical state-
ment. For suppose that the fox looks to me to be but an
indistinguishable object in the distance, but a friend who is nearer
shouts out that it is a fox. I can then make that statement with
propriety and truth.
It may be, however, that there is a third, dimly felt, reason for
persisting in trying to force the immediate/mediate distinction
upon objects of perception. For there may be another distinction
to be drawn in the realm of object-perception, which is not a cog-
nitive one but which gets confused with the immediate/mediate
distinction. In the section below, I try to draw this new distinction.
DIRECT AND NON-DIRECT PERCEPTION
It is necessary first to consider more deeply the nature of object-

perception statements. It is widely, and I believe correctly,
accepted among contemporary philosophers that such statements
entail that 0 is the cause of those perceptions of A that constitute
his perceiving of 0. (The perceptions themselves can, in principle
at least, be exhaustively described by pure-appearance statements.

In the same way that the term "father" attaches to a certain man,
but attaches to him in virtue of his relation to a certain further
thing, a child, so the phrase "perceptions of 0 " attaches to certain
sensory states of A - his perceptions - but attaches to them in
virtue of their relation to the object 0 .) The entailment was can-
vassed for a number of years by C.B. Martin, but to my knowledge
was first argued for in print by H.P. Grice (1961).
The entailment may best be demonstrated by using what I call
"the method of subtraction" .4 If we are interested in the question
whether or not Cis a logically necessary condition for D, we con-
sider a situation that is D-like but where it is explicitly stipulated
that C does not hold. It may then become obvious that, lacking C,
we also lack D. The technique has analogies with reductio ad
absurdum in logical and mathematical proof and with the "method
of difference" in experimental investigation. It has considerable
utility in conceptual analysis.
Consider, then, a case where (a) there is an orange on the table;
(b) A has perceptions as of an orange on the table; but (c) it is not
the case that the orange brings about these perceptions of A 's. We
can imagine a story. Suppose that the light-waves travelling from
the· orange towards A 's eyes are in some way diverted from their
path, but that, at the same time, a probe introduced into A 's brain
at that moment happens, by an incredible coincidence, to produce
in A just those perceptions of an orange that the light-waves would
have produced if they had not been diverted. Given that condi-
tions (a) to (c) obtain, I do not think that anybody would wish to
maintain that A saw the orange on the table. The best which could
be said is that he had a "veridical hallucination" as of an orange on
the table.
Now, however, change the case and allow that the orange acts
upon A 's eyes in the ordinary way and is responsible for A 's per-
ceptions of an orange on the table. We will at once be prepared to
say that A sees the orange. Indeed, we will be prepared to concede
that condition (b) was much too strong. A 's perceptions might not
4. In Belief, Truth and Knowledge (1973), p. 81.

have been orange-like, yet, given the causal condition, we would

still be prepared to say that A saw the orange. 5
So it seems that the "existence-grammar" of "A perceives 0"
is explained by its causa/ grammar. If 0 is to cause A 's percep-
tions, 0 had better exist.
Given this preliminary, I proceed to draw a distinction in the
field of object-perception that bears a shadow-resemblance to the
epistemological distinction between immediate and mediate per-
ception.
Consider the following puzzle cases. (1) A sees an apple. (2) A
sees a half-apple, but the outer skin of the half-apple is turned
towards A so that his eyes are affected just as in case (1). Now con-
sider case (1) again. We would be happy to say, ordinary language
would support us in saying, that A cannot see the back half of the
apple. But this jostles with (1). If A cannot see the back of the
apple, then he cannot see the whole apple.
Perhaps what he sees is only the front half of the apple? But it is
clear that there can be no stopping there. A cannot see the back
half of the apple, and he cannot see most of the front half of the
apple. It seems that we will end up having to say that he can see no
more than the surface of the front half of the apple. And the sur-
face is not even a physical object, although it belongs to a physical
object.
. Faced with this puzzle, it is natural to wonder whether we might
not reintroduce a distinction somewhat like that between immedi-
ate and mediate perception in the sphere of object-perception. I
5. A's perceptions cannot be anything at all, and it still be said that he saw
the orange. At the same time, they can be extraordinary unrepresentative of
the nature of the object seen. If a man looks at Alpha Centauri, then the cause
of his perception, and the thing that he can be said to be seeing, is that huge
object Alpha Centauri as it existed four years ago. But his perception is a per-
ception as of a tiny light source, which exists contemporaneously. If,
however, a poke in the eye with a stick makes A "see stars", we would not
say that his perceptions were perceptions of the stick. How is the line between
these two sorts of cases to be drawn? For the best attempt I know to draw it
see Grice (1961: sect. 5). But I feel sure that Grice would agree that it is
desirable to say more than he says there. (Since writing this note, I think that
the problem has been solved by David Lewis in an as yet unpublished lecture,
"Veridical Hallucination and Prostheti-.: Vision" (n.d.).)
will use the words "direct" and "indirect". Might we not say that
the front surface of the apple is the directly perceived "object",
while the rest of the object is not directly perceived? One might
wonder if this is not the epistemological distinction once again.
But I do not think it can be. For, first, the surface could be utterly
misperceived (a mere blur on the table), so that A could draw no
inference at all from the content of the perception, but A could still
be said to be seeing the apple, and, it would seem, it would still be
the front surface only which he perceived "directly". Second, even
where the surface was correctly perceived, A might still fail to infer
the presence of the whole apple, yet still be said truly to be seeing
the whole apple. Suppose, for instance, that A lived in a world
where half-apples were the rule, but that what he was looking at
was, for once, a whole apple. He might think that it was a half-
apple with its skin turned towards him. But we would be prepared
to say that he actually saw a whole apple.
But if we remember that "A perceives 0" entails "0 is the
cause of A 's perception of 0 ", a suggestion for drawing the
direct/indirect distinction comes to mind. When A sees a (whole)
apple, the bulk of the apple is doing little to bring about A 's per-
ception of the apple. But the front surface is playing an absolutely
vital causal role. So perhaps we can say that the object that is
"directly" perceived is that portion of the perceived object that
plays the causal role in the perception of the object.
Even this, perhaps, is too liberal. When a thing acts, it acts in
virtue of certain characteristics it has. What our philosophy of
characteristics should be is an immense question: it is the problem
of universals. But there is some clear sense in which things have
characteristics or properties, and ordinary thought assumes that,
when things act, they act in virtue of some and only some of these
characteristics: in virtue of the sort of things that the things are.
Hence, given that A sees an apple, we can first distinguish bet-
ween those portions of the apple that do, and those that do not, act
upon A to bring about those visual perceptions of A 's, that are his
seeing of the apple. Furthermore, even when we restrict ourselves
to the causally effective portion of the apple, we can distinguish
between those characteristics of the effective portion in virtue of
which it so acts upon A, and those characteristics that are causally
irrelevant to that effect. Hence, although we can say both with

propriety and truth that A sees an apple, we also could say that
what A directly sees is a portion of the apple having a certain
number of characteristics.
Let me emphasize again that this introduced notion of direct
perception is distinct from the epistemological notion of immedi-
ate perception. A may misperceive what he "directly" perceives,
and even when he perceives it veridically, he may draw incorrect or
no inferences from his veridical perceptions. Statements of direct
perception create a referentially transparent context. It matters not
how that portion of the object perceived that actually works upon
A 's sense-organs is referred to or described. It is still that portion
that is at work upon A 's sense-organs, and so is directly per-
ceived. It matters not how the characteristics of the portion in vir-
tue of which that portion acts be referred to or described. It is still
those characteristics in virtue of which the portions acts, and so
what is directly perceived is still that portion as having those
characteristics. All this is quite different from the epistemological
notion of immediate perception. But perhaps "direct" perception
has a shadow-resemblance to immediate perception, a resemb-
lance that has helped to encourage the confused idea that the
immediate/mediate distinction can be applied to object-percep-
tion.
THE UNIMPORTANCE OF NON-DIRECT PERCEPTION
We noted at the beginning of this paper that there has been a ten-
dency for those who have drawn the immediate/mediate percep-
tion distinction to take mediate perception to be only perception
by courtesy. Yet the only thing that gives support to this notion is
the rather dubious contention that perception and inference are
completely opposed.
But suppose that we now consider the distinction between
direct and non-direct perception as we have drawn it in connection
with object-perception. There really does seem to be some reason
to denigrate and play down non-direct perception. An important
point here is the great arbitrariness there is in the rules that govern
what objects we are said to see. Warnock has pointed out(1954-
55:56) that if all I see of Lloyd George is his left leg, then I can
hardly be said to have seen Lloyd George. Yet a mariner gaining a
distant and indistinct glimpse of a small piece of shoreline might
be said to have been the first Westerner to have seen Australia. Of
course, it is possible to see why there should be these differences
in the two cases, but the reasons do not have much to do with per-
ception. In the Lloyd George case, there is the psychological and
social importance of seeing a man's face, or at least a fair portion
of him (although the portions seen may be clothed). In the
Australia case, there is the fact that we treat the continent as a uni-
tary thing for many purposes. Particularly in the context of discov-
ery, we are naturally led from the thought of the portion sighted to
the thought of the whole. In fact, however, the continent of
Australia really played little part in the mariner's perception. It is
easy to say that "in truth and strictness", the mariner saw but a
tiny portion of Australia.
Hence I am led, not too unwillingly, to say that when A is said
to see an apple, in truth and strictness he really sees a good deal
less. Ordinary speech confesses this because, as we have noted, not
just the philosopher but everybody will admit that, when A is said
to see an apple, it is not the case that A can see the rear of the
apple or, indeed, see any portion of the apple except that portion
of its skin that faces A
Some ordinary language is more ordinary than other portions of
ordinary language. It can be a very ordinary matter indeed what
objects we are said to perceive. Within the sphere of object-per-
ception, we can draw a distinction between what in this paper was
called direct and non-direct object-perception. Only direct object-
perception seems of much theoretical importance. In the case of
immediate and mediate perception, however, both sorts of percep-
tion are important theoretical notions.
9
Perceelion, SenJe-dala
and t..au1alily
INTRODUCTION
This paper attempts to work out a new line of criticism of the

Representative theory of perception. It was originally written for a
volume on A. J. Ayer's philosophy, and so it examines the Repre-
sentative theory in the form that it takes in a recent book by Ayer,
The Central Questions of Philosophy (1973).
It is customary, and I think it is useful, to classify philosophical

theories of perception as Direct Realist, Representative (Repre-
sentative Realist) and Phenomenalist theories. Ayer now definitely
rejects Phenomenalism (1973: ch. 5, sect. D). His reason for this is
that Phenomenalism must give an account of the physical world
largely in terms of unfulfilled conditional statements, but that such
statements are not adequate to this task. He therefore would per-
haps agree that we should look either to some form of Direct
Realist or else to some form of Representative theory. He, at any
rate, now favours a version of the Representative theory (ibid.,
sect E).
If we do reject Phenomenalism, and so take a Realistic view of
the physical world, we must think of the world as acting causally
upon our minds, via a certain causal chain, to produce perceptual
awareness of the world. If we also postulate entities such as sense-
data intermediate between the physical world and perceptual
awareness of the world, then it is natural to consider what place
they have, or do not have, in this causal chain. In this paper, I try
Perception, Sense-data and Causality 133
to show that this question involves a good deal of difficulty for

Representative theories, difficulty that does not arise for Direct
Realism. At the same time, however, I will indicate one other prob-
lem where the Representative theory may perhaps have an advan-
tage over Direct Realism.
I begin by considering the notion of perception of a physical
object and the causal role that the object plays in such a percep-
tion.
PERCEPTION AND CAUSATION
It is clear that many perceptual statements assert that a relation of

perception holds between a perceiver on the one hand, and a
physical object, event or state of affairs on the other. This is not
the case for all perceptual statements. If somebody looks down a
kaleidoscope, it can be said that he sees a symmetrical arrange-
ment of coloured objects. Together with many other philosophers,
Ayer used to, and I take it still does, say that this situation involves
a relation holding between the perceiver and a symmetrical
arrangement of coloured objects. But everybody would agree that
these coloured objects are not physical objects. It can be said truly
that the perceiver seems to see a symmetrical arrangement of col-
oured physical objects. Bt1t seeming to see cannot be a relation in
this context because one term of the "relation" - the physical
objects that seem to be seen - does not exist. However, it can be
said truly that in this situation, the perceiver sees a number of
irregularly disposed pieces of coloured glass. This is a relation of
perception holding between the perceiver and something physical
that is perceived.
What is the nature of the relation that holds when a perceiver
perceives some thing physical? If we abstract from the non-rela-
tional nature of the terms related, it is hard to discern anything
except a particular case of causation. The physical object, event or
whatever acts upon the sense-organs of the perceiver and produces
certain perceptions in the perceiver's mind. These perceptions are
said to be perceptions of the object. But they can be characterized,
and characterized as perceptions, in a way that does not entail the

existence of the physical object. 1
It seems clear, at the very least, that causal relation is a neces-
sary condition for this relation of perception. This is shown by
considering cases where a physical object does not act to produce
perception in a perceiver. In no such case will we admit that that
object is perceived. Suppose that a perceiver's eyes are directed
towards a mirror and, as a result, he is having certain visual
perceptions. Suppose that, behind the surface of the glass, the
physical scene exists exactly as these perceptions present it. The
perceiver's mirror-double, his identical twin except for left-right
reversal, is as far behind the surface of the glass as the perceiver is
in front of it, and so for everything else "seen in the mirror". It is
clear that the perceiver is not perceiving these objects. Rather, he
is perceiving his own body and other objects in front of the surface
of the glass. These objects visually appear to him to be in a place
where they are not and are subject to left-right distortion. Sup-
pose, now, that the mirror is a "distorting" one (that is, suppose it
distorts still further) and suppose that the physical objects behind
the mirror faithfully copy these distortions. It is still the undis-
torted objects in front of the mirror that the perceiver is perceiv-
ing.
It is clear why the objects in front of the mirror are said to be
perceived and the objects behind the surface are not said to be per-
ceived. The objects in front are acting upon the perceiver to pro-
duce his perceptions, the objects behind are not. A particularly
striking case is the perception of stars. The perception we have of a
star almost completely fails to represent the nature of the physical
object that, years before, set in train the process by which we
come to have the perception. Yet we are prepared to say that we
see the star, because the star is the cause of our perception. 2
1. Whatever the "Argument from Illusion" may fail to prove, I take it that it
does prove this. Whether the perceptions can be characterized as perceptions
without using a physical object vocabulary is a further question and one that
will not be examined here.
2. The necessity for a causal relation if there is to be perception of a physical
object is now widely granted. It is granted by Ayer (1973: 87), although he
does maintain that there must be a primitive account of perception that
Once it is granted that causation is necessary for this relation of

perception, it may be asked what more is required to yield suffi-
ciency. It is clear, of course, that not every case of a physical object
causing perception in a perceiver is a case where the perceiver per-
ceives that object. A brain-probe might bring about a visual per-
ception, but we would not say that we saw the probe. Even if it be
added that the physical object must act by stimulating sense-
organs, it is not the case that every object that produces per-
ceptions in this way is perceived. However, all that seems further
required to yield necessary and sufficient conditions for the rela-
tion of perception is (i) a condition that ensures that the perception
in some way reflects, however distortedly, the nature of the object
perceived, and (ii) some restrictions upon the nature of the causal
chain that brings the perception to be. I do not know how to spell
out these restrictions in detail, but it seems that nothing more than
such restrictions are required. 3 If so, the relation of perception that
holds when a perceiver perceives a physical object is funda-
mentally a causal relation.
So far, nothing very new. But I think it is possible to make a lit-
tle further progress in explicating this relation of perception by
reflecting upon the nature of causation. When one object acts upon
another to produce in the latter an effect of a certain nature, it does
not act as an undifferentiated whole but rather in virtue of certain of
its properties. In depressing a scale, for instance, an object acts
solely, or almost solely, in virtue of its mass. Anything closely
similar in mass, acting in the same causal context, would produce
the same, or almost the same, effect.
This trite reflection may be applied to the relation of per-
ception. Consider seeing an apple. Whether the perceptions pro-
duced by the apple be veridical or illusory, it is clear that it is not
the apple as a whole that acts upon the eyes to produce these
perceptions. If, for example, the back half of the apple had been
makes no reference to any causal relation between the percipient and the
objects he perceives. But if these objects are physical objects, the objects that
ordinarily we are said to perceive, then, I maintain, it is not possible to have a
relation of perception without a relation of causation.
3. But see Essay 8 in this volume, "Immediate Perception". note 5.
cut away, this would have had no effect upon the resultant percep-
tion. Although by hypothesis we are seeing a whole apple, the back
half of the apple is playing no causal role in the perceptual situ-
ation. Furthermore, we obviously could pare away a good deal
more of the apple and yet the object left might still have exactly
the same perceptual effect. All that is required for that effect is that
a portion of the apple's surface has certain properties. (They in fact
will be reflectional properties.) Only certain parts of the apple, and
certain properties of these parts, will play any role in bestowing
these reflectional properties. 4
If we consider these facts, then there seems to be a clear sense in
which when, as we say, somebody sees an apple, "in truth and
strictness", as Berkeley would put it, they see something far less.
They see a much smaller object and see only a selection of the pro-
perties of that smaller object. (Note, once again, that this reduced
object does not have to create veridical perceptions in the apple-
seer, even veridical perceptions of the facing surface.) Or, to put
the point as Frank Jackson has put the point in a similar context
(1977: 15-19), when it is true that somebody sees an aprJle, then
he sees it in virtue of seeing very much less than the apple. To use
Jackson's illustration, a car may touch the kerb in virtue .')f the fact
that a portion of the surface of one of its wheels is touching the
kerb. If we remember that perceiving a physical object is a causal
relation, and that things do not act causally as undifferentiated
wholes, we see that an apple is seen in virtue of seeing much less
than the apple.s
4. See "Direct and Non-direct Perception" in Essay 8 in this volume, pages

126-30.
5. It seems to me that this point helps to give further support to Ayer's support
for Russell's claim that ordinary judgements of perception like "This is a
table" entail an inference (Ayer, 1973: 80). If only a small portion of the
table is actually causally responsible for the perception produced in the per-
ceiver, then it is reasonable to think that the non-inferential component of the
perception, or the judgement based on the perception, is far less than "This is
a table". If this is correct, then even a Direct Realist ought to maintain that
most ordinary judgements of perception involve inference.
The point may also remind us of the hypothesis that G. E. Moore was for
ever considering: that visual sense-data are identical with the facing portions
of the surface of physical objects.
SENSE-DATA AND CAUSALITY
So much for the relation of perception that holds between per-

ceiver and physical objects. I now wish to see whether the results
obtained so far can be applied in criticism of the Representative
theory of perception. In a Representative theory, no physical
object or state of affairs is ever immediately perceived. What are
immeidately perceived (or, perhaps, "sensed") are representative
or intermediate entities, variously named. I shall speak of sense-
data. The perceived table, or whatever, stands in a certain relation
to a sense-datum. This is a causal relation. The table brings the
sense-datum into existence. But the sense-datum also stands in
some relation to the perceiving mind, or, if the sense-datum is held
to be within the mind, to the rest of the mind. That a relation is
involved becomes clear when we note that there is never any ques-
tion for sense-datum theorists of the sense-datum being perceived
(or immediately perceived, or sensed) yet no existing. We noted at
the beginning of the previous section that "perceive" may some-
times have the force of "seem to perceive", and that what the per-
ceiver seems to perceive need not exist. But such a possibility is
never admitted in the case of sense-data. If the sense-data always
exist, then they must always stand in some relation or other to the
mind or the rest of the mind. For the Representative theory, then,
perception of a physical object involves two relations, a relation of
the object to the sense-datum and a relation of the sense-datum to
the mind or the rest of the mind.
What is the nature of this second relation? It seems clear that,
however unconsciously, the relation between perceiver and sense-
datum has been modelled on the relation between perceiver and
physical object. If so, then the relation between perceiver and
sense-datum should be a causal one. But in any case, it is a natural
hypothesis that the sense-datum, epistemologically intermediate
as it is between physical object and perceiving mind, is also
causally intermediate. Let us therefore explore the consequences
of this hypothesis. We then shall explore the consequences of
denying that sense-data play this causal role. My contention is that
both hypotheses involve great difficulties for the Representative
theory.
First, then, let us take it that sense-data are causal inter-

mediates. Physical objects act upon the perceiver's sense-organs,
which in turn act upon some portion of his central nervous system.
As a result of this action, sense-data are brought into existence.
(There may be mixed theories according to which in some cases
the physical object causes perception of the physical object in the
perceiver without the causal chain involving intermediate sense-
data.) The sense-data act upon the perceiver's mind. 6 As a result,
the perceiver is caused to have perceptions of (or to sense) the
sense-data. These perceptions in turn bring about perceptions of
the physical object.
We have noted that the perception of a physical object,
although caused by that object, can be characterized, and charac-
terized as a perception, independently of the existence of that
physical object. In the same way, if the relation between the sense-
datum and the perception of the sense-datum is causal, then it
should be possible to characterize the perception of that sense-
datum, and characterize it as a perception, independently of the
existence of the sense-datum.
Sense-data are generally credited with having certain sensible
qualities and sensible relations (to other sense-data): sensible red-
ness, sensible adjacency, and so on. We have noted that a thing
acting as a cause produces its effect in virtue of some, and only
some, of its properties. A.natural hypothesis, therefore, is that the
sense-data act upon the perceiver to bring about perceptions of the
sense-data in virtue of their sensible qualities and relations. The
perception of a sense-datum as having a certain sensible quality
presumably will occur as a causal result of the sense-datum's
possession of that sensible quality. Or, at least, this may be pre-
sumed to be the normal case.
Cause and effect, however, are "distinct existences". It there-
fore must be logically possible for the cause to exist, but not its
customary effect or, indeed, any effect at all. It follows that the
6. Some philosophers will demand that all talk of objects acting as causes be
cashed in terms of events (events involving these objects), which are the true
causes. They may be right in this demand. If so, the event in question here
would appear to be the coming-to-be of the sense-datum or the monotonous
event of the sense-datum persisting.
perceiver's perception or sensing of the sense-datum does not have

to be a veridical perception or sensing. Just as a perceiver may
perceive a physical object, but not perceive the object as it is, so
the perceiver's perception of the sense-datum may misrepresent
the sense-datum. It is true that an effect may be described in such a
way (e.g. as an effect) that it entails the existence of a cause. But
the perception of a sense-datum, described as a certain perceptual
happening in the mind, does not entail the existence of its sense-
datum cause. It follows that the perception of the sense-datum
need not be incorrigible. Still less, the perception of the sense-
datum need not be comprehensive, embracing every feature of the
sense-datum.
This result is immensely important. Once it is granted that the
perception of the sense-datum need not be veridical, it can be
questioned whether there is any particular reason to postulate
sense-data. Historically, one of the major reasons for postulating
them has been to provide a non-physical object that is veridically
perceived in the case of non-veridical perception of physical
objects. When Macbeth seems to see a physical dagger that is not
there, many theorists have thought that a non-physical dagger-like
object must be postulated to be the thing that is veridically per-
ceived. But if it is possible that even the sense-datum should not
be veridically perceived, then this traditional motivation for
postulating sense-data is removed. This is not to say that the
postulation is incoherent, simply that a major reason for making it
is gone.
SENSE-DATA AND THE SECONDARY QUALITIES
If we accept the above sketch of the causal role of sense-data in

perception, a role that, it seems to me, it is natural to give them,
then I can only find one argument for postulating sense-data that
seems to have any force. A case can be made for postulating them
as the bearers of the secondary qualities.
Colour, sound, taste, smell, etc. certainly seem to be perceived
qualities. If they are qualities, then it is reasonable to say that they
are qualities of something. They perceptually appear to be
qualities of physical objects and/or physical states of affairs. But

there is a line of argument, based upon reasonably plausible pre-
misses, that suggests that they cannot be qualities of physical
objects. Sense-data then are introduced as alternative bearers.
There is one view, which I favour myself, according to which
the secondary qualities are nothing but, that is, are identical with,
their physical correlates. Colours, and this means perceived col-
ours, the colours that persons blind from birth are supposed to lack
acquaintance with, can, I believe, be identified with light-waves;
felt heat can be identified with mean kinetic energy; heard sound
can be identified with certain sorts of vibration in the air or other
mediums. This view is, of course, phenomenologically implausi-
ble, but I do not think that phenomenological considerations
should weigh heavily when faced with the systematic considera-
tions in favour of the identification. The identification is the
natural view to take if we want to respect what is known, or plausi-
bly believed, about the causal order of nature, and yet at the same
time want to allow that the secondary qualities bestow causal
efficacy upon the particulars that they qualify. (An important
reason for accepting the second of these propositions is that it is
hard to see how we could ever come to know of the existence of
causally idle properties.)
If this "Realistic reduction" of the secondary qualities can be
carried through, then I can find no argument for inserting sense-
data into the perceptual causal chain. Suppose, however, as Ayer
and many others would maintain, that the reduction cannot be car-
ried through. Can irreducible secondary qualities be plausibly
thought to qualify physical objects?
If physical objects are conceived of as they are conceived of in
what Wilfrid Sellars (1963: ch. 1) calls the "manifest image" of
the world, then there is no particular problem in attributing the
secondary qualities to them. However, the natural sciences, and in
particular physics, now have given us a "scientific image" of the
world that is in certain respects incompatible with the manifest
image. For instance, in the scientific image, the ordinary con-
tinuously solid physical objects of the manifest image are replaced
by swarms of "fundamental particles" in what is, however, largely
empty space. The conflict between manifest and scientific image
would be resolved in favour of the former if an operationalistic or

instrumentalistic view of physics could be accepted. However,
with Sellars and many others, I believe that this strategy is
hopelessly implausible. (A position more cautiously taken up by
Ayer (1973: 110).) In the conflict of the images, the manifest must
give way to the scientific.
However, it is difficult to hold simultaneously that (i) the scien-
tific image of the world is broadly correct; (ii) the secondary
qualities are irreducible; and yet (iii) the secondary qualities are
qualities of physical objects. Granted (i) and (ii), then the sec-
ondary qualities are additional properties to those that appear in
the theories and explanations of physics. Proposition (iii) then
becomes difficult to maintain. First, as J. J. C. Smart has
repeatedly emphasized, we would need to postulate highly
implausible bridge-laws that link the simple or relatively simple
secondary qualities with incredible physical complexities. Second,
there is no evidence that the secondary qualities so conceived
make any contribution to what G. F. Stout called "the executive
order of Nature". As already mentioned, it is difficult to see how
we could ever become aware of causally idle properties of physical
objects. In particular, our perceptions would be the same whether
or not the properties existed. In the light of these two difficulties, it
seems plausible to postulate sense-data to be bearers of the sec-
ondary qualities, and further to make such qualities causally
efficacious in the production of perceptions of the sense-data.7
I have suggested that the argument of the previous paragraph is
the only plausible reason for inserting sense-data into the percep-
tual causal chain. I re-emphasize that the argument depends upon
premisses one of which I do not accept, although I admit its
phenomenological plausibility: that of the irreducibility of the sec-
ondary qualities. Suppose, however, that all the premisses of the
argument are accepted. Are there other ways, besides the postula-
7. It seems that Ayer does not think that the sort of considerations advanced in
this paragraph are very weighty. At any rate, he expresses some preference
for the view that the secondary qualities qualify macroscopic, but not
microscopic, collections of the particles of physics (1973: 110). I suggest
that, if his view is correct, he thereby destroys the most plausible argument
for sense-data!
tion of sense-data, to deal with the problem of the location of the

secondary qualities? Two suggestions come to mind.
First, it might be maintained that the secondary qualities do not
exist, that they are, in Berkeley's (ironically intended) phrase, "a
false imaginary glare". It is clear that the secondary qualities
appear to qualify physical things. It is granted, for the sake of the
argument, that it is impossible that they should be identical with
complex physical properties of these objects. But why must it be
assumed that they qualify something? Perhaps they appear to
qualify physical objects, do not qualify these objects, and do not
qualify anything else. This would be a "Disappearance" theory of
the secondary qualities parallel to the "Disappearance" theory of
the mind advocated by Richard Rorty (1965, 1970) and Paul
Feyerabend (1963).
I think that it would be freely granted that we can form the con-
ception of complex, analysable properties that certain objects
appear to possess, but that in fact nothing possesses. Following
what might be called an "Aristotelian" tradition, I should argue
that such properties do not exist. In the realm of properties, they
resemble the present King of France in the realm of particulars. A
more "Platonic" point of view is that such properties do have
some sort of existence. But it would be generally granted, I take it,
that such complex properties need not be instantiated (at any
time). Why then should we not form the conception of an
unanalysable property that objects appear to possess (we have per-
ceptions as of objects having that property), but that in fact
nothing possesses? There are deep prejudices here, or perhaps I
should say more fairly, deep intuitions. There are philosophies
such as Logical Atomism that enshrine these intuitions ("at least
the simple elements must be real"). But it is not clear to me that
these intuitions are sacrosanct. I suspect that they involve conn-
fusions of meaning with reference and intentional objects with real
ones.
Suppose, however, that this solution is unacceptable on the
ground that, although the secondary qualities cannot be qualities
of the objects that they appear to be qualities of, they must be
qualities of some particulars. Are we forced to postulate sense-data
to be these particulars?
A possibility is that these qualities should be associated, not

with the things perceived, be they sense-data or physical objects,
but with the perceptions themselves, the perceivings rather than
the perceived. It seems implausible to say that perceptions are par-
ticulars. They are, rather, states of particulars, states of mind or
portions of minds. But it might be suggested that the secondary
qualities really qualify these particulars, yet perceptually appear to
qualify the (physical) things that the perceptions are perceptions
of. If I understand him, a view of this sort is favoured by Wilfrid
Sellars. 8
It is possible for there to be unconscious perceptions, percep-
tions that we have but are unaware of having. Normally, however,
we not merely have perceptions but are introspectively (if unself-
consciously) aware that we are having perceptions. The present
hypothesis is that, in being aware of our perceptions, we are aware
of the secondary qualities. But, by a phenomenological mistake,
the secondary qualities appear to be qualities of the objects per-
ceived.
The upholder of sense-data can hardly object in principle to this
wholesale mislocation of properties, for he himself maintains
something very similar. For consider our hypothetical perceptions
(sensings) of sense-data. Phenomenologically, they appear not to
be perceptions of sense-data but of physical things. Hence the
sense-datum theorist must admit that the secondary qualities,
which he takes to qualify sense-data, appear to qualify physical
things. How, then, can he refuse to consider the hypothesis that
they qualify not sense-data or physical things but minds?
W. D. Joske has raised an objection to this suggestion. Consider
the particular case of colour. In visually perceiving a surface, it is
perceived as coloured. Furthermore, if the colour is not perceived,
then the spatial properties of the surface are not perceived. How is
this possible if the colour is merely wrongly attributed to the sur-
face while the spatial properties are rightly attributed?
I do not think that the objection is insuperable. It may be noted,
first, that it seems not to be a necessary proposition that visual
perception of spatial properties demands that the objects having
8. See his "Empiricism and the Philosophy of Mind" in Science, Perception and
Reality (1963), p. 194.
the spatial properties be perceived as coloured. Motion and, in

some degree, size may be perceived out of the corner of the eye
without any perception of colour or even, perhaps, light and
shade. It is a fact that objects in the centre of the visual field are
perceived as coloured. It is also a fact that if no colour differences
are perceived in this centre, no spatial properties are perceived.
But this might be explained by saying that the perception had an
internal structure corresponding to the objects that seem to be per-
ceived, and that this structure was a colour-structure. A change in
the light-waves reflected by the thing perceived is represented in
the perception as a change of colour. When we perceive "the col-
our of a surface", we are really aware of the colour-structure of our
perceptions. We wrongly "project" the colour on to the (genuinely
perceived) surface. The changes induced in perception by coloured
spectacles may furnish some sort of model. Our perceptions them-
selves act as permanent coloured spectacles. A consequence of this
hypothesis would be that unconscious perception would not
involve attribution of the secondary qualities to objects.
This second suggestion for dealing with irreducible secondary
qualities without postulating sense-data has a truly epicyclic
character, particularly when Joske's objection is taken into
account. But perhaps a small epicycle that gets rid of the great
epicycle of sense-data may be justified. If it is unacceptable, a
"Disappearance" account of the secondary qualities may still be
attempted.
SENSE-DATA AND CAUSALITY AGAIN
In the third section of this paper, I tried to spell out the causal role
that it seems natural to attribute to sense-data, if we postulate
them at all. In this section, I consider alternative accounts of this
causal role.
I argued that if sense-data cause perceptions (sensings) of the
sense-data, then it must be possible to characterize the perceptions
or sensings, as perception or sensings, independently of the exis-
tence of the sense-data. It was this that seemed to make postu-
lation of sense-data redundant, unless indeed they are required to
be bearers of the secondary qualities.
Perception, Sense-doto and Causality 145
To get around this, it might be suggested that such perceptions

are "transparent" or "diaphanous". Brought into existence by the
sense-data, the perceptions are differentiated in nature from each
other solely by the differences of the sense-data that are their
causes.
The problem about this suggestion is that perceptions of
different sense-data must have different effects in the economy of
the mind. If traffic-lights are to work, the perception of a green
sense-datum, in general at least, must have a different effect upon
the mind from the perception of a red sense-datum, yet this
differentiation cannot be achieved by perceptions that differ
numerically only. It is clear that the thing with causal effect there-
fore must be green sense-datum + perception of it, red sense-
datum + perception of it, and so on. But then the perception is
seen to be an idle wheel. Why not, like the Neutral Monists, iden-
tify perception and sense-datum, although, unlike the Neutral
Monists, retaining a physical cause of the sense-datum ? 9 I believe
that the position then reached, if thought through just a little
further, turns out to be the true theory of perception: Direct Real-
ism. But I will leave this identification of perception and sense-
datum aside for the moment, because we shall shortly reach it by
another route.
Let us now consider the suggestion that sense-datum and per-
ception are related by some relation that does not involve the rela-
tion of causation. I will first argue against the hypothesis that the
related things are "distinct existences", logically capable of exist-
9. Frank Jackson, to whom I am greatly indebted for commenting on a draft of

this paper, has objected that the perception need not be an idle wheel because
it may be causally necessary for any sense-datum to have its usual effect. It
would not contribute to the difference of effect associated with the difference
between a green and a red sense-datum, but without it, neither could bring
about anything in the mind. My reply to this is to agree that this is a logical
possibility. But there would seem to be no reason to believe that the percep-
tion was playing this role. In default of further evidence that the perception is
playing this "catalytic" role, we therefore should identify it with the sense-
datum. It is perhaps important to say here that I am not arguing that the
various forms of the Represenative theory of perception I have been examin-
ing are logically incoherent. I do not think that they are. I am simply arguing
that we have no good reason to adopt them.
ing in independence of each other as sense-data and perceptions,

and then consider the alternative.
If such distinct existence is assumed, then it seems that the logi-
cal corrigibility of the perception must be granted also. For it must
be logically possible for the perception to exist while the sense-
datum does not. Once this is granted, we have the same reason for
eliminating the sense-datum that we had when considering the
causal version of the theory.
There is also a further difficulty. According to the view now
being criticized, stimulation of the sense-organs brings into exis-
tence both the sense-datum and the perception of the sense-
datum, but the two are brought into existence independently. But
if the sense-datum plays no part in bringing the perception into
existence, what reason is there to think that the sense-datum
exists? In the causal version, the sense-datum was criticized as an
unnecessary link in the chain. But now it is even less. It is a causal
by-product of the process that produces the perception. What
reason can there be to postulate such a by-product? It might be
different if there were independent reasons to postulate the sense-
datum, for instance because it gave rise to further effects, which
had to be explained. But the only reason to postulate the sense-
datum that has ever actually been offered is the perception itself.
Once again, of course, the upholder of sense-data may appeal to
the special problem of the secondary qualities. But it does not
seem that his position has been strengthened in any way by reject-
ing the causal relation. Rather, his position is weakened. If sense-
data are causally impotent, and if the secondary qualities qualify
only sense-data, then the secondary qualities bestow no causal
potence. But if a property bestows no powers, how can it be
known?
Perhaps, then, the defender of sense-data who nevertheless
denies that they bring the perceptions of sense-data into existence
ought to maintain that the two are necessarily connected? The
natural way to develop this hypothesis will be to maintain that the
existence of a perception entails the existence of a sense-datum
corresponding to the content of the perception. (We can leave
unanswered the question of whether the existence of a sense-
datum entails the existence of a perception of it: the problem of
unsensed sensibilia.) The stimulation of sense-organs may then be

thought of as bringing into existence the following complex: per-
ception necessarily connected with the corresponding sense-
datum.
Once again, however, the question arises whether we have any
reason to postulate the sense-data. The problem is not evaded by
invoking a necessary connection between the perceptions of the
sense-data and the sense-data themselves. There is an old link in
philosophical thinking between necessity and self-evidence. But it
is now sufficiently appreciated that necessary truths need not be
self-evident and that the existence of a necessary connection
requires to be argued for. In this particular case, there seems no
necessity to postulate the necessary connection. Why not simply
postulate the perception, and make it a perception of what it
appears to be a perception of: a perception of the physical world?
Since, by hypothesis, the sense-datum is causally inoperative, the
perception of it cannot be "transparent" or "diaphanous". For if
the perceptions are transparent and sense-data causally inopera-
tive, then the different causal powers of different perceptions will
be inexplicable. The perception therefore must have a content that
reflects the nature of the sense-datum. If so, why not let the per-
ception do all the work? Appeal may still be made to the problem
of the location of the secondary qualities. But it does not seem that
the necessary connection does anything to ease the problems
raised by these qualities.
The hypothesis that perception and sense-data are necessarily
connected leads on to a final modification of the theory. Necessary
connection suggests identity. The final hypothesis we might try is
that perception and sense-datum are necessarily connected
because the sense-datum is the perception.
I think that this is pretty nearly correct. But now the theory is
no longer a Representative theory, but a Direct Realist one. For if
the sense-datum is the perception, it cannot be simultaneously the
object of this perception. The perception indeed must have a con-
tent or intentional object: what seems to be perceived. Further-
more, it perhaps should be said of sense-data that they are not
strictly identical with the perception, but are really the content or
intentional objects of perceptions wrongly turned into real
objects. 10 But whether sense-data are perceptions or the content of

perception, they are not something perceived.
So in the end, I think, we arrive at a simple Direct Realist
theory. Physical objects stimulate the sense-organs, giving rise to
what may be called, indifferently, sensations, sensory states, per-
ceivings or perceptions. They are not something perceived,
although we may be, and regularly are, introspectively aware of
them. They are perceptions of physical objects, perceived as hav-
ing certain properties and relations. But the perceptions need not
be veridical, any more than our beliefs need all be true. If what a
perception is a perception of is reified into an object, then we get a
sense-datum theory, in just the same way that, if the content of a
belief is reified into an object, then we get subsistent propositions.
Insoluble problems then arise about the causal role of these
objects.
CONCLUSION
I have tried to criticize Representative theories of perception by

asking what role is played in the causal process by the
epistemological intermediates postulated by such theories. This
line of thought may be intellectually unattractive to Professor Ayer
and others on the grounds that, in their view, causal connections
are not to be found within what, following F.P. Ramsey (1931: ch.
9, sect. A), Ayer calls "the primary system" (1973: ch. 8, sect. B).
If this ground is taken, I can only say that I believe that causal con-
nection is as bed-rock a fact as any in the world, and that any
theory of the world must give an account of how things are
causally interrelated. In particular, I think that causal connection
can be as directly, that is, non-inferentially, perceived as, say, col-
our, shape or heat. To introduce sense-data, or similar objects, but
not to raise questions about their place in the causal nexus, seems
to me to be quite arbitrary. I, however, do find a Humean view of
causality quite incr~dible. One who, like Ayer, accepts a Humean
view, may find it easier to exclude causality from the primary
system.
10. As argued by Anscombe (1965).

10
Na~urali,mfl Ma~eriali1m
and Fiu~ Philo1ophy
In the first section of this paper, I define and defend a spatia-tem-

poral account of the general nature of reality. I call this doctrine
"Naturalism". In the second section, I define and defend the some-
what more specific, although still very general, doctrine of
Materialism or Physicalism. (I take it to be a sub-species of
Naturalism.) However, if we define ontology or "first philosophy"
as the most abstract or general theory of reality, then it seems that
neither Materialism nor even Naturalism is an ontology. In the
third section, I sketch very briefly the ontology I favour. Unlike
that adopted by many Naturalists and Materialists, it admits both
particulars and universals. It is Realistic, not Nominalistic. I main-
tain, in particular, that only by adopting a Realistic (but not
Platonistic) account of universals can the Naturalist and the
Materialist solve the pressing problems of the nature of causation
and of law-like connection.
NATURALISM
Naturalism I define as the doctrine that reality consists of nothing

but a single all-embracing spatia-temporal system. It is convenient
here to distinguish this proposition from the weaker claim that
reality at least contains as a part a spatia-temporal system. I will
say something in defence of the weaker claim first, and then
defend the view that reality is nothing but this spatia-temporal
system.
It is difficult to deny that a spatia-temporal system appears to

exist. But, of course, many philosophers have denied that this
appearance is a reality. Leibniz is an example. He held that reality
consists of the monads and that space and time are illusions, even
if illusions that have some systematic link with reality. Leibniz was
at least a pluralist. But for Parmenides, for Hegel and for Bradley,
reality is not a plurality but is simply one. The spatia-temporal
system is an appearance that completely or almost completely
misrepresents the one.
I will not spend any time considering such views, despite their
importance. The arguments used to establish them are all a priori.
I believe that they can all be answered. But in any case, as an
Empiricist, I reject the whole conception of establishing such
results by a priori argumentation.
But the Naturalist may seem to face a challenge to the view that
there is a spatia-temporal system from a source that he must take
more seriously: from natural science itself. It is the impression of
an outsider like myself that some speculations in fundamental
physics lead to the conclusion that, at deep levels of explanation,
space and time dissolve and require to be replaced by other, more
fundamental, principles.
However, I suggest that such speculations need not perturb the
Naturalist. I believe that he should draw the familiar distinction
between denying that a certain entity exists and giving an account
of that entity in terms of other entities. It is a very extreme view to
deny that the world has spatia-temporal features. I find it hard to
believe that even the most far-fetched speculations in fundamental
physics require such a denial. But it involves no such denial to
assert that the spatia-temporal features of things can be ultimately
analysed in terms that do not involve any appeal to spatia-tem-
poral notions. The Naturalist, as I have defined Naturalism, is com-
mitted to the assertion that there is a spatia-temporal system. But
why is he committed to asserting that spatia-temporality cannot be
analysed in terms of non-spatia-temporal principles? What is not
ultimate may yet be real. I suppose that if the principles involved
were completely different from the current principles of physics, in
particular if they involve appeal to mental entities, such as pur-
poses, we might then count the analysis as a falsification of
Naturalism, Materialism and First Philosophy 151
Naturalism. But the Naturalist need make no more concession than

this.
Consider, as a parallel example, the attitude of Materialists
towards purposes. There are some Materialists who deny that men
and other organisms have purposes. This seems to me to be a quite
foolish position to adopL Materialism may be true - my
hypothesis is that it is true - but it is a speculative doctrine. The
existence of purposes, on the other hand, is a plain matter of fact.
The prudent Materialist therefore will argue in the following way.
There is no reason to believe that what it is for an organism to
have a purpose involves anything more than the operation of
purely physical processes in the organism. (These mechanisms are,
perhaps, very sophisticated cybernetic processes.) In this way, an
account of purposes is proposed in terms of processes that do not
themselves involve purpose. No doubt this is a somewhat defla-
tionary view of what a purpose is. But it is a view of the nature of
purposes, not a denial of them.
Spatia-temporality may be analysed, just as the Materialist
claims that purpose can be analysed. However, in default of some
quite extraordinary analysis of spatia-temporality - say, in terms
of spiritual principles - Naturalism is not thereby falsified. But,
just as it is an incredible view that purposes can be analysed away,
so, I think, it is an incredible view that spatia-temporality could be
analysed away. A priori reasoning should not convince us of the
unreality of space and time. Nor, I have just argued, is it at all
plausible that a posteriori reasoning will ever drive us to the same
conclusion.
So much by way of brief defence of the positive content of
Naturalism. I turn now to its negative contention: that the world
is nothing more than a spatia-temporal system. Here we find that
philosophers and others have postulated a bewildering variety of
additional entities. Most doctrines of God place him beyond space
and time. Then there are transcendent universals, the realm of
numbers, transcendent standards of value, timeless propositions,
non-existent objects such as the golden mountain, possibilities
over and above actualities ("possible worlds"), and "abstract"
classes, including that most dilute of all entities: the null-class.
Dualist theories of mind are interesting intermediate cases,
because they place the mind in time but not in space. The same
holds for some theories of God and also, apparently, for Karl Pop-
per's recently proposed "third world" of theories, which interacts
with the "second" world of mind (Popper, 1973: chs. 3-4).
Despite the incredible diversity of these postulations, it seems
that the Naturalist can advance a single, very powerful, line of
argument that is a difficulty for them all. The argument takes the
form of a dilemma. Are these entities, or are they not, capable of
action upon the spatia-temporal system? Do these entities, or do
they not, act in nature?
In the case of many of these entities, they were at least originally
conceived of as acting in nature. God acted in the world. The
Forms in Plato's Phaedo are causes, and the Forms were
apparently transcendent universals, as well as being transcendent
numbers and transcendent standards of value. Descartes' spiritual
substances interact with matter, and Popper's "third world"
interacts with the "second world" of mind, which in turn interacts
with material things.
Nevertheless, there are very great difficulties involved in hold-
ing that any of these transcendent entities act upon the spatia-tem-
poral system.
First, there are logical or conceptual difficulties. A great many
of these entities are not thought of as capable of change. This
holds for transcendent universals, the realm of numbers and
values, propositions, non-existent objects, possible worlds and
abstract classes. In many theological systems, God is taken not to
change. Now in typical cases of causation, one change brings
about another. It follows that, if these entities work causally in the
world, they do not work in this typical way. How, then, do they
work? Could they be conceived of as sustaining certain features of
the natural world, or as exerting some sort of steady, unchanging,
pressure upon it that, when certain circumstances arise in nature,
gives rise to certain effects? Such a notion is perhaps barely possi-
ble, but the actual identification of such alleged causal operation is
a major difficulty. For instance, where sustaining causes are postu-
lated in nature, hypotheses about such causes can be tested by
observing situations where the alleged sustaining cause is absent. If
the alleged effect is also absent, the hypothesis is supported. But
no such verification is possible, even in principle, in the case of

unchanging entities.
In the Parmenides (133b-134e), Plato goes so far as to raise
logical difficulties for the conception of any relation at all (and so,
a fortiori, a causal one) between the Forms and spatia-temporal
particulars.
Even in the relatively straightforward case of the interaction of
spiritual substance with material body, conceptual difficulties have
been raised. For instance, the impossibility of specifying any
mechanism or other explanation of how the spirit acts upon its
body has been thought to be a problem. Descartes himself, as evi-
denced in particular by his correspondence with Princess
Elizabeth, thought that the action of spirit on matter involved
some conceptual difficulty.
In the case of many of the postulated transcendent entities,
there never was any thought of crediting them with causal power
in the natural world. Possible worlds, for instance, are not thought
to act upon the actual world. But even in the case of entities origi-
nally credited with power in the natural world, considerations of
the sort just sketched have been an important pressure towards
denying that they had such power.
I confess, however, that it is not upon these conceptual
difficulties that I, as a Naturalist, would place the most weight.
Instead, I would appeal to natural science. It seems to me that the
development of the natural sciences very strongly suggests that
Nature, the spatia-temporal system, is a causally self-enclosed
system. We have rather good scientific reasons to believe that,
whatever occurs in this system, if it has a cause at all, is caused
solely by other events (processes etc.) in the spatia-temporal
system. Of cause, this proposition is not susceptible of strict proof.
But in the present state of scientific knowledge, it looks a promis-
ing bet.
In the past, religious thinkers thought of God as intervening
freely in the spatia-temporal world. He might give victory to the
righteous or answer prayers for rain in defiance of the way that
matters would have shaped if the spatia-temporal system had
been left to its own devices. But even those who still believe in a
transcendent God are increasingly reluctant to believe that he acts
upon Nature in this way. They hold that he created it, and created
it for a purpose which is working itself out. But does he ever inter-
vene?
Consider, again, the Dualist theory of the mind. Descartes saw
clearly that, if Dualist Interactionism was to be made plausible,
then he must postulate places in the human brain where physical
events occurred the immediate causes of which were, in part at
least, spiritual happenings. He guessed that this happened in the
pineal gland, but we now know that the pineal gland can play no
such role. Where, then, do spiritual happenings have their immedi-
ate physical effects? Nobody has come up with a plausible sugges-
tion. Most neurophysiologists would be astounded to hear that
what happens to the brain has any other cause except earlier states
of the brain and its physical environment.
Yet the cases of God and the soul are the two most plausible
cases of things outside the spatia-temporal system acting upon it.
(It is noteworthy that they are the two examples that non-
philosophers would be most likely to give of things outside Nature
acting upon Nature.) If the anti-Naturalist case is weak here, it is
far more unpromising in the other cases. Suppose, for instance,
that there is a transcendent realm of numbers. How scientifically
implausible to think that this realm, or members of this realm, can
act on brains!
So let us now explore the other horn of the dilemma. Let us
assume that no transcendent entity acts in nature. I maintain that
this remedy is worse than the disease. The anti-Naturalist goes
from a hot frying-pan into a blazing fire.
The argument is simply this. The spatia-temporal system cer-
tainly exists. Whether anything else exists is controversial. If any
entities outside the system are postulated, but have no effect on
the system, there is no compelling reason to postulate them.
Occam's razor then enjoins us not to postulate them.
Natural science has made spectacular advances as a result of the
postulation of unobservable entities. Consider microbes, genes,
atoms, molecules, electrons, quarks and black holes. The value of
such postulations is a standing reproach to any postivistic concep-
tion of natural science. Now, contemporary analytic philosophers
are deeply affected by the justified reaction against positivism. As
Naturalism, Materialism and Fir.t Philosophy 155
a result, the fashionable defence of transcendent entities is to com-

pare them with the theoretical entities of natural science of the sort
just mentioned. For instance, "abstract" classes (classes over and
above the aggretates of their members) are postulated on the
ground that, by their means, we can explain what mathematics is
about, mathematics which in turn is required for the truth of
physics, which explains the workings of nature. The justification
for the introduction of abstract classes is thus no different from the
justification for the introduction of electrons.
In fact, however, the resemblance is superficial only. There is
this vital difference. Abstract classes, to continue with these as our
example, provide objects the existence of which, perhaps, can
serve as the truth-conditions for the propositions of mathematics.
But this semantic function is the only function that they perform.
They do not bring about anything physical in the way that genes
and electrons do. In what way, then, can they help to explain the
behaviour of physical things? Physics requires mathematics. That
is not in dispute. But must it not be possible to give an explanation
of the truth-conditions of mathematical statements purely in terms
of the physical phenomena that they apply to?
Consider, as a parallel, the dispute in the philosophy of percep-
tion between upholders of the Representative theory and the
Phenomenalists. The former theory postulates physical objects
behind the immediately perceived sense-data, the latter gives an
account of physical reality in terms of sense-data alone. Although I
reject both theories, I have no doubt that the former is by far the
more satisfactory. Suppose, however, that we were to knock away
the central prop of the Representative theory and deny that physi-
cal objects had any power to cause sense-data. The Representative
theory would then be a worthless one. It would be a bad joke to
support it by pointing out that the postulated objects at least pro-
vide truth-conditions for physical-object statements. If, further-
more, the only possible alternative theory was Phenomenalism,
then we would be under intellectual necessity to accept some
Phenomenalist account of physical-object statements. Equally, I
suggest, if the anti- Naturalist fails to endow his transcendent
possibilities, numbers, classes, etc. with any this-worldly powers,
then they explain nothing. We must insist, against him, that state-
ments about possibilities, numbers, classes, etc. be given a this-

worldly interpretation.
What these interpretations are to be is, of course, a very difficult
matter. In the case of the statements of mathematics, for instance, I
do not know that we have a currently satisfactory this-worldly
account of them. But there surely must be such an account. The
incredible usefulness of mathematics in reasoning about nature
seems to guarantee this. It then must be correct to prefer such an
account to one that postulates powerless entities outside that
world.
In Plato's Sophist, the Eleatic Stranger suggests that power is
the mark of being (2740-E). I think he is at least this far correct: if
a thing lacks any power, if it has no possible effects, then, although
it may exist, we can never have any good reason to believe that it
exists. Like all contemporary analytic philosophers, I reject the
Verification principle. But perhaps the Verification principle does
grope for a truth: the Eleatic Stranger's principle weakened in the
way that I have suggested. And if it is only spatia-temporal things
that have power, the principle bids us postulate no other realities.
It seems, then, that the anti-Naturalist who nevertheless admits
the existence of a spatia-temporal reality must try to endow his
extra entities with power to affect that reality, on pain of making
his postulation otiose. Yet, in the present state of scientific
knowledge, it seems implausible to endow these entities with such
power. I conclude that we have rather good reasons for accepting
Naturalism.
MATERIALISM
It seems best to take Materialism as a sub-species of Naturalism.

Contemporary Materialism can, of course, claim to be no more
than the descendent of the Materialism of Leucippus, Lucretius
and Hobbes. I follow J.J.C. Smart (1963) and identify contempor-
ary Materialism (or Physicalism) with the view that the world con-
tains nothing but the entities recognized by physics. Contempor-
ary Materialism takes a Realistic view of the theoretical entities of
physics - molecule, atom, fundamental particle, and so on - and
then asserts that everything there is wholly constituted by such

entities, their connections and arrangements.
The Naturalist, we saw, has first to defend himself against the
objection that, so far from being the sole reality, the spatia-tem-
poral system is not real at all. In the same way, even after Natural-
ism has been accepted, the Materialist must defend himself against
the objection that, so far from the theoretical entities of physics
being the sole realities, all that is real are ordinary macroscopic
objects. Such a view is held by those who take an Instrumentalist
or Operationalist view of physics. They do not deny the truth of
the propositions of physics, but they deny that the truth-conditions
of these propositions require the existence of any of the entities
that the propositions appear to name. The propositions of physics
do nothing but tell us how macroscopic objects behave.
The credit of such doctrines is now deservedly low, as low at the
credit of Phenomenalism about ordinary macroscopic objects. One
striking argument, which I first heard put forward by C.B. Martin,
is drawn from physics itself. According to physical theory,
macroscopic objects consist of fundamental particles associated
with each other in complex ways. But physical theory also allows
that these particles may exist not so associated with each other,
and it is theoretically possible that none of them should be so asso-
ciated. In such a theoretically possible state of affairs there would
be no macroscopic objects, although there would be fundamental
particles. But, if Instrumentalism or Operationalism is correct, this
state of affairs should be logically impossible.
I believe, therefore, that we should take a Realistic view of the
entities of physics. We saw in discussing Naturalism that to accept
the reality of the spatia-temporal system does not preclude the
view that a deeper analysis of that system may yet be given in
terms that do not involve spatia-temporal notions. In the same
way, of course, to take a Realistic view of physics does not rule out
the possibility of reaching a deeper level of analysis in terms of
which a reductive account is given of the entities and principles
currently treated as fundamental.
The main difficulties proposed for contemporary Materialism, at
any rate by contemporary philosophers, are those of the apparent
irreducible intentionality of mental processes, and the apparent
irreducible simplicity of the secondary qualities. A word about

each.
First, intentionality. Given what I take to be the utter
implausibility of Behaviourism, then it seems that a Materialist
must follow Hobbes and identify mental processes with some
subset of processes in the central nervous system. But now con-
sider such paradigm examples of mental processes as purposes and
beliefs. They have the characteristic of intentionality: they point to
a possible reality - the thing purposed or believed - which may
or may not exist. Brentano held that intentionality was a defining
characteristic of mental processes; and, more to the current point,
that it was an irreducible characteristic. If intentionality is irreduci-
ble, then Materialism is false. (irreducible intentionality may be
compatible with Naturalism. ) For such an irreducible charac-
teristic has no place in physics as we now conceive physics. The
Materialist therefore is committed to giving some reductive
account of the intentionality of mental processes. Such an account
is not all that easy to give.
A second objection to Materialism is provided by the alleged
irreducibility of the secondary qualities, or, as Herbert Feigl calls
them, the "raw feels". (His phrase is somewhat tendentious
because it begs the question in favour of a subjectivist account of
these qualities.) If they are irreducible, they fall outside the scope
of physics. They are, in Feigl's famous phrase, "nomological
danglers". They can only be linked to physical states of affairs by
artibrary bridge-laws (Feigl, 1967). (Once again, irreducible sec-
ondary qualities seem to be compatible with mere Naturalism. )
I believe, however, that the contemporary Materialist can argue
against irreducible intentionality and irreducible secondary
qualities in much the same way that, as we have seen, the
Naturalist can argue against transcendent entities. The argument
involves posing a dilemma parallel to the dilemma posed for
Naturalism. "Does intentionality, and do the secondary qualities,
bestow any causal power?"
Suppose first that they do bestow causal power. If they do, and
if this power is to be detectable, then whatever entities have these
properties, will, in suitable circumstances, act according to
different laws from objects that lack these properties. Entities lack-
ing these properties simply will obey the laws of physics. But par-
ticulars that have these extra or emergent properties also will obey
extra or emergent laws.
I argued in the previous section that the anti-Naturalist does
best to bestow causal powers on his transcendent entities. In the
same way, I think that the Naturalist anti-Materialist who believes
in irreducible intentionality and irreducible secondary qualities
does best to treat these extra characteristics as bestowing extra
powers. Nevertheless, as I now proceed to argue, to take this line is
to embrace a scientifically implausible view.
It is, of course, being assumed at this point that a Realist view is
taken of the theoretical entities of physics. Physical objects will be
arrangements of, say, fundamental particles and will obey the laws
of physics. Given certain very complex arrangements of fundamen-
tal particles, however, certain further properties of the complexes
emerge - the property of intentionality and the secondary
qualities. Complexes that have these further properties are sup-
posed to obey further laws besides the laws of physics. Now there
seems no conceptual difficulty in this supposition, but, in the light
of present knowledge, it seems scientifically implausible. I do not
claim that it is as scientifically implausible as the view that the
spatia-temporal system is not causally self-contained. But the
Materialist seems to be placing a good scientific bet if he bets
against these emergent laws. There is little evidence, for instance,
that the brain obeys any different laws from any other physical
object. Yet it is the brain, if anywhere, where emergent laws might
be exp2cted.
At any rate, just as in the case of anti-Naturalism, many anti-
Materialists are unwilling to credit their extra properties with
bestowing any extra power. Intentionality and the secondary
qualities, they conclude, are epiphenomenal, getting a free ride
upon certain configurations of matter but doing no work them-
selves. In this way, anti-Materialists seek to compromise with
Materialism. I think, however, that the result is only to com-
promise their anti-Materialism.
The argument against this second horn of the dilemma is the
same as that against the anti-Naturalist. If these characteristics fail
to endow the particulars that they characterize with causal powers,
then, with regard to the rest of the world, it is as if they did not
exist. The world goes on exactly as if they are not there - and
note that "the world's going on" includes everything that anybody
says or thinks. We can have no more reason to postulate such
causally idle properties than causally idle objects. For instance,
since the causes of the anti-Materialist's beliefs, on this hypothesis,
are something other than these alleged properties, there seems no
reason to hold the anti-Materialist beliefs.
Instead, I suggest, we do better to argue in this way:
1. The cause of all human (and animal) movements lies solely
in physical processes working solely according to the laws
of physics.
2. Purposes and beliefs, in their character of purposes and
beliefs, cause human (and animal) movements.
3. Purposes and beliefs are nothing but physical processes
working solely according to the laws of physics.
Again,
1. The cause of the expansion of mercury in a thermometer is
always a purely physical one.
2. Something's being hot can cause the expansion of mercury
in a thermometer.
3. Something's being hot is a purely physical state of affairs. 1
In the case of Naturalism, we saw that the arguments for it
require to be supplemented by giving a this-worldly account of
statements that make ostensible reference to transcendent entities.
That task is not easy, nor did I attempt it in this paper. In the case
of Materialism, similarly, we require some positive account, at
least compatible with Materialism, of the nature of intentional
mental processes and of the secondary qualities. This task is not
easy either, although I am hopeful that it can be accomplished. 2
FIRST PHILOSOPHY
But if we mean by an ontology, or first philosophy, the theory of
1. For the first of these arguments, see Medlin, 1967: sect. 2.

2. My own suggestions for accomplishing it may be found in my A Materialist
Theory of the Mind (1968) and Belief, Truth and Knowledge (1973).
the most general categories of all - such notions as particularity,

universality, number, substance and causality - then Materialism
is not a first philosophy, nor even is Naturalisn. What is more,
Naturalism and Materialism seem to be prima facie compatible
with various different first philosophies.
Historically, however, there is a link between Naturalism and
Materialism, on the one hand, and Nominalism, interpreted as the
doctrine that nothing exists except particulars, on the other.
Naturalist and Materialists are regularly found denying the reality
of universals.
What these three doctrines appear to have in common is their
commitment to Empiricism, to the method of observation and
experiment, the method of the natural sciences, as opposed to the
attempt to gain knowledge by a priori reasoning. The central
methodological postulate of natural science is that knowledge is
not to be gained a priori. As Popper has insisted, scientific
hypotheses need not be suggested by experience. But the testing
and verification of hypotheses demands experience, observation
and a submission to the facts as found. Since contemporary if not
past Materialism claims to spring out of scientific results and
plausible speculations, it is committed to Empiricism.
It is true that Materialism has sometimes seemed to be an anti-
Empiricist doctrine. Both older and contemporary Materialism are
doctrines that in some degree make rape of the senses, and hold
that immediate observations give but a first and imperfect clue to
the nature of reality. However, Materialism is only anti-Empiricist
if we identify Empiricism with such doctrines as Phenomenalism,
Positivism, Instrumentalism and Operationalism. These doctrines
make immediate observation not merely the first, but the last, word
about the nature of reality. They are excesses of Empiricism.
So the ontologies of Naturalism and Materialism have a natural
link with the epistemology of Empiricism. Nominalism, it is often
felt, has also a natural link with Empiricism. Realism about univer-
sals, on the other hand, is often linked with a priori reasoning (in
particular, with the a priori science of mathematics).
In my view, there need be no such links. Naturalism and
Materialism cannot, of course, have any truck with transcendent or
Platonic Realism. But why cannot Naturalists and Materialists
accept the more moderate doctrine of universalia in rebus? (And,

for relations, universalia inter res? ) Why should they not accept
the view that particulars have objective properties and relations,
properties and relations that are universals? Naturalism and
Materialism then could be interpreted as very general theories
about what properties and relations particulars have.
I cannot take the time here to discuss the difficulties facing
Nominalism, but I am convinced that they are overwhelming. No
version of Nominalism can ever explain the unity of the classes of
particulars said to have the same property, nor give any coherent
account of relations. Still more difficult is the attempt to combine
Nominalism with Naturalism and Materialism, for then there is no
question of calling in new entities,such as abstract classes or
merely possible particulars, to make up for the missing properties
and relations. It seems to me, indeed, that, despite tradition, it is
intellectually most plausible to combine Naturalism and Material-
ism with moderate Realism. (Although the great American
philosopher C.S. Peirce was not a Materialist, he W?S a Naturalist
and a moderate Realist, and I think that he would have accepted
the general stand I take here.)
Why have Naturalists and Materialists been attracted to
Nominalism? It is simply the agreeably hard-headed sound of the
doctrine that nothing exists except particulars? Not altogether,
perhaps. There is one line of argument for Realism about univer-
sals that appears to me to have had the effect of discrediting Real-
ism, at least among Empiricists. It is the argument from the mean-
ing of general terms. This is the argument that general words are
meaningful, meaning is a dyadic relation, hence there must be
entities for such words to mean, these somethings cannot be par-
ticulars, hence they are universals. This argument is very weak,
depending as it does upon the untenable assumption that meaning
is a relation between an expression and the thing it means. But far
worse than this, it has served to destroy the credit of Realism with
Empiricists. For if it is legitimate to move from the meaning of
general terms or predicates to universals in this automatic way, it
is established a priori that for each general term with a distinct
meaning there is a distinct universal to be that meaning. It is this, I
suggest, that Empiricists were unable to swallow. It offended
Naturalism, Materia/ism and First Philosophy 163
against their central epistemological principle that knowledge of

the existence of entities is to be gained a posteriori. Unfortunately,
however, as often happens in such matters, Empiricists mostly
drew what I think was the wrong moral. They should have con-
cluded that the Argument from Meaning is unsound, and rested the
case for universals, as it is easy to rest it, upon other considera-
tions. On particular, upon Plato's "One over Many" argument.)
Instead they rejected Realism altogether.
I wish to combine Naturalism and Materialism with what may
be called a posteriori Realism. Things (particulars) have objective
properties and relations, and these properties and relations are
universals, monadic and polyadic universals. But what properties
and what relations there are is not to be read off from discourse.
Universals are not meanings. It cannot be assumed that because a
general predicate exists, that a universal exists in virtue of which
this predicate applies. Normally it doesn't. Instead we should look
to total science to tell us what properties and what relations there
are. It is the properties and relations of particulars that determine
the causal powers of the particulars.
In this way, a posteriori Realism, Naturalism and Materialism
are seen to rest upon a common intellectual basis. That basis is the
view that the best guide we have to the nature of reality is pro-
vided by natural science. Naturalism and Materialism, although of
course still very general theories, then emerge as specifications of
a posteriori Realism: they are views about the general nature of
those properties and relations that particulars actually have.
Much remains to be said in defence of, and in elaboration of, a
posteriori Realism, but little of that much can be said here. 3 I will
indicate only, very briefly, how I think that such a Realism gives
promise of solving the problem, which, if it is not, at least ought to
be, the central problem that faces a Naturalist and a Materialist
philosophy. This is the problem of the nature of causation and law-
like connection.
3. I have tried to sketch an a posteriori Realism about universals in my

"Towards a Theory of Properties: Work in Progress on the Problems of
Universals" (1975). The position is developed in my book Universals and
Scientific Realism (1978).
That there is a deep problem here is very generally, if some-

times grudgingly, admitted by all Empiricists. It is difficult, I
believe impossible, to make sense of causal connection apart from
law-like connection, or of law-like connection apart from some
sort of universal connection: "constant conjunction" in Hume's
terms. Causation involves law, Jaw involves regularity. So much
seems to be indisputable. But then the question arises whether
causation and law-like connection involve anything more than
regularity. Here the difficulty has been to see what more could be
involved. At the same time, it is a profoundly sceptical doctrine
that nothing more is involved. The universe is surely more of a
unity than Hume thought. Furthermore, it is a scepticism that
seems unable to solve certain technical problems, in particular the
way that statements of law-like connection appear to sustain con-
trary-to-fact conditional statements while statements of mere
regularity do not. Yet the Empiricist who is also a Nominalist is
locked, or, perhaps better, humed into this sceptical position with
all its difficulties.
It has long been recognized, however - it appears that Plato
and Aristotle realized - that the acceptance of a Realistic doctrine
of universals is at least the first step to a solution of the problem. I
would try to develop the solution further in the following way.
I distinguish first between first-order universals, which are pro-
perties and relations of ordinary, first-order particulars, and sec-
ond-order universals. The latter are the properties and the rela-
tions of the first-order universals. Since the argument to universals
from meanings has been rejected, the mere applicability of various
one-place and many-place predicates to first-order universals does
not automatically ensure that they themselves have properties and
relations. But, although I make no attempt to argue the matter
here, I think it can be successfully maintained that they do have
certain properties and relations. However, these second-order pro-
perties and relations are all of the purely formal or topic-neutral
sort.
With regard to the properties of universals; there will be such
things as complexity (including, perhaps, infinite complexity) and
other structural features. Whether certain universals are or are not,
for instance, complex, is a matter to be determined a posteriori. As
for the relations of universals, besides those of inclusion and over-

lap (partial identity), one universal may necessitate, probabilify in
some degree, or exclude another universal.
Let us concentrate upon the relation of necessitation that one
particular universal may bear to another. It is not logical necessita-
tion. The relation would not obtain "in every possible world".
Following in the footst£ps of many other philosophers, we may
call it natural necessitation. It is to be discovered a posteriori, by
the experiential and experimental methods of natural science. But
if such a relation exists between certain universals, then it entails a
"constant conjunction" between the particulars falling under these
universals. A particular fact about the connection of certain
universals logically necessitates a general fact about the connec-
tion of (first-order) particulars. Take an artificially simple example,
and suppose that the universal being F necessitates being G. This
non-logical necessitation entails that, for all x, if xis F, then xis G.
But the reverse entailment does not hold. It might be the case that,
for all x, if xis F, then xis G, but fail to be the case that being F
necessitates being G. In the latter case, there would be no more
than an "accidental" universal conjunction of the two properties in
the particulars.
Nomic or law-like necessity I take to be such a relation between
universals. Causal connection I take to be a particular (and very
complex) case of law-like necessity, and so to involve relations
between universals.
I hope that this mere lightning sketch of a line of thought indi-
cates some of the attractions that a moderate and a posteriori
Realism has for an Empiricist ljke myself. It seems to me to be the
natural first philosophy to combine with Naturalism and Material-
ism. But, more than that, a posteriori Realism, especially when
linked with a doctrine of natural necessitation, furnishes a natural
and fruitful perspective from which to view the whole dispute
about the truth or falsity of these two very general cosmological
hypotheses.
Reference/
Anscombe, G.E.M. (1965). "The Intentionality of Sensation: A Gram-

matical Feature", in Analytical Philosophy (Second Series), ed.
R.J. Butler (London: Blackwell).
Armstrong, D.M. (1961). Perception and the Physical World (London:
Routledge & Kegan Paul).
_ _ _ _ (1968). A Materialist Theory of the Mind (London: Rout-
ledge & Kegan Paul).
(1973). Belief, Truth and Knowledge (Cambridge:
Cambridge University Press).
_ _ _ _ (1975). "Towards a Theory of Properties: Work in Progress
on the Problems of Universals" 50: 145-55.
(1978). Universals and Scientific Realism, 2 vols.
(Cambridge: Cambridge University Press).
Ayer, A.J. (1973). The Central Questions of Philosophy (London:
Weidenfeld & Nicolson).
Beddie, B.D. (1977). Review of Liberty and Politics, ed. 0. Harries,
Pergamon Press (Aust.), Quadrant 21 (May): 78-80.
Berkeley, G. (1948-56). Works, eds. A.A. Luce and T.E. Jessop (London:
Nelson).
Burke, E. (1887). Works (London: Nimmo).
Campbell, K.K. (1969). "Colours", in Contemporary Philosophy in
Australia, eds. R. Brown and C.D. Rollins (London: Allen &
Unwin), pp. 132-57.
Chisholm, R. (1964). "The Descriptive Element in the Concept of
Action", Journal of Philosophy 61: 613-25.
Danto, A. (1965). "Basic Actions", American Philosophical Quarterly 2:
141-48.
Davidson, D. (1963). "Actions, Reasons and Causes", Journal of Philoso-
phy 60: 685-700.
References 167
_____ (1973). "Freedom to Act", in Essays on Freedom of Action,

ed. T. Honderich (London: Routledge & Kegan Paul), pp. 137-56.
Dye, J.L. (1967). "The Solvated Electron", Scientific American 216: 76-
83.
Feigl, H. (1967). The "Mental" and the "Physical" (Minneapolis: Min-
nesota University Press).
Feyerabend, P. (1963). "Mental Events and the Brain", Journal of
Philosophy 60: 295-96.
_ _ _ _ (1975). Against Method (London: New Left Books).
Grice, H.P. (1961). "The Causal Theory of Perception", Aristotelian
Society Proceedings Supp. vol. 35: 121-52.
Gundersen, K. (1971). Mentality and Machines (New York: Anchor).
Jackson, F.C. (1977). Perception: A Representative Theory (Cambridge:
Cambridge University Press).
Kneale, W.C. (1969). Critical notice of A Materialist Theory of the Mind
by D.M. Armstrong in Mind 78: 292-301.
Kuhn, T. (1962). The Structure of Scientific Revolutions (Chicago:
University of Chicago Press).
Lewis, D.K. (1966). "An Argument for the Identity Theory", Journal of
_____ (1969). Convention (Cambridge, Mass.: Harvard University
Press).
(n.d.). "Veridical Hallucination and Prosthetic Vision"
(unpublished lecture).
Locke, J. (1975). Essay Concerning Human Understanding, ed. P. Nid-
ditch (Oxford: Oxford University Press).
Lucas, J.R. (1961). "Minds, Machines and Godel", Philosophy 36: 112-
27.
Medlin, B. (1967). "Ryle and the Mechanical Hypothesis", in The Iden-
tity Theory of Mind, ed. C.F. Presley (St Lucia: Queensland Univer-
sity Press), pp. 94-150.
Moore, G.E. (1925). "A Defence of Common Sense", in Contemporary
British Philosophy (Second Series), ed. J.H. Muirhead (London:
Macmillan) .
Oakeshott, M. (1962). Rationalism in Politics (London: Metheun).
Place, U.T. (1956). "Is Consciousness a Brain Process?", British Journal
of Philosophy 4 7: 44-50.
Popper, K.R. (1969). "Towards a Rational Theory of Tradition", in Con-
jectures and Refutations (London: Routledge & Kegan Paul), pp.
120-35.
- - - - - (1973). Objective Knowledge (Oxford: Oxford University
Press).
Ramsey, F.P. (1931). The Foundations of Mathematics (London: Rout-

Rorty, R. (1965). "Mind-Body Identity, Privacy and Categories", Review
of Metaphysics 19: 24-54.
_____ (1970). "Incorrigibility As the Mark of the Mental", Journal
of Philosophy 67: 399-424.
Russell, B. (1912). The Problems of Philosophy (London: Home Univer-
sity Library).
Ryle, G. (1949). The Concept of Mind (London: Hutchinson).
Sapir, E. (1930). "Custom", in Encyclopaedia of the Social Sciences, ed.
E.R.A. Seligman (New York: Macmillan), vol. 4, pp. 658-62.
Sellars, W. (1963). Science, Perception and Reality (London: Routledge
& Kegan Paul).
Smart, J.J.C. (1959). "Sensations and Brain Processes", Philosophical
Review 68: 141-56.
_____ (1963). Philosophy and Scientific Realism (London: Rout-
Taylor, R. (1966). Action and Purpose (Englewood Cliffs, N.J.: Prentice
Hall).
Unger, P. (1972). "Propositional Verbs and Knowledge", Journal of
Warnock, G.J. (1954-55). "Seeing", Aristotelian Society Proceedings
55: 201-18.
D. M. Arm1lron~: Wrilin~l up lo and includin~ 1979
This bibliography includes books, articles, discussion notes and major

reviews, but does not list any subsequent reprinting, translation, etc. of
any item.
1954
"Berkeley's puzzle about the water that seems both hot and cold",
Analysis 32: 44-46.
1955
"Illusions of Sense", Australasian Journal of Philosophy 33: 88-106.
1956
"Discussion: Berkeley's New Theory of Vision", Journal of the History of
Ideas 17: 127-129.
1958
"Critical Notice: A.J. Ayer's The Problem of Knowledge", Australasian
Journal of Philosophy 36: 128-45.
1959
"Discussion: Mr Arthedeva and Naive Realism", Australasian Journal of
1960
Berkeley's Theory of Vision (Melbourne: Melbourne University Press).
1961
Perception and the Physical World, International Library of Philosophy
and Scientific Method (London: Routledge & Kegan Paul).
1962
Bodily Sensations, Studies in Philosophical Psychology (London: Rout-
1963
"Absolute and Relative Motion", Mind 72: 209-23.
"Discussion: Max Deutscher and Perception", Australasian Journal of
"Discussion: Vesey on Sensations of Heat", Australasian Journal of
"Is Introspective Knowledge Incorrigible?", Philosophical Review 72:
417-32.
1964
"Discussion: Vesey and Bodily Sensations", Australasian Journal of
1965
Berkeley's Philosophical Writings, edited, with an introduction (New
York: Collier-Macmillan).
"Freedom of the Will", The Pluralist (Sydney) 2: 21-25.
"A Theory of Perception", in Scientific Psychology, eds B. Wolman and
E. Nagel (New York: Basic Books), pp. 489-505.
(with R.N. Spann)
"The Knopfelmacher Case", Minerva 3: 334-44.
1966
"Critical Notice: Richard Taylor's Action and Purpose", Australasian
Journal of Philosophy 44: 231-40.
"The Nature of Mind", Arts (Proceedings of the Sydney University Arts
Association) 3, no. 1: 37-48.
1967
"Review: Brain and Mind, ed. J.R. Smythies", Philosophical Review 76:
246-49,
1968
"The Headless Woman and the Defence of Materialism", Analysis 29:
48-49.
Locke and Berkeley: Collection of Critical Essays, edited with C.B. Mar-
tin (New York: Doubleday, and London: Macmillan) .
.A Materialist Theory of the Mind, International Library of Philosophy and
Scientific Method (London: Routledge & Kegan Paul).
"The Secondary Qualities", Australasian Journal of Philosophy 46: 225-
41.
1969
"Colour Realism and the Argument from Microscopes", in Contemporary
Bibliography of Writings 171
Philosophy in Australia, eds R. Brown and C.D. Rollins (London:

Allen & Unwin), pp. 119-31.
"Dispositions are Causes", Analysis 30: 23-26.
"Does Knowledge Entail Belief?", Proceedings of the Aristotelian
Society 70: 21-36.
1971
"Meaning and Communication", Philosophical Review 80: 427-47.
1972
"Materialism, Properties and Predicates", The Monist 56: 163-76.
1973
"Acting and Trying", Philosophical Papers 2: 1-15.
Belief, Truth and Knowledge (Cambridge: Cambridge University Press).
"Change and Continuity in Philosophy", Quadrant 17: 19-23.
"Epistemological Foundations for a Materialist Theory of the Mind",
Philosophy of Science 40: 178-93.
"Language and Mind", in Linguistics and the Mind, ed. D. Douglas
(Sydney: Sydney University Extension Board), pp. 1-10.
1974
"Infinite Regress Arguments and the Problem of Universals", Australa-
sian Journal of Philosophy 52: 191-201.
1975
"Beliefs and Desires as Causes of Action: A Reply to Donald Davidson",
Philosophical Papers 4: 1-7.
"Towards a Theory of Properties: Work in Progress on the Problem of
Universals", Philosophy 50: 145-55.
1976
"Immediate Perception", in Essays in Memory of Imre Lakatos, Boston
Studies in the Philosophy of Science, Vol. 39, eds R.S. Cohen, P.K.
Feyerabend and M. Wartofsky (Dordrecht-Holland: D. Reidel), pp.
23-35.
"Incorrigibility, Materialism and Causation", Philosophical Studies 30:
125-27.
"Modern Materialism and the Mind", Proceedings of the Russel/ian
Society (Sydney) 1: 44-55.
"The Nature of Tradition", in Liberty and Politics, ed. 0. Harries
(Sydney: Workers' Educational Association of N.S.W. and
Pergamon Press (Australia) ), pp. 7-19.
1977
"The Causal Theory of the Mind", in Neue Hefte fiJr Philosophie, Heft 11
(Gottingen: Vandenhoek & Ruprecht), pp. 82-95.
"Naturalism, Materialism and First Philosophy", in 1st Systematische
Philosophie Moglich ?, Proceedings of the Stuttgarter Hegel-
Kongress, - 1975 (Bonn: Herbert Grundmann), pp. 411-25; re-
published 1978, in Philosophia 8: 261-76.
"On Metaphysics", in "Fifty Years of John Anderson", Quadrant 21
(May): 65-69.
"The Problem of Universals", Proceedings of the Russel/ian Society
(Sydney) 2: 11-22.
1978
"On Passing the Buck": Commentary on Roland Pucetti and Robert W.
Dykes, "Sensory Cortex and the Mind-Brain Problem", The
Behavioural and Brain Sciences 3: 346.
"Review: The Self and Its Brain by K.R. Popper and J.C. Eccles", Times
Literary Supplement (17 February): 183-84.
Universals and Scientific Realism (vol. 1: Nominalism and Realism; vol.
2: A Theory of Universals), (Cambridge: Cambridge University
Press).
"What is Consciousness?", Proceedings of the Russel/ian Society
(Sydney) 3: 65-76.
1979
"Laws of Nature", Proceedings-of the Russel/ian Society (Sydney) 4: 46-
61.
"Perception, Sense-data and Causality", in Perception and Identity:
Essays presented to A.J. Ayer with his replies to them, ed. G.F.
Macdonald (London: Macmillan), pp. 84-98.
"Three Types of Consciousness", in Brain and Mind, Ciba Foundation
Series 69 (new series), (Amsterdam: Excerpta Medica), pp. 235-
41.
Index
abilities, 82 certainty, search for, 50, 54, 120

action-recipes, 78-81 Chisholm, R., 76-77, 79, 81
Anscombe, G.E.M., 148n classes, 155
Argument from Meaning (to univer- colour, 104-18 passim, 143-44;
sals), 162-63 standing versus transient senses,
Aristotle, 164 115-18
Attribute theories of mind, 34 commonsense, 40, 44
Aveyron, Wild Boy of, 64 computers, 57, 65-66, 84-85
Ayer, A.J., 132-48 passim concept revision, 45-49
conceptual analysis, 26-27, 52-54
basic actions, 77, 80-81 consciousness, 11-15, 24, 55-67;
Beddie, B., 103n minimal, 55-58; perceptual, 58-
Behaviourism, 4-7, 8-10, 11, 13, 59; introspective, 59-67. See
17, 22, 52, 53 also introspection
beliefs, 23-24, 56-57, 160 customs, 91, 92, 95-96, 99, 100-
Bennett, D., 82-83 101, 103n
Berkeley, G., 104, 106, 119-21,
123, 126, 136, 142 Danto, A., 80
bodily sensations, 28, 29 Davidson, D., vii, 74, 82-85
Bradley, F.H., 150 Descartes, R., 4, 8, 55, 120, 152,
Brentano, F., 25, 158 153, 154
Burke, E., 99n, 103 Dialectic, 10
Dickens, Charles, 66
Campbell, K., 105, 114n, 115 Direct Realism, 104-18 passim,
capacities, 9, 13, 86-87 132-33, 145, 147-48
Cartesian theories of mind. See disappearance theory of mind, 43-
Dualist theories of mind 49, 142
causal quiescence, 57-58 dispositions, 5-6, 8-10, 79, 84, 91
causal theory of the mind, 7-10, 19- Double-aspect theories of mind, 34
30, 35-36, 52-54 dreams, 67
causation, 135, 148, 152-53, 163- Dual-attribute theories of mind, 34
65 Dualist theories of mind, 25, 57,
174 Index
151-54 passim Leucippus, 156

Dye, J., 113 Lewis, D., vii-viii, 35-36, 128n
Locke, J., 14, 61, 117-18.
Eleatic Stranger, 156 Logical Atomism, 142
emotions, 28, 29 Lucas, J.R., 43n
Empiricism, 150, 161-65 Lucretius, 156
Feigl, H., 158 Macbeth, 124, 139

Feyerabend, P., 4n, 43-44, 49, 142 "manifest image" of the world, 140-
freedom, 93-95 41
Martin, C.B., 127, 157
God, 151, 152, 153-54 Marx, K., 94
Godel, K., 42-43 Medlin, B., 160n
Grice, H.P., 71, 127, 128n memory, 66-67
Gundersen, K., 54n mental concepts, causal analysis of.
See causal theory of the mind
habits, 91, 92, 95-97, 100 mental images, 24
hallucination, veridical, 127 minimum visible, 107-9
Harvey, W., 2 Moore, G.E., 40, 42, 136n.5
Headless Woman illusion, 51 murder, concept of, 46-4 7
Hegel, G.W.F., 10, 150
Hobbes, T., 2, 156, 158 necessary truth, 54
Hume, D., 16, 23, 64, 164 neutral Monism, 145
Nominalism, 34, 149, 161-62
Illusion, Argument from, 134n
imitation, 97-100 Oakeshott, M., 99
inference, in perception, 120, 125, Occam's razor, 43, 154
136n Operationalism, 5, 141, 157, 161
"inner sense", 61, 64. See also
introspection Paralysis, unexpected, Argument
Instrumentalism. See Operationalism from, 68-73, 85-86
intentionality, 25-26, 157-60 Parmenides, 150
introspection, 24, 25, 37, 49-51, Parmenides, 153
60-67. See also consciousness Peirce, C.S., 92, 162
perception: appearance statements,
Jackson, F., 136, 145n 124-25; and causation, 48, 126-
Joske, W.O., 143-44 30, 133-36, 137-39, 144-48;
and consciousness, 58-59; direct
Kant, 1., 14, 16, 61, 64 vs indirect, 126-31; as mapping,
kinaesthetic perception, 62 23-24, 26; mediate, 120; object
Kneale, W.C., 34 statements, 121-23; proposi-
Kuhn, T., 4n tional, 123-24; and secondary
qualities, 27-28, 37; and selec-
Lakatos, 1., 24 tive behaviour, 12-14. See also
Laws of Nature, 163-65 Direct Realism, Phenomenalism,
Leibniz, G., 150 Representative Realism
Index 175
Phaedo, 152 "scientific image" of the world, 140-

Phenomenalism, 106-7, 132, 155, 41
161 scrutinizing, 63
phenomenological method, 40-41, secondary qualities, 27-31, 36-38,
49-52 49-52, 117, 139-44, 146, 158-
philosophy, 41; as conceptual 60
analysis, 16-19 self, 64-67
pineal gland, 154 Sellars, W., 140-41, 143
Place, U.T., 35, 52 sense-data, 120, 126, 137-48, 155
Plato, 152, 153, 156, 163, 164 sense impressions. See sense-data
poison, concept of, 20-21, 22, 25- skills, 79, 82
27 Smart, J.J.C., 35, 37, 42, 43n, 52,
Popper, K.R., 18, 42, 94, 152, 161 141' 156
practical reasoning, 77-82, 83, 85 Sophist, 156
practices, 91, 92, 93, 99 stars, perception of, 128n, 134
properties, 34-35, 129-30, 142 Stout, G.F., 141
proprioception, 61-62, 65 subtraction, method of, 127
purposes, 22-24, 26, 151, 160 Suttee, 97
Ramsey, F.P., 148 Taylor, R., 68-69, 75-76

rational consensus, 3, 39-43
Realism, about universals, 149, 162- Unger, P., 123
65 universals, 129, 149, 152, 161-65
Representative Realism, 104, 132-
33, 137-39, 145-48, 155 Verification principle, 156
representative theory of perception. Vienna Circle, 16
See Representative Realism volition, 70, 73
rituals, 91, 92
Rorty, R., 43-44, 49, 142 Warnock, G., 122, 130-31
rules, 91-92 Watson, J.B., 4
Russell, B., 64, 136n.5 Whitehead, A.N., 98
Ryle, G., 4-7, 9, 16-18, 74-75, 91n Wilkins, R., 99n
Wittgenstein, L., 16-17
Sapir, E., 96n
science, authority of, 2-4, 38-43,
52-54

D.M. Armstrong - Nature of Mind-Harvester Press (1981)

Uploaded by

Copyright:

Available Formats

You might also like

D.M. Armstrong - Nature of Mind-Harvester Press (1981)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

D.M. Armstrong - Nature of Mind-Harvester Press (1981)

Uploaded by

Copyright:

Available Formats

T~e

TLe Harve1~er Prell

Publishers: john Spiers and Margaret A. Boden

© University of Queensland Press, 1980

British Library Cataloguing in Publication Data

Manufactured in Hong Kong by Silex Enterprise & Printing Co.

All rights reserved

1 The Nature of Mind 1

These papers deal with the philosophy of mind, the philosophy of

that certain social concepts can be analysed in terms drawn from

and Identity: Essays presented to A.J. Ayer with his replies to

set towards the physico-chemical hypothesis. And we have

THE AUTHORITY OF SCIENCE

us what a watershed in the intellectual history of the human race

superseded. Must we not therefore appeal to these relative certain-

DEFINING THE MENTAL

my addressing strong words to you, striking you, turning my back

the Behaviourist are, for the most part, of this temporary

2. Ryle, 1949: 43; emphasis added.

thing is obvious that there is something actually going on in me

brittleness is the fact that it gives rise to the characteristic

cumstances. But to do this is to abandon the central inspiration of

THE PROBLEM OF CONSCIOUSNESS

behaviour. Suppose we want to decide whether an animal can per-

Now how is this discussion of perception related to the question

As I have emphasized before, I have done no more than sketch a

IS PHILOSOPHY JUST CONCEPTUAL ANALYSIS?

What can philosophy contribute to solving the problem of the rela-

tein's. After all, it retained a special, if much reduced, realm for

philosophy is, the consideration of the methods of science must be

THE CONCEPT OF A MENTAL STATE

same way, it has been thought, a perception or a feeling of sorrow

die. I This is but a rough analysis of the concept the structure of

1. "Any substance which, when introduced into or absorbed by a living organ-

food is a state of a person or animal that characteristically brings

to assert that it is the notion of a cause within which drives, or

poses.) The logical dependence of purpose on perception and

May not the intentionality of mental states, therefore, be in

poisons kill. It is at the centre of our notion of what poisons are

THE PROBLEM OF THE SECONDARY QUALITIES

No discussion of the Causal theory of the mental concepts is com-

smells. These phenomenal qualities, it may be argued, endow

being red in Causal terms would lead us to produce such analyses

fact complex, and perhaps they are complex characteristics of a

The doctrine of mental states and of the secondary qualities

In this paper, I shall be arguing for the general plausibility of a

THE GENERAL NATURE OF A MATERIALIST THEORY OF THE MIND

A slogan of this sort is not to be despised. It is like a view of terrain

1. The term "Double-aspect" is misleading because even a Central-state

cess or state.He proposes that all mental concepts are concepts of

"secondary qualities". Whether these qualities be treated as

3. See, in particular, his Philosophy and Scientific Realism (1963).

identifying the properties that they could classify perceptually only

Suppose that we want to cast light upon some great metaphysical

tion is also fairly uncontroversial, although the answer, if accepted,

exist," "The moral law exists," "There is no moral law") we find

5. Most famously, in his "A Defence of Common Sense" (1925).

theories distort even the most scrupulous attempts to make

also expands, if much more slowly. In the present, at least, the