An Introduction to the Philosophy of Mind - Mental states

3
Mental states
In the previous chapter, we focused on two important meta-
physical questions in the philosophy of mind. One was the
question of whether persons or subjects of experience are
identical with their physical bodies, or certain parts of those
bodies, such as their brains. The other was the question of
whether the mental states of persons, such as thoughts and
feelings, are identical with certain physical states of their
bodies, such as states of neuronal activity in their brains.
Many materialists would endorse positive answers to both of
these questions, although later in this chapter we shall
encounter a species of materialism which denies that mental
states, as we ordinarily conceive of them, really exist at all.
But before we examine that position, it is worth remarking
that, so long as one is a realist about mental states – that is,
so long as one considers that states of thinking and feeling
really do exist – one can, for many purposes, afford to remain
neutral with regard to the question of whether or not mental
states are identical with physical states. There are many
issues in the philosophy of mind which we can usefully discuss
without presuming to be able to resolve that question. And
this is just as well, knowing as we now do how thorny a ques-
tion it is. One of these issues is that of how we can best
characterise and classify the various different kinds of mental
state which, if we are realists, we believe to exist. So far we
have been talking about mental states quite generally, with-
out differentiating between them in any significant fashion.
But in a detailed description of the mental lives of persons
we need to be able to distinguish, in principled ways, between
39

An introduction to the philosophy of mind40
sensations, perceptions, beliefs, desires, intentions, fears, and
many other kinds of mental state: and providing a satisfact-
ory account of these distinctions is no easy matter. It is to
the difficulties besetting that task that we shall turn in this
chapter.
PROPOSITIONAL ATTITUDE STATES
Let us begin by considering those mental states that philo-
sophers like to call propositional attitude states. For brevity, I
shall refer to them as ‘attitudinal states’. These include
beliefs, desires, intentions, hopes and fears, to name but a
few. A common feature of such states is that we may ascribe
them to subjects of experience by using statements of the
form ‘S φs that p’. Here ‘S’ denotes a particular subject or
person, ‘p’ stands for some proposition – for example, the
proposition that it is raining – and ‘φ’ represents any so-
called verb of propositional attitude, such as ‘believe’, ‘hope’
or ‘fear’. Such verbs are called ‘verbs of propositional atti-
tude’ because each of them is considered to express a particu-
lar attitude which a subject may have towards a proposition.
(What propositions are and how subjects can have ‘attitudes’
towards them are matters which we shall take up more fully
in the next chapter.) Thus, the following sentences can all
be used to ascribe attitudinal states to subjects: ‘John
believes that it is raining’, ‘Mary hopes that she has an
umbrella’, and ‘Ann fears that she will get wet’. In each case,
the ‘that’ clause expresses the propositional content of the atti-
tudinal state which is being ascribed. A number of questions
immediately arise concerning attitudinal states and our
knowledge of them. For instance: what, exactly, makes a

belief, say, different from a hope or a fear? One and the same
subject may simultaneously possess two different attitudinal
states with the same propositional content: for example, Ann
may both fear that she will get wet and believe that she will
get wet. What does this difference of ‘attitude’ to one and
the same proposition amount to? Again, how do we know what
attitudinal states another person possesses – indeed, how do
we know what attitudinal states we ourselves possess? Notice
Mental states 41
here that knowledge itself is an attitudinal state, so that
knowledge of someone’s attitudinal states appears to be a
‘second-order’ attitudinal state – as in the case of John’s
knowing that Ann fears that she will get wet, or John’s know-
ing that he himself believes that it is raining.
We may be inclined to think that there is no special prob-
lem about identifying our own attitudinal states, at least at
the times at which we possess them. Surely, if I have any
beliefs, desires, hopes or fears right now, then I, at least,
must know what they are, even if no one else does. Descartes
certainly seems to have assumed that this is so. But how do I
know? Perhaps it will be suggested that I know by a process
of ‘reflection’ or ‘introspection’, which somehow reveals to
me what I am thinking and feeling. After all, if someone asks
me whether I believe or desire such-and-such, I am usually
able to reply fairly spontaneously, so it seems that I must
have some sort of direct access to my own attitudinal states.
On the other hand, perhaps the feeling of having such a
direct or ‘privileged’ access to one’s own attitudinal states is
just an illusion. Perhaps I find out what I believe, for
example, by hearing myself express my beliefs in words,

whether out loud to other people or just sotto voce to myself.
This would make my knowledge of my own beliefs no differ-
ent in principle from my knowledge of other people’s beliefs,
which I commonly discover by hearing them express them.
Of course, sometimes I can form a good idea of someone’s
beliefs on the basis of their non-verbal behaviour, as when I
judge that John believes that it is raining when I see him
open his umbrella. But perhaps the same is true, at times,
with regard to my knowledge of my own beliefs – as when I
surprise myself by the fact that I have taken a certain turning
at a road junction and realise that I must believe that it will
take me to my desired destination. (We shall explore the
nature of self-knowledge more fully in chapter 10.)
BEHAVIOURISM AND ITS PROBLEMS
There are some philosophers who are extremely sceptical
about the reliability – or, indeed, the very existence – of
An introduction to the philosophy of mind42
introspection as a source of knowledge about our own mental
states. At its most extreme, this scepticism finds expression
in the doctrine known as behaviourism. Behaviourists hold that
the only sort of evidence that we can have concerning any-
one’s mental states, including our own, lies in people’s out-
wardly observable behaviour, both verbal and non-verbal.
‘Scientific’ behaviourists take this view because they think
that a science of mental states – which is what scientific psy-
chology in part claims to be – ought only to rely upon object-
ive empirical evidence which can be corroborated by many
independent observers, whereas introspection is necessarily
a private and subjective affair. But more radical behaviour-
ists – those who are sometimes called ‘logical’ behaviourists –

would go even further than this. They maintain that what it
is to ascribe a mental state to a person is nothing more nor
less than to ascribe to that person some appropriate behavi-
oural disposition.
1
A ‘behavioural disposition’, in the sense
understood here, is a person’s tendency or propensity to
behave in a certain way in certain specified circumstances.
Thus, for instance, a logical behaviourist might suggest that
to ascribe to John a belief that it is raining is simply to
ascribe to him a disposition to do such things as: take an
umbrella with him if he leaves the house, turn on the wind-
screen wipers if he is driving the car, assert that it is raining
if he is asked what the weather is like – and so on. The list
must, of course, be an open-ended one, since there is no limit
to the ways in which someone might evince the belief in ques-
tion. This sort of account is intended to apply not only to
what we have been calling ‘attitudinal’ states, but also to
what we might call sensational states, such as pain and nausea.
To be in pain – for example, to feel a sharp twinge in one’s
big toe – is not to have some attitude towards a proposition:
pains do not have ‘propositional content’. But the logical
behaviourist will once again contend that to ascribe to John
1
A sophisticated version of logical behaviourism is developed by Gilbert Ryle in
his The Concept of Mind (London: Hutchinson, 1949). See especially p. 129, for
Ryle’s account of belief.
Mental states 43
a feeling of pain in his big toe is simply to ascribe to him a
disposition to do such things as: wince and groan if the toe

is touched, hobble along if he has to walk, assert that his toe
hurts if he is asked how it feels – and so on. The list is, once
again, an open-ended one.
Now, the very fact that such lists of behaviour are necessar-
ily open-ended ones presents a problem for logical behaviour-
ism. Take the list of activities associated with John’s belief
that it is raining. What can possibly be meant by saying that
John has a disposition to do ‘such things’ as are on this list?
What unifies the items on the list, other than the fact that
they are the sorts of things that one might reasonably expect
a person to do who believes that it is raining? It seems that
one must already understand what it means to ascribe to some-
one a belief that it is raining in order to be able to generate
the items on the list, so that the list cannot be used to explain
what it means to ascribe to someone such a belief. Even set-
ting aside this difficulty, however, there is another and still
more serious problem which besets the logical behaviourist’s
account. Take again John’s belief that it is raining. The prob-
lem is that someone may perfectly well possess this belief
and yet fail to do any of the things on the associated list in
the appropriate circumstances. Thus John may believe that
it is raining and yet leave his umbrella at home when he goes
out, fail to switch on the windscreen wipers when he is driv-
ing, deny that it is raining when he is asked about the
weather, and so on. This is because how a person who pos-
sesses a certain belief behaves in given circumstances does
not depend solely upon what that belief is, but also upon what
other attitudinal states that person may happen to possess at
the same time, such as his or her desires. John may, for
instance, be unusual in that he likes to get wet in the rain

and so never takes an umbrella with him when he believes
that it is raining. Indeed, whatever behaviour the logical beha-
viourist attempts to represent as being characteristic of
someone who believes that it is raining, it is possible to envis-
age a person who holds that belief and yet is not disposed to
behave in that way, if only because such a person might have
An introduction to the philosophy of mind44
a very strong desire to deceive others into thinking that he
or she does not hold that belief. The same is true with regard
to a person’s sensational states: one may feel a pain in one’s
big toe and yet suppress the kinds of behaviour which the
logical behaviourist says are definitive of feeling such a pain,
because one is determined to be stoical about it or does not
want to appear weak. So we see that there can in fact be
no such thing as behaviour that is uniquely characteristic
of someone who possesses an attitudinal state with a given
propositional content or a sensational state of a certain type.
And consequently it is impossible to explain what it means for
someone to possess such a state in terms of his or her sup-
posed behavioural dispositions. ‘Logical’ behaviourism is
clearly doomed to failure.
However, the failure of logical behaviourism still leaves
untouched the weaker form of behaviourism mentioned earl-
ier, which is espoused by ‘scientific’ behaviourists. This merely
maintains that the only evidence on whose basis attitudinal and
other mental states can be ascribed to subjects is behavioural
evidence, that is, publicly available evidence of how people
behave in various circumstances. We could perhaps call this
‘epistemic’ behaviourism, because it concerns our knowledge of
mental states rather than the nature of those states them-

selves. Logical behaviourism entails epistemic behaviourism,
but not vice versa – which is why I have described epistemic
behaviourism as being the weaker doctrine. However, epi-
stemic behaviourism, unlike logical behaviourism, does not
offer us any account of what mental states are, so it needs to be
supplemented by such an account – an account which must, of
course, be consistent with what the epistemic behaviourist has
to say concerning our evidence for the existence of mental states.
Is a satisfactory account of this sort available? Many contem-
porary philosophers of mind would say that it is, in the form of
a doctrine known as functionalism.
FUNCTIONALISM
Functionalism acknowledges the fact that it is impossible to
identify types of mental state with types of behavioural dis-
Mental states 45
position, but it still wishes to characterise mental states by
reference to behaviour, albeit only indirectly. It tries to
achieve this by characterising mental states in terms of the
causal roles they are thought to play in determining how a
subject behaves in different circumstances.
2
(Another term
for ‘causal role’ in this sense is ‘functional role’: hence the
name ‘functionalism’.) According to functionalism, three dif-
ferent kinds of causal relationship can be involved in the
make-up of a mental state’s causal role. (Here, it should be
emphasised, we are primarily talking about types of mental
state, although by implication also about particular
instances, or tokens, of those types: I shall say more about the
type/token distinction in due course.) First, there are charac-

teristic ways in which states of a subject’s environment can
cause that subject to have a certain type of mental state, as
when injury to one’s leg causes one to feel pain in it, or when
light falling upon one’s eyes causes one to experience a visual
sensation. Secondly, there are characteristic ways in which a
certain type of mental state can interact causally with other
mental states of the same subject, as when a feeling of pain
in one’s leg causes one to believe that the leg has been
injured, or when a visual experience causes one to believe
that one is seeing something. And third, there are character-
istic ways in which a certain type of mental state can contrib-
ute causally to the bodily behaviour of its subject, as when a
belief that one’s leg has been injured, together with a desire
to relieve the consequent pain, can cause one to rub the affec-
ted part of the leg or withdraw the leg from harm’s way.
Thus, to a first approximation, the functionalist might say
that the causal role of a feeling of pain in one’s leg is to
signal the occurrence of a physical injury to the leg, cause
one to believe that such an injury has occurred, and thereby
help to bring it about that one acts in such a way as to repair
the damage done and avoid any further injury of a similar
kind.
2
One of the founders of functionalism was Hilary Putnam: see his ‘The Nature of
Mental States’ (1967), reprinted in W.G. Lycan (ed.), Mind and Cognition: A Reader
(Oxford: Blackwell, 1990).
An introduction to the philosophy of mind46
We see, then, that functionalism is committed to regarding
mental states as really existing states of subjects which can
properly be referred to in causal explanations of the overt

behaviour of the persons whose states they are. But it reco-
gnises that how a person behaves in given circumstances can
never be accounted for simply in terms of the fact that he or
she possesses a mental state of this or that type, since an
adequate explanation of behaviour will always need to take
into account the causal interactions which can occur between
different mental states of the same subject. Consequently,
functionalists acknowledge that the behavioural evidence for
ascribing mental states to a subject is always open to many
alternative interpretations and that often the best that we
can hope to do, in the light of a given subject’s circumstances
and pattern of behaviour, is to assign a reasonably high
degree of probability to the hypothesis that that subject pos-
sesses a certain combination of mental states of various types,
on the grounds that such a combination of mental states
would most plausibly explain the subject’s behaviour in these
circumstances. Suppose, for example, that John exhibits the
following behaviour when, on a particular occasion, it starts
to rain as he is walking to work: he unfolds his umbrella and
puts it up. Part of an explanation of this behaviour might be
that John believes that it is raining. However, it will be reason-
able to ascribe this belief to John in partial explanation of
his behaviour only if it is also reasonable to ascribe to John
certain other mental states which are suitably causally
related to that supposed belief of his – for instance, a sensation
of wetness on his skin, a desire to remain dry, a memory that
he has brought his umbrella with him, and a hope that putting
up an umbrella in these circumstances will help to keep him
dry. However, in the light of John’s behaviour in other cir-
cumstances, it might well be the most likely hypothesis in the

present case that his current behaviour is indeed the result of
his possessing this particular combination of mental states.
In many respects, mental states as characterised by func-
tionalism are rather like ‘software’ states of a computer,
which may likewise be characterised in terms of their rela-
Mental states 47
tions to the computer’s ‘inputs’, other software states of the
computer, and the computer’s ‘outputs’. By a ‘software’ state
of a computer, I mean, for instance, its storing of a particular
piece of information – which, on the proposed analogy, may
be likened to a subject’s possession of a certain belief. Such
a software state may be contrasted with a ‘hardware’ state of
the computer, such as an electromagnetic state of certain of
its circuits, which might correspondingly be likened to a
neural state of a person’s brain. The computer’s ‘inputs’ are,
for example, keystrokes on its keyboard, while its ‘outputs’
are, for example, patterns displayed on its video screen: and
these again might be likened, respectively, to stimulations of
a subject’s sensory organs and movements of his or her body.
Indeed, for many functionalists, this is more than just an
analogy, since they think of the human
brain as being, in
effect, a biological computer fashioned through eons of evolu-
tion by processes of natural selection. According to this view,
just as the biological function of the heart is to circulate
blood through the body and thereby keep it oxygenated and
nourished, so the biological function of the brain is to gather
information from the body’s environment, ‘process’ that
information in accordance with certain ‘programmes’ which
have been ‘installed’ in it either by genetic evolution or else

through processes of learning, and finally to use that informa-
tion to guide the body’s movements about its environment.
How seriously or literally we can take this as an account of
what the brain really does is a matter we shall discuss more
fully in chapter 8. But as a model for the functionalist’s con-
ception of mental states the analogy with software states of
a computer is certainly an apt one. Indeed, as a matter of
historical fact, it would appear that the computational model
provided an important source of inspiration for functionalism.
3
3
See again Putnam, ‘The Nature of Mental States’, and also Alan M. Turing, ‘Com-
puting Machinery and Intelligence’, Mind 59 (1950), pp. 433–60, reprinted in
Margaret A. Boden (ed.), The Philosophy of Artificial Intelligence (Oxford: Oxford
University Press, 1990).
An introduction to the philosophy of mind48
FUNCTIONALISM AND PSYCHOPHYSICAL IDENTITY
THEORIES
It is important to appreciate that functionalism is noncom-
mittal on the question of whether mental states can be identi-
fied with physical states of a subject’s brain – and this is often
held to be one of its advantages. Identity theories, in this
context, fall into two classes: those which maintain that every
type of mental state can be identified with some type of phys-
ical state (type–type identity theories) and those which main-
tain only that every token mental state can be identified with
some token physical state (token–token identity theories). The
type/token distinction, which is of perfectly general applica-
tion, can be illustrated by the following example. If it is asked
how many letters I write when I write the word ‘tree’, the

correct answer is either three or four, depending on whether
one is concerned with letter-types or letter-tokens, because
what I write includes one token of each of the letter-types ‘t’
and ‘r’, but two tokens of the letter-type ‘e’. Now, in contem-
porary philosophy of mind, type–type identity theories have
been called into question on the grounds that types of mental
state are, plausibly, ‘multiply realisable’. For instance, it is
urged that it is implausible to suggest that pain of a certain
type is identical with a certain type of neuronal activity,
because it seems conceivable that creatures with very differ-
ent types of neural organisation might nonetheless be cap-
able of experiencing pains of exactly the same type. The com-
puter analogy once again seems apt, for in a similar way it is
possible for two computers to be in the same ‘software’ state
even while possessing very different ‘hardware’ states – for
example, they might be running the same programme but
implementing it on very different computing machinery. But
rejection of any type–type theory of mental and physical
states is consistent with acceptance of a token–token theory,
that is, a theory according to which any token mental state,
such as a pain that I am now feeling in my left big toe, is
identical with some token physical state, such as a certain
state of neuronal activity now going on in my brain. Such a
Mental states 49
combination of views – rejection of type–type identity
together with acceptance of token–token identity – is com-
monly referred to as (a species of) non-reductive physicalism and
is usually advanced in conjunction with the doctrine that
mental state types, while not being identical with physical
state types, do none the less supervene upon physical state

types.
4
What this doctrine asserts, roughly speaking, is that
no two people exemplifying exactly the same physical state
types could differ in respect of the mental state types which
they exemplify, although it is allowed that two people exem-
plifying exactly the same mental state types could differ in
respect of the physical state types which they exemplify (in
line with the thesis of the multiple realisability of mental
state types). More specifically, it may be held that a person’s
mental state types at any given time supervene upon that
person’s brain-state types at that time.
Functionalism is fully consistent with non-reductive phys-
icalism of this kind. But it is not committed to that position,
since it is equally consistent with a type–type identity theory
and even, indeed, with a thoroughgoing dualism of mental
and physical states. This is because functionalism only
attempts to characterise mental states in terms of the pat-
terns of causal relations which they need to bear to other
states, both mental and physical, in order to qualify as
mental states of this or that type: it consequently leaves
entirely open the question of what, if any, intrinsic properties
mental states must have. (An intrinsic property is one which
something has independently of how it is related to other
things: for example, in the case of material objects, shape is an
intrinsic property, whereas position and velocity are relational.)
According to functionalism, the states of an immaterial soul
may qualify as mental states, provided that they exhibit a
4
For an excellent discussion of the notion of supervenience, see John Heil, The

Nature of True Minds (Cambridge: Cambridge University Press, 1992), ch. 3. See
also the important work of Jaegwon Kim in this area, collected in his Supervenience
and Mind: Selected Philosophical Essays (Cambridge: Cambridge University Press,
1993). The topic of supervenience is too complex and large, however, for me to
do it justice in the present book.
An introduction to the philosophy of mind50
suitable pattern of causal relationships, as may the states of
a bundle of human neurones, or the states of a piece of com-
puting machinery, or even the states of a pile of pebbles on
a beach. But this extreme liberality, although initially
attractive, does harbour difficulties for functionalism, as we
shall soon see.
It may be wondered how functionalism could really be con-
sistent with a type–type identity theory of mental and physical
states: if all that is required of a state, in order for it to
qualify as a mental state of a given type, is that it should
exhibit a suitable pattern of causal relationships with other
states, both mental and physical, then it is hard to see how
a certain type of mental state could simply be identical with
a certain type of physical state – for what could rule out the
possibility of another type of physical state having tokens
which exhibited the same pattern of causal relationships?
Here, however, we need to distinguish between two different
possible ways of interpreting functionalism. Functionalism,
we have seen, characterises mental states in terms of their
distinctive ‘causal roles’. Thus, for example, it may be said
that pain states are typically caused by physical damage to a
subject’s body, cause the subject to desire relief, and give rise
to appropriate avoidance behaviour. But here one can ask:
should we identify the type of mental state to which a token

pain state belongs in terms of the causal role which charac-
terises pain, or should we instead identify it in terms of the
type of state which ‘plays’ that role in the person whose pain
state this is? An analogy which may help us to understand
this distinction is the following. Consider the role of the
bishop in chess, as defined by the rules governing its move-
ments. And suppose that we have before us a particular
wooden chessman of a certain shape, which is a bishop. We
might think of identifying the type to which this particular
chessman belongs in terms of that role – in which case we
must allow that other chessmen, made of different materials
and possessing different shapes, may none the less qualify as
chessmen of exactly the same type as this one, provided only
that they are all played with in accordance with the rules