1. Why a Functional Account of Intentionality is not Possible

This claim of Kim's expresses a view which is widely held. The standard materialist view seems to be that the real problem for the physicalist (the ‘hard’ problem) is presented by ‘qualia’ rather than by intentionality itself. For reasons I shall explain, I find this position quite incredible. But first, let us note that Kim's supporting consideration expressed in the second sentence in fact offers no support for a physicalist conception of intentionality at all. It is surely obvious that an interactionist dualist will accept that there cannot be a duplicate world lacking intentionality, since to remove the intentional causes of behaviour will result in a world which cannot be a duplicate of this one. Kim offers another supporting consideration for his view, which runs as follows:

This equally offers no support for a physicalist position, since, once again, it is something with which any interactionist dualist can readily agree. If these creatures really are behaviourally indistinguishable from us, we would ascribe to them much the same sort of inner life as we have ourselves, and which we ascribe to other people on the basis of their behaviour. It is difficult to understand how Kim could have supposed that either of these points offers support for physicalism.

Now let us look at Kim's claim that there are no principled objections to a functional construal of intentionality. The obvious objection to Kim's claim is that, while it is indeed possible to give what might look like a functional construal of intentional states in terms of their typical causal inputs and outputs, those inputs and outputs themselves are irreducibly mental. Indignation, to take one example, might be defined as that state which is brought about by the perception of a wrong or an insult, and leads in turn to the desire to protest. And gratitude might be defined as that state which is brought about by the recognition that one is the recipient of something which one values and which causes in its turn the desire to express one's thanks. And so on. But, of course, these states are irreducibly mental, or so I would claim.

Kim, by contrast, while allowing that no one has yet produced full functional definitions of believing, desiring and intending, and that it is ‘perhaps unlikely that we will have such definitions any time soon’,Footnote ³ sees no problem in the idea that such definitions will eventually be produced. We surely need to ask what such definitions could possibly look like, even in their most abstract form. One obvious obstacle that stands in the way of the possibility of such definitions is that an intentional state such as that of the experience of gratitude or indignation must involve the here-and-now or categorical direction of consciousness to its object: my thought is here and now directed to the benefit I have received and to a way of conveying to the benefactor how much I appreciate this benefit, or to the wrong I have suffered and to the need to make this clear. Such states are necessarily events in consciousness and necessarily involve the categorical, non-dispositional, directedness of consciousness to an object of thought. It is utterly unclear how this essential property of intentional states could be realised in an assembly of physical elements; at any rate, I know of no explanation of how this is possible And it is impossible to tell from Kim's discussion whether, in claiming that full functional definitions of states such as believing and desiring will eventually be found, he supposes that such definitions will show that these intentional states do not, after all, involve occurrent states of consciousness which are here-and-now directed to their objects, or, on the other hand, that this essential property of non-dispositional directedness to an object can be accommodated within the complete physicalist-functionalist story. The former option is, in my view, clearly false, and the second a bewildering claim.

2. Why Intentional Patterns are not in the Physical World

Dennett has argued for a position with which Kim may be in agreement, and examining it will bring the issue more into focus. In a departure from his earlier instrumentalist position, Dennett claims that there is a sense in which we can accept that intentional concepts pick out real patterns in the world. These patterns, however, are not visual patterns, but, one might say, intellectual patterns. To adapt one of his examples slightly, a range of chess games all played to a conclusion will reveal to the novice a completely haphazard collection of physical events. Eventually, however, one might expect that a certain pattern will begin to become evident: something about the way a selection of pieces of one colour is grouped around a particular piece of the other colour – checkmate. Similarly, the Martian peering through a telescope at the Superbowl game might eventually begin to understand what is going on, particularly if he adopts ‘the intentional stance’. And in an earlier paper Dennett considers the Martian (again) looking at a stockbroker placing an order for 500 shares in General Motors, and goes on:

Adopting the intentional stance in these cases, Dennett tells us, hugely increases one's predictive power, and enables one to discern the relevant patterns far more easily. These patterns, at any rate, are real, though perceiving them is not a matter of our being confronted with some easily perceptible visual pattern, but of discerning a high-grade ‘intellectual’ pattern, a pattern which may be difficult to detect through the surrounding ‘noise’. And I suppose that this must also be Kim's view, though, for him, the noise is such as to have prevented us up to now from perceiving the underlying intellectual patterns, or algorithms. To discern this in any particular case is to be able to pick out the relevant paths through the physical world, the paths which constitute the different possible realisations of the intentional state.

The suggestion that ‘the reality of intentional patterns’, to use Dennett's phrase, can be accounted for in this way is deeply flawed. What primarily undermines the suggestion is that all Dennett's examples involve either games (chess, the Superbowl game) or an activity defined by certain rules or conventions (placing an order for shares in General Motors). In these examples, it is true that understanding what is going on is a matter of grasping the underlying pattern or algorithm, a matter of coming to understand the rules which govern the activity in question. And it is true that, while adopting ‘the intentional stance’ will help one to grasp what is going on much more easily (or so we can allow), it is not essential that one adopts this stance. The relevant paths through the physical world might be grasped without assuming intentions on the part of any imagined person.

But grasping that what one is witnessing is an expression of indignation or gratitude cannot be like this. There is no pattern to be discerned, not even the most high-grade or ‘intellectual’, no underlying rule or algorithm, something which might initially be discerned without presupposing any underlying intentional attitude. The only thing common to a range of possible expressions of indignation or gratitude or remorse is that they are all seen by us to be such expressions. Such understanding is not a matter of discerning some pattern which obtains in the physical world, a pattern which we might eventually perceive through the surrounding ‘noise’. It is not an understanding which might be achieved without adopting the intentional stance at all. It is essentially intentional understanding, subjective or first-personal. It is understanding which is achieved initially by bringing the template of one's own conscious experience of what it is to have intentions and emotions to bear. And, to repeat a point made earlier, it is knowledge of intentional states as involving the immediate, non-dispositional directedness of consciousness to its various objects. I take all this to show, contrary to what Kim and others suppose, that the expectation that we will eventually have a functionalist account of intentionality which is compatible with physicalism is totally misconceived.

3. Why Mental Causation Rules out Physicalism

Since intentional concepts do not pick out a pattern of pathways through the physical world, a pattern determined by some underlying algorithm, the question must arise, what is the relation between the intentional state and the physical events which are its expression? On the face of it, it is causal: my gratitude for the receipt of a gift, or my remorse for a wrong I have committed, leads me to act in a certain way. The course of events is determined by the nature or the logic of the intentional state. It seems to me incredible that a complete explanation in intentional terms of a certain course of behaviour might be paralleled by an equally complete explanation of the same course of behaviour couched in the terminology of the physical sciences, and one which involved no recourse to intentional notions. That would be simply a miraculous coincidence.

I take the term ‘miraculous coincidence’ from Adrian Cussins, who has forcefully described the challenge that the physicalist has to answer. However, his attempt to answer this challenge seems to me very puzzling. Here is a relevant passage:

I cannot see that this point does anything to meet the challenge to the physicalist that I outlined above. What is clear, surely, is that the mother's behaviour is determined by her love for her child and her concern for its safety. These are the intentional states which determine her behaviour, and which thus determine the neurophysiological processes which underlie that behaviour. It is not, as Cussins claims, a matter of our physiology meshing with our psychology or our intentional states, but of our intentional states determining the course of our neurophysiological processes.

Imagine a mother who entirely lacks concern for her child, and is quite prepared to use her child as a way of making money. Standing at the edge of the canyon, she remembers that her holiday insurance covers accidental death, and her child then has an unfortunate accident. This mother has the same neurophysiology as the rest of us, but the neurophysiological processes that take place in this case are very different from those that take place in the loving mother, because the intentional states which determine her behaviour are very different.

I am equally puzzled by Galen Strawson's attempt to meet this problem. Here is the relevant passage:

I cannot see how this passage does anything to meet the problem which Strawson outlines. To claim, as he does, that part of the answer is to recognise the ‘qualitative-experiential character’ of consciously entertaining a proposition is to emphasise something which the epiphenomenalist readily accepts. For to insist that consciously entertaining a thought has a qualitative character no more shows that it is not an epiphenomenon than pointing out that pain is a sensation, something having a qualitative character in consciousness, shows that the sensation is not a mere epiphenomenon. Epiphenomenalism readily accepts that such mental events do have a qualitative character, but that does not in any way indicate that they are not mere epiphenomena. Strawson's point does nothing to meet the central difficulty: if there is a complete explanation of human behaviour in terms of the categories available to the physical sciences, then the mental is epiphenomenal.

I have looked only at two responses to the problem of mental causation, and I will confess that, on looking back at much of what has been written, and continues to be written, on this issue, after a while I felt the life-force draining out of me and had to give up. So let me just say that I found nothing which threatened to overturn my view that the supposition that a complete explanation in intentional terms of one's acting from gratitude, jealousy, remorse, etc., can be paralleled by an equally complete explanation couched in the terms of the physical sciences would be to posit something utterly miraculous.

4. Why the Claim that Intentional States are Multiply Realisable Must be Rejected

One further suggestion that needs to be looked is the claim that what I have said would be utterly miraculous, viz. a complete explanation of a stretch of behaviour in intentional terms paralleled by an equally complete explanation of that same stretch of behaviour in physical terms, is not in the least miraculous, since intentional states and operations are realised in the purely physical world. We see this, for example, in the operations of a computer playing chess. Intentional concepts, we can allow, are irreducible to purely physical concepts, but that is only in the sense that the same intentional operation can be realised in different physical set-ups. An intentional activity such as that of playing chess might be undertaken by computers of different designs, for example.

This suggestion cannot stand. The game of chess is governed by certain rules, and, as I said above, grasping the underlying algorithm allows one to see the essential similarity common to a number of games played to a finish. And that is to say that that algorithm, those rules, can be computed. The algorithm determines what possible paths through the physical world are permissible. But the irreducibility of most intentional states is quite different from this. What counts as a possible expression of remorse or gratitude, of jealousy or indignation, is not something determined by an algorithm. To repeat: the only paths through the physical world which count as possible expressions of any of these intentional states are those which, in the first instance, our own experience of emotion enables us to see as such.

Papineau, by contrast, has indeed suggested that intentional states are multiply realisable in much the same way as, for example, the operation of thermostatic heaters is multiply realisable. There are thermostatic heaters of different designs, and that means that there is no single pathway to the end of heating water to a required temperature.Footnote ⁷ The suggested parallel is quite untenable, however. First, the various possible expressions of gratitude or indignation are not different pathways to some physically identifiable end, or an end defined by some set of rules or an algorithm, but to an end which our own subjective, first personal experience allows us to see as an end. Second, talk of intentional states being realised in some pathway through the physical world must be rejected, as I have already claimed. Various pathways through the physical world may indeed be expressions of an intentional state, but their being so is a matter of their being ways of achieving the end to which the thought of the subject is directed in that intrinsic, non-dispositional way which is an essential feature of intentionality, but which cannot be a feature of any aspect of the physical world. Simply to be confronted with a complex pattern of pathways through the physical world, just that, would leave one without any insight at all into the nature of an intentional state such as what it is to express sympathy for someone or what an expression of gratitude might be.

Recognising the fact that intentional states are causes of behaviour, and that the intentional cannot be reduced to the physical, or seen to be realised in sequences of physical events, leads inexorably to the conclusion that the principle of causal closure has to be abandoned. I think this in turn leads to interactionist substance dualism. I shall defend this claim by looking first at a well-known attempt to undermine the very notion of a mental substance, and then at an argument recently re-invoked by Kim which it is claimed shows that the notion of causal interaction between the physical and the non-physical is incoherent.

5. Why Kant's Rejection of the Notion of a Mental Substance is Misplaced

Here is a very well-known footnote from The Critique of Pure Reason:

A considerable number of commentators have endorsed this conclusion, though none more enthusiastically than P. F. Strawson, who writes:

Strawson takes this to be ‘the coup de grâce to Cartesianism’.Footnote ¹⁰ It is no such thing. It rests on a major misinterpretation of Descartes which quite undermines it. To see this, we need first to remind ourselves of Descartes' definition of ‘substance’: ‘By substance we can understand nothing other than a thing which exists in such a way as to depend on no other thing for its existence’.Footnote ¹¹ And it is absolutely clear that for Descartes thought itself is substantival. The idea of thought, that is, is an idea of that which can be understood completely, or as a complete thing. Descartes does indeed acknowledge that there is a distinction between modes of thought on the one hand and the objects which have them, but insists that this distinction is merely a conceptual one. Here is the crucial passage:

You would think that nothing could be clearer than that: the distinction between all the various modes of thought on the one hand and the mind which has them on the other is merely a conceptual one. Thought, then, is not something that inheres in some underlying substratum. Indeed, if that were the case then we could not know of that substratum that it is not also the bearer of corporeal properties, and Descartes would have no argument at all for the real distinction between mind and body, for which he argues in the sixth Meditation.

What Kant appears to have done is to foist on Descartes an ontological distinction between thought on the one hand and the object in which thought inheres, ignoring Descartes' clear assertion that this distinction is merely a conceptual one. Kant's claim that the Cartesian view of mind would allow the possibility of ‘representations and the consciousness of them’ to be passed from substance to substance as motion might be passed from ball to ball is in fact doubly confused. First, the idea of thought is an idea of something which exists in such a way as not to depend on anything else for its existence, and as such cannot be compared to motion, which is clearly not an idea of something which exists in the relevantly independent way, but is something which cannot be understood apart from the idea of the substance of which it is a property – the ball, in Kant's example.Footnote ¹³ Second, Descartes' claim that ‘thought … is nothing else but thinking substance itself’ means that the idea, entertained by Kant, that on the Cartesian view ‘representations and the consciousness of them’ might be passed along from one substance to another is simply incoherent, for the flow of ‘representations and the consciousness of them’ is itself substantival. Footnote ¹⁴

Strawson's attempt to press Kant's point even further by suggesting that acceptance of the Cartesian view would mean having to accept the possibility that each thought of mine might be the thought of a thousand souls simultaneously thinking this thought seems to add absurdity to misconception. No sense whatever can be made of the idea that my thought ‘It's quite a nice day’ might be a thousand such thoughts. In any case, this claim of Strawson's clearly rests on the fundamental misinterpretation of Descartes that I have just emphasised. What remains to be considered, I think, is a concern about the identity of the thinking self through time, and the idea that the Cartesian conception of a thinking substance is so peculiarly vulnerable to doubts about continuing identity as to render the whole idea unusable.

I have two points to make about this suggestion. The first is that it seems to me a mistake to suppose that doubts about the continuing identity through time of the thinking substance (res cogitans) are of an essentially different order from the doubts that might be raised about the continuing identity of any physical object. I can always raise a doubt about whether the object in front of me continues as the same object, or whether it is replaced by an exactly similar object each time I blink. This sort of doubt is certainly ‘hyperbolic’ to use Descartes' word, but it is not logically absurd. Doubts about the continuing identity of the self when one is going through a complex argument, say, or being held by a piece of music, seem to be of the same order. We ought to see that they are indeed of the same order once we free ourselves of the Kantian misconception that for Descartes thoughts inhere in something which is not itself thought, but an unknowable substratum underpinning thoughts. That conception certainly invites the sceptical response expressed by Kant, Strawson and others. There is a very great deal more that I might say about personal identity through time, but I cannot enter more deeply into that issue here.Footnote ¹⁵

The other point that needs to be made is this. I have argued that the relation between the intentional and the physical is causal: intentional mental events are causes of behaviour, and the intentional does not reduce to the physical, nor can it be seen to be realised in the physical. Given this, we must ascribe to the mental the status of substance. To go back to Kant's example, it is certainly true that the motion of one ball might be passed on to the next in some sense, but it is not some theoretically separable property, motion, which has this causal power, the power to cause the next ball to move, but the object-in-a-state-of-motion, and that object is certainly a substance. To ascribe causal power to the mental, and to reject the idea that the mental is a property of the physical, or is reducible to the physical, or is realised in the physical, is to be committed to the view that the mental is substantival. So we have interactionist substance dualism.

6. Why Kim's Rejection of Immaterial Minds is Unsuccessful

It has been a stock objection to Cartesian dualism that the notion of a causal interaction between the material and the immaterial is unintelligible. Yet, as Kim acknowledges, it is very difficult to come up with a conclusive demonstration that the idea of trans-substantival causal transactions is incoherent, and, as he mentions, Descartes actually talks of ‘an association between thoughts and bodily motions or conditions so that when the same conditions recur in the body they impel the soul to the same thought; and conversely when the same thought recurs, it disposes the body to return to the same conditions’.Footnote ¹⁶ What is notable about this passage, as Kim points out, is that Descartes actually posits, not a Humean conjunction between the material body and the immaterial mind, but the operation of causal power across the substantival divide. The challenge to Descartes' critics is, then, to show just why this is incoherent, ‘to put up a real argument or shut up’, as Kim puts it.

Kim claims that he has such an argument.Footnote ¹⁷ ‘Descartes' trouble’, he claims, ‘has nothing to do with the bruteness or primitiveness of causation or whether causation is merely a matter of Humean regularity, and it has everything to do with the supposed nonspatiality of Cartesian minds’. Here is the argument:

Thus the claim that an immaterial, or non-spatial, mind causally interacts with a body in space is incoherent, Kim would have us believe. In fact the argument is quite unsuccessful. What undermines it is the claim that ‘to say that this is the body with which this mind is united is to say [only] that this body is the only material thing that this mind can directly affect’. This claim is simply false. To see this, let us compare Kim's claim with what A. J. Ayer said about this issue in his paper, ‘Privacy’. In that paper, Ayer lists three conditions in virtue of which some particular body is mine. The body that is mine is the body which (a) is under the control of my will in a way in which no other body is, (b) is delineated by my sensations, and (c) provides, as it were, the centre from which I view the world.Footnote ¹⁹ This seems to me exactly right. And of these three conditions, only the first is causal; that my body is the one which is the locus of my sensations, and that it provides the centre from which I view the world, are conditions which do not involve causality in any way. I therefore find no obstacle to the claim that when I will my arm to go up, it is my arm, not simply in virtue of the fact that it is a thing which I can directly affect, something which, in Kim's speculation, would also be true of an arm attached to another body, but in virtue of that arm being a part of the body which is delineated by my sensations. And it is also, crucially, that arm to which my volitional thought is directed.

You would think then that there can be nothing which might prevent Kim's acceptance of causal interaction. For surely the notion of a body's being one's own in virtue of being thus delineated cannot present a problem. But if Kim can accept that, then his fundamental claim that what makes an arm Smith's is just that there is a specially direct causal connection between Smith and that arm must be rejected. My volition is directed to the particular right hand, say, which is mine in virtue of being part of the body delineated by my sensations. What is the problem?

Two points of confusion seem to be evident in Kim's discussion of this issue. First, it is very difficult to make any sense of what Smith's willing to raise a hand could be in the example as presented by Kim. Are we to suppose that Smith's willing is not directed to one particular hand, or what? Even if we can accept Kim's speculation that Smith's willing is always followed by the hands of two different bodies rising, the question remains, which hand did Smith in fact will to rise? Was it both of them, or was his willing a sort of objectless mental operation?

The second problem for Kim is what seems to me to be his extraordinary misconception as to the nature of intentionality. Kim claims that ‘we need causal relations to understand intentionality’. If this were the case, then there might indeed be the problem that Kim supposes that he has highlighted. For in Kim's imagined case there appears to be an equally direct causal connection running from Smith both to Arm A (Smith's arm, as we would want to say) and to Arm B (Jones' arm). But the claim that intentionality rests on causality is simply false. Kim is led to make this claim by generalising from the case of perception. We can allow that what it is for me to perceive a certain tree rather than a qualitatively indistinguishable one hidden behind it is a matter of the causal impact of that tree on my visual experience. But to suppose that what is true of perception is true of intentionality in general is a clear mistake. I may, for example, be looking at two trees, but only of the one on the left do I think that it is blocking out too much light and may well have to come down. That such a thought is directed to the one tree rather than the other is not a matter of causal connection between that tree and my consciousness at all. I may be thinking, not about one of the trees in front of me, but about the Battle of Hastings or the Big Bang, and no reference to causality is required to understand these examples of intentionality. Once this (surely obvious) point is seen, that is, once it is seen that intentionality is not a matter of a causal process running from an object in the world to the subject, but of the directedness of thought from the subject to an object in the world,Footnote ²⁰ then surely there can be no obstacle to accepting that, as I said above, I think volitionally of that right arm which is part of the body delineated by my sensations, and which provides the centre from which I view the world.

7. Why Kim's Rejection of the Possibility of Immaterial States Being Connected to the Body is not Coherent

Kim has another concern, which seems to me quite independent of the particular problem about causality which he discusses. This is the problem of how we can make sense of something which is immaterial and non-spatial being connected to or situated in some particular body in space. If the soul is non-spatial, how can it be located in a particular body, or even be connected with some particular body?

We must ask how seriously Kim can press this difficulty, if such it be, in the light of his own view of the nature of qualia or sensations. Kim admits that it is not possible to give a functionalist or physical reduction for sensations. They are themselves, then, not part of the physical world. But it would be a bold person who claimed that they have no position in space at all. On the contrary, we feel them in various areas of the body. The fact that Kim regards them as epiphenomena, as the effects of processes in the body, does not affect the claim that they themselves are not events or objects in the physical world. The fact that they are caused by locatable events in the body is not enough to secure their immaterial effects such a location. But they are, or seem to be, located in the physical world.Footnote ²¹ Given this, I do not see how Kim can press the objection that if the mental is conceived of as immaterial then there is no sense in which it can be located in space.

Further than this, Kim's claim that sensations such as that of pain, which are clearly locatable in the body, are mere epiphenomena, having no effect on behaviour at all, is wholly unconvincing. Pains and itches, he says, have a motivational/behavioural aspect in addition to their qualitative aspect, and ‘it is clear that the motivational/behavioural aspect of, say, pain, can be given a functional account’.Footnote ²² I think this attempt to sever the motivational aspect of pain from its qualitative aspect is totally incredible. Is it seriously claimed that a world in which the qualitative aspect of pain was totally absent, a world in which no one had ever felt pain, would have been just like this one, a world in which people over the centuries have devised hideously painful ways of putting men and women to death, such as burning at the stake and being hanged, drawn and quartered? What is the current debate on whether torture is ever admissable about, if it is not about whether it can ever be right to inflict such horribly unpleasant experiences on anyone? And in contrast with the case of pain, what can we be doing on Kim's view when we recommend a certain sensory experience – a scent, or a taste, say – as particularly pleasant? How can this be reconciled with the claim that sensations are mere epiphenomena, having no place in the causally determined order of the world? The attempt to hive off qualia as items which have no causal relevance is clearly quite misconceived. Such qualia, then, are, by Kim's own admission, not items in the physical world, but are clearly locatable, or so it seems, and clearly have effects on our behaviour.

I conclude, then, that, however much it may grate against contemporary positions and assumptions, substance dualism remains the most persuasive solution to the mind-body problem.