Kirk.SISP

Strict Implication, Supervenience, and Physicalism

Robert Kirk

In the world as it is, the mental involves nothing beyond the physical. That statement - however it may be elaborated - commits me to minimal physicalism¹. But of course it needs to be made clearer. I think it is best clarified with the aid of a thesis of strict implication: it is impossible that the physical truths about us should have been exactly as they are, and the mental truths different. However, some philosophers think the broad character of the mental-physical relation can be clarified in terms of supervenience. Supervenience has been defined in many ways, but I think its usual varieties are badly designed to capture the commitments of minimal physicalism. Strict implication helps to do the job better. I have discussed some aspects of the Strict Implication thesis elsewhere.² Here I want to show how it makes the core commitments of (non-eliminative) physicalism reasonably clear and explicit, and avoids certain objections which have been raised to formulations in terms of supervenience.

I. Strict Implication and minimal physicalism
Pretend we have an idealized version of today's physics, and let P be a conjunction of all true statements formulable in terms of the vocabulary of that theoretical complex. As well as all truths about the distributions and states of elementary particles throughout space and time, P includes statements of all physical laws. So if we assume the idealized version of physics is itself true, P specifies the whole physical universe past, present, and future. Notice that P does not include statements of any psychophysical laws or true generalisations because psychological words are excluded from the narrow vocabulary of our ideal physics - or so we may assume. Let Q be a conjunction of all actually true statements ascribing mental states to the organisms whose existence is provided for by P. The Strict Implication thesis is:

P strictly implies Q. That is, it is impossible that P should be true and Q false.

It is crucial that the impossibility is of the strongest kind: `absolute', not just nomic. The reason, in a nutshell, is that any (non-eliminative) physicalist will agree that when we are talking about the mental states of actual organisms, we are talking only about the physical. We may not realise that that is what we are doing; and of course we are not always talking in physical terms. But all physicalists share the view that (in fact) there is nothing there to talk about but the physical. (The question of the existence of numbers and other abstract entities can be left on one side for purposes of this discussion - though it is worth noting that their existence too will be strictly implied by P.) So any physicalist ought to agree that there is no possible world at all where P is true and Q is false. Such a world would not differ physically from the actual world, yet it would differ in respect of the mental states of organisms whose existence was provided for by P. In that case there would be more to those mental states than is involved in the physical. And that is incompatible with the assumption that when we are talking about the mental (in the actual world), we are talking only about the physical (I will expand this argument in the next section). However, not all those who call themselves physicalists have recognised that commitment; and that goes especially for certain supervenience theorists: see below.
We need to distinguish two main varieties of non-eliminative physicalism. One is just an up-dated version of traditional materialism: it is a purely ontological thesis. The other is more ambitious, and includes claims about physical explicability. What concerns me here is the first variety - what may be called minimal physicalism. Modern physics has overthrown the ancient conception of the ultimate constituents of matter as solid, impenetrable, and imperishable; but the old motivation remains. The underlying thought is still that what actually exists (abstract entities aside) involves no more than whatever sorts of things are involved in the existence of inanimate things such as sticks and stones. For this approach the Strict Implication thesis is nearly all that is required. By itself it expresses the minimal physicalist thought that the mental facts about human beings and other sentient animals are not only determined by and dependent on the physical facts, but involve nothing beyond the physical.
It does so because the impossibility involved is not merely nomic, or even `metaphysical' in any weak sense, but absolute. The thesis implies that a contradiction would result from asserting P and not-Q. (The contradictions need not be easily detectable: philosophical work is needed on top of scientific work.) That gives the right sense to the idea that the mental facts involve nothing beyond the physical - that there is no more to them than there is to the physical. For the right sense can only be one in which our talk about the mental is just a special way of talking about the physical. If the Strict Implication thesis is true, our true descriptions of mental states are redescriptions of exactly the same things and events as could also be described in the austere vocabulary of physics.^*
Although I reject dualism, I know of no good a priori refutation of it, and think some version of substance dualism might have been true. So I assume there are dualistic possible worlds where non-physical items are involved in mental life. In each of those worlds the thesis which corresponds to the actual Strict Implication thesis is false, at any rate if physicalists are right about our world. (Let the sentence used above to state the Strict Implication thesis be S. The `corresponding' thesis is the one stated by using S to apply to that world, so that `P' and `Q' refer respectively to the physical and mental truths which hold there.) If our world is one where what I am calling minimal physicalism holds, P and Q are such that P does strictly imply Q, and by uttering S one makes a necessarily true statement. If on the other hand the dualists are right about our world, the utterance of S makes a different statement, which is necessarily false.³
The Strict Implication thesis is not sufficient even for minimal physicalism. Two further conditions are necessary. One is that the converse thesis is not true: Q does not strictly imply P. This makes explicit the view that while the mental is fixed by the physical, the physical is not fixed by the mental. It gives one clear sense to the thought that the physical is somehow basic, and blocks any awkward moves by people like George Berkeley. It also reflects the idea, which any physicalist is likely to accept, that there is a large area of reality in no way dependent on the mental. But a further condition is needed in order to rule out gods, angels, thought-rays, and other putatively possible non-physical items. The Strict Implication thesis (if true) certainly rules out dualism with respect to the mental states of human beings and other physical organisms, since it ensures that all truths about our mental states (which are by definition included in Q) are strictly implied by P. But by itself it doesn't rule out non-physical beings with their own non-physical kinds of thinking and feeling, assuming there could be such entities. So minimal physicalism needs a further clause to the effect that nothing exists other than things whose existence is strictly implied by P. (You might wonder what this version of minimal physicalism has to say about how other physical world-specifications than P are related to mental facts. I will deal with that important question in the course of discussing supervenience: see section IV below.)
To summarise, minimal physicalism commits you to the following.
1. P strictly implies Q. (The Strict Implication thesis.)
2. Q does not strictly imply P.
3. The only things that exist are those whose existence is strictly implied by P.

II. More on non-eliminative physicalism
I mentioned a second, more ambitious, variety of physicalism. Minimal physicalism doesn't go far enough for some people. They hold strong views on questions of explanation and existence. Minimal physicalists will probably concede that every event which is explicable at all has some explanation in physical terms. (If the Strict Implication thesis is true, any true explanation of an event in non-physical terms will be strictly implied by a subset of austerely physical truths. Arguably the latter will explain the event; but I will not pursue that complex issue.) But they are not also committed to the view that physical explanations are necessarily the most illuminating. It is entirely consistent with minimal physicalism to hold that the best explanations of events such as economic recessions, for example, are not physical, but economic or psychological. For some stronger varieties of physicalism, however, the best explanation of any event will be in physical terms. So if the only route to such explanations is via the concepts of economics or psychology, physicalists who adopt this approach will maintain that the necessary economic and psychological concepts are analysable in physical terms: that predicates expressing them have logical equivalents in terms of the austere physical vocabulary. In this way they can maintain that physics tells us what there is: that physics has ultimate ontological authority.⁴ I regard it as a virtue of the Strict Implication thesis that it does not commit you to any of these stronger and (as I believe) wrongheaded varieties of physicalism. By helping to make explicit the minimal commitments of any kind of physicalism, it forces those who wish to go further to spell out just what their additional commitments are.
But the Strict Implication thesis goes much too far for some other people. A common objection is that a minimal physicalism based on it overlooks all those `physicalists' who maintain that the link from the physical to the mental need be no more than nomically necessary.⁵ I will now try to make clear that their position is inconsistent - or else that they are physicalists in name only. I assume all non-eliminative physicalists hold at least this: that what actually exists involves nothing beyond the physical, so that when we talk about the mental we are (in a way) thereby talking about the physical, regardless of whether or not we realise it. Suppose, then, you maintain that it is only by nomic necessity that P implies Q: you say that in every nomically possible world, if P then Q. In other words, you hold that what's doing the work of binding the mental to the physical is certain laws. What sort of laws might these be? Are they just the laws of physics (which are included in P, remember)? Or do they perhaps include or entail special psychophysical laws (which are definitely not included in P because they introduce non-physical terms)? It is hard to see how the mental could be regarded as being bound to the physical by any laws other than psychophysical ones: laws on the lines of `Wherever physical state U occurs, mental state V occurs'. So suppose those are the laws which matter. The crucial question is, are those psychophysical laws themselves strictly implied, in my strong sense, by P?
Suppose they are not strictly implied by P. In that case the nomically-based thesis doesn't rule out a possible (that is, a non-contradictory) world w in which P holds but Q does not hold. By definition w contains all the physical items that exist, have existed, or will exist in the actual world. In particular it contains particle-for-particle counterparts of all human beings who will ever have existed. P also ensures that all the physical laws and generalisations which hold here also hold there, although P itself includes no psychophysical laws. However, in some respect the mental life of our counterparts in w is, on the present supposition, different from that of their originals in this world. And the trouble is that in that case the truths in Q are not after all just different ways of describing components or aspects of the very same state of affairs that is captured by P. Instead, the truths in Q – and remember these are truths about the actual world – involve something more than what is purely physical. So interpreted, then, the nomically-based thesis implies that there is more to the mental in the actual world than actual physical reality supplies. Indeed, that thesis could be endorsed by epiphenomenalists. In other words, it commits you to some kind of dualism. So those who maintain that nomic necessity is all that is needed for physicalism must reject the assumption made at the beginning of this paragraph. Like it or not, they are forced to suppose that all psychophysical laws are themselves strictly implied by P in my strong sense.^* But then the claim that it is only by nomic necessity that the mental depends on the physical is false.
I had better add that this argument applies to the contingent property identity theory. Any true statements of psychophysical identity there may be will belong to Q. So if contingent property identity theorists want to avoid commitment to the Strict Implication thesis they must maintain that such statements are not strictly implied by P. But then something other than the purely physical facts about the actual world must play a part in ensuring that the identities hold here. For if the identities involve nothing beyond the physically statable facts, they will hold in any world where P holds - P will specify everything required for those identities - in which case the Strict Implication thesis holds: even though P does not actually include the relevant identity statements. So to deny that psychophysical identities are strictly implied by P is after all to imply that mental properties involve something over and above the purely physical.⁶ Nor can contingent property identity theorists escape this reasoning by emphasising that the properties in question are not analysable in physical terms. I entirely agree. My argument is quite consistent with there being possible worlds where Cartesian dualism reigns.
So nomic necessity doesn't provide an adequate basis for even a minimal physicalism - in spite of a widespread assumption to the contrary. Exponents of that assumption may object that what happens in possible worlds which are not also nomically possible is irrelevant: they are interested only in nomically possible worlds - no doubt only in physically possible worlds.⁷ But that way out is not open to them. Their interest in the actual world compels them to take account of those barely possible worlds too. It is only by taking them into account in the way described in the preceding paragraphs that we can get sufficiently clear about the actual world. What we have seen, in effect, is that physicalism leaves no logical room for a nomic necessity which prevails in the actual world yet is neither straightforwardly physical nor strictly implied by the physical. The point is made vivid by Kripke's image of creation. Having created a universe as specified by P, did God have `further work' to do to fix the details of our mental lives? If so, the purely physical facts stated by P leave something out. Since P is assumed to state all the actual physical truths, anything P leaves out must be non-physical.⁸ -
Even if we knew the Strict Implication thesis was true, that could not be regarded as a solution to the mind-body problem. By itself it merely clarifies what has to be done to produce a solution. We might become convinced it was true by means which fell short of explaining how it could be true. That is an uncomfortable epistemic situation, though common enough. For a long time I was baffled by the problem of how sewing machines worked. In effect I knew that `Sewing machines knot stitches' was strictly implied by P, but didn't know how that could be so until I saw some diagrams. We need the equivalents of such diagrams for the mind-body problem: explanations of what it takes for something to have thoughts and feelings which remove the bafflement we feel even when we think we know the Strict Implication thesis is true. It is pertinent to add that minimal physicalism, defined in terms of the Strict Implication thesis, is entirely compatible with, though it doesn't entail, functionalist accounts of the mental. If some such account works, it will provide the right sort of explanation of how the Strict Implication thesis could be true - an explanation which shows how, given the expected physical facts, the mental facts could not possibly have failed to be as they are.⁹
Explanations of how the Strict Implication thesis could be true must show how a purely physical system could be a subject of thoughts and feelings - without appealing to merely natural or nomic links from the physical to the mental. They must make clear that there is not even a bare possibility that organisms exactly like us in all physical respects should differ from us in mental respects. In that sense they must show how the physical facts could determine not only that such organisms have thoughts and feelings, but that they have precisely the same thoughts and feelings as we have - that my counterpart has exactly this experience of the blue word-processor screen, for example (except that an exact physical replica of the actual world need not be one where Twin-London was London, and so on). It is not always recognised that physicalism must provide explanations satisfying these requirements.
The Strict Implication thesis does not imply that physics provides a conceptual basis for all other true ways of describing the world. To be sure, the totality of physical truths P strictly implies all truths about our possession of concepts. But knowledge of the former doesn't automatically provide a grasp of the latter. For example, a creature could know any number of physical facts about a certain community without thereby acquiring a knowledge of the bearing of those facts on that community's understanding of a language. So getting from P to - Q would at any rate need work. Now, you might at first assume that if truths about concepts are strictly implied by physical truths, knowledge of the physical facts must somehow or other provide the means to acquire mental concepts, so that although getting from P to Q would take a lot of work, it could be done. I would argue that even that assumption is mistaken. It is not just that physics and everyday psychology are two very different ways of talking about the world, devised for very different purposes. More decisive is the consideration that not every conceivable kind of intelligent system is automatically guaranteed capable of acquiring just any sort of concept. Not every intelligent creature with a grasp of the purely physical concepts involved in P would necessarily also be able to acquire the concepts of everyday human psychology. Everyday psychology is accessible only to those whose nature, capacities, tendencies, and interests are sufficiently like those of human beings. It seems clear that no amount of knowledge of physical truths could by itself endow the knower with that nature or those capacities or interests. To take one example, it couldn't equip anyone with colour vision. If, as many believe, a full grasp of human colour concepts depends on the ability to see colours as we do, that consideration alone shows that a knowledge of P would not necessarily enable the knower to acquire the concepts needed for knowing Q. (Arguably it would not necessarily supply an alien being even with the means to detect which psychological sentences were held true, hence which were true.) If that is right, physics does not provide a conceptual basis for all other true ways of describing the world. So the Strict Implication thesis is not objectionable on the ground that it forces us to deny a reasonable amount of autonomy to psychology and other sciences.¹⁰

III. Troubles with supervenience
Perhaps the Strict Implication thesis conveys what some people have meant by the supervenience of the mental. But usually supervenience is taken to be a significantly different relation. The rough idea is that there can be no mental difference without a physical difference. There have been numerous attempts to make this idea more precise.¹¹ But from the restricted point of view of an interest in arriving at a reasonably clear statement of minimal physicalism, I find them less than satisfactory. I want now to show how the Strict Implication thesis helps to bring out the core commitments of physicalism in a way that tends to be obscured by formulations in terms of the usual varieties of supervenience.
Jaegwon Kim has usefully clarified some different versions of supervenience, although he scrupulously points out their disadvantages.¹² One is defined as follows, where A is a set of properties supposed to supervene on a set of properties B.

Weak supervenience: Necessarily, for any object x and any property F in A, if x has F, then there exists a property G in B such that x has G, and if any y has G, it has F.

That is widely agreed to be unsatisfactory as a basis for stating physicalism. There is more than one reason; but what is decisive for present purposes is that it is consistent with epiphenomenalism (and parallelism) about the actual world. For it requires only that there be some physical property correlated with the mental one, not that the latter involve only the physical. In contrast, my kind of minimal physicalism rules out both those dualistic doctrines. Statement (3) of minimal physicalism explicitly denies the existence of anything not strictly implied by P, and that includes epiphenomena. But also, on the assumption that conscious experiences are factors in the causation of some physical events, truths to that effect are strictly implied by P. Let us move on to a more promising form of supervenience.

Strong supervenience: Necessarily, for any object x and any property F in A, if x has F, then there exists a property G in B such that x has G, and necessarily if any y has G, it has F.

We get different theses from this schema depending on how we read each of the two modal operators. (We get even more if we distinguish on the basis of whether the domains of the variables include only actual individuals or possible ones as well; but we need not pursue that thought.) Since what concerns us is supervenience as a possible basis for formulations of physicalism, we need consider only two varieties of necessity: `absolute' in the sense discussed earlier, and physical. That yields four theses.
Consider first the two theses where the initial `necessarily' is read in the absolute sense. On that reading strong supervenience entails that Cartesian dualism could not have been true of any of us. That entailment holds because of the crucial existential assumption embodied in strong supervenience: there absolutely must be some physical property underlying every mental property. Now that assumption may indeed be correct. However, I know of no good argument for it. For that reason (among others) those two theses seem unsatisfactory as bases for a minimal physicalism.¹³
In the remaining two theses the necessity invoked by the initial modal operator is construed as physical. Consider first the one where the second modal operator invokes physical necessity too. I think this version of supervenience misrepresents what even a minimal physicalism commits you to. It implies that what binds the mental to the physical is physical ties. But not even a minimal physicalism can consistently make that claim. Any kind of physicalism whatever is committed to the view that it is absolutely impossible that the physical facts should be as they are, and the mental facts other than they are. (At any rate that must hold for those individuals whose existence is guaranteed by the physical facts: they may be contrasted with putative non-physical individuals such as angels.) As I tried to show earlier, the commitment to this view springs directly from the simple thought that in talking about the (actual) mental we are talking about the physical. So no physicalism can imply that the mental is linked to the physical by merely physical ties. (Compare the statement, `By physical necessity, any box containing just three apples contains an odd number of apples'. Here, as in the case discussed, physical necessity has nothing to do with the matter.) I conclude that the second necessity operator has to invoke `absolute' necessity.
That leaves the case where the first operator invokes physical necessity and the second `absolute' necessity. Paraphrasing, `By physical necessity, whatever has a mental property has a physical property such that, by absolute necessity, whatever has that physical property has that mental property'. Now, you might at first think the first part is too strong. After all, physics says nothing explicitly about mental properties. However, on the reasonable assumption that the laws of physics imply that the physical world is closed under physical causation, the physical facts do ensure that anything with a mental property has a physical property satisfying the stated condition - simply because causal closure under physics rules out both interactive dualistic beings and non-physical beings which act on the physical world. So apparently this version of strong supervenience fills the minimal physicalist bill. (You could still object that causal closure under physical laws doesn't rule out non-physical beings which do not interact with the physical world; but perhaps that is not a serious objection.)
Let us see how this particular version of the strong supervenience thesis compares with minimal physicalism based on the Strict Implication thesis. First, consider the effect of omitting the initial occurrence of `necessarily'. If we confine the domain of the variables to actual individuals, the resulting thesis, paraphrased, is that all actual occurrences of mental properties involve nothing more than physical properties. And that seems equivalent to minimal physicalism.¹⁴ So what is the point of adding the initial operator? The obvious answer is that it provides for different physical realisations of the (actual) mental properties of (actual) individuals. Pretty well all physicalists are likely to agree that the same mental state is capable of being realised in more than one physical state. By the device of inserting the (physical) necessity operator at this point, those possibilities are taken care of.
That device certainly works, which means that this version of supervenience also works. However, I think it leaves the situation less clear than the conjunction of theses (1) (the Strict Implication thesis), (2), and (3). It is not just that exponents of the supervenience thesis risk confusion by assigning two different kinds of necessity to the modal operator on its two occurrences.¹⁵ The real trouble, I suggest, is that it obscures the crucial point which is conveyed by the Strict Implication thesis: that the actual physical facts strictly imply the mental facts. We have seen that omission of the initial modal operator leaves a thesis essentially equivalent to the Strict Implication thesis. So I am making a virtue of the absence of necessity.

IV. What about other worlds?
You may now want to press an objection I put off earlier. We can usefully consider it in connection with the third main variety of supervenience discussed by Kim: `global'. The psychophysical version of the global supervenience thesis is

Worlds that are indiscernible in all physical respects are indiscernible in mental respects; in fact, physically indiscernible worlds are one and the same world.¹⁶

Kim thinks global supervenience fails to yield `an appropriate relation of dependency between the mental and the physical, one that is strong enough to qualify it as a physicalism' [9, p. 277]. He objects that it does nothing to rule out pairs of worlds which, though different in some physical respects, are indiscernible in all the physical respects one might suppose relevant to their mental properties, yet still differ mentally. For example, he asserts that global supervenience doesn't rule out a world differing physically from the actual world only in that Saturn's rings contain one more ammonia molecule, yet `which is entirely devoid of consciousness, or has a radically different, perhaps totally irregular, distribution of mental characteristics over its inhabitants'. Nor, he claims, does it rule out there being in the actual world `two physically indistinguishable organisms with radically different psychological attributes'.¹⁷ You may also suspect that a physicalism based on the Strict Implication thesis is vulnerable to the same objection. For the thesis specifies strict implication only by the (presumed) actual physical facts, and says nothing explicitly about possible worlds where the physical facts are different - even if only slightly - from the actual ones. I think both objections are fundamentally mistaken.
Certainly, neither global supervenience nor Strict Implication explicitly rules out the possibilities envisaged. But they don't explicitly rule out the possibility of square circles either; and we don't hold that against them. Neither the Strict Implication theorist nor the global supervenience theorist can properly be reproached for not spelling out every necessary truth, or making explicit every necessary relation. And the key consideration, as I will try to show, is that the global supervenience theorist is committed to maintain that what Kim assumes is possible is in fact absolutely impossible. Similarly the physicalist is committed to maintain that the possibilities which may seem to be left open by the Strict Implication thesis are not genuine either.
We can easily see that the global supervenience theorist is committed to maintain that Kim's assumed possibilities are not genuine. For suppose, if possible, that the physical truth about a certain world w1 is less than sufficient to strictly imply the mental truth about w1. In that case there is another possible world w2 which shares exactly the same physical properties as w1 but differs in its mental properties, and the global supervenience thesis is false for that reason alone. The case would be a counter-example to the claim that any pair of possible worlds indiscernible in their physical properties (which must bring with them conformity to all physical laws) are also indiscernible in their mental properties. So the global supervenience theorist is committed to the view that in every possible world the thesis which corresponds to the Strict Implication thesis holds for that possible world (and not just for the actual world). If that is right, the global theorist can simply deny that what Kim assumes to be logical possibilities are genuine. And here the global theorist is joined by the Strict Implication theorist.
For if the physical truth about the actual world strictly implies the mental truth, that must be on account of physical truths about organisms with thoughts and feelings, not on account of absolutely all physical truths whatever. At any rate no one who accepts the Strict Implication thesis can be forced to go so far. (That is so even on strong externalist assumptions about the contents of mental states: the ammonia molecule example seems intended to be one where no one has any relevant propositional attitudes about it.) Exponents of the Strict Implication thesis can maintain that there are necessary and sufficient conditions for having thoughts and feelings which are satisfied only by certain kinds of system; and that these conditions are satisfied by a whole range of possible systems, not just those actual physical systems which have mental states. Arguably, necessary and sufficient conditions of the required sort can be formulated at a rather high level of generality, and do not depend on the fine details of, say, particle physics.¹⁸ Physical differences in the world outside the right kinds of organisms will not affect the nature or distribution of mental states (or only to the extent that they bear on the contents of mental states). And if I am right about the relatively high level of generality of the necessary and sufficient conditions for thoughts and feelings, considerable physical differences in the organisms themselves will not affect their mental states either. So although neither the Strict Implication thesis nor the global supervenience thesis explicitly rules out the possibilities Kim mentions, the right line for the Strict Implication theorist to take is that they are not genuine possibilities at all. Showing they are not genuine is of course not a trivial task; but actually carrying out that task is not itself part of stating what physicalism is committed to.¹⁹ Contrary to Kim's assumption, then, neither the Strict Implication thesis nor the global supervenience thesis can be objected to on the ground that they don't rule out those alleged possibilities explicitly. I conclude that the main versions of supervenience have no discernible advantages over strict implication as a basis for defining minimal physicalism. And supervenience in general suffers from the additional disadvantage that it comes in so many varieties that it is hard to - use without confusion.²⁰

V. Strict implication and reduction
Defining minimal physicalism in terms of strict implication has a substantial advantage not yet mentioned: it carries no commitment to any stronger sort of reduction. Now Kim has attacked nonreductive physicalism in general, arguing that `a physicalist has only two genuine options, eliminativism and reductionism' [9, p. 267]. On the contrary, I think minimal physicalism based on the Strict Implication thesis is an attractive third option. Detailed discussion of this complicated issue would be out of place here. But it is appropriate briefly to point out two ways in which physicalism based on strict implication seems to escape certain kinds of objection, notably those advanced by Kim.
First, Kim raises the question of why it should be the case that, as global supervenience has it, `physical truths determine all the truths'. He rightly comments that this is a legitimate question to raise, and says, `as far as I can see the only answer, other than the response that it is a brute, unexplainable metaphysical fact, is in terms of local correlations and dependencies between specific mental and physical properties'. But that, he urges, involves psychophysical laws, which `raise the spectre of unwanted physical reductionism' [9, p. 278]. I will not discuss whether this latter claim is correct. The point is that he has apparently overlooked a quite different kind of answer to the question raised: a philosophical one. For if the Strict Implication thesis holds, one way to explain why it holds is by means of philosophical explanations of how a purely physical organism can be a subject of mental states. Such explanations must show how certain (presumably very general) conditions are at the same time logically sufficient for having mental states, and satisfiable by a purely physical system. It is not necessary (failing reasons which Kim, for one, does not offer) for the explanations to rule out absolutely the possibility that the same conditions should have been satisfied by a system with some non-physical features. So the explanations would not imply that there were laws, or even true biconditionals, connecting mental states essentially with physical ones. If that is right (and of course there is much more to be said), one of Kim's reasons for claiming that espousing supervenience leads to reductionism is undermined - at any rate it leaves minimal physicalism untouched.
The second point is this. Kim offers a general argument for the conclusion that `If nonreductive physicalists accept the causal closure of the physical domain, . . . they have no visible way of accounting for the possibility of psychophysical causation' [9, p. 284]. They cannot adopt the otherwise attractive option of identifying mental properties with physical ones, he maintains, since `These property identities would serve as bridge laws par excellence, enabling a derivational reduction of psychology to physical theory'. On the other hand, global supervenience doesn't enable us to speak of the `supervenience of specific mental properties on specific physical properties, since it only refers to indiscernibility holding for worlds'. So it is unable to accommodate his preferred solution, which is in terms of causal relations supervening on micro-causal physical processes. Minimal physicalism based on the Strict Implication thesis offers a promising way out. It implies that when we talk about the mental we are in fact talking only about the physical. The philosophical explanations which (we hope) will show how this is so can be expected to show in particular how in talking about actual mental properties we are in fact talking about physical ones, without this entailing that mental properties in general (in all possible worlds) are identical with physical ones. All it commits us to is that actual mental properties are realised or constituted by physical ones. In this way minimal physicalism does not commit us to there being laws correlating psychological properties with physical ones.
There is much more to be said about both the points I have just noted. But perhaps they help to show how minimal physicalism based on the Strict Implication thesis provides opportunities that have not yet been fully recognised.

VI. Conclusion
Defining minimal physicalism in terms of strict implication has four advantages over defining it in terms of supervenience. First, the Strict Implication thesis is reasonably clear, straightforward, and readily intelligible, while definitions of supervenience are confusingly diverse. Second, it enables us to spell out the commitments of minimal physicalism perspicuously, without confusing it with stronger varieties. Third, although it by no means provides a solution to the mind-body problem, it does make clear what has to be done by any solution: explain how it is that the Strict Implication thesis holds. And fourth, it retains an advantage which has induced people to resort to supervenience: it involves no commitment to any strong variety of reductionism.²¹

FOOTNOTES

REFERENCES

1. Churchland, P. M., Scientific Realism and the Plasticity of Mind (Cambridge: Cambridge University Press, 1979).

2. Churchland, P. M., `Eliminative Materialism and the Propositional Attitudes', The Journal of Philosophy 78 (1981) pp. 67-90.

3. Crane, T., 1991: `All God has to Do', Analysis 51 (1991) pp. 235-244.

4. Davidson, D., `Mental Events', in Experience and Theory, ed. L. Foster and J. Swanson (London: Duckworth, 1970), pp. 79-101, reprinted in his Essays on Actions and Events (Clarendon Press: Oxford 1980) pp. 207-25.

5. Horgan, T., `From Supervenience to Superdupervenience', Mind 102 (1993) pp. 555-86.

6. Kim, J., `Concepts of Supervenience', Philosophy and Phenomenological Research 45 (1984) pp. 153-176, reprinted in his [8], pp. 53-78, to which page references refer.

7. Kim, J., `"Strong" and "Global" Supervenience Revisited', Philosophy and Phenomenological Research 48 (1987) pp. 315-326, reprinted in his [8], pp. 79-91, to which page references refer.

8. Kim, J., Supervenience and Mind (Cambridge: Cambridge University Press, 1993).

9. Kim, J., `The Myth of Nonreductive Materialism', in his [8], pp. 265-84.

10. Kirk, R., `Zombies v. Materialists', Proceedings of the Aristotelian Society, supp. vol. 48 (1974) pp. 135-52.

11. Kirk, R., `From Physical Explicability to Full-Blooded Materialism', Philosophical Quarterly 29 (1979) pp. 229-37;

12. Kirk, R., `Consciousness and Concepts', Aristotelian Society Proceedings supp. vol. 66 (1992) pp. 23-40.

13. Kirk, R., Raw Feeling: a philosophical account of the essence of consciousness, (Oxford: Clarendon Press, 1994).

14. Kripke, S., Naming and Necessity, (Oxford: Blackwell, 1972).

15. Lewis, D., `Radical Interpretation', Synthese 27 (1974) pp. 331-344.

16. Lewis, D., `New Work for a Theory of Universals', Australasian Journal of Philosophy 61 (1983) pp. 343-377.

17. Mellor, H., `Supervenience? No Chance! Reply to Menuge', Analysis 53 (1994) pp. 236-39.

18. Poland, J., Physicalism: the Philosophical Foundations, (Oxford: Clarendon Press, 1994).

19. Quine, W. V., Word and Object (Cambridge, Mass., New York and London: MIT and Wiley, 1960).

^1. David Lewis introduced the phrase `minimal materialism' for essentially the same view. See his [15, p. 334], also his [16] and footnote 18 below.

^2. See Kirk [13, pp. 71-86]. Of course eliminativists maintain that what I have just called `the job' doesn't have to be done at all, since the mental side of the mental-physical relation is empty. In what follows I will ignore eliminativism: see [13, pp. 66-70], for some arguments against it. (This paper and [13] together correct errors in my earlier attempts to clarify the relation in [10] and [11].)

3footnote^*.Since physicalism is essentially linked to physics, the issue of what to count as `the physical' is largely irrelevant for our purposes. We are interested in today's physicalism, which can perfectly well be stated by reference to today's physics. See Poland [18] for some excellent proposals for defining the physical: his whole approach is congenial to the one adopted here, although he does not use the notion of strict implication.

^4.That ought not to cause any worries. Let A be the set of all statements in this article whose first letter is `A', and B the set whose first letter is `B'. Suppose I say that A strictly implies B. Then although the truth value of my statement depends on contingent facts about what statements there are in this article, the statement is perfectly well-formed, and has a determinate truth value.

^5. Cf. for example Quine [19, especially p. 221]: `If we are limning the true and ultimate structure of reality, the canonical scheme for us is the scheme that knows no quotation but direct quotation and no propositional attitudes but only the physical constitution and behaviour of organisms'. Cf. also Paul Churchland's [1] and [2].

^6. For example Davidson [4]. Cf. Kim [6, p. 66].

7footnote^*. By maintaining that any true psychophysical laws are strictly implied by P, physicalists can sidestep an argument of Tim Crane [3]. His argument depends on the claim that `if psychophysical laws exist, the physical alone will not determine everything' (p. 239). Certainly P does not determine everything, which is why minimal physicalism needs statement (3) above. But physicalists can happily accept that not only all psychological truths, but all true psychophysical laws too, are strictly implied by P.

^8. Some contingent identity theorists may be content to accept that these mental properties are non-physical in the relevant sense (that is, it is not just that the predicates ascribing them don't belong to the narrow physical vocabulary, but that in talking about those properties we are not talking, even indirectly, about the physical). But I don't see how such a doctrine can count as a form of physicalism.

^9. Cf. T. Horgan: `Supervenience theses of interest to materialists ... seem more plausibly construed as involving all physically possible worlds' [5, p. 566].

^10.As he says: `Materialism, I think, must hold that a physical description of the world is complete description of it, that any mental facts are `ontologically dependent' on physical facts in the straightforward sense of following from them by necessity' [14, pp. 154, 155].

^11.Hence the Strict Implication thesis does not require mental properties to be analysable in physical terms. Of course it requires the physical facts to include features in virtue of which the mental descriptions apply; but that is consistent with the possibility of non-physical things having such features. For discussion see [13, pp. 82-6].

12footnote^*.That claim is subsidiary to my main aim in this paper. For further discussion of the matters touched on in these two paragraphs see Kirk [13, pp. 82-6, 216-32].

^13. See Horgan [5] for many examples. Superficially the `rough idea' of supervenience is like my initial rough statement of the Strict Implication thesis: that it is impossible that the physical truths about us should have been exactly as they are, and the mental truths different. But there are significant differences: see below.

^14.The definitions which follow are taken from Kim [7, p. 80].

^15. The Strict Implication thesis does not rule out dualistic possible worlds. It says only that the actual world is such that the physical facts about actual individuals strictly imply the mental ones; it is neutral about whether the physical facts about other possible worlds also strictly imply the mental ones.

^16. I assume properties are in a one-many correspondence to predicates, including many-placed predicates. So the Strict Implication thesis and statement (3) in section I (that the only things that exist are those whose existence is strictly implied by P), jointly entail that for any object x and any mental property F, if x has F, then there is at least one physical property G such that x has G, and necessarily, if anything has G, it has F. For one such property is the complex relational one of being the individual occupying such-and-such a space-time location in the universe specified by P. The Strict Implication thesis guarantees that anything with that property has F.

17footnote^*.Confusion continues to be caused by disagreements over the brand of necessity invoked. The point emerges clearly in Mellor [17]. See also Kim: `For psychophysical supervenience it is possible to interpret the first occurrence as metaphysical necessity and the second as nomological necessity; it is also possible to interpret both as metaphysical, or both as nomological. In the case of mereological supervenience the most plausible construal may be that the first occurrence signifies metaphysical necessity and the second nomological or physical necessity.' [6, p. 66.] (The one combination he omits to mention in this passage is the one I have argued is best for expressing a minimal physicalism.)

^18. Kim [9, p. 276]. Cf. his [7, pp. 82-91]. Note that global supervenience is too strong for my purposes because, unlike minimal physicalism based on the Strict Implication thesis, it rules out Cartesian possible worlds. A subtler version, which avoids that objection, is David Lewis's: `M5. Among worlds where no natural properties alien to our world [i.e. not instantiated in our world] are instantiated, no two differ without differing physically' ([16], p. 364)'. In spite of considerable surface differences, the net effect of this definition is close to that of my clauses (1)-(3) in Section I above. The condition that there be no `alien' properties has essentially the same effect as clause (3) (that the only things that exist are those whose existence is strictly implied by P). We can take it that `natural' properties bring laws with them (Lewis says `the discovery of natural properties is inseparable from the discovery of laws', [16], p. 365.); which has the same effect as my provision that P includes statements of physical laws. Although my definition is less elegant than Lewis's, it avoids the notion of `natural' properties and is arguably more perspicuous.

^19. These quotations are from Kim [7, pp. 85, 86]. Cf. his [9, pp. 277f.].

^20.For example, given that mental life is constituted by a system of interacting processes (on which see [13, pp. 93f.]) we may conclude that any purely physical realisation of a mind must have components that are within causal reach of one another. And given that the system must have a `basic package' of capacities (op. cit., 106-173) there will be some necessary conditions at least strong enough to rule out the relevance of ammonia molecules in Saturn's rings.

^21. For attempts at this task, see Kirk [12] and [13, pp. 106-20].