Nonreductive Physicalism and Strict Implication


Robert Kirk





4,698 words




97 Westhorpe
Southwell
Notts NG25 0NE
England
Tel: 01636 813436

[University address:
Department of Philosophy
University of Nottingham
Nottingham NG7 2RD
England]

Robert.Kirk@nott.ac.uk



Nonreductive Physicalism and Strict Implication

Robert Kirk


Do physicalists have to be reductionists? I have argued that a strong kind of physicalism based on the ‘Strict Implication thesis’ can consistently reject both eliminativism and reductionism (in any nontrivial sense: see II below) [8, 9, 10]. Andrew Melnyk disagrees. He faces me with this dilemma: either my formulation doesn’t actually entail physicalism, or it must be interpreted in such a way that the mental is after all reducible to the physical [12]. His alternatives depend on two interesting assumptions, each worth discussing for its own sake. One is about the metaphysical implications of my formulation of physicalism, the other concerns what it takes to establish the strict implication thesis. Each assumption turns out to be mistaken. The outcome, I think, is that this kind of nonreductive physicalism becomes both clearer and more clearly defensible.


I. The Strict Implication thesis


The Strict Implication thesis is stated on the assumption that we have an idealized version of today’s physics, including a set of laws and a special vocabulary. Let P be a conjunction of all truths statable in terms of that theory, including statements of physical laws together with all truths about the distributions and states of elementary particles, or whatever the fundamental items may be, throughout space and time. So P describes the entire physical universe (assuming our idealized physics is successful). By definition the special physical vocabulary does not include psychological expressions, so P does not include any statements of any psychophysical laws or true generalizations. Now consider whatever truths in psychological language there may be (scientific and everyday, particular and general), and let Q be a conjunction of all such actual truths.
1 The Strict Implication thesis is:

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

It is crucial that the impossibility involved here is the strongest kind there is: absolute. If you suspect there is something not quite as strong yet still not merely natural or nomic – ‘metaphysical’ necessity, perhaps – it is not what is in question here.
2 Note also three further points. First, the Strict Implication thesis alone is not supposed to be sufficient for physicalism. It has to be supplemented by two other theses:

(2) Q does not strictly imply P;

(3) Nothing exists other than what is strictly implied by P.

(2) and (3) combine with (1) to ensure that it is the physical realities which determine the mental, not the other way round. (3) is needed to rule out any conceivable nonphysical items – ghosts, thought-rays and so on, on top of the purely physical world.
3 (More on this later.)
Second, although theses (1)-(3) are individually necessary and jointly sufficient for minimal physicalism, they cannot be regarded as providing a solution to the mind-body problem. They leave it unexplained how the Strict Implication thesis can be true, assuming it is. I see that as a virtue of the formulation (1)-(3). It helps to make clear what philosophical work still has to be done.
Third, theses (1)-(3) are not intended to say all there is to be said by way of explanation of minimal physicalism. They are not even intended to express directly the basic idea of physicalism. The basic idea is probably best conveyed by statements like ‘Nothing exists but the physical’, or ‘Mental language is just a special way of talking about certain parts and features of the physical’. Given that basic idea, (1)-(3) are intended to make reasonably clear what minimal physicalism is committed to. I will argue that theses (1)-(3) entail the basic idea because they entail that the contingent psychological descriptions in Q are redescriptions of certain parts and features of what’s described by P, and apply in virtue of the latter (section III).


II. Reduction


Kim argues that ‘a physicalist has only two genuine options, eliminativism and reductionism’ [6, p. 267]. But ‘reduction’ is a slippery word. Some philosophers – David Chalmers for example – seem to use it in such a way that any position that I would count as even minimally physicalist would be ipso facto reductionist.
4 That would trivialize Kim’s claim and make my position untenable from the start. Obviously I am not maintaining that physicalists can avoid being reductionists in that minimal sense. Kim is trying to show that physicalists are committed to a variety of reductionism that is problematic for them (indeed embarrassing: something that warrants the characterization of the reductionist as one over whom there hangs ‘an aura of doctrinaire naiveté’ [6, p. 266]). Accordingly, in [10] I attempted to show how noneliminative physicalists could avoid any kind of reductionism which required the terms or explanations of the reduced science to be correlated with those of the reducing science. Both Kim’s and Melnyk’s arguments seem to presuppose that what is in question is that ‘classical’ type of reductionism. Later I will try to make clear that there are no reasons to suppose (1)-(3) commit one to other nontrivial kinds of reductionism either (section IV).
But I am not here concerned to attack reductionism. My line is that no good reasons have been offered why physicalists should be forced to accept it in any nontrivial sense, while there are some good reasons to reject it (see [10] and section IV below).
It may seem misleading to describe the physicalism represented by theses (1)-(3) as ‘nonreductive’, since the expression ‘nonreductive physicalism’ is sometimes regarded as equivalent to ‘property dualism’.
5 But I think theses (1)-(3) provide a neat way to distinguish between genuine physicalism and property dualism. If the argument in the next section succeeds, then if you accept those theses your position can properly be described as full-bloodedly physicalist – even if you reject nontrivial kinds of reduction. If on the other hand you reject (1)-(3) – in fact, if you reject only the Strict Implication thesis – you are committed to dualism or some more extreme anti-materialist position, since you imply that psychological properties involve something over and above the physical – you imply that P ‘leaves something out’.6 If I am right, theses (1)-(3) offer a genuinely physicalist alternative to property dualism.


III. Why theses (1)-(3) entail physicalism


Melnyk argues that if I reject reductionism I am forced to construe (1)-(3) in such a way that they do not entail physicalism. That is the first horn of his dilemma. In his attempt to impale me on it he points out that there would be no problem with an absolutely necessary link from P to Q in (1) if it depended on either syntactic entailment or bridge laws, but notes that both such explanations would entail reduction, contrary to my claim. The only other possibility he can think of is that the relevant notion of necessity is ‘primitive and unanalysable’. His key thought is that we could construe the strict implication relation as ‘the holding of some primitive and unanalysable necessary connection between properties of two otherwise quite distinct kinds’ ([12], p. 324). He thinks it follows that on that interpretation theses (1)-(3) do not entail physicalism because he assumes that:
(M1) . . . what makes ascriptions of mental properties true is simply the distribution of mental properties and those alone; on the current interpretation, nothing in the conjunction of (1) through (3) guarantees that the distribution of physical properties is in any sense doing so ([12], p. 325).
That is the first of the two assumptions I mentioned at the start. Melnyk attempts to reinforce the appeal of this assumption (M1) by remarking that ‘There seems no logical difficulty in the idea of the weirdest, most intuitively non-physical spook-stuff you like still being absolutely necessitated, in the sense of the current interpretation, by the physical’ (pp. 325f.). That remark might have seemed justified if my position had been a variety of property dualism. However, even in that case I don’t see how the necessity could have been absolute. Surely the dualist cannot claim more than that mental properties are nomically linked to physical ones. Since, in contrast, the necessity of the link from P to Q is absolute, (1)-(3) entail that it is not even logically possible for the properties specified by P to be present and those specified by Q to be absent. It is not as if any arbitrary pair of properties could consistently be claimed to be linked by absolute necessity.
Against assumption (M1) I will argue that (1)-(3) entail:

(K) The contingent truths in Q are redescriptions of certain parts and features of what’s described by P, and hold in virtue of the latter.

(K) means, I take it, that (1)-(3) entail that the contingent psychological truths in Q (in contrast to any logical or conceptual truths there may be) are not just different ways of talking about some of the realities (things, events, properties, whatever) described by P, but made true by the latter. If that is right, the necessity involved is far from being blankly ‘mysterious and unanalysable’. It is just the familiar sort involved in any case where what is described in one way also qualifies – purely because it fits that first description – for being described in another way. Consider for example a small piece of quartzite, worn smooth in a river-bed. P includes its own descriptions of this piece of rock, together with accounts of the processes that resulted in its coming into existence and rolling around in the river; but P’s narrow vocabulary does not include the word ‘pebble’ itself. Obviously, though, the facts that P states in its own terms make the description ‘pebble’ true in this case. The necessity by which that description applies, given P, is surely not mysterious; and if what follows is right, theses (1)-(3) ensure it is by the same necessity that Q holds, given P.
What Melnyk needs to support his assumption (M1) is cases of mental property-instances that are absolutely necessitated by physical ones (as follows from theses (1) and (2)), while their descriptions in Q are not just redescriptions of some of the physical realities described by P. That is, he needs the following two conditions to be capable of being satisfied simultaneously:

(A) The realities (things, properties, whatever) described by P – the P-realities – absolutely necessitate those described by Q;

(B) The P-realities do not make Q true.

(Recall that the P-realities make up a purely physical universe. If they alone sufficed to make Q true, then given (2) and (3), physicalism would be true.)
But it is impossible for (A) and (B) to be satisfied simultaneously. Since Q includes contingent statements about creatures’ mental states, something must make it true – the ‘Q-realities’. But (B) entails that those Q-realities include something additional to the purely physical world described by P. We could put this by saying that (B) entails that at least some of what Q describes is extra stuff – which will of course be nonphysical (‘spook-stuff’). The crucial point now is this. If the Q-realities include nonphysical extra stuff over and above the P-realities, then P alone, which is in purely physical terms, cannot logically imply either what that extra stuff is, or even that it exists at all. A purely physical description of the world cannot by itself settle whether there is anything nonphysical. Hence (B) entails that no contradiction would be implied by saying that the world could, logically, have been such that P was true and Q false: there would have been the same purely physical P-realities, but some or all of the actual Q-realities would have been absent. But that contradicts (A). If it is so much as logically possible for P to be true when Q is false, then P doesn’t strictly imply Q. So the situation envisaged is impossible.
If theses (1) and (2) hold, (A) holds; but we have just seen that in that case (B) cannot also hold. Since by thesis (3) there is no extra stuff, (K) follows: Q not only redescribes the P-realities, but is made true by them. So theses (1)-(3) do after all entail physicalism.
7



IV. Strict implication without reduction


As remarked earlier, I see (1)-(3) not as a solution to the mind-body problem but as a useful way to make reasonably clear what philosophical work has to be done on it. We still need explanations of just how and why the Strict Implication thesis holds; and such explanations are not themselves part of (1)-(3). Melnyk concedes that such explanations ‘promise to go a useful way toward explaining (1) [the Strict Implication thesis]’ [12, p. 326]. He also notes that I claim they would require no appeal to psychophysical laws and would not entail reductionism, but argues that I am wrong. I think his reasoning fails because it rests on a second undefended assumption, which I will try to show is also mistaken. This is:

(M2) The relevant explanations of why P strictly implies Q must be ‘valid derivations of what is to be explained’ [p. 327].

Specifically he assumes that if P strictly implies Q, then an explanation of the kind I envisage must consist of a sequence of conditionals of the form:
A[P —> S]
A[S —> Q]
(where ‘A’ is to be read as ‘It is absolutely necessary that . . .’), from which a conclusion of the form A[P —> Q] can be seen to follow. He also assumes that the first premiss in such a sequence must be ‘expressed, consequent and all, exclusively in the proprietary vocabulary of physics’ (p. 328). Given those assumptions, he argues that sooner or later a psychophysical bridge law will appear in the sequence of conditionals, from which he reasonably infers a commitment to a correlation-based reductionism.
It may be true that such reductionism follows from my position given his assumption (M2). But I see no reason to accept (M2) (see [8, 9, 10]). To illustrate the main points, consider the case of shapes. Recall that P is a conjunction of all the true statements that are expressible in the narrow vocabulary of physics, including statements of physical laws, and let S be a conjunction of all truths about the shapes of things, including both particular and general statements. Does P strictly imply S? Well, we have a rich vocabulary for describing shapes, much of it depending on comparisons, but some of it exploiting geometrical concepts. ‘Spherical’ and ‘regular tetrahedron’, for example, belong to the narrow physical vocabulary, so many truths about things’ shapes will already be included in P and thus be trivially strictly implied by P. But vague shape-descriptions such as ‘jagged’, ‘gnarled’ and ‘spoon-shaped’ are common, and surely occur in true statements. Such statements will belong to S but not to P. Yet surely they too are strictly implied by P, since it is pretty obvious that:

(C) A thing’s having a certain shape involves nothing over and above what is strictly implied by P.

In contrast, it is far from obvious that:

(D) All truths about the shapes of things, including those using expressions like ‘jagged’, ‘gnarled’, and so on, are correlated with truths expressible in narrowly physical terms.

Yet if Melnyk is appealing to a general principle, as he seems to be, it would have to be one according to which the only way to establish (C) would be via (D). In fact it is easy to see that we can establish (C) without establishing (D) – indeed even if (D) is false. If anyone were so confused as to doubt (C), we need only point out that:

(E) The shapes of things are just the arrangements of their surfaces, which depend (contingently) on the distributions and states of the physical particles which compose them.

It is hard to see how anyone could seriously dispute (E) (waiving general Quinean qualms about ‘contingently’). As to what makes (E) true, No more need be said. (E) doesn’t require us to formulate detailed particular statements of the form A[P —> S]. It doesn’t even entail that some relevant statements of that kind are true, though hard to discover. (E) is entirely sufficient to make clear that the strict implication in question holds – and it does so without providing a definition of ‘shape’ in physical terms, and without appealing to correlations, derivations, or reductions of any (nontrivial) kind.8
That should not be surprising. The vocabulary of shapes leaves open how things come to have the shapes they do; so when we describe things as having such and such shapes, we are not thereby committing ourselves to the underlying physical details. Indeed, treating shapes as if they were essentially linked with the physics of the actual world would overlook that things could perfectly well have had shapes even if the laws of physics had been different. Nevertheless, it is the actual physical facts which fix (nomically) the arrangements of things’ surfaces, and the facts about their surfaces which fix (logically or conceptually) the facts about their shapes. That example illustrates the point that there is no universal principle requiring cases of strict implication by P to involve derivations of the type Melnyk assumes.
Nor does there seem to be any reason why the psychophysical case should be an exception. Some versions of behaviourism and functionalism would, if successful, explain how it is that P strictly implies Q – without recourse to derivations. To achieve that result, of course, they must avoid claiming that each individual mental statement has necessary and sufficient conditions in terms of behaviour or its causes: that claim entails reductionism. But it has long been recognized that there is no need to insist on such conditions: nondefinitional versions are available (see for example Armstrong [1], pp. 84f.). Such nondefinitional varieties of behaviourism or functionalism are analogues in the psychophysical case to (E) in the case of shapes. If successful, they would explain why the Strict Implication thesis holds without dependence on correlations.
You might suggest that unless explanations of the deductive form in question were at any rate available, we could not be justified in claiming that the strict implication held. But why must that be so? (E) fully explains innumerable particular instances of strict implication, and can be seen to cover them regardless of the fine details of different particular shapes. To see that P strictly implies all truths about shapes we do not have to consider each particular ascription of a shape to an individual thing.
9 We can know (E) even though we know neither all the relevant physical facts, nor all the facts about shapes. It would be pointless to try to discover the detailed physical facts which held in particular cases, and it would probably obscure what mattered.10
Although Melnyk seems to have been thinking of the ‘classical’, correlating type of reductionism, the above considerations have wider implications. Instead of looking for bridge-laws or sentence-correlations, some types of reductionism aim to construct an image or model of the reduced theory in the reducing theory – a model deducible from the reducing theory conjoined with certain special conditions (see e.g. [2], [4]). The idea is that once a reduction of that kind has been achieved, ontological questions can be decided on the basis of how ‘smooth’ or ‘bumpy’ it is: how closely the categories of the reduced theory are modelled in the reducing theory. But there is no more reason to suppose that (1)-(3) commit us to that type of reductionism than to the classical type. Nondefinitional functionalism has no more need to engage in the project of constructing a model of psychology in physics (or in neurobiology) than it has to take on the unpromising task of specifying necessary and sufficient conditions for each individual type of mental state.
11
Certainly physicalists owe explanations of why everyday and/or scientific psychology work as they do, and also of why the Strict Implication thesis holds. And of course reductionism of whatever kind – provided it does not just eliminate psychology – would be one way of providing such explanations. But the examples of nondefinitional behaviourism and functionalism show it is not the only way. Neither Kim nor Melnyk, nor anyone else to my knowledge, seems to have provided sound reasons for thinking that physicalists have to be reductionists (or else eliminativists) in any nontrivial sense.
Melnyk ends by arguing that theses (1)-(3) do not adequately formulate physicalism about nomic mental facts. He rightly notes that I must hold that P strictly implies every mental law (p. 331), but alleges that I give no indication ‘either how or even that these explanations [viz. the philosophical explanations which perform the role taken by (E) for the special case mentioned above] could be used to explain the strict implication by P of mental laws’. He goes on to claim that I am again ‘forced back to the interpretation of the strict implication of mental laws by P’ in terms of ‘some primitive and unanalysable relation of absolute necessitation of mental laws by the physical facts’ (p. 331).
I can now reply very briskly. In offering (1)-(3) as a way of helping to make the commitments of minimal physicalism reasonably clear, I did not also set out to provide explanations of the kinds needed to establish that the Strict Implication thesis does in fact hold. That is a different project (although I have offered a contribution towards it in [8]). Sure enough, Q includes all true laws and lawlike statements of scientific and everyday psychology. Explaining how the facts specified by P make those statements true is a nontrivial project; but if I am right, there are no good reasons to suppose it cannot be done without resorting to reduction. When Melnyk claims I am ‘again’ forced back to a blankly mysterious type of necessity, that is because he again overlooks that we can be justified in accepting that a relation of strict implication holds without having to provide explicit derivations.
I have argued that theses (1)-(3) entail that the contingent truths in Q are redescriptions of, and made true by, parts and features of the world described by P, so that they entail full-blooded physicalism, not mere property dualism. (1)-(3) offer a reasonably clear alternative to property dualism. Melnyk’s contrary assumption (M1) turned out to have inconsistent implications. I have also argued (in [10] supplemented by this) that (1)-(3) do not commit us to reductionism in any nontrivial sense: Melnyk’s objection depends on a second mistaken assumption (M2). I think those conclusions enhance the usefulness of (1)-(3) for making the commitments of minimal physicalism reasonably clear.
12


FOOTNOTES


1. In earlier formulations I confined Q to truths about the organisms whose existence is provided for by P. In view of (3) below that restriction seems unnecessary.

2. For more detailed exposition and discussion see [8], pp. 71-86, and [9]. For a similar approach see Frank Jackson’s [5].

3. You might think (3) alone would be enough for physicalism, but there are two reasons for including (1) and (2). One is that (3) by itself would not rule out, as (2) does, a variety of idealism according to which all physical truths were strictly implied by psychological truths. The other is that just asserting that nothing exists but what is strictly implied to exist by P fails to make explicit, as (1) does, that all psychological truths – including e.g. any laws there may be – are strictly implied by P. You might now be tempted to suggest that (3) would be redundant if (1) were revised by substituting something like ‘all truths’ for ‘Q’. The trouble is that P alone does not strictly imply that there are no nonphysical items unless, like Hobbes, you think materialism is in any case necessarily true. If, like me, you think physicalism/ materialism is not absolutely necessary, the suggested revised version of (1) would be false. There is a notable contrast with one type of supervenience (though certainly not with other types): the one Chalmers calls ‘logical’ supervenience ([3], ch. 2). If in his sense you say ‘All facts/ entities/ properties logically supervene on physical ones’, you automatically rule out nonphysical facts etc. (I have argued that essentially that type of supervenience is the only one consistent with physicalism: see [9], especially pp. 250-2.)

4. Chalmers says, ‘A phenomenon is reductively explainable ... if the property of exemplifying that phenomenon is globally logically supervenient on physical properties’ [3], p. 48. Since his notion of logical supervenience is to all intents and purposes equivalent to theses (1)-(3), that seems to make exponents of (1)-(3) reductionist by definition. However, he distinguishes between reduction and reductive explanation, so perhaps my interpretation is open to question. Others are more explicit: e.g. Robert McCauley states: ‘Physicalism in the philosophy of mind anticipates a comprehensive reduction . . .’ [11], p. 712. Note that in [8] and [9] I have argued that all physicalists are committed to something close to theses (1)-(3).

5. ‘Opposed to reductive physicalism is nonreductive physicalism, also called property dualism’ – Kim [7], p. 645. Cf. also Bickle: ‘. . . contemporary nonreductive physicalism is identical to property dualism, traditionally conceived. Much current “nonreductive physicalism” is not physicalism at all’ [2], p. 8.

6. For reasons, see [8], pp. 71-80, [9], pp. 244-9.

7. Here I must mention another objection Melnyk alleges against my version of minimal physicalism. He says, ‘Kirk’s physicalism is ... consistent with the possibility of the world’s being exactly as it is physically, save in some utterly trifling respect ..., but entirely lacking in mentality’ [12], p. 325, n. 3. That is the ‘rogue atom’ objection advanced by Kim e.g. [6], p. 277. In [10] I explain why it fails. Briefly, it overlooks the likelihood that what goes on in remote regions of space is mostly irrelevant to mentality. If that is right, the possibility Melnyk alleges is not genuine: many different conjunctions of statements other than P will also strictly imply Q, including vast numbers which assign different positions to atoms in deep space.

8. See also [8, 9, 10]. I need hardly add that there is no guarantee that all explanations of strict implication which do not include derivations of the sort Melnyk assumes will be as straightforward as (E).

9. Melnyk bases his reasoning on the assumption that the ‘valid derivations’ he envisages will be about such particular ascriptions.

10. These points are closely related to those made by Putnam about causal explanations in [13].

11. It is worth noting that Bickle, in his exposition of ‘new-wave reductionism’, regards functionalism, though unattractive, as a nonreductionist option for physicalists (see e.g. [2], p. 48), thereby committing himself to the view that physicalism does not actually entail either reductionism or eliminativism.

12. Thanks to Robert Black, Ed Lindon, and Andrew Melnyk for comments on earlier versions, and special thanks to Bill Fish for extended discussions. Thanks too to two anonymous assessors for the AJP.


REFERENCES


1. Armstrong, D. M., A Materialist Theory of the Mind (London: Routledge and Kegan Paul,1968).

2. Bickle, J., Psychoneural Reduction: The New Wave (Cambridge, Mass., and London: MIT Press, 1998).

3. Chalmers, D., The Conscious Mind: in search of a fundamental theory (New York and Oxford: Oxford University Press, 1996).

4. Churchland, P. S., Neurophilosophy: Toward a Unified Science of the Mind-Brain (Cambridge, Mass. and London: MIT Press, 1986).

5. Jackson, F., From Metaphysics to Ethics: A Defence of Conceptual Analysis, (Oxford: Clarendon Press, 1998).

6. Kim, J., ‘The Myth of nonreductive Physicalism’, in his Supervenience and Mind: selected philosophical essays (Cambridge: Cambridge University Press, 1993) pp. 265-284.

7. Kim, J., ‘Physicalism’, in [14], p. 645.

8. Kirk, R., Raw Feeling: a philosophical account of consciousness (Oxford and New York: Clarendon Press, 1994).

9. Kirk, R. ‘Strict Implication, Supervenience, and Physicalism’, Australasian Journal of Philosophy 74 (1996) 244-56.

10. Kirk, R., ‘How Physicalists Can Avoid Reductionism’, Synthese 108 (1996) 157-170.

11. McCauley, R., ‘Reductionism’, in [14], 712-14].

12. Melnyk, A., ‘The Prospects for Kirk’s nonreductive Physicalism’, Australasian Journal of Philosophy 76 (1998) 323-332.

13. Putnam, H., ‘Philosophy and Our Mental Life’, in his Mind, Language and Reality, Philosophical Papers, ii (Cambridge: Cambridge University Press,
1975) 291-303.

14. Wilson, R. A. and F.C. Keil (eds.) The MIT Encyclopedia of the Cognitive Sciences (Cambridge, Mass. and London: MIT Press, 1999).



#