In this paper, I wish to argue that Kim's putative solution to the exclusion problem rests on an equivocation between instantiations of properties as bearers of properties and instantiations as property instances. On the former understanding, the causal inheritance principle is too weak to confer causal efficacy upon mental properties. And on the latter understanding, the identification of mental and physical instantiations is simply untenable.
Recall: the exclusion problem arises for views according to which mental properties/events are (i) causally efficacious vis-à-vis physical events and (ii) non-identical to physical properties/events. If the physical domain is causally closed, then every (caused) physical event has a physical cause. And this physical cause threatens to exclude any mental cause of the physical effect in question.Foot note 4_5
According to Kim, functional properties are 2nd order properties.Foot note 4_6 A 2nd order property is the property of having a (1st order) property which satisfies some specification or other. For example, my shirt has the 2nd order property of being my favourite colour because it has a colour property -- blueness -- which meets the condition of being my favorite. Functional properties are 2nd order properties for which the specification is that of occupying a certain causal role. So, for example, a functionalist might claim that being in pain is the property of instantiating a property which is caused by tissue damage and causes winces and groans. In humans (as the philosophical literature would have it) the realizing property is that of C-fibre activation.
Kim's solution to the exclusion problem relies on the Causal Inheritance Principle and what might be called «Instantiation Identity.» Kim characterizes the Causal Inheritance Principle as follows:
If a second-order property F is realized on a given occasion by a first-order property H ...then the causal powers of this particular instance of F are identical with (or are a subset of) the causal powers of H (or this instance of H).Foot note 4_7
According to Instantiation Identity, each instantiation of a (functionalizable) mental property is identical with an instantiation of the 1st order physical property which realizes it on the occasion in question; «...each instance of M is an instance of P1, or of P2, or ..., where the P's are M's realizers.»Foot note 4_8 Kim utilizes these theses as follows. From Instantiation Identity it follows that mental causation does not involve two causes competing for efficacy. There is only one cause -- the physical cause -- of any given physical event. Nevertheless, the Causal Inheritance Principle entails that mental properties inherit the causal powers of their 1st order realizing properties.Foot note 4_9 The upshot is that «...functional mental properties turn out, on account of their multiple realization, to be causally heterogeneous but not causally impotent.»Foot note 4_10
`Instantiation' talk is equivocal between talk of bearers of properties and talk of instances of properties. A particular shirt might be said to be a bearer of the property blueness, whereas the blueness-of-the-shirt is an instance of the property. Kim's use of the term `instance' suggests that the latter is what he has in mind. But at one point, he remarks, «[we] may take «instances» or «instantiations» of properties as events, states, or phenomena,»Foot note 4_11 which suggests that property bearers are what are at issue. In order to avoid terminological confusion, I will henceforward articulate the distinction as one between property bearers and property tokens.
Now it might be thought that, given his analysis of events, Kim would argue that being a property bearer and being a property token amount to the same thing. After all, on Kim's view, an event just is a property token, a complex consisting of a substance x, a property P, and a time t -- [x, P, t] -- which exists just in case x has P at t.Foot note 4_12 Although tempting, this interpretation of Kim's background metaphysic should be resisted. Events cannot be identified with property tokens because events and property tokens differ in their modal profiles. Events are the relata of the causal relation, and need (or, perhaps better, have) essences commensurate with this role. Even if one rejects a counterfactual analysis of causation, the causal relation remains counterfactual supporting (except in cases of causal over-determination, pre-emption, etc.). And Kim concedes that this requires treating the constitutive time of an event, and perhaps the constitutive property, as inessential to it.Foot note 4_13 At the end of the day, Kim is best thought of as identifying events with functions from possible worlds to substance-property-time complexes existing at those worlds, rather than with substance-property-time complexes per se (although Kim might eschew the possible-worlds formulation I've given here).
For expository purposes, I will assume that property tokens also are functions from worlds to substance-property-time complexes.Foot note 4_14 In my view, however, they are (typically) distinct functions. More to the point, what I want to argue is that if the constitutive property of an event is an inessential feature of it then that event is distinct from a token of the constitutive property. Even though the property token and the event share a manifestation at at least one world -- that is, the value of each of the corresponding functions at that world is the same substance-property-time complex -- they will have distinct manifestations at other worlds.
There are a couple of reasons for thinking that the constitutive property of a property token is an essential feature of it. First, intuitively, although a blue shirt, for example, is presumably only accidentally blue, the blueness of the shirt is essentially blue. Or again, suppose it is true of a walking event that was in fact a strolling that it might have been a striding.Foot note 4_15 Nevertheless, the strolling-ness of the event is, again intuitively, essentially a stroll. Second, in contrast to events, the modal profiles of property tokens are governed (in part) by the role(s) they play in property metaphysics. For example, in order for various reductive programmesFoot note 4_16 in metaphysics to be even minimally promising, the modal relation between property tokens and property bearers needs to systematically co-vary with the modal relation between properties proper (types) and their bearers. This means that at any world w1 at which an object x has a property P at time t there must exist the corresponding property token -- a complex consisting of x, P, and t. And at any world w2 at which x lacks P at t, no such complex can exist. One might rejoin that this is compatible with the existence of the w1-token at w2, as long as we do not insist that being P is an essential feature of the w1-token. But if type-token modal co-variance is to be sustained, this would seem to require that we take x to bear P at w2, contra hypothesis.Foot note 4_17,Foot note 4_18
Suppose we formulate Kim's putative solution to the exclusion problem in terms of property tokens. This would require that we reformulate the Causal Inheritance Principle and Instantiation Identity as follows:
CIPT: If a second-order property F is realized on a given occasion by a first-order property H then the causal powers of this token of F are identical with the causal powers of this token of H.
IIT: Every token of a functional mental property is identical with the token of the 1st order physical property which realizes it on the occasion in question.
Now CIPT does seem to yield a promising basis for the causal efficacy of functional mental properties.Foot note 4_19 After all, if each token of a property has causal powers then the property itself does, even if its powers are highly heterogeneous. But it is hard to see why the F-token would inherit the causal powers of the H-token unless they were identical. As a result, on the token-formulation, the truth of IIT is required to underpin the transfer of causal power from realizing to functional properties, as well as to ensure that mental causes do not compete for efficacy with physical causes.
But IIT is simply untenable. The connection between functional mental properties and their 1st order realizing properties is contingent: not only could a functional property be realized by a number of distinct 1st order properties, a substance could bear one of these 1st order properties without bearing the functional property -- Lewis's madman is a case in point.Foot note 4_20,Foot note 4_21 More to the point, a functional mental token cannot be identical to a realizing physical token because there are possible circumstances in which that very token occurs but fails to occupy the functional role at issue. If, as argued above, the constitutive property of a token is essential to it, the functional mental token in question could not exist in such circumstances. And since identity holds of necessity, this entails that the mental and physical tokens are distinct.Foot note 4_22,Foot note 4_23
Perhaps Kim's solution does better if formulated in terms of property bearers. This would require that we again reformulate the Causal Inheritance Principle and Instantiation Identity:
CIPB: If a second-order property F is realized on a given occasion by a first-order property H then the causal powers of this bearer of F are identical with the causal powers of this bearer of H.
IIB: Every bearer of a functional mental property is identical with the bearer of the 1st order physical property which realizes it on the occasion in question.
Now IIB, unlike IIT, is more or less uncontroversial.Foot note 4_24 Moreover, as long as we assume that the only mental causes are the bearers of mental properties, IIB may well suffice to ensure that mental causes do not compete for efficacy with physical causes. Moreover, if IIB is true, then CIPB trivially follows from it; if the bearer of a mental property just is the bearer of its realizing physical property, then of course the causal powers of the former and the causal powers of the latter coincide.
The trouble that arises for the bearer-formulation is that the truth of IIB and CIPB does not suffice to confer causal efficacy upon functional mental properties. In order to do so, these theses would need to be supplemented with a principle to the effect that if all bearers of a property are causally efficacious, then the property itself is causally efficacious. However, this supplementation is untenable: it implies that any property whatsoever with at least one (causally efficacious) bearer is itself causally potent. But not only would this undercut Kim's own criticisms of anomalous monism,Foot note 4_25 it would entail that highly extrinsic/ relational properties are efficacious. The property of being seventy-three million light years distant from me as I write this sentence may well have a (causally efficacious) bearer -- if it does not, I'll just choose another distance. Nevertheless, this property is not (thereby) efficacious in its own right.Foot note 4_26
One final possibility would be to endorse a mixed formulation of Kim's solution to the exclusion problem, perhaps by combining the uncontroversial IIB and the promising CIPT. One difficulty with this suggestion is that the falsity of IIT brings the truth of CIPT into question. If a 2nd order token is distinct from its 1st order realizing token, it is unclear why the former should inherit the causal powers of the latter. Moreover, there seem to be clear counter-examples. Consider again my shirt, which has the 2nd order property of being my favourite colour because it has the 1st order property of blueness which meets the condition of being my favorite. Included among the causal powers of the blueness token would be certain photochemical powers. But insofar as the favourite-colour token is distinct from the blueness token, it would be implausible to suppose that it inherits these photochemical powers.Foot note 4_27
There is, however, a natural rejoinder to this last point. Kim did not claim that instances of 2nd order properties inherit all of the causal powers of their realizing 1st order instances. He claimed instead that they inherit either all or a subset of these powers.Foot note 4_28 As a result, he could easily assert that the photochemical powers simply fall outside of the subset of powers that the favourite-colour token inherits from of the blueness token. But this response immediately raises the following question: exactly which of the causal powers of its realizing property token does a 2nd order property token inherit? And the answer is not hard to find: the powers specific to the condition that the 1st order token has to meet in order to realize the 2nd order property. The favourite-colour token inherits from the blueness token exactly those powers it has in virtue of being a token of my favourite colour. And a functional mental token inherits from its realizing token those powers it has in virtue of meeting the functional specification at issue.
Now strictly speaking, this manoeuvre may well render the truth of CIPT consistent with the falsity of IIT, even if a certain puzzlement remains as to why any inheritance of causal powers should occur, let alone such a tidy pattern of inheritance. But even so, it remains far from clear that the mixed-formulation yields a satisfactory solution to the exclusion problem. An adequate solution requires establishing that (i) functional mental properties are causally efficacious and (ii) mental causes do not compete for efficacy with physical causes. On the mixed-formulation, the latter is established by taking the bearers of mental properties to be the only mental causes, and identifying mental and physical bearers. But if mental bearers are the only mental causes, then mental tokens are not causes. Any suggestion that they are nonetheless causally efficacious borders on unintelligibility.
Department of Philosophy
University of Lethbridge
4401 University Dr.
Canada T1K 3M4
<peter.alward [at] uleth.ca>