Last Friday the SMU philosophy of mind reading group read Sydney Shoemaker’s “Realization and Mental Causation” (in Gillet and Loewer (eds.) Physicalism and Its Discontents, reprinted in the 2nd edition of Shoemaker’s collection Identity, Cause, and Mind (2003). I am something of a fan of Shoemaker’s work, having written about and followed his work on phenomenal content very closely. I had not read his more metaphysical work in the philosophy of mind as carefully. But I found this paper of Shoemaker’s to be insightful in the ways that characterize the rest of his work.
According to Shoemaker, what it is for a property X to realize a property Y is for the conditional powers bestowed by Y to be a a subset of the conditional powers bestowed by X. A conditional power of a thing is one that, together with other properties the thing has, will determine a power simpliciter.
In the case of multiple realizability, “the conditional powers bestowed by it [the realized property] will be a proper subset of the sets bestowed by each of the realizer properties.”
There are at least three ways of understanding this view, and it was not clear to me which was Shoemaker’s intended view. It is quite possible that a more careful reading might eliminate my uncertainty about which view was intended. But I think all three views may in fact have difficulties, so that my worries about Shoemaker’s proposal might be expressed as a dilemma.
Interpretation 1: If mental property M is realizable by three distinct physical properties P1, P2, and P3, then there is a non-disjunctive set of conditional powers bestowed by M that P1, P2, and P3 all have in common. These conditional powers are specifiable purely in physical terms.
Interpretation 2: If mental property M is realizable by three distinct physical properties P1, P2, and P3, then for each instance of M the conditional powers bestowed by M is a proper subset of the conditional powers bestowed by the realizer. However, these conditional powers need not be identical for each realization of M. The conditional powers that are shared by all instances of M may be specifiable only as a disjunction. These conditional powers are specifiable purely in physical terms.
Interpretation3: If mental property M is realizable by three distinct physical properties P1, P2, and P3, then there is a non-disjunctive set of conditional powers bestowed by M that P1, P2, and P3 all have in common. These conditional powers are not, however, specifiable purely in physical terms.
Problem with Interpretation 3: Interpretation 3 seems to fail to deliver a view according to which the physical causes of behavior do not compete with the mental causes of behavior. The physical realizer P1 of a pain has causal powers that are sufficient to cause the screaming, for example. These causal powers are specifiable purely at the physical level. If the causal powers of the mental are not specifiable in purely physical terms, then they would seem to be distinct causal powers.
I don’t think that interpretation 3 is the intended one, since it does not adequately fit the following statement by Shoemaker: “It is qua realization of pain that P1 made its contribution to causing behavior; for the conditional powers it conferred that are independent of those conferred by the property of being in pain were irrelevant to its making this contribution” (p. 433 of ICM). This statement makes sense only if the conditional powers bestowed by being in pain are identical to a subset of the physically-specifiable conditional powers of P1. For there clearly are physically-specifiable conditional powers of P1 that are relevant to the production of the behavior.
Problem with Interpretation 1: It is not clear that there are any physical conditional powers that all realizations of a mental property will have in common. Neural realizers of pain, for example, have certain physical conditional powers to cause screamings. But they do not, presumably, have a conditional power to cause screamings (in the relevant way) in Martians or in robots. The relevant, non-disjunctive and physically specifiable conditional powers of neural realizers of pain may be entirely distinct from the conditional powers of silicon realizers of pain.
Shoemaker recognizes something like this point when he says that the different physical realizers of mental properties will fall into “families” of properties that are jointly instantiable. But it is not clear to me that this is quite the right way to handle the issue. Neural properties and silicon properties are jointly instantiable. The problem is that for any type of mental property, the physically-specifiable conditional powers of a realizer of that property can be different powers from those of a different physical realizer.
Problem with Interpretation 2: In light of the problem with Interpretation 1 we are lead naturally to interpretation 2. But this view has its own problems. This view identifies the property of being a pain with a disjunction of conditional powers. Now it seems to me that the causal efficacy of a disjunctive property will be preempted by the causal efficacy of one of its disjuncts. For instance, if my pain is realized by P1, and if the property of being a pain is the disjunction of P1 and P2 and P3, then my instantiation of P1 preempts my instantiation of {p1 or p2 or p3}. Intepretation 2 thus seems to fail to give an account according to which my pain causes me to scream “ouch” in virtue of being a pain. [Perhaps it can deliver a view according to which my pain causes me to scream "ouch" in virtue of being a human pain, assuming that there is a single set of physical conditional powers in common between all instances of pain in human beings. But it is not clear to me that this is quite what we want in an account of mental causation.]
Dear Brad,
I think that the key to dealing with your questions is in the distinction I make in section IV of the paper between physical causal features and mental causal features of properties. Correponding to these will be two sorts of conditional powers bestowed by physical realizers – those such that the effects of their activation can be specified in purely physical terms, and those such that the effects of their activation are specified in mental terms. E.g., if C-fiber firing is a realizer of pain, it will have the mental causal feature of contributing to the production of the belief that one is in pain, and the physical causal feature of contributing to the production of whatever physical property is the realizer, in creatures of the relevant sort, of the belief that one is in pain. And it will bestow conditional powers corresponding to each, the mentally specified conditional power being realized by the physically specified conditional power.
It is the mental causal features of a realizer that are shared by the realized mental property and by all the other possible realizers. Of these your interpretation 3 is true. I think this is compatible with the passage you quote from p. 433. P1 qua realizer of pain has the mental causal features of pain, but it has these in virtue of having certain physical causal features (dispositions to cause physical states that realize mental states pain is disposed to cause). The conditional powers associated with these causal features are not independent of those bestowed by pain – they are not like the power to activate a C-fiber firing detector – since it is the bestowal of these that constitutes the instantiation of pain in this sort of creature.
What interpretation 2 says of the conditional powers bestowed by the realizers is true of the physical conditional powers – there won’t be any set of these that is bestowed by all of the realizers. But the conditional powers bestowed by an instance of M won’t be a subset of the physical conditional powers bestowed by its realizer – rather they will be mental conditional powers that are realized by a subset of those physical conditional powers.
Interpretation 1 has to be rejected because (as I say in the paper) the different possible realizers of a given mental property in different kinds of creatures will have different causal features and bestow different physical conditional powers – although, of course, the conditional powers they bestow will realize the mental conditional powers bestowed by the mental property.
I stick to the claim that the different realizers in different sorts of creatures are not jointly instantiable. The realizers I am talking about are total realizers, not just “core” realizers, and I take it that in a given creature on a given occasion it cannot be that its being in pain has different total realizers of different sorts.
You complain against interpretation 2 that it identifies the property of being in pain with a disjunction of conditional powers, and that the causal efficacy of a disjunctive property will be preempted by the causal efficacy of one of its disjuncts. Now I don’t want to identify a property with a disjunction of conditional powers – for one thing, this leaves out of account what I call (in later work) the “backward looking” causal features of properties, those having to do with how the instantiation of the property can be caused. I do, however, think that multiply realized properties are identical with disjunctions of their realizers (that’s analytic: a multiply realized property is one something has just in case it as one or another of a set of realizers, and so is equivalent to the disjunction of the members of the set). But I deny that disjuncts of a disjunctive property preempt the causal efficacy of the disjunctive property. That is true only when the disjuncts can only be given by a list – it is not true when the realized (disjunctive) property has a causal profile which all of the disjuncts must realize. All properties that we can know about are multiply realized and so disjunctive – and it had better not be the case that all properties lack causal efficacy.
Left by Sydney Shoemaker on September 6th, 2006