Glossary and Notational Conventions
For the purposes of this handout, lower-case letters will name particular events (i.e. event tokens). Capital letters will represent properties or event types. So m might be a particular mental event (say, my having a headache at a particular time), and M might be a general type of mental event (say, heaving a headache).
Nomological. Involving laws (from nomos, which means law). So e.g. "the nomological requirement on causation" is the requirement that, if one event causes another, then there is a universal law that events of the same type as the first always lead to events of the same type as the second.
Determinate/determinable. A determinable property is a property like being red or walking, where there are many different more specific ways of having the property. Something can be red by being crimson, or scarlet, or . . .. Someone can walk by walking rapidly, walking slowly, shuffling, etc. (Recall the old "walk this way" joke.) So crimson and scarlet are determinates of the determinable red; walking rapidly and shuffling are determinates of the determinable walking.
Supervenience. Let H be higher-level properties, and L be lower-level properties. H properties supervene on L properties if and only if there cannot be a difference in H properties unless there is also a difference in L properties. (Another way to say the same thing: H properties supervene on L properties if and only if, necessarily, if all the L properties are the same, the H properties are also the same.) Uncontroversial examples (I think): the shapes on my monitor supervene on the R, G, and B components of the individual pixels. Whether a plant is alive or dead supervenes on the physical properties of the plant. More controversial example: mental properties supervene on physical properties.
[Kim's definition of supervenience on pp. 174-175 is a little more complicated but embodies essentially the same idea. We can distinguish many different varieties of supervenience. For example, global vs. local; metaphysical vs. nomological. See the Horgan essay in our text for lots more about this notion.]
Dualism and Mental Causation
Mental causation has always posed a problem for dualism. (As Yablo reminds us, Princess Elizabeth of Bohemia corresponded with Descartes about the issue.) It has always seemed mysterious how something nonphysical could possibly have causal effects on something physical. Kim suggests that some versions of materialism (in particular, the "nonreductive physicalism" defended by Fodor, which seems to be very close to functionalism) may also face problems about how mental causation is possible. Yablo offers a way to respond to this apparent problem.
Let's start by seeing how physicalism at first seems to solve dualism's problem with mental causation.
Consider a token mental event m (my intention to raise my arm at a particular time) which causes a physical event p (my arm going up). So we have:
m --> p
On the other hand, if we believe in the causal closure of the physical, then we will also think that there was a complete physical cause c of p. But now the idea that m caused p seems superfluous: if p has a complete physical cause, what's the role of m in bringing it about? It seems silly to think that p has two different complete causes, m and c. So it looks like we have:
c --> p
Physicalism seems to resolve this problem. According to physicalism, m = c. The mental event is identical with a physical event, so there is no danger that the existence of a physical cause will "exclude" the mental cause. Instead, the physical cause just is the mental cause!
Kim: Three Problems of Mental Causation
Kim surveys three problems related to mental causation, but only discusses one in depth (namely the third, the "problem of causal exclusion").
1. The problem of anomalous mental properties.
1. There are no exceptionless laws involving psychological notions.
2. If event e1 causes event e2, then there must be an exceptionless law connecting events of the same type as e1 with events of the same type as e2. (The "nomological requirement on causation.")
3. If e1 causes e2 because e1 is of mental type M, then the law connecting e1 and e2 must involve events of type M. (That is, the "events of the same type as e1" mentioned in premise 2 must be events of type M.)
4. Mental events do not cause physical events because of the kind of mental event they are.
2. The problem of intrinsic mental properties.
1. The semantic properties of mental states (i.e. what they are about) are not intrinsic. (They depend on facts about the individual's history and environment.)
2. What an event causes depends only on the current, intrinsic properties of the event.
3. Mental events do not cause other events by virtue of their semantic properties. (So, for instance, the content of my beliefs does not play a role in determining what they cause.)
3. The problem of causal exclusion.
Consider a case in which a mental event m of mental type M causes a physical event p. (Recall that lower-case letters represent events, upper-case letters represent types or properties.)
1. p has a complete physical cause c. (causal closure of the physical)
2. c causes its effect by virtue of its physical type C.
3. Even if m = c, M ≠ C. (This follows from nonreductionist physicalism, e.g. functionalism. For an identity theorist, of course, M = C.)
m does not cause p by virtue of its mental type M.
Yablo's Response to the Problem of Causal Exclusion
Yablo: physical events are determinates of mental determinables. That is, the relation between a mental property and its physical realizers is like the relation between red and its determinates such as scarlet, crimson, or like the relation between drinking and its determinates such as guzzling, sipping, etc.
If it is true that the relation between mental properties and their physical realizers is an instance of the relation between determinables and their determinates, then we would expect there to be a similar exclusion problem for other determinate/determinable pairs. For example, since the property of guzzling the hemlock is sufficient to ensure Socrates' death, it seems that this would exclude the property of drinking the hemlock from being responsible for his death. But that seems nuts.
Yablo argues for the following two things:
1. a determinable and its determinate on a particular occasion can both be causally relevant to an effect: causal relevance of the determinate does not exclude causal relevance of the determinable. The drinking/guzzling example seems to show this.
2. Even though a determinable and one of its determinates may both be causally relevant to a a given effect, it may be that we can't regard both of them as being "the cause" of the effect. But here the determinable may have a better claim in some cases than its determinate. The cause of an effect is such that, if it had not occurred, the effect would not have occurred either. Socrates' guzzling the hemlock does not have this property (since if he had not guzzled it, he would have drunk it in some other way, still resulting in his death). His drinking the hemlock, however, does have this property. So the drinking is a better candidate to be the cause than the guzzling.
[note to myself: I should be a little more careful about the distinction
between two ways of describing these issues. Kim treats token events m and c as
being identical (m = c), but talks about events causing effects by virtue of
properties, so the exclusion is an exclusion of property M by property C. Yablo
starts off talking about properties, but then argues that the events themselves
are distinct. My last two sentences above presuppose this view, and really
should be reformulated to make them compatible with Kim's way of describing the
situation. But it's hard to do that without some pretty awkward formulations.]