Lewis attempts to solve this difficulty by combining functionalism and the identity theory in the following way: he gives a functionalist account of pain for a population, and gives an identity theoretical account of pain for individual members of a population. According to Lewis, a state S is pain for a population P if and only if, with few exceptions, whenever a member of P is in S, her being in S has the sorts of causes and effects given by the pain role. (Lewis, 1980, p. 113) As a result, C-fibre activation (or what have you) is pain for the human population, and foot-cavity inflation is pain for the Martian population. And an individual, X, is in pain if and only if X is in a state which is pain for the appropriate population. (Lewis, 1980, p. 113) Thus, any individual for whom the appropriate population is the human population is in pain just in case she is in a state of C-fibre activation. And any individual for whom the appropriate population is the Martian population is in pain just in case she is in a state of foot-cavity inflation.
The problem that arises for Lewis' view concerns his account of under what conditions a population P is appropriate for a given individual X. He provides us with the following four criteria:
Lewis does not give much explicit guidance as to how to apply these criteria, so it is best to look at how he does so in particular cases. Consider the following:
The suggestion seems to be that each criterion is of relatively equal weight and, so, the population that satisfies the greatest number of criteria for a given individual is the appropriate population for that individual.
But note that given this characterization of the procedure for applying the criteria, Lewis' application of them is in both instances slightly askew. Consider case (a). The Martian population does satisfy criteria (2), (3), and (4) for a Martian for whom foot-cavity inflation occupies the pain-role. After all, it is a population which is a natural kind and of which our ordinary Martian is an unexceptional member. But the Human population does not merely satisfy criterion (1), but criterion (4) as well. And so while Lewis's theory does imply that the appropriate population for our ordinary Martian is the Martian population, it has to be because (2), (3), and (4) outweigh (1) and (4) together, and not simply (1) by itself as Lewis suggests. Similarly, in case (b), Lewis's theory implies that the appropriate population for the mad human is the human population rather than the mad human population. But this has to be because (1), (2) and (4) outweigh (2) and (3) together, and not (3) by itself as Lewis claims.
The real trouble arises, however, when we consider the case of the mad Martian. Lewis gives the following analysis of this case:
Now I agree that the Martian population does satisfy criteria (2) and (4) for a Martian for whom foot-cavity inflation fails to occupy the pain-role. But note: the human population does not merely satisfy criterion (1) for our extraordinary Martian, but criteria (1) and (4). And the mad Martian population does not merely satisfy criterion (3) in this case, but criteria (2) and (3). The upshot is that Lewis view does not imply that the appropriate population for our extraordinary Martian is the Martian population as opposed to the Human or mad Martian populations. As it stands, each of the three populations has an equal claim to be the appropriate one in this case. And so it seems that Lewis's theory fails to accommodate the possibility of mad Martian pain.
The source of this difficulty is the fact that criteria (1) and (3) have been inappropriately given the same weight as (2) and (4). The reason less weight ought to be given to criterion (1) is that, given that the human population is a natural kind, it automatically satisfies two of the criteria for being the appropriate population for any arbitrary individual. And the reason less weight ought to be given to criterion (3) is that for any physical realization of any given causal role, there will be a population of individuals who are similar with respect to the realization of the role, but many of whose (at least possible) members will be otherwise physiologically very different.Foot note 1_2 And some such motley population will always satisfy two of the criteria for being appropriate for any given individual.
In light of this, one possible emendation to Lewis' view would be simply to dispose of criteria (1) and (3) altogether. This approach, however, ought to be rejected. Suppose, for example, humans were an exceptional species with respect to the physical realization of pain within a larger a larger genus, itself a natural kind. The wholesale dismissal of (1) and (3) would have the effect of giving the human population and the more inclusive genus equal claim to being the appropriate population for any given ordinary human.
A better suggestion, at first glance, would be to retain (1) and (3) but give them lower weight than (2) and (4). This would yield the desired results regarding both sane human and mad Martian pain. But optimism here may well be premature. This is because there seem to be good reasons for demanding both that criterion (1) outweigh criterion (3) and that criterion (3) outweigh criterion (1), surely an untenable result. First, unless (1) outweighs (3), there will be no grounds for taking the appropriate population for Lewis' madman to be the human population as opposed to a more inclusive genus in which C-fibre activation is the occupant of the same causal role it occupies for the madman. But second, unless (3) outweighs (1), there will be no grounds for taking the appropriate population for an artificially intelligent robot to be the population of similarly designed robots (no natural kind) as opposed to the human population. Lewis' view, it seems, may ultimately be unsalvageable.
[Foot Note 1_1]
Thanks are due to the student in my philosophy of mind class at the University of Lethbridge in the fall of 2001 whose struggles with Lewis helped me «see the light.»
[Foot Note 1_2]
Note: I am assuming that what criterion (3) requires is that an individual have the same physical realization of (relevant?) causal roles as members of a population.