Over the semester break, I took some time to look at Peter van Inwagen's paper "Materialism and the Psychological-Continuity Account of Personal Identity" (Philosophical Perspectives 11 (1997): 305-319) and, as I realized that I don't have a good candidate for submission to Monday's Philosopher's Carnival, I thought this would be a good time to write down some thoughts that I had in connection with this paper and (very) broadly Lockean "psychological continuity" accounts of personal identity in general.
The aim of van Inwagen's paper is to show that these kind of psychological continuity accounts require the existence of immaterial substances, and so are incompatible materialism. He takes aim at what I see as the primary flaw of contemporary analytic philosophy: the attempt to isolate from one another philosophical issues that are in fact inextricably connected. In this case, this is primarily the attempt separate personal identity from the rest of ontology, which, I agree with van Inwagen, is not a good idea, and is probably nearly impossible.
However, I do think there is a version of the psychological continuity account that IS very nearly ontologically agnostic, and is at least agnostic as to the existence of material and immaterial substance, which is precisely what Locke claims about his theory. I will not spend time here making any arguments that the theory I'm presenting is Locke's (although I think it probably is), but I do want to show that there is a broadly Lockean account here that avoids van Inwagen's argument. This account makes a highly unusual assertion: according to it, persons are not in fact substances but events.
According to any psychological continuity account of personal identity, persons remain the same over time because their mental states bear a certain relationship to one another. In Locke's case, memory is emphasized, but I do not think this is necessary. It would be just as easy to say that a person A existing at time T1 is the same as a person B existing at a later time T2 if and only if B's mental state can be explained by a series of previous mental states leading back to A. This is of course not a rigorous formulation, and cannot handle all objections, but you get the idea. At any rate, the mental state of person B is connected to the mental state of person A in some relevant way, which leads us to assert that they are the same person.
Now, Locke is committed to a theory of what is called relative identity (see van Inwagen's brilliant paper "And Yet There are Not Three Gods, But One" in the collection Philosophy and the Christian Faith, ed. Thomas V. Morris for a rigorous account of relative identity). Relative identity is the claim that in at least some cases we cannot answer the question of whether A is the same as B without first asking, "The same what?" In particular, Locke thinks that although one being might be simultaneously a material object, a human, and a person, the question of whether it is the same material object, the same human, and the same person may in some conceivable cases have different answers. In the everyday case, it will not be the same material object over time because of the constant intake of food, elimination of waste, sloughing of dead skin cells, etc., that cause it not to be made up of the same matter. It will be the same human over time because it has continuity of organization as a living organism. It will be the same person over time because is has psychological continuity in the form of memory. What is curious about Locke's account is he envisions the case where two entities are compared and turn out to be the same person while NOT being the same substance. This would occur, for instance, if two souls (assuming, for the moment, that souls exist) were to swap memories.
The core of the strangeness of this account comes, I believe, from the fact that we are comparing things that we think of as being substances (persons) based on events that occur within them (mental states). We wouldn't normally do this. Now, perhaps we want to say that the mental state is the collection of properties of the person, and the person's having those properties is an event. This is where van Inwagen's attack comes in, since the materialist (allegedly) cannot say that the person is distinct from his body. Now, I'm not familiar with the particular accounts of materialism van Inwagen attacks here, but it seems to me that the materialist is only committed to saying that the person does not exist as a substance independent of the body. Van Inwagen presents a further attack by pointing out that if the person is identical with the physical body, then whatever is predicated of the person can be predicated of the body. However, if it is possible for persons to switch bodies, this will lead to the breakdown of the transitivity of the (absolute) equality relation. By way of illustration (an illustration different than van Inwagen's) let NBx (where x is a numeric subscript) be the new body at time x, and Px be the person at time x. The transfer occurs at time 2. We have:
NB1 != P1
NB3 = P3
P3 = P1
NB3 = NB1
I believe that both of these objections can be escaped, on a reasonable definition of materialism. That is, it may be reasonable to define materialism as the view that only material substances exist, but it is also reasonable to define it as the view that there are no substances other than material ones. This difference is significant, because many ontologies posit that substances are not all that are "real" - events are also real in an important sense. If persons are viewed as complex events consisting of series of psychological states, then they can exist without being identical with any physical objects; they simply need physical objects to "inhere" in. We would say, then, that each mental state is a "time-slice" of a person, a simple event consisting in (according to the materialist) some brain having certain properties. The person is a complex event consisting of a collection of such simple events which are related to each other in some relevant way, such that we say that any one mental event in the collection is "continuous" with all the others. These events need not inhere in the same material object. This could be defined rigorously by making explicit what types of properties are relevant to psychological continuity.
For the record, I do believe this theory of persons and events, although I am not a materialist. I think that it is the best account of what it is we mean by the word person; for the mental states that we speak of in terms of personhood are clearly events and, as Locke's arguments show, what substance they happen in doesn't seem to be relevant. This possibility is completely absent from van Inwagen's paper, and I'm really not entirely sure why.