A simple thought experiment, originating with Locke and repeated in numerous science fiction stories since, seems to show that the persistence criteria for persons is not identical with persistence criteria for human bodies. Locke asks us to imagine that the bodies of a prince and a cobbler swap souls at time t, so that after t, the person inhabiting the body of the prince has the memories, knowledge, and personality of that the person in the body of the cobbler had prior to t, and that the person in the body of the cobbler has the knowledge, memories, and personality that the person in the body of the prince had prior to t. Intuitively, the person in the cobbler's body prior to t = the person in the prince's body after t, and the person in the prince's body prior to t = the person in the cobbler's body after t. This, Locke claims, shows that the criteria for the persistence of persons are not the same as the criteria for the persistence of masses of matter (the sum of the atoms that make up the two bodies), nor the same as the criteria for the persistence of living bodies (which may, like persons, undergo changes in their material composition over time). A single person may be comprised, over time, of two different substances, and a single substance may comprise two different persons over time.
The fact that Locke describes this in terms of soul-swapping is irrelevant to the thought experiment. Locke claims that it is at least conceivable that the manifest changes in memories, knowledge, and personality at time t happen without any exchange of the immaterial thinking substance "attached to each body." (Locke concedes that it might be, as we would now put it, nomically impossible for such changes to occur without swapping of immaterial substance. But what concerns us here is a broader sense of possibility, what we might call metaphysical possibility.) To avoid this irrelevant issue in the thought experiment, we should alter it so that the hypothetical prince and cobbler undergo at time a process that Shoemaker calls a Brain State Transfer, or BST. Without any exchange of material or immaterial stuff, the memory, personality and knowledge of the prince and cobbler are read off their brains (or immaterial souls, if indeed that's the part of them that "houses" their mental life), and their brains (or immaterial souls) are respectively imbued with the memory, personality, and knowledge of the other.
Locke proposes the following as the criterion of persistence of persons over time. A person at time t1 (a "person-slice") is the same person as a person at a later time t2 iff the person at t2 is "conscious of", i.e. remembers, the experiences of the person at t1. Since it is at least conceivable, or metaphysically possible, that a person in body A at t2 remember the experiences of the person in body B at t1, the criteria for persistence of persons are not the same as the criteria for the identity of living human bodies.
Criticism of this theory of personal persistence immediately cropped up in works of Butler and Reid. Butler charges that the criterion proposed by Locke is implicitly circular, for the concept of memory presupposes the concept of personal persistence. It is a conceptually necessary condition of one mental state m2 being a memory of another mental state m1 that m1 and m2 belong to the same person. If I suddenly came to have what appears to me to be a memory of witnessing the Kennedy assassination, which occurred seven years before I came into existence, this would not be a genuine memory, but a delusion.
This defect in Locke's theory may be easily fixed by appeal to a technical concept that Parfit calls "quasi-memory": mental state m2 is a quasi-memory of m1 iff (1) m1 seems, to the person who has it, to be a memory of such and such an experience, (2) someone did in fact have such an experience, and (3) m2 is caused by state m1. We may then restate Locke's theory without circularity: A person at time t1 (a "person-slice") is the same person as a person at a later time t2 iff the person at t2 quasi-remembers the experiences of the person at t1.
Although this revised theory is no longer circular, it clearly will not do. Suppose I came at time t to have a quasi-memory of the experience shooting President Kennedy, caused by Lee Harvey Oswald's experience of shooting President Kennedy. This by itself does not seem to make me the same person as Lee Harvey Oswald on that fateful day. At the very least, a later person-stage should be, in Parfit's words, "strongly connected" in order to qualify. It should be the case that *many* or even *most* of the experiences of the earlier stage be quasi-remembered by the later stage. (This raises a counting problem, for it is not obvious how to divide the sum of a person's experience at a time into "individual experiences", but I will herein ignore this problem.) With this emendation, we might try to state the memory theory of personal persistence as follows: A person at time t1 is the same person as a person at a later time t2 iff the person at t2 quasi-remembers many of the experiences of the person at t1.
But this theory seems too strong. By any reasonable standard of counting, I don't remember many of the experiences that I had at a time exactly a week before this one. This fact points to a deeper problem. The relation of strong-connectedness of quasi-memory is not a transitive relation, whereas the relation "is the same person as" is supposed to be a transitive one. This was first pointed out by Reid, in his famous "brave officer" counter-example to Locke's theory. "Suppose a brave officer to have been flogged as a boy for robbing an orchard, to have taken a standard from the enemy in his fist campaign, and to have been made a general in advanced life; suppose, also, which must be admitted to be possible, that, when he took the standard, he was conscious of [remembered] his having been flogged at school, and that, when made a general, he was conscious of [remembered] his taking the standard, but had absolutely lost the consciousness [memory] of his flogging." In this case, the revised memory theory yields the result that the general is the same person as the person who took the standard, and the person who took the standard is the same person as the boy who was flogged, but the general is not the same person as the boy who was flogged.
In light of this example, the memory theory must clearly be revised. The revision needed is an appeal to an ancestral relation of the strong connectedness of quasi-memory used in the most recent version. Here goes: A person-stage at time t0 is a stage of the same person as a person-stage at time tn iff there are person stages t1...tn-1, s.t. t1 quasi-remembers many of the experiences of t0, ... and tn quasi-remembers many of the experiences of tn-1.
But there are further sorts of counter-examples involving transitivity that this theory cannot avoid. Suppose that at time t, the brain states of the person in body A are transferred into two distinct bodies, B and C. (This is Parfit's "branch-line case.") According to the current version of the memory under consideration, the person in body B after t is the same person as the person in body A prior to t; the person in body A prior to t is the same person as the person in body C after t; but the person in body B after t is not the same person as the person in body C after t.
If this problem is to be avoided, the theory will have to be revised so that if there is more than one person stage at t2 appropriately related to a person-stage at t1, the "best candidate" at t2 is selected as co-personal with the stage at t1, or if there is a tie for best candidate, no stage at t2 is co-personal with the stage at t1. (An arbitrary choice would presumably be intolerable.)
Williams has proposed a thought experiment to undermine this sort of theory. (Given that this is the only sort of memory theory that will avoid the previous problems, it undermines the memory theory altogether.) Williams asks you to imagine that someone informs you that at a future time t, you will be tortured. The appropriate reaction, is, of course, fear. But then you are told that prior to t, all your memories will be erased. This, Williams thinks, should do nothing to quell your fears of torture. You are then also told that in addition to having your memories erased, they will be replaced with some false memories. Still, Williams thinks, this will not comfort you. Then you are told that these false memories will be implanted via a BST from another person (i.e. these mental states will be quasi-memories of another person's experiences). Feeling less worried about the upcoming torture? Williams thinks that you should not be. At this point, you are also told that your memories will be similarly transferred into the body of the person from whom you will receive the false memories. But what difference could this make? You should still fear the upcoming torture.
But the situation described at the end of this thought experiment is analogous to the original Lockean thought-experiment that seemed to support the possibility of body-swapping (same person in two different bodies). Given that it is rational to fear the upcoming torture, even when all the details about the preceding BST procedure are given you, and given that it is rational to fear torture only if it is going to happen to you, it must be the case that the person in the body to be tortured is you.
Locke, John. "Of Identity and Diversity," in Perry, op. cit.
Parfit, Derek. Reasons and Persons. Oxford, 1986.
Reid, Thomas. "Of Identity," in Perry, op. cit.
Williams, Bernard. "The Self and the Future," Philosophical Review 79 (1970). Reprinted in Perry, op. cit.
Back to Brock's Philosophy Page