Williams on Personal Identity In "The Self and the Future

     1. Bernard Williams attempts to show that the thought experiments, tracing back to Locke, that purport to show that "body-swapping" is conceptually possible do not support this thesis as well as they might at first seem to. Williams intends to do this by presenting a second thought experiment, which, although identical to the first (Lockean) thought experiment in all its features except personal identity, seems to lead us to the opposite conclusion, that radical difference in character traits and quasi-memories is not sufficient for there being two different persons.     Furthermore, Williams claims, in spite of the fact that the second thought experiment is presented in question-begging terms, it actually supports its conclusion better than the first thought experiment supports its conclusion, since it rests on a certain "straightforward principle," "that one's fears can extend to future pain whatever psychological changes precede it" (198). After presenting the two thought experiments Williams utilizes, I will present some considerations that suggest that the second thought experiment supports its conclusion at best no better than does the first one, and that this "straightforward principle" is not as self-evident as Williams takes it to be.

     Williams begins by presenting what would seem to be a straightforward case in which we would describe the outcome as one of "body-swapping." Imagine, he asks, that two persons, A and B, are to undergo a process after which the person with the A-body (whoever that may be) will have the quasi-memories and character traits that B had before the process, and the B-body person (whoever that may be) will have the quasi-memories and character traits that A had before the process. After the process, we may ask the A-body person who she is, and she will say that she is B. We may also ask the B-body person who she is, and she will say that she is A. The A-body person exhibits the character traits, interests, and knowledge that B had before the process, and vice-versa. Given this description of the process, it seems quite intuitive to say that A and B have changed bodies: after the process, the A-body person is B, and the B-body person is A.

     Williams then asks us to imagine that A and B are told before the process that one of the two persons that emerge from the process will be horribly tortured, and the other one will be given $100,000; and that before the process we ask both A and B to choose, on selfish grounds, whether the A-body person or the B-body person will receive the torture after the process (the other one will receive the $100,000). First suppose that A chooses that the A-body person will receive the torture and the B-Body person the $100,000, and that B chooses that the B-body person receive the torture and the A-body person the $100,000. (This would indicate that both A and B regard "swapping bodies" as the proper description of the outcome of the process.) Since we cannot do both what A has chosen and what B has chosen, we choose to allocate the torture and the money in accord with A's expressed preference. After the process, the B-body person will be quite pleased at receiving the $100,000, and, since the B-body person has A's memories, will feel that she got what she chose. The A-body person, on the other hand, will feel quite upset that she is getting the torture, and will believe, since she has B's memories, that this is not the outcome she chose. If it really is the case that the A-body person is B, and the B-body person is A, then we should say that both A and B chose wisely (or rationally).

         Suppose now that A chooses that the A-body person gets the $100,000 and the B-body person the torture, and that B chooses that the B-body person gets the $100,000 and the A-body person the torture. (This would indicate that they do not believe that "body swapping" is a correct description of the outcome of the process.) Since we cannot act in accord with the expressed wishes of both A and B, we choose to act in accord with B's expressed preference, giving the torture to the A-body person and the $100,000 to the B-body person After the process, the A-body person will of course be quite upset at the allocation of the torture and the money. The A-body person, since she has B's memories, will believe that this is what she chose (that the A-body person receive the torture), but will feel that she chose unwisely. The B-body person, on the other hand, will of course be pleased at getting the $100,000, and will feel lucky, since she did not get what she believes she so unwisely chose. And if "body swapping" is the correct description of the outcome of the process, then we should say that both A and B chose unwisely (irrationally).

       So given our initial intuition that after the process the A-body person is B and the B-body person is A, we should say that it would be wise, or rational, for A to choose (on selfish grounds) that the A-body person be tortured and the B-body person receive the $100,000. This is reinforced by our consideration that in the first scenario above, in which A chooses that the A-body person be tortured and B that the B-body person be tortured, both the A-body and the B-body person would say ,after the process, "I chose wisely"; and that in the second scenario, in which A chooses that the B-body person be tortured and B chooses that the A-body person be tortured, both the A-body person and the B-body person would say, after the process, "I chose wisely." It seems as though A has nothing to fear (for herself) if the A-body person is tortured after the experiment. On the contrary, it seems that A may well look forward to getting the $100,000.

Having described this first thought experiment and noted that we would intuitively describe this situation as one of body swapping, Williams then takes us through another thought experiment, designed to lead us to an incompatible conclusion. Suppose, Williams asks, that you are told that you will be tortured tomorrow. Unless one has rather strong inclination toward Stoicism, fear would seem to be the appropriate attitude to take toward such a revelation: that is to say, it would be rational to fear the upcoming torture. You are then told that when the torture comes, you will no longer remember that you were told of it beforehand, because a powerful amnesiac will be administered between now and the torture. This would surely not comfort you very much. Furthermore, you are told that not only will you not remember that you are to be tortured, but you will also not remember many of the things which you now remember.

This would not alleviate your fear in the least, claims Williams, for you can certainly imagine waking up, not remembering who you are, where you are, or why you are there, and finding oneself undergoing great torture. If anything, this would increase your fear; and more importantly, it would still be rational to fear the torture. Then you are told that not only will you not remember many of the things which you now remember, you will have a completely different set of apparent memories, as well as a new set of character traits, identical to those of another person now alive. In fact, those apparent memories and traits will be read right off of this other person's brain. Would this somehow serve to comfort you in the face of the upcoming torture, to alleviate your rational fear of the horrible things which will happen tomorrow? Williams claims that it would not. It would still be rational to fear the upcoming torture.

If this assessment of Williams' second thought experiment is correct, then something is amiss. For the situation in which you are described as being in the second thought experiment 2 is exactly the situation which A is in prior to the process in the first thought experiment. The only detail which the informant (the person describing what is going to happen) has left out in the second situation is that the person whose quasi-memories and character traits you will be given will in turn be given your quasi-memories and character traits (and $100,000 on top of that, which would seem to be adding insult to injury). "But," Williams asks, "why should he mention this man and what is going to happen to him? My selfish concern is to be told what is going to happen to me, and now I know: torture, preceded by changes of character, brain operations, changes in impressions of the past" (189). The second thought experiment suggests that the rational, self-interested choice for A to make in the first thought experiment would be to choose that the A-body person receive the $100,000 and the B-body person the torture. And it is rational for A to make this choice only if A has good reason to think that the A-body person will be A. Therefore, it is rational for A to think that the A-body person will be A, contrary to the conclusions of traditional thought experiments, such as the first one described.

This conclusion rests on two crucial assumptions. The first assumption is the principle that the properties of the B-body person 3 are not relevant to the determination of the identity of the A-body person. (This principle should, of course, be stated in a more general fashion.) Suppose that rather than putting A's memories and character traits into the B-body in the first thought experiment, the ones running the process merely put B's memories and character traits into the A-body, leaving B as she was before the process. Who are we to say that A-body person is after the process? The A-body person cannot, Williams claims, be B, for the B-body person is still B. And if we say that the A-body person is A after this process, why shouldn't we say that the A-body person is A in the original thought experiment, for the only difference between this scenario and the original thought experiment are properties of the B-body person? For by the principle stated at the beginning of this paragraph, properties of the B-body person are irrelevant to determination of whether the A-body person is A. I am not concerned with disputing this principle in this paper, although it certainly deserves some examination.

The second principle on which Williams' conclusion rests is the principle mentioned in the opening paragraph of this paper: that one's (rational, self-interested) fears can extend to future pain regardless of the psychological changes which precede it.

My parenthetical addition of "rational" to this principle, which Williams himself never makes, is essential, for without it, Williams cannot draw the conclusion which he wishes to draw. For a person might have irrational, self-interested fear of some event in the future which would, intuitively, not happen to her. Suppose that I believe (perhaps based on my studies of the prophecies of Nostradamus) that there will be a hideous nuclear holocaust one hundred years from now, and I fear this event. Someone might, after having failed to persuade me that this belief is not well-supported, try to comfort me by pointing out that I will not be alive in one hundred years to experience the horror of nuclear holocaust. I, however, happen to also believe in reincarnation, and I hold that at the moment this body dies, I will re-enter the world in another body. There is thus no way, I believe, that I could possibly escape being around during the holocaust. Now my self-interested fear can, in Williams' words, "reach through" (191) to this future event. And this indicates that I believe that I will be around in one hundred years. However, unless this is rational fear, there is no evidence for reincarnation. One's irrational, self-interested fear may "reach through" to any event ("Don't chop that old oak tree down after I die: my soul will enter it"), but this does not provide any evidence that one will be around to experience the event. And the question of the rationality of the self-interested fear of torture in the second thought experiment is exactly what is in question. I suppose that I am merely here pitting my intuitions against Williams', but if I were in the situation described in the second thought experiment, I might be at a bit of a loss as to whether I should fear the upcoming torture, any more than I should fear torture that will happen to some body other than my present one, after this one is dead.

My second point about this principle concerns what Williams claims is the conceivability in the second thought experiment (thinking of it in first-person terms) of the proposition 'The torture will happen to me.' During his setting out of the second thought experiment, Williams gives his reaction to be told that before the torture he will have severe amnesia induced:

This does not cheer me up, either, since I can readily conceive of being involved in an accident, as a result of which I wake up in also in great pain; that could certainly happen to me, I should not like it to happen to me, nor to know that it was going to happen to me. (186)And when he is told that not only will he forget everything about his past, he will also have other apparent memories and character traits implanted in his body, Williams says:

         I do not think that this would cheer me up, either. For I can at least conceive the possibility, if not the concrete reality, of going completely mad, and thinking perhaps that I am George IV or somebody; and being told that something like that was going to happen to me would have no tendency to reduce the terror of being told authoritatively that I was going to be tortured, but would merely compound the horror. (ibid.) Now it is important to distinguish between my conceiving of being in such-and-such a state, and conceiving that I am in such a state, so to speak. The first type of conceiving is imagining, to use Nagel's phrase, what it would be like to be something. I may well imagine what it would be like to be Napoleon, to feel the thrill of crowning myself emperor of France, the sense of loss as the empire collapsed after the defeat at Waterloo, and the humiliation of dying in exile. In this sense, I may conceive of being Napoleon. However, given that I am not Napoleon, I cannot conceive of the following state of affairs: I (Brock Sides) am Napoleon. This is very similar to conceivability of other statements involving identity. I can imagine finding out that Samuel Clemens was not Mark Twain, but merely someone who said that he was, in order to become famous for writing such terrific stories and novels. Given that Samuel Clemens is identical with Mark Twain, I cannot conceive of a state of affairs such that Samuel Clemens is not Mark Twain.

        And it is only this latter sort of conceivability that is, as it were, closely linked with metaphysical possibility. If the A-body person after the process is not A, then there is no way for A to conceive of the following state of affairs (from a first-person perspective): the torture of the A-body person is torture of me. This is not to say that A cannot conceive of what it would be like to be tortured, having such-and-such a set of memories, but merely that A is not conceiving of this in a way that entails the metaphysical possibility that the A-body person is A. The third point of trouble with this supposedly straightforward principle is that Williams' discounts the possibility that part of A's fear of the upcoming torture is, in part, due to epistemic uncertainty about whether the A-body person, who will receive the torture, is in fact A. Williams' imagines the following case: I "have good reason to expect that one out of us five is going to get hurt, but no reason to expect it to be me rather than one of the others." Certainly in this case one has the attitude of fear, and rational fear, toward an upcoming event, but Williams holds that such uncertainty is not present in the second thought experiment:

It is important that the expectation of S [the torture] is not indeterminate in any of the ways we have just been considering. . . . It is [not] like the expectation of the man who expects one of the five to be hurt; his [A's] fear was indeed equivocal, but its focus, and that of the expectation, was that when S came about, it would certainly come about in one way or the other." (194-5) I see no reason to say that A's fear should not be tempered with exactly the same kind of uncertainty found in the case of a person who knows that she is a member of a group of five, one of whom will be hurt. Admittedly, the question of whether the A-body person is A is a metaphysical question, one which is necessarily true or false, but this does not mean that A will not be in the same sort of epistemic uncertainty about the future torture: A is not sure whether the person who will be tortured is her. For it has been widely accepted since Kripke pointed this out, that certain necessarily true propositions may be only discovered a posteriori, by use of empirical evidence 4 (the classic example of this is the necessary truth that Hesperus is Phosphorus). I do not have any good argument that the conclusion Williams draws from the second thought experiment is outright wrong, but I think the considerations I have given show that the second thought experiment, due to the weakness of Williams' principle that one's fears can reach through to future pain regardless of psychological changes which precede it, does not support Williams' conclusion as well as he seems to think it does. At the very most, it seems that the second thought experiment supports Williams conclusion no better than the first thought experiment supports the conclusion that body-swapping is indeed metaphysically possible.

1. In Personal Identity, ed. John Perry. Berkelely, CA: University of California Press, 1975, pp. 179-98.

2. Williams himself describes the second thought experiment in the first person: "Someone in whose power I am tells me that I am going to be tortured tomorrow" (185, emphasis added). I feel that describing it in the second person acheives Williams' goal better, i. e. to get the reader to view the thought experiment in first-person rather than third-person terms.

3. What counts as a "property" for the purposes of this principle will have to limited in some way, in order to rule out properties of the B-body person such as being such that the A-body person is A.

4. I do not wish to suggest that empirical evidence will is necessarily the correct court of appeals for deciding whether the A-body person is indeed A. I'm not at all sure how one would would go about determining this, but I think the uncertainty involved is not relevantly different from the uncertainty involved in an empirical matter.