In Descartes’ famous slogan, “I think therefore I am,” what does the “I” refer to? For Descartes it definitely does not refer to my body; rather, it refers to my mind, the mental substance that constitutes the essential me. We have now seen good reason to suppose that Cartesian dualism is not a philosophically acceptable account of the nature of the mind. But for those of us who reject dualism there is still a serious question left over: What exactly is the self? What fact about me makes me me? Many contemporary philosophers, including myself until fairly recently, think that Hume had more or less the last word on this issue. In addition to the sequence of experiences, and the body in which these experiences occur, there is no such thing as the self. Hume says, when I turn my attention inward and try to discover some entity that constitutes the essential me, all I discover are particular experiences; there is no such thing as the self in addition to these experiences.
There are several more-or-less separate questions about the self, and I will now distinguish, for the purposes of this chapter, three different families.
A persistent traditional question in philosophy has been, What fact about a person makes that person the same person through the various changes that he or she undergoes in the course of a lifetime? In my own case, for example, I have undergone a rather large number of changes over the past decades. My body looks somewhat different, I have learned some new things and forgotten some old ones, my abilities and tastes have changed in various ways, but all the same there is no question that I remain exactly the same person through these changes. I am identical with the person who bore my name and lived in my house decades ago. But what fact about the sequence of events and changes that I have been describing makes it the case that they are all events in the life of one and the same person?
In addition to the sequence of psychological events that constitute perception, action, reflection, and so forth, and the body in which these psychological events occur, do we have to postulate something in addition to the body and its sequence of psychological events?
I have not stated this question very precisely and I will try to make it more precise later. The point for present purposes is to pose a general question, In addition to the sequence of my actual thoughts and feelings and the body in which they occur, do we need to postulate a thing, an entity, an “I” that is the subject of all of these events? Let us suppose that we can all agree, as I have been assuming throughout this book, that I am constituted at least in part by a physical body, and that this body contains a sequence of mental phenomena—conscious states and unconscious brain processes capable of producing conscious states. The question is, Is there anything else we have to postulate? And if so, what is it? As far as I can tell, most contemporary philosophers follow Hume in thinking that we do not have to postulate anything more; but I have been reluctantly forced to the conclusion that we do, and I will explain why in the course of this chapter.
This question is often thought of in contemporary life as a matter of the social, psychological, cultural, and biological forces that shape my particular personality and make me the sort of person that I am. There is, in popular speech, a use of the notion of “identity” in such expressions as “identity politics” or “cultural identity,” which concerns the sources, both cultural and biological, that shape one’s personality. I think that this is a different sense of the notion of personal identity from those expressed in questions 1 and 2. It has more to do with character and personality than it does with the metaphysical problem of the existence and identity of a self across time.
This chapter will be concerned with the family of questions that surround questions 1 and 2 above. We will see that they give us enough difficulty without going into questions of personality.
Questions about identity are as old as philosophy, but there does seem to be a special problem about the identity of persons. Probably the most famous puzzle about identity in the history of the subject is the example of the “Ship of Theseus.” A ship made of wood is entirely rebuilt gradually over a period of time. It continues to function, it has a crew that sails it around the Mediterranean, but one by one the boards that constitute it are gradually replaced until finally there is not a single board left from the original construction. Now, is it still the same ship? Well, I think most of us would feel that it is the same ship, that the spatial and temporal continuity of functioning was sufficient to guarantee its identity as a ship, because the notion of a ship is, after all, a functional notion. But now suppose somebody gathers up all the discarded boards and constructs a ship out of them that contains all and only the parts of the ship that was originally launched, so that every single board in the second ship is identical with a board that was in the ship as originally launched. Which is the ship we originally started with? Is it the one that has the continuity of function or is it the one that has the continuity of parts? The mistake in these discussions, as is so often the case in philosophy, is to suppose that there must be some additional fact of the matter about identity beyond all of the facts that I have just told you. It seems to me there isn’t any further fact of the matter. It is up to us to say which is the original ship. This might be a matter of some importance, for example, who owns which ship? Who is responsible for the taxes? Which ship has docking rights? But there is no additional factual question left over as to which ship is identical with the original ship beyond all of the facts that I just told you.
Some of the questions about personal identity are like the example of the Ship of Theseus, but in the case of personal identity, we feel there is a special problem that is not present in these traditional examples. We tend to feel that each of us is presented to himself or herself in a special way, and that these first-person experiences are essential to our identity in a way that the third-person phenomena are more or less incidental. We feel, for example, that we all understand what it would mean to say that we might wake up one morning to find ourselves in a different body. Like Gregor Samsa in the story by Franz Kafka, our physical external appearance would have changed totally, yet we feel somehow that we would know, even if no one else could be convinced, that we were the same person who before occupied a different body. To make this example seem more concrete, let us suppose that brain transplants become possible and that my brain is transplanted into the body of Jones and Jones’s brain is transplanted into my body. It seems to me there is no question from my point of view that I will now think that I am exactly the same person as before, but my brain (and hence I) now occupy a different body. I might have trouble convincing other people of this, but we feel, at least from the first-person point of view, there would be no question that I would think of myself as the same person as the person who once occupied a different body and who now occupies Jones’s body.
A more puzzling case is this: imagine that all of my mental capacities are equally realized on each side of my brain. Now imagine a case of brain bisection where the two halves of my brain are transplanted each into a different body. The original body, we will suppose, is discarded and now there are two halves of my brain in two different bodies. Which of the resulting characters, if I may so describe them, is me? This case seems to me like the Ship of Theseus in that there is no fact of the matter beyond what I have just told you. That is, it seems to me that we have equal reason for saying that I am number one or that I am number two or, I think what we would be more likely to say is that there are now two people where there was previously only one. This case is like fission cases, where one amoeba splits into two. Yet even in this case, from the first-person point of view, one feels there must be a fact of the matter. If I now am one of the offspring of this fission, I am likely to feel, “I am still me, the same unique individual I always was. I don’t care what anybody else says.” The problem is that my twin will have exactly the same conviction with the same justification and we can’t both be right.
It is typical of our concepts that their application to the real world presupposes certain sorts of regularities. This is as much true of the concepts of a ship, or house or tree or car or dog, as it is of such fancy concepts as personal identity. We normally are able to apply the concept of personal identity because the first-person criteria and the third-person criteria tend to come together. They do not come apart in radical ways. But it is easy to imagine sciencefiction worlds in which they come apart radically. Suppose that fusion and fission became common, that is, suppose that it was quite common when several people were walking down the street that suddenly they would coalesce into one body. Or, to take the fission case, imagine that a single person might branch out into five identical people as a result of the fission of her original body. If such cases became common, then we would have very serious problems with our notion of personal identity. It seems to me that it would probably no longer apply.
If we actually look at the criteria that people employ in ordinary speech for deciding which person today is identical with which person in the past we find that there are at least four conditions that constitute our notion of personal identity. Two of these are from the third-person point of view, one is from the first-person point of view, and one is mixed. Let us review them.
My body is continuous in space and time with an infant born seven decades ago. It is this spatio-temporal continuity more than anything else that the public relies on in regarding me as the same person. Notice that the spatio-temporal continuity of my body does not imply the spatio-temporal continuity of the micro parts of which the body is composed. At the molecular level, the parts of my body are constantly being replaced. The molecules that compose my body are now totally different from the molecules with which I began life, but all the same, yes, it is still the same body, in large part because it is spatio-temporally continuous with the original body of the infant.
Though my structure changes over the decades—I grow bigger and I grow older—all the same, I am recognizably a human being. If, like Gregor Samsa, I woke up one morning metamorphosed into the body of a large insect, or if I should suddenly turn into an elephant or a giraffe, it is not at all clear that other people would be willing to say that it is still JRS. So, in addition to the sheer brute continuity of a continuum through space and time, it seems we need also to acknowledge certain sorts of structural regularities in the changes that this spatio-temporal object undergoes.
The reason there is a special problem about personal identity is that these two conditions do not seem to be enough for my first-person point of view. Even if other people refuse to recognize a certain object as me, all the same, I have a confidence that I would know from my insider’s first-person point of view who I was, even if I were in the body of an elephant or a giraffe or even if I had shrunk to the size of my thumb, all the same, I feel confident I would be able to identify myself. But what are these criteria supposed to amount to?
The next criterion is a first-person criterion.
From my inner point of view, it seems that there is a continuous sequence of conscious states bound together by my capacity at any given point to remember conscious experiences occurring in the past. To many philosophers, most famously Locke, it has seemed that this is the essential element of personal identity. The reason we need this in addition to bodily identity is that it seems easy to imagine cases where I might wake up in a different body, but from my point of view there is no question that it would still be me. I still have my experiences as part of the sequence. It includes memory experiences of past conscious states. Locke claimed this was the essential feature of personal identity. He called it “consciousness,” but the standard interpretation is that he meant memory. Hobbes and Hume thought they could refute it by pointing out that the memory relations were intransitive. That is, the old general might remember events that occurred when he was a young lieutenant and the young lieutenant might remember events that occurred in his childhood, but the old general might have forgotten the childhood. They were surely right about this, but the fact that one forgets things does not seem to count seriously against the claim that from the first-person point of view, the sequence of my conscious states, bound together by memory, is essential to my sense of my existence as a specific individual.
This is perhaps less important than the other three, but nonetheless there is a certain relative continuity of my personality and my dispositions. If I woke up tomorrow morning feeling and behaving exactly like Princess Diana shortly before her death then we might wonder if I was “really the same person.” Or to take a case from real life, in the famous clinical example of Phineas Gage, Gage’s brain was damaged when he was working on a railway construction crew and a steel bar went entirely through his skull. Miraculously, Gage survived but his personality was totally changed. Whereas before he had been a cheerful and pleasant person he became mean, suspicious, vicious, and nasty. In some sense we might feel that Gage was “a different person.” Notice, however, that in describing these cases, we continue to use the same proper name as before. For practical purposes there is no question that it is still Phineas Gage that we are discussing. The sense in which he is a different person is not one that we regard as essential for practical purposes, such as figuring out who owes his income tax or owns his house. Still, his friends and family might feel that he is “not the same person.”
As I remarked earlier, it is typical of our concepts that we often have a variety of criteria that enable a concept to function and the tacit background presupposition is that all of these go together. And in the cases we are familiar with in normal life, these do go together. All the same, there are some puzzles that arise.
I have said that memory plays an essential role in our first-person conception of personal identity. Here is how. I now have conscious memories of earlier conscious experiences in my life, and I have a capacity to call up a very large number of other conscious memories of earlier conscious experiences in my life. My sense that I am exactly the same person over time, from my first-person point of view, is in a large part a matter of my ability to produce conscious memories of earlier conscious events in my life.
I think this is what Locke meant when he said that consciousness functions essentially in our conception of personal identity, but regardless of whether or not Locke meant this, continuity of memory is at least an important part of our concept of personal identity. Leibnitz made a similar point: imagine that you become Emperor of China, but that you lose every trace of every kind of memory of your past. There is no difference, says Leibnitz, between imagining this and imagining that you cease to exist and a new Emperor of China comes into existence.
There is a stock objection to Locke’s account, which many people think is decisive and that I now want to state and answer. Here is how it goes. The account is circular. We can truly say of an agent that he can remember events in his earlier life only on the presupposition that he is identical with the person to whom those events occurred in the earlier life. But we cannot therefore explain personal identity in terms of memory, because the memory in question presupposes the very identity that we are trying to explain. We can put this more formally as follows:
A person P2 at time T2 is identical to an earlier person P1 at time T1 if and only if P2 at T2 remembers events occurring to P1 at T1, where the events in question are conscious experiences and the experience of remembering is itself a conscious experience.
The claim that this is circular is justified as follows: in order that P2 at T2 should really remember an event occurring to P1 at T1, as opposed to just thinking that he remembers it, P2 has to be identical with P1. But if that is true, then we cannot use the memory to justify the claim of identity or the criterion of identity, because we require identity as a necessary condition on the validity of the memory.
We can illustrate these points with examples. Suppose I now say, sincerely, I remember writing The Critique of Pure Reason. This does not in any way establish or tend to support the view that I am identical with Immanuel Kant, because we know that I could not have written The Critique of Pure Reason, since I am not identical with Immanuel Kant, and he wrote The Critique of Pure Reason. But by exactly the same token, if I say I now remember writing Speech Acts, this does not by itself go any way toward establishing that I am identical with John Searle, the author of Speech Acts, because we would have to know that I am John Searle before we could know that I correctly remember writing Speech Acts. The two cases are exactly parallel. Is this argument decisive against the theory that memory is an essential part of personal identity? I think the answer depends on which question we take the theory as trying to answer. If we take it as answering the question, What are the criteria of personal identity such that if those criteria are satisfied, then person P2 at T2 would be identical with person P1 at an earlier time T1? then the criterion fails. No matter how many of Kant’s putative memories I have, I am still not Kant. However, there is a different question, which it seems to me this theory answers, and that is the first-person question: What is it about me, about my personal experiences, that makes me sense myself as a continuing entity through time, which is in addition to the continuity of my body? And to this question, it seems to me that the continuity of my memory experiences is an essential part of my sense of myself as a continuing self. Someone who is not me might have type-identical personal experiences that give him a sense of himself which is type identical to my sense of myself. All the same, we are not identical, and yet each of us has a sense of himself as a continuing self.
All of these discussions leave open the question of whether or not we need the notion of a self in addition to the notion of particular psychological states and dispositions at all. I think most philosophers agree with Hume in his criticisms of both Locke and Descartes that there is no self or personal identity beyond the sequence of our actual experiences. Hume’s skepticism about the self is like his skepticism about necessary connection and causation. He looks around to see if he can find some unifying impression that unites all of his various perceptions together and, not surprisingly, he fails to find any such unifying impression. When I turn my attention inward, he tells us, what I find are specific experiences. I find this or that desire for a drink of water, or a slight headache, or feeling of the pressure of the shoes against my feet, but there is no experience of the self in addition to these particular experiences. Consequently, any identity that I might attribute to myself must be a result of the sequence of particular experiences. It is an illusion, Hume tells us, to suppose that there is something over and above the specific experiences that constitute my self. As with necessary connection, Hume talks as if it were some lamentable failure on our part that we fail to find the experience of the self, just as we fail to find the experience of necessary connection. But, as in the earlier case, Hume is making a logical point, not a psychological point about the absence of a certain kind of experience. The point is, nothing could count as an experience of the self, because any experience we have, even an experience that lasted an entire lifetime, would simply be just another experience. Suppose I had a continuous yellow spot in my visual field that was with me my entire conscious life, always present. Would that be a self? No, it would just be a yellow spot. Nothing could satisfy the conditions necessary for something to be an experience of the self, that is, an experience that bound all of our other experience together. I think that Hume’s arguments at the level they are directed at are quite convincing; and I believe that many, perhaps most, philosophers agree with me about the power of Hume’s argument.
But I have reluctantly come to the conclusion that Hume left something out; and this leads to our second set of questions: Do we need to postulate something in addition to our bodies and the sequence of our experiences? I have come to the conclusion that, yes, we absolutely must postulate a self in addition to the sequence of experiences and I will now give you an argument for this postulation.
Let us go back to our original supposition that I consist of a body and a sequence of experiences. This sequence will include such things as the taste of coffee, the sight of the color red, the view of the San Francisco Bay from my window, etc. Is anything left out? I think there is. The first thing to notice is something I have remarked on earlier. We do not just have disordered experiences; rather, all of the experiences I have at any instant are experienced as part of a single, unified conscious field. Furthermore, the continuation of that conscious field throughout time is experienced by the possessor of that conscious field as a continuation of his or her own consciousness. That is, I do not experience my consciousness of five minutes ago or even five years ago as disconnected from my present consciousness; rather, I have the experience of a continuous consciousness interrupted by phases of sleep. (It is a fascinating fact, insufficiently appreciated in philosophical discussions, that one maintains a sense of the passage of time even during sleep, in at least this sense: when one wakes up, one has a sense of greater or lesser time having passed while one was asleep. This apparently is not true of people who have been knocked unconscious or have had a general anesthetic.)
The arguments that convinced me that we need to postulate at least a formal notion of the self (and I will say later what I mean by “formal”) have to do with the notions of rationality, free choice, decision making, and reasons for action. We noticed in chapter 7 that intentionalistic explanations of rational human decision making and acting have a peculiar logical form that differs from the standard form of causal explanations. The contrast is between saying, for example:
1. I made an X on the ballot paper because I wanted to vote for Bush and
2. I got a stomachache because I wanted to vote for Bush.
Now we will suppose, for the sake of the argument, that both of these are true, and that both give adequate explanations. All the same, their logical form is quite different. On a standard interpretation, number 2 states causally sufficient conditions. In that context, my desire to vote for Bush was sufficient to produce in me a stomachache. But on a standard interpretation, number 1 does not state causally sufficient conditions. Yes, I did make an X on the ballot paper for that reason, but all the same, I might not have. I might have decided not to vote for Bush after all, or to leave the room, or do any number of other things. But now we seem to have a puzzle. How can the explanation of my behavior in terms of reasons be an adequate explanation if it does not give causally sufficient conditions? Without such conditions, it does not explain why I did what I did, rather than any number of other things that I could equally well have done, all other conditions remaining the same. It seems that if the explanation does not state causally sufficient conditions, then it does not adequately explain the phenomenon that it was supposed to explain. But the decisive answer to that point is that the explanation is perfectly adequate from my point of view. It is my behavior that I am explaining, and I can explain why I did what I did by giving my reasons for doing what I did, without in any way being committed to the view that the reasons state causally sufficient conditions. Indeed, I may be perfectly well aware that they do not state causally sufficient conditions.
But how, then, are we to interpret statements of form 1, indeed, how are we to interpret any statement that gives an explanation of my free voluntary behavior by giving my reasons for acting? And the answer, I believe, is that we have to suppose that in addition to the “bundle of perceptions,” as described by Hume, there are certain formal constraints on the entity that makes the decisions and carries out the actions. We have to postulate a rational self or agent that is capable of acting freely and capable of assuming responsibility for actions. It is the complex of the notions of free action, explanation, responsibility, and reason that give us the motivation for postulating something in addition to the sequence of experiences and the body in which they occur. To be more precise, in order to account for free, rational actions, we have to suppose there is a single entity X such that X is conscious (with all that consciousness implies), X persists through time, X formulates and reflects on reasons for action under the constraints of rationality, X is capable of deciding, initiating, and carrying out actions under the presupposition of freedom, and (already implicit in what I have said), X is responsible for at least some of its actions.
Hume thought he had a decisive objection against any such postulation. I have no experience of this self, this X. If I turn my attention inward and examine all the experiences I am now having, none of them would I call my “self.” I feel the shirt on my back, the aftertaste of coffee in my mouth, a slight hangover headache from last night, and the sight of the trees outside my window, but none of these is a self, and none of them would count as a self. So what then is this self? I think Hume is absolutely right; there is no experience of this entity, but that does not mean that we do not have to postulate some such entity or formal principle, and I will now explore further what sorts of reasons compel us to that and what sort of entity the self in question might be.
One way to think of these issues is to think of them as engineering problems. If you were designing a conscious robot, and you wanted a robot that would duplicate the full range of human rational capacities, that is, it would be able to reflect on reasons for action, make decisions, and act under the presupposition of its own freedom, then what would you have to put into the robot?
The first and obvious requirement of any such robot would have to be that it is conscious. Furthermore, the form of its consciousness would have to be cognitive, in the sense that it would have to take in perceptual inputs, consciously process the information derived from perception, and reason on the basis of that information toward action.
The second feature that it would have to have would be the capacity to initiate action, a capacity sometimes called “agency.” This is a capacity additional to conscious perceptions. It is a peculiar capacity that humans and many animals have. It is a feature of certain sorts of consciousness, but not of all. The third step is, I believe, the crucial one. The conscious rational agent that we have created must be able to engage in something that in English we call acting on reasons. Now, this is important because the notion of acting on a reason is different from the notion of having something happen to one causally. That was the point of the illustration I gave earlier about the difference between the claim that I got a stomachache because I wanted to vote for Bush, and the claim that I performed a free action, I acted on my desire to vote for Bush. The notion of “acting on” presupposes the gap of free will that I have described earlier. So far then, we have put into our robot consciousness, with conscious perceptual experiences and other intentional states, the capacity to reflect on its intentional states and rational agency, which is the peculiar capacity to undertake actions on the presupposition of freedom. But if we have done that much, we already have a self. The self as I am describing it is a purely formal notion; it does not involve having a particular type of reason or a particular type of perception. Rather, it is a formal notion involving the capacity to organize its intentionality under constraints of rationality in such a way as to undertake voluntary, intentional actions, where the reasons are not causally sufficient to fix the action.
Why is such a notion of the self “formal” rather than “substantive”? To answer that question, I want to draw an analogy between the self and another formal notion. In order to understand my visual perceptions, I have to understand them as occurring from a point of view, but the point of view itself is not something that I see or otherwise perceive. The point of view is a purely formal requirement necessary to render intelligible the character of my experiences. The point of view itself has no substantive features other than this one formal constraint, namely, it has to be that point from which my experiences take place. Now, similarly, the notion of a self that I am postulating is a purely formal notion, but it is more complex. It has to be an entity, such that one and the same entity has consciousness, perception, rationality, the capacity to engage in action, and the capacity to organize perceptions and reasons, so as to perform voluntary actions on the presupposition of freedom. If you have got all of that, you have a self.
Now we can account for a whole lot of other features, of which two in particular are central for our notion of the human self. One is responsibility. When I engage in actions I undertake responsibility, and thus such questions as desert, blame, reward, justice, praise, and condemnation make a kind of sense that they would not make otherwise. Second, we are now able to account for the peculiar relations that rational animals have toward time. I can organize time, I can plan for the future, because one and the same self that makes the plans will exist in the future to execute those plans.
In this chapter I have been mostly concerned with two issues, first the criteria of personal identity, or in other words, what fact about a person makes that person the same person across time and change. And second, I have tried to provide an argument to the effect that though Hume was right that there is no self as the object of our experiences, nonetheless there is a formal or logical requirement that we postulate a self as something in addition to the experiences in order that we can make sense of the character of our experiences. As far as the argument goes, I am not dissatisfied with it. But I am very dissatisfied by the fact that it does not seem to me to go far enough, and I do not really know how to complete it. I have two related worries. First, the underlying difficulty with Hume was his atomistic conception of experience. He thought that experiences always came to us in discrete units that he called “impressions” and “ideas.” But we know that that is wrong. We know, as I have tried to emphasize, that we have a total, unified, conscious field and that in this conscious field our experiences are organized both at any given point and across time into quite orderly and complex structures. The Gestalt psychologists gave us a lot of evidence for this nonatomistic but rather holistic character of our perceptual experiences. A second worry that I have is that I do not know how to account for the fact that an important feature of our experiences is what one might call a “sense of self.” One way to put this is to say that there is definitely something that it feels like to be me. And one way to get yourself to see that there is something that it feels like to be you is to try to imagine what it must feel like to be someone totally different. Imagine what it felt like to have been Adolf Hitler or Napoleon or George Washington. And it is important when you do this imaginative exercise that you not cheat and imagine yourself in the situation of Adolf Hitler, etc.; rather, you have to imagine not yourself playing the role of Adolf Hitler, but what it is like to be Adolf Hitler. If you do that I think you see that you imagine an experience that is quite different from the experience where you normally have a sense of your self as this self and not some other self. But of course the existence of the sense of self does not solve the problem of personal identity. Granted that there is something that it feels like to be me, that is not sufficient to guarantee that anybody who has that experience must be identical with me, because it is quite possible that any number of other people might have this same type-identical experience that I call the “sense of what it is to be me.” My sense of self definitely exists, but it does not solve the problem of personal identity, and it does not yet so far flesh out the purely formal requirement that I said was necessary to supplement Hume’s account in order to account for the possibility of free rational action. So, though this chapter is a beginning of a discussion of the self, it is not more than a beginning.