Raindrops and Persons
Consider a single raindrop: it exists for a certain period of time, eventually meeting its end through evaporation. But suppose instead that it divides into two before that happens—by wind or human intervention. Now we have two raindrops where before we had one: what is the relationship between the original raindrop and the resulting raindrops? The first thing to say is that the case is not like evaporation: the raindrop is in some sense still around after division, unlike with evaporation. It survives division, though now existing as two separate drops. It could easily be reconstituted, simply by joining the two drops together again. It has not simply gone out of existence or been totally destroyed. Yet we cannot say that the original drop is identical to the resulting drops, since they are not identical to each other; nor can we choose one of them as identical to the original and not the other. We might try saying that the original is identical to both drops together, but then we have to admit that one raindrop can exist as two. 
Evidently there is a bit of a puzzle about division and identity—identity seems too simple to handle the case. So we might be inclined to say that raindrop survival does not depend on raindrop identity: a raindrop can survive even though no future raindrop (or collection of raindrops) is identical to the original. Still, we should note that a weaker identity claim is true: each of the resulting raindrops is identical to a part of the original raindrop. It had parts and they became separated, but they are the very same parts that existed earlier. And the survival of the original raindrop seems bound up with the fact that parts of it are identical with raindrops that exist in the future; the raindrop still exists because certain future raindrops are its parts. So identity is playing a role in survival, though a slightly more complex role than when the raindrop continues to exist in an undivided state.
Now consider cell division. A cell can either divide or perish: if the latter occurs it has not survived, but if the former occurs we are inclined to say that it has survived—as two cells. Division is not the same as death. Again, we have a bit of a puzzle about identity, since the future cells are not simply identical with the original, but one thing is clear: the future cells are identical to parts of the original cell. So identity is playing a role in survival, even if it is not the role it plays in cases of undivided survival. We can admit that in cases of cell division there is survival without identity between the original and any of its progeny, but still we have identity at the level of parts. Certainly, there is no reason to abandon identity altogether in accounting for survival in such cases, replacing it with some such notion as causal continuity or mere similarity. Future cells are literally identical with parts of earlier cells, and that is why we distinguish division from sheer destruction. The general principle seems to be that if the parts of an object are identical with any future objects then that object can be said to survive in those future objects. No doubt this principle would need to be qualified, but it roughly fits our intuitions about certain cases. If I disassemble a car engine and scatter its parts, we may suppose that the engine survives, even though it is not identical to any of these parts; still they are identical with parts of the original. Or if I saw an eight-legged table into two, thereby creating two tables, it may be said that the original table survives, despite not being identical with either table (and possibly not with the sum of both). The table certainly has more of a claim to continued existence than a table that is burnt to ashes.
Then we come to brain division. Here again it is awkward to say that the brain survives in virtue of identity with the future brains, but we can still say that the future brains are identical with parts of the original brain—its two hemispheres. If the brain is a mouse brain and I put the two halves in separate mouse bodies, then we have two mice as a result–neither of which is identical to the original mouse. But it is still true that each mouse has a brain that is identical to a part of the original mouse brain. This is why we are inclined to say that the original mouse survives—certainly more than if it had been incinerated. An intelligent mouse would have a reason to favor brain division over brain incineration. We have survival without identity of mice—no future mouse is identical to any past mouse—but not survival without identity to any past part of a mouse. The part lives on, identically, though now separated from the other part, which also lives on, identically. It is not that identity is playing no role in survival, so that we must resort to some other notion, such as causal continuity. So far as these cases show, survival is still conceptually demanding—it doesn’t collapse into some weak notion of causal connection between successive states. It is not that these entities survive merely by causing future things to have certain states; they survive by virtue of there being future objects that are identical to original objects—but parts not wholes.
It is the same with human persons. Fission cases show that survival doesn’t depend on identity with any future person, but they don’t show that survival is possible without identity between person parts. The cases are convincing precisely because half the brain of the original person sits in another body—and that half brain is identical to part of a past whole brain. The person survives because his two brain hemispheres are identical to the brains of two descendants—that, at any rate, is the intuition tapped into by brain fission cases. I would survive in one hemisphere because it is my hemisphere, and doubling up doesn’t undermine that; but if the future person did not possess a brain literally identical to half of my brain, I would not be so sanguine. I would certainly not be so sanguine if I was assured merely that the future person would be causally continuous with me in certain ways. Survival, intuitively, does require identity, but it can be identity at the level of parts not wholes.
Does any of this conflict with our commonsense view of persons? No, it is all part of common sense. It no more conflicts with common sense than the analogous view of raindrops, cells, and mice: we quickly see that in cases of division we need to tinker with identity to explain our sense that there is survival in such cases. But we are not metaphysically committed to a mistaken view of the nature of persons—any more than with raindrops and cells. Nothing radical is demonstrated by fission cases in any of these areas. If personal fission were as common as other kinds, we would have no hesitation in describing fission cases as survival without identity (of whole persons); but we quickly see the point once an imaginary case is constructed. Our concept of a person prepares us for the possibility of fission cases, and our knowledge of the brain makes this concrete because of the facts of brain anatomy (two separate hemispheres). There is no revisionary metaphysics here, no deep error about the ontology of persons. We don’t need to replace our ordinary notion of the self with a new one that weakens it to something like mere continuity of psychological state. Brain division cases by themselves have no such dramatic consequences.
It is sometimes claimed that we have an unduly simple view of the self and its survival because of the word “I”: the word looks simple, so we suppose that its bearer must be. This is a preposterous suggestion: who would commit such a gross non sequitur? Simple things can have complex names and complex things can have simple names. Do we have overly simple views of time and space because of the simplicity of “now” and “here”? And people often have long proper names, as well as lengthy descriptions, not just short personal pronouns. It would be absurd to infer a simple view of the self merely from the syntactic simplicity of some terms for selves.
Nor is it true that we naturally have an all-or-nothing view of personal survival, failing to recognize the possibility of degrees of survival. We don’t have an all-or-nothing view of survival in general (sand dunes, cities), and we recognize that the person of childhood may not fully survive into adulthood, as well as that dementia can weaken personal survival. We are well aware that personal survival requires psychological overlap and that survival can be a matter of degree. Just consider pushing fission cases further so that we halve each hemisphere and retain only certain aspects of the original person—we rapidly conclude that full survival is not guaranteed in such cases. There may be partial survival, which is something, but not survival of everything that matters. We don’t have an all-or-nothing view of the survival of persons (or animals). We know that in certain cases there is no survival at all, but we don’t have the naïve idea that survival is always either complete or completely absent. Maybe there are some people who do, but it is not part of general common sense. So common sense has no need of revision in this regard.
If there is anything especially puzzling about personal fission, it is that we don’t have a clear idea of a person having parts in the way a physical object has parts. A person’s brain has literal parts and that is what guides our intuitions in fission cases, but it doesn’t follow that persons have parts: I don’t divide into two person-like hemispheres. Thus we cannot easily say that the resulting two people are identical to parts of the original person, because we don’t normally think that persons have persons as parts. But split brain cases suggest that we really do have parts that are persons—that our sense of the unity of the self is mistaken. In any case, puzzles about the divisibility of the self should not be taken to show that fission cases undermine the basic model supplied by raindrops and cells, namely that survival arises from identity as to parts. Raindrops and cells, like brains, are complex entities that have parts; these parts can be detached from each other and the results are also raindrops, cells, and brains; these results count as the survival of the original entity, in contrast to more drastic changes. We have no more reason to revise our common sense view of persons than we do of raindrops and cells in the light of fission cases. No doubt the self is philosophically puzzling, and skepticism about its existence can be pressed, but brain fission cases show nothing remarkable about personal survival. Survival without identity is commonplace, intelligible, and non-revisionary; and it involves identity anyway (as to parts). By judicious deployment of parts you can make two persons from one: in such a case the original person can be said to survive—in virtue of those parts. Similarly, you can make two raindrops from one, and you also get survival from the persistence parts. Logically, the two cases are on a par. 
 I won’t explore this possibility further in this essay, but I don’t think it is absurd. For purposes of argument I will suppose that it is ruled out; we can still maintain an identity view of survival, as I shall argue.
 No just any parts: reducing an object to its atoms looks a lot more like destruction than other sorts of division into parts. I won’t here consider what notion of part is needed to make the right distinctions.
 As will be obvious to many readers, I have been discussing views associated with Derek Parfit without going into questions of precise attribution. If Parfit’s views differ from those I discuss, my criticisms don’t apply to him. However, I think the views I discuss correspond closely with what he has maintained. I have not considered whether it is possible to motivate a continuity view of the persistence of the self on the basis of considerations independent of fission cases; my point is that such cases fail to take us in that direction.