If, as Dawkins reminds us in "The God Delusion", our cellular self is completely renewed over time, should we absolve the criminal of his crimes after time has passed on the grounds that he is no longer the person that committed the crime - for example, the rapist who is not caught until decades after his crime, or the aging general who committed war crimes. If not, does this prove that there is more to the self-hood of a person than just a collection of cells?

It's an interesting question, and to answer it, I'm inclined to turn things around. Let's start with what's clear: the fact that the rapist committed the rape seven years ago (supposing for the moment that this is the magic number) isn't a reason to let him off. In fact, the very way you pose the question makes the point. You ask about "the rapist" who committed "his crime" long ago. You've already take it for granted that we can say: this man is the one who committed the crime. And we can say it without worrying about how many cells have come and gone.

So yes: there is something more -- or something other -- to the notion of a person than just the idea of a collection of cells. The something needn't be anything spooky. After all, a corporation can exist for a hundred years, even though all the people have changed and all the buildings and equipment it owns have gradually been replaced. Although saying exactly what sameness amounts to here is complicated, it won't call for talking about anything strange. It will be a matter of various continuities and connections among perfectly ordinary agents, entities and events.

In fact, the word "agent" gives us a clue to the case of persons. John Locke famously observed that 'person' is a forensic notion. What he meant was that personhood has to do with accountability, agency and the like. To decide whether it makes sense to hold this human being accountable for what a certain human being did twenty years ago isn't a question that we approach by reviewing the history of the cells in his body, and someone who thought it did would show that they don't understand what was being asked.

Once again, detailed account will be complicated. But we can say at least this: when we ask whether a human being at one time is the same person as a human being at another time, we're asking about various kinds of connections and continuities. And what seems clear is that there can be enough of the right sorts of connections and continuities even if the cells that make up the present body are not the ones that made up the one from long ago.

And one might add that the cells themselves are hardly immune from "renewal" at the molecular level. So the short version is: If identity requires complete coincidence of matter, then essentially nothing but sub-atomic particles survive over any reasonable stretch of time. That does rather suggest, though the contrary view is certainly held, that identity over time simply does not require complete coincidence of matter. What it does require is not very clear, but that is no reason to despair.

Of course, the question didn't ask about complete coincidence of matter. But it's unclear why anything less might suffice. And, if it does, then you run into issues about transitivity: A might share much of its matter with B, which shares much of its matter with C; but A and C do not share much of their matter.

Read another response by Allen Stairs, Richard Heck
Read another response about Biology, Existence