QUESTION: If history plays such an important part in our development, how come human beings keep making the same mistakes time and time again? How do you predict the future? You've got all that history-can you not see the future with that information?

ROSE: The whole point is the future is radically unpredictable. It's unpredictable because we can only track change. We can't predict futures. Humans can do a little better than other species in predicting futures, but because of the rate of change of technology in human society, constantly throwing out new problems because of the complexity of the social changes that are occurring, then predicting the future becomes extremely hard. That is why I say in many respects it's radically unpredictable. What I do insist is that we have the freedom to make choices about it, which is a different argument. But we don't have infinite flexibility in making those choices. Steve Pinker and I would both agree that we are constrained by our evolutionary past, by our biological givens-none of us can walk on water, any more than we can grow wings. What we can do is find technology that can solve those problems. Those constraints are there. We see and understand the world through spectacles that are given us by our biology-the fact that we are somewhere between one and a half and two and a half meters high, most of us, rather than a couple of centimeters high radically transforms the way that we understand the world. If we were those small creatures we'd see the world-we'd have quite different biological problems and social problems to resolve. So our past is indeed in many ways the key to the present.

PINKER: I have nothing to add to that; I agree with it.

QUESTION for Pinker: I wanted to ask Professor Pinker again about Cartesian dualism. Although your book does argue that you want to approach understanding consciousness in physical terms, in a materialist way, in the book at one point you talk about your materialist work being the project you do during the day, and in the evening when you're talking with your friends and so on you acknowledge that human beings are sentient and have free will and so on. You acknowledge that it's a non- trivial problem to bridge that gap. You say it might not be possible to do that, whereas elsewhere in the book you talk about the computational theory of mind, I assume as a way to bridge that gap. But I wasn't convinced by that, because it seemed to me that it was just relocating the problem. Social categories like desires and beliefs were just being relocated in the heads of individuals. So when your Bill gets on the bus, his belief that the bus is going to his granny's can just be re-represented as a physical symbol in the brain, and that fills that gap. There seems to be some flip-flopping between being a physicalist on the one hand, and on the other saying that you can approach the same subject in two completely different ways. PINKER: There's no flip-flop in my discussion of mental states such as beliefs and desires, which doesn't call for any kind of substance dualism-the idea that there is some kind of stuff different from neural interactions that accounts for how we behave and how we perceive the world. As a nonreductionist I think there are different levels of analysis, and that the information-processing level of analysis gives rise to psychological regularities and generalizations that can't easily be captured directly in terms of the neurophysiology. Take the simple example that our short-term memory can hold only five or so items. We have no neurophysiological explanation of that, but we can characterize it in computational terms. Eventually it will be tied to the neurophysiology because they're two different levels of analysis of the same phenomenon. In terms of morality, I believe that there is a role in our discourse for moral judgments and for a concept of free will that is not dualistic but that simply is part of a different system of reasoning, in the same way that mathematics is a system of reasoning that differs from science. We don't actually believe that there are perfect circles or infinite straight lines or Euclidean planes, but we can still perfectly well reason within mathematics. Like many moral philosophers I believe that there's a sphere of moral reasoning we can engage in that makes use of idealizations like free will but without making any commitments that there's actually a different kind of stuff in the physical world. It's an assumption that makes that system of reasoning possible. We can't have ethics unless we hold someone responsible for their behavior; we can't hold them responsible for their behavior unless we believe that the behavior is not directly caused. That's how we make moral judgments, but it doesn't obligate us when we shift to a scientific mode of explanation to believing that there's a ghost in the machine.

ROSE: Very briefly, there's a book which has just been published called The Number Sense by Stanislas Dehaene. It makes a very interesting point about this question about whether you can hold more than five things in your mind at the same time, which is a classical piece of data which appears in every student psychology textbook. Dehaene points out that it is entirely culture-bound. Chinese culture, for example, which has a different way of counting and representing numbers, can hold many more than five items in their mind at the same time. So it's got again this beautiful interaction between culture, society, biology and history, which I think we have to again take into account whenever we try to say these are universals about the way the mind works.

PINKER: The Dehaene finding is part of a set of phenomena that's been known for as long as the five-item constraint has been known, namely that a particular item in memory can point to a much larger data structure, a phenomenon called chunking. The difference between the Chinese memory span and the American one is simply a difference in chunking; the underlying constraint in memory, according to my memory of Dehaene's work, is the same as it is in American children.