In the first place, what is the theory presume to say? Well, what it presumes to say is to turn the problem on its head and say, instead of being like a Turing machine... which is what a computer is – a finite state automaton with a program with a tape head that can read or write a zero or a one and then move one step or move in the other direction one step and change its program again, and in fact even expunge a zero and one and replace it, which Turing showed, in fact. If you have an effective, unambiguous algorithmic procedure, you can have a universal Turing machine; you can show mathematically that there is not problem you can't solve except certain ones which he showed were insoluble. So this idea pervaded; I mean, people still in cognitive science consider this the central idea: namely, if that's what you can do with a computer, then why not the brain as a computer?

Well, it turns out, when you look at the actual biology and you go back, and there is a suggestion here to embryogenesis, things happen uniquely. And then you have the problem of... well, that means you have a kind of ambiguous situation – you have mistakes, so-called. Now what do you do with a computer when it makes mistakes? Well, you put in error correcting programs. But when you look at the biology, the number of possible error corrections would become unbelievably large. So this theory, which was called the theory of neuronal group selection or neural Darwinism in honor of the great man, said the following. That there were three fundamental stages: the first was the stage in which you lay down the neuroanatomy which was following those kinds of rules I mentioned about in morphogenesis – namely you have cell division, cell motion, cell adhesion, cell death, and... and the interaction amongst different parts, embryonic induction, in which one set of cells influences another by signaling. All of this is true in the nervous system, but you have to remember the nervous system is made of very special kinds of cells which are extremely polar, which have long processes called axons which go from one cell to connect to another through a synapse or connection, and also have other kinds of processes called dendrites which receive that connection from axons. When you look at it, it sort of scares you. The numbers are truly scary. Let me give you a feeling for that.

The cortex of the brain has about 30 billion neurons. It has a million, billion connections. If you count one second... one connection per second you will just finish counting 32 million years later. That's just counting them. If you calculate how many different ways they can connect, it's ten followed by a million zeros. There are ten followed by 83 zeros, give or take, of particles in the known universe. Gives you a lot of respect. So here you have this problem, that in each embryo, when you're making the nervous system and its connections and its neuroanatomy, it's true that these genes we mentioned before, Hox genes and Pax genes, constrain the fact that say it's a human brain rather than a chimp brain. But in very short order the whole process becomes epigenetic; it becomes that neurons that are connected, even at a distance, that fire together will... will influence which neurons wire together. That first step has to necessarily introduce variation. So even though all brains are alike in the way the faces are alike, they're also different in the way faces are different, even in twins.

The second step is related to the first: namely that, even after you have the anatomy, that there is differences in the strengths of the connections. There are differences in the strengths of the connections. So that the influence of particular events on a synapse are such as either weaken it or strengthen it. That's essentially like putting a traffic cop on each of the highway points, to let it go through fast, to let it go through slow, or to stop it completely. And neurons come in two flavors – excitatory and inhibitory – and all of the things I'm mentioning refer to both. The fact is this second process is occurring right now in our brains and is the basis, in large extent, a necessary basis at least, for things like memory, learning and what have you. So that is the second stage. So the first stage is what I call developmental selection, where you change amongst the different embryonic processes; the second stage is experiential selection, because obviously your experience of the input to the various neurons changes which ones fire together and wire together. But neither of those really covers the essential point that you... you're faced with when you give up the idea of a computer. A computer works by logic and a clock. Now I've given up logic and I've given up a clock, and I've said brains don't work that way – they work by pattern recognition – so what gives you the equivalent of time and space? The answer to that is the hard part of the theory, I'm afraid to say.