The small size of the nematode - the adult is only 1 millimetre in length and about 40 microns in cross section, and the larva of course is much smaller - meant that these could be fit well into the window of the electron microscope. And the idea then evolved, well, what we would do is we will determine the wiring diagram of an... of an animal. And at the same time we will make mutations, and then we'll see what alterations we will get in this wiring diagram, and then we will be able to relate this to the genes. And at the same time it had become clear from... from this that if you had a wiring diagram - that is if you had the structure of the nervous system, the anatomy - you could begin to ask questions about how it worked. And that I think is very, very important. Because in one of his papers, one of his lectures, von Neumann made a number of statements which, I think, are very, very interesting and in which he said that for things at a high level of complication - like the example he used always, pattern recognition - von Neumann said there may be no way of saying... of giving a theory about pattern recognition other than in the form of a prescription - of a device or an algorithm that would be the same today - that actually performs pattern vision. And this theory of explanation that he gave... and, of course, he said this at the high level of complication because as someone who discussed it with him later, and it is recorded, asked him what he meant by this, and he said what he meant is it wasn't a Gödel problem, it was that there's no language exists higher than the... the language of the system itself to talk about it... to do this. And, of course, he was comparing it with the other thing, which is quantum mechanics, in which there's another paradox there. So, but leaving aside what was in von Neumann's mind there, it was very clear that algorithmic explanations might have to be used for things that have a very high level of complication. And I would contrast that not so much with anything... with a level of explanation within the terms of the object itself, but with a level of explanation much used in physics; namely that of the equation. The equation is the correct meta-language, if you like, in that case. And I mean, it was in no sense that we could write down a set of equations for behaviour - that we didn't expect - or for pattern recognition. What we could write down though was an algorithmic explanation. Now, you can't just make algorithmic explanations out of nothing. And so our idea... no, you can't just make algorithmic explanations out of nothing, and our idea was you've got to say how the wiring diagram works. Let us just contrast this with a normal degree, a normal mode of behavioural explanation in... in neurobiology. What you argue is that you find certain conditions in which the animal performs in a certain way, and you make a theory that there's some... there's an oscillator inside the brain that when that oscillator is tuned to something else, then a switch is thrown and off it goes in this direction. And then you find another bit of behaviour which you need to postulate some other sub-system. And there's no end - in the Turing sense - there's no end to this explanations, because for every new case you have to come and give another explanation. And so if someone says, 'What happens if I glue this animal to the bottom of a satellite?' you say: 'Well, you can't compute that from this knowledge because it's so fragmentary, you'd have to go up to the satellite, see what happens and then provide a mechanism to explain it.' On the other hand, if you've got a complete wiring diagram you could compute from this, given certain inputs, what the outputs would be. And the argument is that if you could do this and prove it was a unique explanation, then you have the total explanation, nothing more, because if someone says this, you say: 'Well, let us go and put the model on the bottom of the satellite and run it, because it'll explain to us what is happening there.' And I think that that would be the model of all complex system explanations, especially in the biological field, where they are not... they are not instances of some very general theory but each one has been evolved in order to... to make a good fit between what's inside and what's outside. So the big job was: find the wiring diagram.