Sperry had put forward what was called the chemoaffinity. What he proposed is that each recipient had a molecule and each donor had a molecule, and only when they matched could... would the nerves wire up, so that you had these matching sets which had to meet. And of course if you were to think of... of how that... that would work, you would have to have a very large number of specific molecules. Now, of course at the time many people thought that there is another thing where we seem to have a lot of specific molecules; it's in the immunological system. You know, everybody said something we called the Beilstein paradox. Beilstein is a big, is a big book of organic chemistry and describes all the organic chemicals that have been found. And people have said well, we're sure that if you take a chemical out of Beilstein you can make an antibody to it. And of course we have to think not only there... are there a 122 volumes of Beilstein now, but there's going to be, you know, 1022 volumes of Beilstein written over the next 20 years, and of course this statement will be equally true. So how does the antibody first of all know what's been published in Beilstein and secondly, a more, a deeper question: how's it going... how does it know what is not yet published in Beilstein but will be in the future? So what is in fact the landscape of this area of chemical... of molecular recognition? And it becomes very clear that it has to be a rather more complicated thing than, you know, one antigen, one antibody, and it means that everything must be pluralistic in a sense. That is, one antigen can be recognised by more than one antibody, and perhaps one antibody can recognise more than one antigen, and then we can, we can think of how that will become a code. Szilárd, in fact, wrote a paper about how the nervous system would be wired up by this form of chemical recognition, and what is more, he used the paradigm of immunisation as a model of learning. In other words, if you wanted to do new things you kind of formed a molecule, an antibody, if you like, that... that learnt what this new thing in, and therefore that would allow you to have learnt it. In that sense the antibody system does show a kind of molecular learning thing; once exposed to a chemical configuration it learns what that is and produces an antibody. Now of course it doesn't learn it in the sense of doing some big interpretation of the structure then computing what it has to do. Because that would be one way of learning, and of course it learns by being totally ignorant, which is the cheapest way. Which it says, if it works do it, if it doesn't work forget about it. And so I mean, it is learning under the regime of total ignorance that distinguishes the antibody system from what we might speculate as learning. This is quite important because many people now can't make the clear distinction between selection and acquisition in this sense.