Here I am in 1954, and I can’t find a map of the neurons. And if you take a slide of a piece of nervous system and stain it so that all the neurons turn black, then the whole slide is black, and you can’t see anything. So the only information available was from something called the Golgi Stain, which was a fascinating way of treating a piece of nervous system with… I think it was Osmium – some Osmium compound – and it has the property that if there’s a… a leak in a neuron the stain gets into it and the whole neuron turns black, but the other ones around it don’t.  And so this enabled people to get very good portraits of what an individual neuron was, but you could never get any picture of how they were all connected with each other because the thing is so full of dark objects that no light gets through.

So I decided to fix that, and I invented something called the confocal microscope, which was a new way of putting light through a specimen so that if the light bounced off… if you tried to see what was happening at a given point, you only look at light that gets to that point and comes out of it again. And the way to do that is that if some other light bounces into the path of your… that you’re picking up, it gets rejected.  And so I built this gadget which won the Rank Prize – there’s a picture of the Rank Prize up there – and now every biological laboratory has a confocal microscope, and they’re used all over the world. And it was… I wrote a paper on it, and a patent; and I didn’t really have the idea of publishing very clearly in my head then so the patent was issued, but the second confocal microscope was built 20 years later.  So anyway, that was a very exciting adventure, but I sort of showed people how it worked and they said, oh, that’s amazing, but nobody built another one for many years.  That was my fault, I guess, for not getting it published openly.

Where was I? Oh, anyway, but right in the middle of that, when I began to understand Solomonoff’s ideas about making a high level theory of learning and inference I decided to drop all that stuff, and I realized it wouldn’t help to know how the nervous system was wired into you, because it’s so… it's such a mess that until you had an idea of how it worked you wouldn’t be able to decode looking at that diagram, even for a… C.elegans; unless you’d do separate experiments to see what each neuron does, I don’t think anybody could learn much from looking at the diagram. Besides, those neurons aren’t very much connected to each other, anyway.