"When you look at the cerebellum it's a godawful mess" says Randy Gallistel, and its circuitry is very regular. Some remarks at blogginheads.tv:
BTW, the whole discussion is worthwhile. Gallistel makes the point that the human genome project was driven by a good computational model of what the genome does. We have no such model for the nervous system. In particular, we don't know how read-write memory is implemented in the nervous system. Read-write memory is essential to all computation.
And THAT's what the discussion is about, memory. The action potential ("spike") carries information from one place to another in the nervous system. But what carries information forward in time?