Central Booking

Wow.

I finally finished reading Godel, Escher, Bach: An Eternal Golden Braid by Douglas R. Hofstader, suggested by Wolfe from Wolfe's Musings. The book is 742 pages long, and it's not exactly light reading. It's basically an exploration of the idea of intelligence, artificial and otherwise. Hofstader's goal is to shed some light on how intelligence / consciousness / self-awareness happens. I would call him a materialist, in the sense that he believes that there is a physical basis for thoughts, feelings and emotions. He is dismissive of "soulists," who believe that there is some sort of inexplicable metaphysical aspect to consciousness.

The question, in Hofstader's mind, is, "If the human brain is made of essentially the same stuff as a kitchen table or a pocket calculator or a tree, why does the first have a sense of of self -- of being an 'I' -- whereas the others do not? Hofstader begins with the assumption that there is a physical corollary for all the activity that goes on in the brain, and that it is this physical activity that is responsible for thoughts and the sense of "self." He goes on to explore how this physical activity, which seems completely mechanistic and completely unlike the process of thought that we experience, can in fact give rise to a qualitatively different sort of activity occurring at the "higher levels" of the brain. One example of such a system is an ant colony: the individual ants are stupid, acting in response to basic stimula, but the colony as a whole is much smarter.

This kind of qualitative difference between the different levels of a system is key to Hofstader's thesis that critics of the possibility of artificial intelligence have misinterpreted the implications of such limitative notions as Godel's theorem. Godel's theorem states, in essence, that any sufficiently powerful system will contain truths that are not provable within the system. The problem with computers, these critics charge, is that they are stuck within a particular system -- there is no way to program them to realize that there is no solution to a particular problem within the system, even though such a fact would be perfectly obvious to an intelligent person who can "jump out of the system." So the computer is stuck, going round and round (remember the episode of Star Trek where Captain Kirk and crew wreak havoc with a race of robots by acting in a completely illogical manner?), trying to solve the problem with a method that is doomed to fail.

Hofstader argues, on the other hand, that there is nothing magical about being able to jump out of the system and reflect on whether it is the appropriate system to be using. In fact, in doing this one is still "in" a system -- it's just a bigger system, one that has the ability to think about lower level systems. And one can jump out of that top level system and reflect on it as well -- but then of course one has entered a new system again. There's always another system, at a higher level, no matter how high up you go.

One could question whether you really are jumping "up" a level by thinking about a system of thinking, or thinking about thinking about a system of thinking, or thinking about thinking about thinking about... but I think that's part of Hofstader's thesis. At a certain point these levels blur together, because they are recursive -- meaning they reflect back on themselves. Hofstader uses the works of Escher and Bach to illustrate the concept of recursivity (among other things). For example, there is Escher's stairway that goes up and up until you are back where you started, or the two hands, each of which is drawing the other.

Hofstader believes that self-awareness -- the "I" -- arises from this kind of recursivity. To put it very simply, at the highest levels the brain is a system that deals with symbols, and the "I" is the symbol for the system itself.

There is much, much more to this book. There are lengthy tangents into mathematics, philosophy, biology (the section about the recursion that takes place in the copying of DNA is particularly fascination), etc. I'll admit to skimming some of the sections that dealt with proving some abstract mathematical/logical concept. One nice thing about the book is the way it's organized into sections that alternate between straight exposition of some concept and a fictional dialogue that illustrates the concept. Still, it's not light reading, and I did wish at points that there was a Reader's Digest version that would just give me the main points without going off on a tangent for 30 pages about wasps or something.

Godel, Escher, Bach made me rethink a lot of my preconceptions about consciousness and artificial intelligence, and is well worth reading the next time you have three months to spare.

Next up: Foreigner by C.J. Cherryh suggested by Claire from Notes from a WildCat Fan.