josh@cs.rutgers.edu (01/06/90)
Penrose, Roger, "The Emperor's New Mind; Concerning Computers, Minds, and the Laws of Physics", Oxford University Press, 1989 To a scientist, or anyone who takes what a philosopher would call a "naive realist" view of the world, the question of consciousness represents a serious problem. Occasionally a brilliant scientist will confront the problem head-on. When he does, stand back-- almost anything is likely to occur. Penrose is such a scientist, and he has so confronted, and something decidedly wild is going on. The same thing has happened before, of course. Julian Jaynes, a Princeton (I think) psychologist, once wrote a book entitled "The Origin of Consciousness in the Breakdown of the Bicameral Mind" which was another attempt to confront the question head-on. I asked a friend in the Psychology Dept whether Jaynes was sound, and he replied that he was eminently qualified and highly respected "but that book is strictly off the wall." My judgement is that basically the same can be said of TENM. That said, however, I will differ with Ralph Merkle's recent review in this forum and say that I find the book valuable and worthwhile to read. As Ralph mentioned, Penrose is intellectually honest in his investigation in a way that sets him far apart from such fanatical AI critics as Searle. Penrose is honest enough to be thought-provoking even though he is wrong. Basically, Penrose's mistake is the "superhuman human" fallacy. For example, he trots out the Godel's Proof argument that there is a statement undecideable in the formal system that we can clearly see by "insight" to be true. So what? There are statements we can't decide either, including the very Liar's Paradox that Godel derived the Proof from in the first place. Godel showed that the formal system can't be proven complete and consistent; the human mind isn't complete OR consistent! Why read the book then? Well, Penrose takes the reader on a magnificent intellectual tour, and because of the aforementioned honesty, it is relatively easy to pass up the "mentalistic" souvenirs he sells along the way. In particular, of greatest interest to nanotech fans, he spends a fair amount of time on the subject of quantum mechanics, general relativity, and a possible theory of quantum gravity. This is of major interest to nanotech because Einstein and Heisenberg could not see eye to eye. Schroedinger's cat is watching our little hole, waiting to pounce on our micromechanical mouse. What's the problem here? We know that the reactions we intend to use "work" in chemistry, and in the molecular machinery of cells. Here's the catch: chemistry and molecular biology use probabilistic mechanisms. The veil Heisenberg is trying to draw between us and our machines is probabilistic; chemistry wouldn't look different, but a molecular computer might just have problems. (Oh by the way: this has nothing to do with consciousness at all. Nothing. Forget it. Completely. Forget it.) The problem is that quantum theory gives us no clue as to when the wave function, evolving under the dictates of Schroedinger's Equation, is going to "collapse", cashing in its chips to the tune of P equals psi squared. And the trouble with *that* is that it makes a difference, in terms of what physics predicts about the process being described, when it collapses! Well, physicists have blithely ignored this hole in their theories for a good fifty years now, and here's why. Whenever they are calculating about something *microscopic*, by which they mean electrons going through slits and whatnot, the wave function is used; whenever they describe something *macroscopic*, like looking at the film afterward, the wave function is "cashed in" and the probabilities taken. This has led to some really screwball speculation but largely is taken on faith. When does it really collapse? Nobody knows. You can do experiments in which several quantum things happen in sequence, and the results reflect a single unbroken propagation of the wave function, collapsing at the very end. If you try to measure what is going on at the intermediate stages, THE ACTUAL RESULT CHANGES, to reflect that the wave function collapsed everywhere you made a measurement! Penrose speculates that there is a theory of quantum gravity that specifies the collapse of the wave function as a function of the amount of mass the function "delocalizes". I'm betting that Penrose's quantum gravity intuition is better than his AI intuition, partly because he *is* a highly respected general relativist (collaborator with Hawking, for example), and partly because I've heard the same intuition before. John Bell (of Bell's inequality fame) once visited here and spoke on this very question, and gave some indications of his own intuitions of what a reasonable answer might look like. Very similar. So, there is a nice explanation of this "hole in physics" that happens to be of great interest to nanotechnicians, and a possible cork for the hole. Read the book, ignore the AI (or wallow in it for all I care), and wonder how big a "graviton" is! --JoSH