X-Message-Number: 23950 Date: Wed, 21 Apr 2004 22:09:33 -0700 From: Mike Perry <> Subject: Machines and People This is a response to Thomas Donaldson's message #23939 of yesterday. He writes in part: >I said specifically that human beings were not hyperadvanced computers. >However I did NOT say that we could not be seen as machines. How >can a thinking machine not be subject to Goedel's theorem? By >not thinking according to any fixed formal system (I've found this >hard to explain to some people before, and this point may not make >what I say any clearer, but there it is). Our brains don't work >like computers. Quantum mechanics may itself be a "fixed formal system" according to which brains as well as computers and everything else operate (though it is also probabilistic, from the standpoint of the embedded observer, that is, ignoring many-worlds). And QM seems fundamentally computational in nature, dealing with discrete jumps in states for instance, which are Turing computable. This may be inescapable. However, it is worth remarking that even ordinary computers today have a way of escaping the constraints of a formal theory and are not simply "prisoners of Godel." How? Through the use of *heuristics*, which are procedures or algorithms not guaranteed to give desired results, but which are used nonetheless because they are found to be useful. (A good discussion of this idea will be found in Daniel Dennett, *Darwin's Dangerous Idea*, Ch. 15; I also consider it in my book.) We see a tradeoff then. We could insist that the computer get results which are guaranteed correct because, in essence, they arise as mathematical proofs in some formal theory we trust. The price we must pay--one price--is that some questions then will not be answerable, as Godel's results demonstrate. Another is that we could pay heavily in terms of efficiency, even when the results are obtainable in this formal way. Or we could take the alternate route, often chosen, of having the machine get answers that are not guaranteed correct, but might well be, and are also obtainable quickly. The machine, in any case, really has no limitations in this area (ignoring possible efficiency considerations) that a human does not. >As for Mike Perry's comments, given that Mike thinks that we DO >think like computers, they partly provide an illustration of >what I said. I'll acknowledge that the brain is very complicated and, on the face of it, does not seem to function much like our computers. Yet it too is a physical object subject to physical laws. Those in turn seem fundamentally computational, as I've noted. We appear to be computational, at the very least, based on basic physics, QM specifically. Of course theories come and go, and there is no guarantee this particular one will not be overturned in the future. One reason for my confidence, though, is that QM has held up amazingly well so far. Even if a successor must eventually supplant it, it too could embody the computational principle. So the question is not whether QM is unassailably true but whether the computational principle is deeply enough embedded in the fabric of reality to be relevant in the way I think it is. It is also worth remarking, though, that efficiency considerations could preclude any practical demonstration of human-level functioning in a classical computational device, however sophisticated it may become. I doubt if the brain actually embodies a quantum computer. But in subtle but significant ways it may depend on the same sorts of effects, invoking multiple, simultaneous states of matter, for instance. If so, it could mean that, while a classical computer could isomorphically reproduce the brain's processing, it could only do so inefficiently and would never be a practical competitor. This remains to be seen. And even if it proves true, the possibility will exist of artificial devices which do make use of quantum effects and are able to compete with and outperform natural brains. This possibility we must see not as a threat but as something that could help us in numerous ways in our quest for immortality. Mike Perry Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=23950