X-Message-Number: 14961 Date: Sun, 19 Nov 2000 18:19:05 -0700 From: Mike Perry <> Subject: Machines and Brains A lot has been said lately on whether a brain could be considered a "Turing machine," or could be realized as such a device, but I see a need for clarification.The point is made, and well made, that in practical terms we may never see such an implementation. Certainly this is true if taken literally, but come to think of it, what thing in our world today *is* a Turing machine (ever see one?), and by extension, do we expect to see one in the future, except as some curiosity? By this I am referring to the literal notion of a finite-state device with an infinitely long (or finite but indefinitely extendible), 1-dimensional tape it runs back and forth on, marking and/or erasing symbols of a fixed, finite alphabet. (It might be fun to try to implement this, just to show how it could be done. Maybe you could use something like a model train set. Have the machine itself be an electrically powered boxcar that runs back and forth on a track divided into squares each of which is equipped with a liquid crystal display to show different symbols. The car would sit on a square and have some way of electronically detecting what symbol was showing, and also making changes to both the symbol and its own state, then would move right or left one square or halt. Probably straightforward to do, and I think it would make a good instructional tool for computer science classes, but I've never heard of it being done.) But the point is made that Turing machines and real computers can be considered equivalent devices, inasmuch as computations in either form can be done in equivalent form in the other device. In either case the behavior of the system is "quantized"--over finite time and distances you have a finite number of discrete jumps or sudden changes, such that whatever is happening in between the jumps is unimportant, so long as the jumps occur. So in this more general sense we could ask if the brain too is such a device, i.e. a "Turing machine." At the quantum level it seems we must concur with that judgment; the Bekenstein bound puts a limit of around 4 x 10^53 (discrete) state changes per second in an object the size of a human brain. In other ways the possible changes conform to a finite-state device, with allowance for probabilistic effects. But the number of changes is so large that in practical terms many possible ways of realizing them are effectively ruled out, or maybe rendered as curiosities that, assuming immortality is possible, we still won't encounter except very, very marginally and rarely. Thus the Turing machine "running" the human seems most unlikely, if again you mean the device with its 1-dimensional tape. But the question remains of what other non-meat devices might make suitable substitutes for your brain and mine, if well configured. For this we'll have to wait and see. My hunch is we will find such devices, as our miniaturization approaches the nano scale, and also, that consciousness is really not so mysterious, being mainly a way to limit the attention of a sensing, acting system or entity, so it can make important decisions more efficiently. Mike Perry Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=14961