X-Message-Number: 9287 Date: Sun, 15 Mar 98 16:58:17 From: Mike Perry <> Subject: Re: CryoNet #9283 Bob Ettinger has raised concerns about whether an emulation of a person or persons, in some type of computational device, would be able to experience feelings and consciousness like "real" persons. The emulation might be done in any of a large variety of devices, all assumed to be of an information-processing nature, i.e. computers. To satisfy me, though, it would have to be a "sufficiently good" emulation. I'm not sure what would be "sufficiently good" but would provisionally accept modeling, particle by particle, at the quantum level. This, I think, could even be done by a classical Turing machine on a simple, 1-dimensional tape, slowly working its way back and forth over who-knows-how-much time, keeping track of a multidimensional environment of enormous complexity. (This, of course, is presented as a thought experiment, not a practical project!) The whole emulated system, it appears could adequately be described, for an arbitrary finite time interval, by a very long, finite but growing, string of bits. The machine would "emotionlessly" simply change one bit (at most) at a time, sequentially. I submit that persons modeled in such a system would experience genuine emotions, again, if the modeling was accurate enough. I don't see that individual particles--electrons, quarks, and photons--have "emotions" either, yet we who are made up of them do. Emotions and consciousness seem clearly emergent properties, that can be supported by an emotionless substrate. The emulator and emulatee are quite different things. Especially, I think, when we are talking about absolutely enormous amounts of processing--perhaps 10^60 or so bit-changes per emulated second! In particular, I should be able to talk to an emulated person, by bringing about appropriate changes on the Turing tape and waiting to see what further changes the patient emulator-machine makes over time. (We could also allow other changes in the tape inscription to be made from the outside, to model other "outside" interactions, and the Turing machine itself might be probabilistic to model randomness.) If I talked to an emulated person, say, and asked "How do you feel, George?" and the reply was "Not bad, though the weather has been rainy of late here, and I have a slight cold, which is beginning to annoy me," and over a long period asked many other such questions and got reasonable answers, and was asked reasonable questions in turn, and found certain other appropriate features, it would seem reasonable that I was addressing a being with real feeling. I would think then that "there is a quantum of consciousness in every bit that flips"--and in fact this is how I do think. Bob asks, somewhat in relation to the emulation scenario, if a description of a picture would BE the picture. But I think this overlooks who is assumed to be looking at the picture. A picture in one isomorphic domain will be represented very differently in another one. If you "belong" in one of those domains, and you "see" a representation from the "wrong" domain, it may not look or seem at all like what you think it should. Though it would take more argument than I reasonably have space for here, I don't think it's foolhardy to commit to the "info view" and have done so, provisionally of course, in my recently completed book. Mike Perry Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=9287