X-Message-Number: 22308 Date: Fri, 08 Aug 2003 01:20:21 -0700 From: Mike Perry <> Subject: Re: More on Simulation Robert Ettinger, #22303, and my responses >Mike Perry asks again how we could be sure a system isn't conscious (if it >has functions isomorphic to those of feeling). > >This is just another way of asking: "How can we know that isomorphism isn't >everything?" > >One answer that makes sense to me is that in other areas of life and thought >we reject the idea that isomorphism (or partial isomorphism) is sufficient. >For example, a mechanical analog computer can do definite integrals which >could >be interpreted as the charge accumulated on a capacitor--but we do not >therefore say that the mechanical computer "is" a capacitor accumulating >charge. The reference here, if I understand it right (and I thank Hugh Hixon for clarifying this), is to *two* types of analog computer. The mechanical computer represents a numerical value (definite integral) as a position of a physical pointer--essentially, a physical distance, and the electrical computer as a charge on a capacitor. >It seems rather odd that upmorphists are willing--even eager--to accept a >simulation as a person, but not (for example) willing to accept a >disk-and-stylus >analog computation as an accumulation of charge. True, a physical distance is not the same as an electrical charge, but both computers are doing integration. Similarly, a suitable robot brain would be non-meat but both the robot and organic brain would at least *appear* to be conscious. Are consciousness and feeling basically computational in nature? If so, then an imitation of consciousness would itself be consciousness, in the same way that an imitation of a computation is itself a computation. It is yet to be demonstrated but plausible, I think, that consciousness at least could be well-imitated computationally--this would involve internal as well as external congruences with natural brains. If so, we would be confronted with an artificial system that seemed to be conscious with, I think, no way in principle to demonstrate that it was not conscious. Consciousness, in other words, is not a property that is testable in the same way as being a physical distance versus an electrical charge is testable. >Or you could put the >computation on an ordinary digital computer, with symbols for charge as >well as time >and current and anything else you deem important. What it boils down to, >again, is the ASSUMPTION by the upmorphists that information processing is >everything. As I believe Mike has acknowledged, this is an untestable >assumption, >hence many would regard it as meaningless. That I survive sleep (rather than being replaced by another, similar individual) is an untestable assumption, but far from meaningless. >As part of this, remember that only a fraction of the stored bits in the >computer correspond to coordinates of the simulated system in phase space. >MOST of >the bits relate to intermediate calculations, and which is which is >arbitrary, a matter of labeling and understanding the labels. But this >understanding is >in the mind of the beholder or programmer, who tells the computer which items >he wants highlighted or displayed. It may be that many simulations would be inefficient, as a price to be paid for accuracy. As a practical concern, though, I would ask about the feasibility of replacing certain parts of the brain with artificial components that do equivalent things but are more durable. I understand there is work being done on an artificial hippocampus right now. If successful it could cure a severe forgetting problem known as Korsakov's syndrome. More generally, the use of artificial brain components, or outright uploading of the personality elements into some sort of programmable device, could be a fast track to physical immortality, not to mention a means of facilitating cryonic resuscitation. Something to think about. Mike Perry Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=22308