Path: utzoo!attcan!uunet!ncrlnk!ncrcae!hubcap!gatech!rutgers!elbereth.rutgers.edu!harnad From: harnad@elbereth.rutgers.edu (Stevan Harnad) Newsgroups: comp.ai Subject: Re: Question on Chinese Room Argument Summary: Out of Sight, Into Mind... Message-ID: Date: 11 Mar 89 06:27:06 GMT References: <4298@pt.cs.cmu.edu> <399@censor.UUCP> Organization: Rutgers Univ., New Brunswick, N.J. Lines: 83 jeff@censor.UUCP (Jeff Hunter) of Bell Canada, Business Development, Toronto wrote: " I'm curious as to the bounds you put on the TTT. Does a candidate have " to look exactly human even under X-rays, etc... or does it just have to " be able to pass the LTT, look vaguely humanoid, and be able to pick up " a glass? The TTT requires that a candidate be able to DO everything a human can do. It must be indistinguishable from a person in its behavioral capacity. How it LOOKS does not matter in principle (though it might in practice bias a human judge -- that was the motivation [not just an over-riding faith in the functional sufficiency of pure symbol-crunching] for Turing's "out-of-sight/into-mind" constraint, leading to the LTT in preference to the TTT). Both physical appearance and observable details of the physical structure and function of our bodies (including our brains) are irrelevant to our informal, intuitive, everyday solutions to the other-minds problem; I also think they will turn out to be just fine-tuning variables in the construction of a successful TTT-passing model. Most of the real problems will have been solved before we get around to the last bit of fine-tuning. (This is not to imply that brain function may not give mind-modelers some functional clues.) " it is possible, in principle, to reconstitute a living human from the " symbolic information of the position of the atoms... Do you agree " that a re-embodied simulation can understand? Of course I do. I have even agreed that a simulation of a plane or a mind can symbolically encode (and aid us in discovering and testing) ALL of the relevant causal and functional principles of flying and understanding, respectively, that need to be known in order to successfully implement a real plane that flies and a real mind that understands. I simply deny that the simulation flies or understands. Now in the case of flying it's perfectly obvious why symbol-crunching alone will never get you off the ground; but because of (1) the power of Turing Machines and natural language as well as (2) the inherent ambiguity of the LTT (symbols-in/symbols-out) on this question, some of us seem to have made the mistake of thinking that in the case of the mind the simulational medium and the implementational medium can be the SAME: mere symbol-crunching. Searle's arguments and my own are intended to show that this is incorrect. (The rest of your list of hypothetical examples is again irrelevant.) " You repeatedly ridicule the notion that "Searle + rules" can understand " Chinese... [but] "Searle + rules + laser + interface hardware + body" can " pass the TTT, and therefore you should believe that this can " understand. Do you? I find it hard to believe that adding a few " peripherals to the processor... magically adds understanding somehow. " Please try to explain again. I too find it hard to believe that adding on peripherals to a symbol-cruncher will make it magically understand -- in fact I give reasons why this won't even make it magically pass the TTT. It takes more (I never tire of saying, and my interlocutors never tire of ignoring) to ground symbols than simply hooking peripherals onto a symbol-cruncher. " Well [if Searle's Argument is] so simple that dozens of messages later " there still is no clear agreement on what you see as the difference " between real and simulated understanding... I have to remind you that the actual number of opponents I've had in this discussion is rather small relative to the size of the readership of the Net (in fact it's often the same individuals coming back round after round); and their repertoire of arguments is even smaller. Not that I think I would win if a poll were conducted (even if opinion polls were a rational way to decide such matters); after all, on "comp.ai" a critic of symbol-crunching is not exactly preaching to the converted... Refs: Searle, J. (1980) Minds, Brains and Programs. Behavioral and Brain Sciences 3: 417-457 Harnad, S. (1989) Minds, Machines and Searle. Journal of Experimental and Theoretical Artificial Intelligence 1: 5 - 25. -- Stevan Harnad INTERNET: harnad@confidence.princeton.edu harnad@princeton.edu srh@flash.bellcore.com harnad@elbereth.rutgers.edu harnad@princeton.uucp BITNET: harnad@pucc.bitnet CSNET: harnad%princeton.edu@relay.cs.net (609)-921-7771