Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!decwrl!labrea!polya!geddis From: geddis@polya.Stanford.EDU (Donald F. Geddis) Newsgroups: comp.ai Subject: Re: Question on Chinese Room Argument Message-ID: <7431@polya.Stanford.EDU> Date: 5 Mar 89 21:35:13 GMT References: <4298@pt.cs.cmu.edu> <7408@polya.Stanford.EDU> Reply-To: geddis@polya.Stanford.EDU (Donald F. Geddis) Organization: Stanford University Lines: 78 In article harnad@elbereth.rutgers.edu (Stevan Harnad) writes (in reply to a posting of mine): >There's no other entity, no >other eligible candidate for having a mind in the Chinese Room; nobody >home! "Searle + rules" is a piece of cog-sci-fi. Do you believe that I >could fail to understand, and alpha centauri could fail to understand, >but "I + alpha centauri" could compositely understand? As a matter of fact, I do. So if you are appealing to intuitions here, it is obvious that we disagree about what is commonsense. So try another tack. >It's not that there CAN'T be a systems reply in principle: Searle COULD >have been a non-understanding part of an understanding system. He could >have been standing in, say, for the input and output of one neuron in a >real brain. Then the system WOULD have understood and Searle would not >have. But then neither would Searle have been performing ALL THE >FUNCTIONS that were the substrate of the understanding. So it would be >no surprise that he didn't understand. One of Searle's premises is that >he himself must do EVERYTHING the candidate mental model does, yet not >understand. (This is why my "robotic functionalist" counterargument >works, and why the TTT is immune to Searle's Argument.) OK, so a crucial point for you is that Searle can do it all himself, I assume by memorizing the rules. So now we have a very complex (and improbable, but we'll ignore that for the moment) entity in Searle's body. When we ask it, in Chinese, if it understands, the body replies "Yes, I do". When we ask, in English, if it understands Chinese, the body (using the small part of it that used to be Searle, before it memorized the rules), the body replies "Of course not! And I wrote a paper telling you this a long time ago!". Now I believe each reply to the same extent. Namely, that the whole entity (Searle + memorized rules) *does* understand Chinese, although a section of it (Searle by himself) doesn't. To try again: Now there are *two* minds in Searle's brain, just as there were two minds in the old Chinese Room (Searle and Searle + rules). I don't consider the lone Searle to be an authority on what Searle + Memorized Rules understands, although you seem to. Why? >See earlier replies on "speed and complexity." This is just >hand-waving. It's equivalent to taking a dumb toy model and saying >"Just more of the same will pass the TT and will have a mind." I think >the gap is not one of speed and complexity but missing, >yet-to-be-discovered substantive functional concepts (and not just >symbolic ones!). I agree that complexity arguments are not important for this thought experiement. But it is useful to be aware that our intuitions about how small rule sets function probably don't scale up well to large rule sets. In no way am I claiming that this is sufficient to create intelligence; I just don't want you to appeal to the triviality of small symbol processing in order to claim that large symbol processing won't work either. >You've missed the point. Of course the Lingustic Turing Test is a >subset of the Total Turing Test, and of course the whole Test would be >harder to pass. But my point (and it had supporting arguments) was >that it may well be that the only kind of device that could pass the >LTT in the first place would have to be a device that could likewise >pass the TTT, and that IN BOTH CASES it would have to draw essentially >on nonsymbolic internal functions. Probably true. (I agree.) But if that's the case, then just using the LTT is sufficient for judging intelligence, as Turing originally claimed. The point of the Turing Test was to eliminate non-cognitive things from the test, like "oh, it is colored green, and no human beings are colored green, so this must be the computer". We only want to judge cognitive ability. Whether this requires TTT ability is a problem for the engineers, not the judges. (As another point, while the Turing Test may be sufficient, it is not necessary. Turing made it too hard to pass. A hypothetical entity has to foolishly duplicate human errors, like arithemetic errors and long pauses for thinking. This needlessly eliminates intelligences that are comparable but not identical to human performance.) -- Don Geddis P.S. Thanks for the responses to my other questions. -- Geddis@Polya.Stanford.Edu "We don't need no education. We don't need no thought control." - Pink Floyd