Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!rutgers!elbereth.rutgers.edu!harnad From: harnad@elbereth.rutgers.edu (Stevan Harnad) Newsgroups: comp.ai Subject: Re: Question on Chinese Room Argument Summary: On the Pitfalls of a Contemporary Cog-Sci-Fi Education Keywords: Understanding, Comprehension, Learning Message-ID: Date: 5 Mar 89 18:32:31 GMT References: <4298@pt.cs.cmu.edu> <7409@polya.Stanford.EDU> Organization: Rutgers Univ., New Brunswick, N.J. Lines: 80 geddis@polya.Stanford.EDU (Donald F. Geddis) of Stanford University wrote: " the Systems Reply is exactly that Searle-plus-chalk-plus-blackboard " -plus-rules *does* understand... Rather than present arguments against " the Systems Reply, Stevan seems to be appealing to our intuitive sense " of "well, come on, *everyone* knows that this is foolish". But a good " many well-educated people who have thought hard about the problem don't " consider it to be foolish, so a little more effort on the rebuttal is " required... Look, the "Systems Reply" and the reasoning, evidence and intuition behind it, were precisely what Searle's Argument was formulated to refute. It was already anticipated and explicitly discussed in Searle's original target article. There's nobody home in the Chinese Room except Searle! The rest is just chalk and blackboard. What do the "Systems" enthusiasts reply? "So be it. Well then Searle + chalk understands." Now try to think back -- way, way back: Were you in the habit of thinking that a person plus chalk could be understanding where the person alone was failing to understand? (Sometimes certain kinds of "education" can be a handicap...) " Strangely enough, I agree with your prediction [that the premise that " the LTT (sic) could be successfully passed (till doomsday!) by symbol " manipulation alone is false], and with your solution. But that is not " important for the Chinese Room argument. You could start with a robot " that is embedded in the world, and after it achieves full understanding " (the same way humans do: learning within the context of a society), " then you can disconnect the sensors and effectors and leave only a " teletype to the outside world. Sounds a lot like Stephen Hawking in " real life, no? Suddenly, the LTT (i.e., the original Turing Test) " returns. You can't agree with my solution, because unfortunately you have not understood it. If nonsymbolic functions are ESSENTIAL to passing both the LTT and the TTT then it's not just a matter of sensors and effectors connected to a symbol-crunching core. And Stephen Hawking is not just a symbol-crunching core. " while you may be correct that understanding (getting the proper set of " rules) is impossible without being embedded in the world, why make this " part of the test? It's just an implementation issue... Because understanding may NOT be just a matter of "getting the proper set of rules." If not, then even if a savvy robot jumped fully educated out of the head of Zeus it could not pass the LTT or TTT without internal NONSYMBOLIC functions; and if these were removed, it would no longer be able to pass the LTT or TTT (and would no longer have a mind). (Moreover, removing them would NOT be just a matter of yanking off sensors and effectors from a symbol-cruncher.) " ["Searle is all there is to the `system.'"] Completely, 100%, false. " Wrong. Incorrect. The Chinese Room contains Searle AND THE RULES. And " the system as a whole DOES understand, as evidenced by the Chinese " answers to Chinese questions. Unshakeable conviction. Who am I to try to counter the effects of an education that gave rise to this? (I only timidly remind you again that the possibility of successfully passing the LTT by symbol-crunching alone was a hypothetical, so-far-counterfactual PREMISE that Searle simply carried over from Strong AI for the sake of argument. It may well be false; I've given reasons why it may be false. In fact, Searle's Argument itself can be taken as one of the reasons for concluding that it's false. Simply reiterating the premise cannot serve as a logical counterargument against the VERY untoward conclusion that FOLLOWS from the premise.) Refs: Searle J. (1980) Minds, Brains and Progams. 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