arm@ihlpb.ATT.COM (Macalalad) (02/24/89)
Being naturally curious why some people, notably Stevan Harnad, see only one system in the Chinese room, where as others, notably me, see two distinct systems, I decided to dig up Searle's response to the systems theory: "My response to the systems theory is quite simple: Let the individual internalize all of these elements of the system. He memorizes the rules in the ledger and the data banks of Chinese symbols, and he does all the calculations in his head. The individual then incorporates the entire system. There isn't anything at all to the system that he does not encompass. We can even get rid of the room and suppose he works outdoors. All the same, he understands nothing of the Chinese, and a fortiori neither does the system, because there isn't anything in the system that could understand because the system is just a part of him. "Actually I feel somewhat embarrassed to give even this answer to the systems theory because the theory seems to me so implausible to start with. The idea is that while a person doesn't understand Chinese, somehow the conjunction of that person and bits of paper might understand Chinese. It is not easy for me to imagine how someone who was not in the grip of an ideology would find the idea at all plausible." Being one of those "in the grip of an ideology," I find it remarkably easy to recognize two systems, and Searle's reply of "internalizing" the second system only clouds the issue. His process of internalization does not get rid of the second system; it merely transfers the medium from pieces of paper to someone's brain. (Didn't someone say that memorization was the lowest form of learning?) What I would argue is that for true internalization to take place, the rules must be converted from one system to the other. In other words, the person has to just sit down and learn Chinese. Perhaps this example will help. Let's take a variation of the Chinese room where the purpose of the room is to interpret BASIC instead of to understand Chinese. To make it interesting, and to prevent the person inside from taking any shortcuts, let's make the BASIC Chinese, with the person inside knowing no Chinese or BASIC whatsoever. As the thought experiment progresses, authors of Chinese BASIC programs get the expected interpretation of their programs (although they may remark at how slow the implementation is). Can the person interpret Chinese BASIC? Obviously not. She's only performing calculations and moving bits of paper around. Can the person interpret Chinese BASIC even if she memorizes all of the bits of paper? In a sense, yes, in the same sense that the person in the Chinese room who has memorized his bits of paper can understand Chinese. Both would appear to the objective observer to interpret Chinese BASIC or understand Chinese, depending on which set of papers were memorized. But introspectively, the person in the Chinese BASIC room is not doing any interpretation of Chinese BASIC, and I'm sure that she would be the first to admit it. No, she's still only performing calculations and remembering bits of information. Is it fair to conclude, then, that the system of the person in the Chinese BASIC room and her bits of paper is not really interpreting BASIC? Those Chinese BASIC programmers would surely be astonished at the deception, were this the case. In fact, the processor of your computer probably does not directly interpret BASIC, so doesn't it surprise you to learn that your computer was fooling you every time it ran one of your BASIC programs? Now who's in the grip of whose ideology? Hopefully, we all see now that there indeed are two different systems of understanding in the Chinese room. One has bits of information being moved around and changed on bits of paper by human hands. The other has bits of information being moved around and changed by neurons and synaptic firings. Searle questions how understanding can arise out of the first system. I question how understanding can arise out of the second. What I'm waiting for is an explanation of how understanding can arise out of the second but not the first. Searle's Chinese room thought experiment tends to cloud this issue rather than shed any light. Searle's thought experiment is certainly thought provoking, but it certainly isn't compelling. I see no serious threat to strong AI here. What I do see is how ridiculous it is to argue that one system can understand and another system cannot without having a firm grasp on what understanding is all about. Despite Stevan's strident pointing and categorizing those systems which do understand (and woe to he who MIScategorizes), I still think that such pointing is a bit premature. Alchemy is a dead science, but it took advances in science, not clever philosophizing, to kill it. (Of course, further advances in science showed that the transmutation of elements was indeed possible, but the resulting study in nuclear physics bear little, if any, resemblance to alchemy.) Until similar levels of understanding on understanding are reached, I reserve judgment and stick to wondering.
harnad@elbereth.rutgers.edu (Stevan Harnad) (02/24/89)
This is a reply to two successive postings by arm@ihlpb.ATT.COM (A. R. Macalalad) of AT&T Bell Laboratories, who wrote: " Now for the sake of argument, let's assume that there is a " distinction between Searle and (Searle + rules)... the only entity able " to decide if (Searle + rules) really understands Chinese is (Searle + " rules). Not you or me or any outside observers or even Searle himself. " Only (Searle + rules). Of course, if we are assuming that much for the sake of argument -- namely, a separate entity that exists and understands -- then of course there is no argument. You've assumed it all. " The issue I now want to take up is your justification of the " Total Turing Test... [tap-dancing, etc.]... The justification for the Total (robotic) Turing Test (TTT) in preference to the Language-In/Language-Out Turing Test (LTT) is fourfold (and has nothing to do with arbitrary calls for tap-dancing): (1) The TTT is what we already use with one another in our everyday, practical "solutions" to the other-minds problem -- not the LTT, which we only use, derivatively, with pen-pals. (2) The TTT (fine-tuned eventually to include neuronal "behavior" too) encompasses all the available empirical data for the mind-modeler. (The only other data are subjective data, and I advocate methodological epiphenomenalism with respect to those.) The LTT, on the other hand, is just an arbitrary subset of the available empirical data. (3) The LTT, consisting of symbols in and symbols out, is open to a systematic ambiguity about whether or not everything that goes on in between could be just symbolic too. (I conjecture that the LTT couldn't be passed by a device that couldn't also pass the TTT, and that a large portion of the requisite underlying function will be nonsymbolic.) (4) Evolution, the symbol grounding problem, and common sense all suggest that robotic (TTT) capacities precede linguistic (LTT) capacities and that the latter are grounded in the former. " Of course, if you'd rather offer an objective definition of... " understanding, please feel free.... As stated many, many times in this discussion, and never confronted or rebutted by anyone, this is not a definitional matter: I know whether or not I understand a language without any need to define anything. " Conduct on the net... I think that a few other apologies are due. I'm trying to criticize views and arguments, not people. If I have offended anyone, I sincerely apologize. (It seems not that long ago that *I* was the one preaching against intemperate and ad hominem postings on the Net as not only ethically reprehensible but an obstacle to the Net's realizing its full Platonic potential as a medium of scholarly communication.) " Being one of those "in the grip of an ideology," I find it remarkably " easy to recognize two systems, and Searle's reply of "internalizing" " the second system only clouds the issue... for true internalization to " take place, the rules must be converted from one system to the other. " In other words, the person has to just sit down and learn Chinese.... One of the tell-tale symptoms of being in the grip of an ideology is that one can no longer tell when one is begging the question... " Let's take a variation of the Chinese room where the purpose of the " room is to interpret [Chinese] BASIC instead of to understand " Chinese... Is it fair to conclude, then, that the system of the person " in the Chinese BASIC room and her bits of paper is not really " interpreting BASIC?... Now who's in the grip of whose ideology? The suspicious reader who might think I stacked the cards by clipping out the ARGUMENTS in pasting together the above excerpt will be surprised to see, upon reading the entire original posting, that there ARE no arguments: The Chinese Room has simply been reformulated in Chinese Basic, and voila! (There's also a double-entendre here on the syntactic vs. the mentalistic meaning of "interpret.") Mere repetition of credos is yet another symptom of ideological grippe. -- 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