kck@g.gp.cs.cmu.edu (Karl Kluge) (03/14/89)
From: roelw@cs.vu.nl Subject: Chinese room argument > 2. Put the UTM in a box equiped with video cameras, wheels and a motor and > let it drive in the street. Again there is no difference to its computation > if you interpret "green" as red etc. (You may put any other > symbol-manipulating device in a box with the same result.) Fine. Let's suppose that the presence of a red traffic light facing the box in the camera's field of view causes the symbol "xyz" to be generated, while the presence of a green traffic light doesn't. Let's suppose that the behavior of the UTM in driving the box is perfectly satisfying -- it avoids accidents, follows traffic rules, etc. Let's pretend that the programmer decides to change the denotation of the symbol "xyz" from "there is a red traffic light facing me" to "there is a green traffic light facing me" (me = the box), and does this without touching the box to change it's program or wiring. Fine. The box still drives in the same way as it did before the programmer did this. The programmer has chosen to do something bizarre. The games formal logicians play allow the programmer to do this. That doesn't make what the programmer has done sensible, since the formal games logicians play are exactly that, and you can't get more out of them conceptually than went into them in the first place. > The general result is that a symbol-manipulating device is not affected by > the denotation you or anyone else give to the symbols it manipulates. If you are talking about the process by which strings of output symbols get produced, this is true but irrelevant. The process by which *you* generate the strings of symbols that form your posts (regardless of whether your mind is describable by a formal system) is not affected by the denotation I or anyone else give to the symbols you produce. That is not a demonstration that you do not "know" the denotations of those words, yet you want us to believe that it is a demonstration that a UTM does not (and cannot) "know" the denotations of its symbols. > 3. Talk with someone, using the word "green" to mean red and "red" to mean > green, without the other person's knowing about this change in denotation. > S/he quickly will find something weird about your conversation;... And if you persist in acting as though I mean green when I say "red", I will quickly find something weird about the conversation. And if you persist in acting as though a robot means green when it says "red" *simply because you programmed the robot, and despite everything the robot says and does to prove that it means red when it says "red" (picking the right crayon from a box, etc)*, then I will also quickly find something odd about *your* conversation -- not the robot's conversation. > What this shows, I think, is that the denotation of public symbols is > publicly known and that if you make a private change in what is > conventionally denoted by a symbol, you will get social problems. > Conventions about what the denotations of symbols are, are not (merely) in > the head of one individual but are social institutions. First, let's acknowledge the distinction (popping up explicitly for the first time) between the "public" symbols produced by a system (the words sent along the teletype to the subject in the Turing Test, for instance), and it's "private" symbols (gensyms produced by LISP, for instance). Neither the programmer or the system can reasonably insist on ignoring the social conventions as to the denotations of the system's "public" symbols -- if the program insists on describing the American flag as "red, white, and cream cheese", then there is something odd about the program; and if the programmer insists that the program means "orange" when it says the word "blue" in describing the American flag as "red, white, and blue", then there is something odd about the programmer. There are no social conventions, and hence no privileged denotations for the system's "private" symbols. Second, to talk of an entity knowing "the" denotation of a symbol implies that a symbol *has* an intrinsic denotation, and as you have been so persistent in pointing out that just isn't true. Third, to talk of "knowing" the denotation of a symbol is dubious. If you insist that it's possible, then I'm afraid there are several million deconstructionists who'd like to have a word with you in the hall. As far as I can tell, the only indication that I "know" the denotations of the words I use is 1) I have some goal in mind (I'd like a bowl of ice cream) 2) I emit a string of symbols ("John, would you get me a bowl of ice cream while you're getting yourself one?"), and 3) Things in the environment react in such a way that my goal in emitting the string of symbols I did is satisfied (my apartmentmate brings me a bowl of ice cream). Since this happens fairly consistently, I assume that the denotations I have for the words I use roughly overlaps the denotations of those words in the minds of those I talk to, but that's the most I can conclude. What I say is not what you read. What you say is not what I read. Fourth, to talk of knowing the "denotation" of symbols is also questionable. If I have some robot whose sensors produce a symbolic description of the world which is transformed by a formal system into a symbolic description of actions for the robot's effectors (the view Steve Harnad objects to so vehemently), * the syntax of the symbols in the formal system is real, in that it describes the causal relationship between successive states of the physical machine, and * the semantics of the symbols are also real in that the input symbols are causally related to the patterns that the world creates in the sensor data, and the effects of the system's actions are causally linked to the symbols that result in those actions. The denotations of the symbols have no corresponding reality. They are a matter of arbitrary social consensus (in the case of symbols emitted by the system to communicate with others, including humans), and may not be defined at all (who's to decide what denotation some symbol "X" has in some production "yXzz --> yXyzz" buried deep in the symbol system away from public inspection? There is no privileged, unique denotation). I could define other arbitrary extrinsic properties of symbols ad nauseum (for instance, "blurbness", which is the property of all symbols which remind me of a hamburger -- I clearly know the blurbness of an arbitrary symbol, and I know that I have a mind if anyone does, therefore since neither you nor an arbitrary formal system can know the blurbness of the symbols you use (the English words you emit if it happens that you aren't just a formal system anyways) only I have a mind. Q. E. D.) You want the supporters of AI to accept the bizarre notion that Entity A can insist on defining the denotations of symbols produced by Entity B, regardless of the behavior of Entity B. I see no reason to accept that at all. I certainly don't accept it when the two entities are humans. I see no reason to accept it when one of the entities is a computer, and you have provided no reason other than "When formal logicians play with symbol systems, this is what they do." By the way, *who* is "making a private change in what is conventionally denoted by a symbol" when the programmer insists that the robot means green when it says "red" (and acts in a way that indicates that it does)? Certainly not the robot. Karl Kluge (kck@g.cs.cmu.edu) --