Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!usc!apple!amdahl!kp
From: kp@uts.amdahl.com (Ken Presting)
Newsgroups: comp.ai
Subject: Re: Sci Am AI debate: Searle's 3rd Axiom and the (a+b)%16 room
Summary: Syntax determines a class of semantics
Keywords: Searle, Chinese Room, syntax, semantics
Message-ID: <23Q302Ee7d0m01@amdahl.uts.amdahl.com>
Date: 9 Jan 90 22:59:36 GMT
References: <7502@pt.cs.cmu.edu>
Reply-To: kp@amdahl.uts.amdahl.com (Ken Presting)
Organization: Amdahl Corporation, Sunnyvale CA
Lines: 32

In article <7502@pt.cs.cmu.edu> kck@g.gp.cs.cmu.edu (Karl Kluge) writes:
>It occurs to me that Searle's argument (in particular his 3rd axiom, "Syntax
>by itself is neither constructive of nor sufficient for semantics") depends
>on the reader accepting as plausible and natural a rather bizarre claim.
>Specifically, in defense of this axiom on page 31 he says, "But now imagine
>that as I am sitting in the Chinese room shuffling the Chinese symbols, I
>get bored with just shuffling the -- to me -- meaningless symbols. So,
>suppose that I decide to interpret the symbols as standing for moves in a
>chess game. Which semantics is the sysytem giving off now? Is it giving off
>a Chinese semantics or a chess semantics, or both simulaneously?"
>
>Well, this is pretty strange if one assumes that once Searle decides to
>start interpreting the symbol shuffling as a chess game, the Chinese speaker
>outside the room continues to see what looks like a coherent conversation in
>Chinese. Are we to suppose that you can impose an interpretation on the
>symbols such that they form both a coherent conversation in Chinese *and* a
>legal and coherent chess game? ...

  (cogent example deleted)

> ...  (Remember, we're not talking about "semantics" in
>the sense that someone who does formal logic does - in fact, if Searle is
>correct, we *can't* be talking about the same kind of "semantics".)

I think Karl has a crucial point here.  Searle's claim that "syntax is
neither constructive of nor sufficient for semantics" is true when applied
even to formal semantics, but does not have the strong implication Searle
needs.  What's true is that no syntax determines a unique semantics, even
when axioms and inference rules are specified.  That's how we get non-
standard models of the real numbers, and other non-standard interpretations
of mathematical theories.  But Karl is right to point out that there are
limits to the interpretation of a formal grammar.