From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!cs.yale.edu!mcdermott-drew Thu Feb 20 15:20:05 EST 1992
Article 3659 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca comp.ai.philosophy:3659 sci.philosophy.tech:2117
Newsgroups: comp.ai.philosophy,sci.philosophy.tech
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!cs.yale.edu!mcdermott-drew
>From: mcdermott-drew@CS.YALE.EDU (Drew McDermott)
Subject: Re: Reference (was re: Multiple Personality Disorder and Strong AI)
Message-ID: <1992Feb12.040025.14716@cs.yale.edu>
Keywords: consciousness,functionalism,meaning
Sender: news@cs.yale.edu (Usenet News)
Nntp-Posting-Host: aden.ai.cs.yale.edu
Organization: Yale University Computer Science Dept., New Haven, CT 06520-2158
References: <1992Feb4.214433.9121@psych.toronto.edu> <1992Feb7.162533.4653@cs.yale.edu> <1992Feb7.232150.8611@husc3.harvard.edu>
Date: Wed, 12 Feb 1992 04:00:25 GMT
Lines: 63

  In article <1992Feb7.232150.8611@husc3.harvard.edu> zeleny@zariski.harvard.edu (Mikhail Zeleny) writes:
  >In article <1992Feb7.162533.4653@cs.yale.edu> 
  >mcdermott-drew@CS.YALE.EDU (Drew McDermott) 
  >
  >>However, I acknowledge that the theory does depend on a
  >>"correlationist" theory of reference and meaning.  That is (as we've
  >>hashed out before), it depends on meaning being objectively given by
  >>correlations between model and thing modeled, and not being dependent
  >>on "original intentionality" or the like.
  >
  >Recall your earlier claims about semantics.  What determines the meaning's
  >being objectively given by correlations, etc.?

I assume the apostrophe is a misprint, and the question is "What
determines the meanings being objectively given by correlations...?"

I guess I'd like to say the meanings *are* the correlations.

I don't claim to have a full theory of meaning, but what we need for
the present purpose is an explanation of "System P has a model M of
system Q."  (In particular, we're interested in the case where P=Q.)
It seems to me to be sufficient if 

   M is a part of P
   M goes through transitions that fairly reliably track or anticipate
the properties of Q.  (I.e., the states of M are "approximately"
homomorphic to the states of Q.)
   P's behavior with respect to Q is caused to be "appropriate" by the
states of M (or, if you prefer, the outputs of M)

If a state S_M of M tracks or anticipates a state S_Q of Q in this
way, we say that S_M "means" S_Q.  (Often it will be more appropriate
to speak of a property of a state, which is the same as an equivalence
class of states.  E.g., a symbol may be said to denote an object type
if states of M in which tokens of the symbol are present are
correlated with states of Q in which objects of that type are present.)

I don't think the occurrence of terms such as "approximate" and
"appropriate" should be held against the theory.  For any given
candidate M, we replace these terms with more precise specifications
that fit the particular case.  The notion of symbols and objects being
"present" will also require fleshing out in different ways for
different applications.  Ponder the way in which "unicorn" refers to
unicorns in a model containing a representation of "There are no
unicorns."  It's entirely possible that there is no general theory of
how things mean.  And I suppose that there will inevitably be cases
where there is just no answer to the question, What does this state or
symbol of M refer to, if anything?  

I'm sorry for the weird combination of formality and vagueness in this
article, but I realize that this is the wobbliest part of my theory,
and I want to avoid begging any questions.

What I really want to say is that a the symbols used by a system mean
"X" if the system behaves as if the symbols meant "X".  But I'm trying
to avoid saying that it's a matter of taking an intentional stance
toward the system, as if it were up to an external observer to choose.
That is, they really do have the meanings they appear to have, because
there's ultimately nothing else for meaning to mean.

                                             -- Drew McDermott




