Re: Contexts and quantifiers in KIF

sowa <sowa@turing.pacss.binghamton.edu>
Reply-To: cg@cs.umn.edu
Date: Thu, 15 Apr 93 00:27:50 EDT
From: sowa <sowa@turing.pacss.binghamton.edu>
Message-id: <9304150427.AA09445@turing.pacss.binghamton.edu>
To: jfulton@atc.boeing.com, sowa@turing.pacss.binghamton.edu
Subject: Re: Contexts and quantifiers in KIF
Cc: cg@cs.umn.edu, interlingua@isi.edu, sowa@turing.pacss.binghamton.edu,
        srkb@isi.edu

Jim,

I agree completely that the criteria for truth depend on "what we
learn from the real world and extrapolate beyond it."  I also agree
that "we have criteria for exploring contrary-to-fact hypotheticals...,
for distinguishing truth from falsity, consistency from inconsistency,
likely from unlikely."  And I even agree that "what is expressed by a
contrary-to-fact conditional... is a real fact about the world."

But the basic point that I have been insisting on all along is
Aristotle's three-way distinction between words, "experiences
in the psyche", and things in the world.  The fundamental flaw
in the work by Kripke, Montague, and many others is that they
have tried to relate words directly to things while ignoring
the concepts, "experiences in the psyche", or whatever you want
to call them.  Since the term "experiences in the psyche" is
too anthropomorphic for computer systems, I prefer to call them
"data structures" when I implement them.  But I believe that such
a middle level is essential for explaining all the facts about
language use, planning for as yet nonexistent futures, reasoning
about hypotheticals, recalling the past, etc.

I am not arguing against model theory, modal logic, or counterfactual
conditionals.  The only point I am objecting to is the "identification"
of the abstract models of model theory with the real world.  I do believe
that for a model to be useful, there must be a mapping from the model
to the world, but that mapping is never an isomorphism.  It is at
best an approximation that may be more or less adequate for one purpose,
but probably inadequate for many other purposes.

Example:  During the early 20th century, one physicist "proved"
that it was impossible for anything to move through the air faster
than the speed of sound.  It turned out that his "proof" was based
on certain simplified equations of fluid mechanics that were derived
on the assumption that the velocity v was much less than the speed
of sound.  Those equations may have been fine for their original
purpose, but they were egregiously wrong for another purpose.

Perhaps if we were all omniscient and could define models that 
captured all the details of reality in absolute precision, then
our models would be truly isomorphic with the real world.  But no
model has ever attained that degree of precision, and I don't believe
that the universe could ever contain enough paper or computer chips 
to model the entire universe with absolute precision.  Therefore,
we must acknowledge our limitations and admit that no model can
ever be perfect and no truth can ever be absolute.  There are
more things in heaven and earth than any model could ever represent.

John