Path: utzoo!attcan!uunet!cs.utexas.edu!usc!apple!rutgers!mcnc!rti!ntpdvp1!sandyz
From: sandyz@ntpdvp1.UUCP (Sandy Zinn)
Newsgroups: comp.ai
Subject: Re: Chess, Reductionism, Probablistic Determinism.
Summary: sorting out "interpretation" & value
Keywords: interpretation, intentional, Chinese Room
Message-ID: <379@ntpdvp1.UUCP>
Date: 14 Apr 90 00:13:16 GMT
References: <491fffd5.1a4d7@cicada.engin.umich.edu> <2080@aipna.ed.ac.uk> <c6rR02FT99Vj01@amdahl.uts.amdahl.com>
Organization: Northern Telecom DMS-10 Div., Raleigh, NC
Lines: 81

(Ken Presting) writes:
> 
> What I had in mind [is something that] I
> (eventually) want to call "Intentional" behavior, or more precisely,
> behavior _qua_ intentional.  This distinction is based on what abstraction
> is used to describe the behavior.  Physical, chemical, biological, or
> medical abstractions are non-intentional, while descriptions which
> refer to "propositional attitudes" such as belief and desire are
> intentional.
> 
> Whereas non-intentional descriptions are justified (or criticised) by
> citing observations *within* the vocabulary of a single abstraction,
> intentional descriptions are conditional on INTERPRETATIONS of cited
> observations in the vocabulary of *another* abstraction.  So intentional
> predicates are very much like normative predicates.  The only difference
> is in connotation - normative predicates involve value judgements, while
> intentional predicates do not.
 
Wait a minute, Ken.  There are some mixed referents here:  if intentional
descrs. are dependent on Interpretations, then it's clear that they MUST
involve value judgments.  Value is precisely the interpretation of an 
observation's (a behavior's or a property's) location in a scheme which is
held to be relevant.  The choice of a relevant scheme is a value judgment,
as is the activity of locating the observation in that scheme: interpreta-
tion.  VALUE is linked in the Socratic sense to *sense*.  I would even say
that *value* is a measure of order, of information (which means that money-
grubbers, since currency currently carries so little information, are indeed
grasping at straws -- just what I'd want to have operating in MY paradigm.)

What _particular_ frame of reference (or value) are you NOT wanting to apply
to intentional predicates which you think IS applied to normative predicates,
and why do you feel it is not *sensible* to do so?

I'll guess that you are talking about a sociocultural frame, in which case
the question becomes, why isn't it sensible to relate intentionality to
these values?  If you want to suspend the individual's *beliefs and desires*
from reference within a larger social context, I'm going to jump all over you,
so you'd better have a high-information-quotient reason.

(On second thought, maybe you're going to suspend *logic*, which you might
be able to get away with.)

(Ken Presting):
> >> Even a Universal Turing machine can apply only one interpretation to
> >> the data on its tape.
> 
> (Steven Daryl McCullough) writes:
> >... Are you saying that Searle's argument *does* prove that a Turing
> >machine can't think? . . .
>
> Yes.  ...observe that a 
> compiler will take any garbage input file, try to parse it, and barf if
> it hits a syntax error.  That's what I mean by a program "applying an
> interpretation".  The compiler acts as if every file it gets its hands on
> is a source program, and if the file doesn't fit the compiler's
> "conceptual scheme"  (ie parsing algorithm) then the compiler just dies.
> An AI program needs to manipulate its own "concepts" along with its input.
 
A very clear explication, which I agree with.  This means not only recur-
sive transformation, but also *excursive* transformation -- ie, the string 
of symbols itself needs a second pass.  One of the things that bugs me
about the TM model is its reduction of interaction to a serial stream
going merrily past.  TOO SIMPLE.  And therefore tasteless. (IMO, natch.)
 
> [ about the Symbol Grounding Problem. ]
> Hilary Putnam [ with reference to problems of unique interpretation ]
> mentions on the very last page.  He thinks it entails Behaviorism.  On
> Functionalist assumptions, it may, but on Implementationist assumptions,
> this is not so.
 
I have a friend who's trying to convince me that Skinner advocated a
very different view of grounding and interpretation than is commonly
assumed; more along the lines of contextual frame orientation and
analogic reference than the usual serial S => R model.  When I have a
little better grip on it, I might bring it up here in a grounding/
interpretation context, to see what flies.  (or what flies it attracts!)

@ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @ @
    Sandra Zinn              |   "The squirming facts
    (yep these are my ideas  |      exceed the squamous mind"
     they only own my kybd)  |         -- Wallace Stevens