From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uunet!tdatirv!sarima Thu Feb 20 15:22:21 EST 1992
Article 3890 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!uunet!tdatirv!sarima
>From: sarima@tdatirv.UUCP (Stanley Friesen)
Newsgroups: comp.ai.philosophy
Subject: Re: Determinism precludes truth?
Message-ID: <430@tdatirv.UUCP>
Date: 19 Feb 92 21:19:32 GMT
References: <407@tdatirv.UUCP> <1992Feb12.205814.75156@spss.com> <417@tdatirv.UUCP> <1992Feb17.224820.7895@spss.com>
Reply-To: sarima@tdatirv.UUCP (Stanley Friesen)
Organization: Teradata Corp., Irvine
Lines: 45

In article <1992Feb17.224820.7895@spss.com> markrose@spss.com (Mark Rosenfelder) writes:
|In article <417@tdatirv.UUCP> sarima@tdatirv.UUCP (Stanley Friesen) writes:
|>However, what I was suggesting, very tentatively, was that real-world
|>ecosystems ae so complex, and contingent, and changeable that by and
|>large only true beliefs are actually useful (to the extent that they
|>contribute to reproductive success).
|
|I suppose it's not fair to criticize a tentative suggestion, but a few
|problems spring to mind.
|
|1. Human beliefs are so various, not to say rococo, that it would seem hard
|to maintain that the most useful beliefs also tend to be true.
|
It is true that this approach would need considerable tuning to fit with
observed facts.

It is also true that in a deterministic universe this seems to be the
only way of getting reliable knowledge.
|
|2. Your theory should address more than mere reproductive success, since
|it's hard to see this applying to beliefs at all except in the very long
|term.  If Chomsky's innatism superseded behaviorist theories of language,
|this was not because Chomsky's followers reproduced more efficiently.

I was not talking about individual ideas or beliefs, I was talking about
the occurance of reliable mechanisms for learning about the world.

Thus, I see evolution (which equals reproductive succes for our purposes)
as a filter that throws out variants that cannot learn about their environment.

Remember, in a deterministic universe you and I 'learn' what the initial
configuration of particles determined we would 'learn'.  There is no a
priori reason to suppose that such 'knowledge' is anything except a random
collection of bits.  The question is then, is there any mechanism by which
even a trace of reliability can be introduced into the system?  Any way in
which what we 'learn' has *some* relationship to what actually is (no matter
how tenuous)?

This is the question I am trying to propose a possible answer to.
[Actually, I tend to prefer a non-deterministic universe, but that
is just prejudice].
-- 
---------------
uunet!tdatirv!sarima				(Stanley Friesen)



