From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!qt.cs.utexas.edu!yale.edu!spool.mu.edu!uunet!trwacs!erwin Tue Mar 24 09:57:23 EST 1992
Article 4604 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!qt.cs.utexas.edu!yale.edu!spool.mu.edu!uunet!trwacs!erwin
>From: erwin@trwacs.fp.trw.com (Harry Erwin)
Newsgroups: comp.ai.philosophy
Subject: Re: 'goals' vs. 'features'
Message-ID: <517@trwacs.fp.trw.com>
Date: 19 Mar 92 16:43:07 GMT
Article-I.D.: trwacs.517
References: <ksfc1gINNkvc@exodus.Eng.Sun.COM>
Organization: TRW Systems Division, Fairfax VA
Lines: 44

silber@orfeo.Eng.Sun.COM (Eric Silber) writes:

> Although Evolution proceeds as a statistical phenomenon mediated
> by Natural Selection, the way in which Natural Selection chooses goals
> may properly be viewed as a kind of 'local-teleology'.

> Example:
> In the primordial world of unicellular organisms , some mutations
> lead to photosensistive membrane-regions , there was a competititve
> advantage in being differentially reactive to light, therefore
> Natural Selection selected FROM THAT POINT IN TIME ON for 'vision'.
> Thus random , selected primordial 'features', became CONTINUOUSLY
> selected for 'goals'.

What you are implying is that since the dynamics of selection converge,
you have teleology. I suppose that is a _possible_ interpretation. However,
there is no evidence of design involved in the process, just random
variation, winnowed by natural selection. We know of cases where the
direction of evolution has reversed, and that is hardly what would expect
if there were an optimal design to which the system were converging. Also
we know that most genes are at the local optimum almost all of the time.
What variation appears is random and disadvantageous for the most part.
Let me give you an example.

A few years back, I demonstrated that the ESS for a specific non-zero sum
game could be expected to evolve chaotically (given certain assumptions on
the mechanism of evolution--specifically that it exhibited the same sort
of local teleology that you suggest). Gordon Hines was interested in
whether this implied that there was an optimal, non-zero rate of evolution
for the population strategy. I simulated Darwinian evolution for the
system and found that random variants were so frequently deleterious that
the lowest possible rate of evolution was advantageous, even though that
meant that the population strategy fixated on the best available strategy
in the initial set, which was not the _best_ possible strategy at any
time. The point is that although the _optimal_ population strategy would
evolve, the system actually fixated on a suboptimal strategy. No evidence
of teleology here--the system does not converge to an optimal design--
rather evidence that fitness will be maximized, whether or not it
corresponds to an optimal design.

Cheers,
-- 
Harry Erwin
Internet: erwin@trwacs.fp.trw.com


