Newsgroups: sci.nonlinear,sci.cognitive,comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!gatech!EU.net!news.sprintlink.net!noc.netcom.net!netcom.com!kovsky
From: kovsky@netcom.com (Bob Kovsky)
Subject: Re: Chaos and Computation
Message-ID: <kovskyD8HBx3.I15@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <3opic3$kbb@uuneo.neosoft.com> <kovskyD8Dwu7.AAG@netcom.com> <dougD8FM8y.GI6@netcom.com> <800300035snz@longley.demon.co.uk>
Date: Fri, 12 May 1995 19:10:15 GMT
Lines: 42
Sender: kovsky@netcom7.netcom.com
Xref: glinda.oz.cs.cmu.edu sci.nonlinear:3126 sci.cognitive:7580 comp.ai.philosophy:28003

David Longley  <David@longley.demon.co.uk> wrote:

[deletions]

> ...  When  some
>say that Artifical Meural Networkers are not on the right track,  I  wonder
>*what* the alternative could *possibly* be like, since the whole  neuroaxis
>looks like a lot of layers between sensory systems, sensory systems & motor
>systems, and motor systems and motor systems. Extracting invariant patterns
>out of what otherwise seems like random noise seems to be *exactly* what we
>are doing all the time, and over time - ie identifying strange  attractors.
>The hard part surely is building systems which might approximate those that
>natural selection has had such an advantage in building and tuning over so
>many eons.
>
>I should imagine that just about  *anyone*  working  in  neuroscience would
>accept the main themes previously discussed in this thread- they'd probably 
>just say, 'yes - well, that's what we started with...

	Neuroscientists who deal with the <chemistry> of the brain 
(instead of its electrical interactions) say that the brain is more like 
a gland than like a computer.  The brain incorporates a rich broth of 
chemicals, highly variable, which effect activity at synapses, overall, 
and, apparently, over regions everywhere in-between.  The ANN's I've seen 
have all involved only point-to-point interactions.

	Nor do I see that we are "[e]xtracting invariant patterns ... all 
the time..."  There are certainly some invariants, even variations 
treated as invariants (like the letters of the alphabet), but there is 
also responsiveness to and creation of enormous diversity.  "Theme and 
variations" is not of music alone.  Likewise, the notion of "strange 
attractors" simply does not correspond to the (non-hyperbolic!) 
infinities of gradations that can be accomodated by so constrained a 
system as hand and eye (drawing, using machine tools, flower arranging).  
Then consider the ways in which we can describe a visual image.  Very 
"strange" attractors indeed!

-- 

*   *    *    *    *    *    *    *    *    *    *    *    *    *    *    *   * 
    Bob Kovsky          |  A Natural Science of Freedom 
    kovsky@netcom.com   |  Materials available by anonymous ftp
                        |  At ftp.netcom.com/pub/fr/freedom
*   *    *    *    *    *    *    *    *    *    *    *    *    *    *    *   * 
