From newshub.ccs.yorku.ca!torn!cs.utexas.edu!uunet!trwacs!erwin Wed Aug 12 16:52:08 EDT 1992
Article 6546 of comp.ai.philosophy:
Xref: newshub.ccs.yorku.ca comp.ai.philosophy:6546 sci.cognitive:273
Path: newshub.ccs.yorku.ca!torn!cs.utexas.edu!uunet!trwacs!erwin
>From: erwin@trwacs.fp.trw.com (Harry Erwin)
Newsgroups: comp.ai.philosophy,sci.cognitive
Subject: Chaos on the Brain
Keywords: chaos cognition neurobiology
Message-ID: <683@trwacs.fp.trw.com>
Date: 3 Aug 92 01:23:26 GMT
Followup-To: sci.cognitive
Organization: TRW Systems Division, Fairfax VA
Lines: 22


I'm starting to see a pattern--biological systems approximating a functional
capability and then relying on tuning. Instead of holography, pseudoholography
implemented in distributed, associative memory, that comes close enough
once tuned by learning. Instead of wavelets, filter banks, tuned by learning,
that come close enough. There's likely to be something similar for some of
the applications of chaos. For instance, a chaotic process is an efficient
starting point for pattern matching since it has a continuous frequency
spectrum--thus any desired periodic function can be constructed by
selective direct amplification/deamplification. But it's unlikely the
brain is chaotic in a structurally stable sense. Instead, it is probably
reset to a chaotic process whenever it drifts away. Think of Lou Pecora
and Tom Stafford's algorithm for controlling chaos and reverse it. Instead
of capturing the process and controlling it, try to spot when the process
has become predictable, capture it, and then drop it near a hyperbolic
fixed point. That's how we do problem-solving...


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



