Re: Hawkins ideas on building AI's

From: bkaz (bkaz__at_hotmail.com)
Date: 10/22/04 

Date: 22 Oct 2004 16:56:55 -0700

> B: That is pattern recognition, & that's what cortex does on every level.
> Memory/association is just another term for it.
 
> GS: Not exactly. Although it is true that much of what is called "memory" is
> simply the effects of exposure to repeated presentations of stimuli
> (habituation), "pairings" of stimuli (classical conditioning), and "3-term
> contingencies" (operant conditioning), there is a bit more. That is, when
> animals are exposed to, for example, delayed match-to-sample procedures,
> they learn to do subtle things that mediate the temporal gap. The
> acquisition of such behavior is, to be sure, the result of conditioning, but
> it is not just one simple operant response class.

There're probably many redundant mechanisms in the brain, evolution is
dumb but we can do better.
 
> B: There're lots of redundant terms in AI, philosophy, neuroscience..., but
> "Entities must not be multiplied unnecessarily" (Occam)
 
> GS: But, unfortunately, cognitive psychology's raison d'etre is the
> promiscuous invention of explanatory fictions. And cognitive psychology has
> corrupted AI, neuroscience, etc.

I don't know who's corrupted whom, it seems human brain just isn't
good at abstractions.

Boris.