Re: The Emotional Computer

From: Andrea Gallagher (drea@alumni.stanford.org)
Date: Wed Apr 02 1997 - 19:10:23 MST


At 08:39 AM 3/30/97 -0800, John K. Clark wrote:

>Nature found it much easier to come up with feeling than the ability to
>reason, it's certainly came up with it first. The most ancient part of the
>brain, the spinal cord, the medulla and the pons is similar to the brain of
>a fish or amphibian and first made an appearance on the earth about
>400 million years ago. According to Paul MacLean of the National Institute of
>Mental Health it deals in aggressive behavior, territoriality and social
>hierarchies.
        [snip]
>If nature came up with feeling first and intelligence much later, I don't
see
>why the opposite would be true for our computers. It's probably a hell of a
>lot easier to make something that feels but doesn't think than something
that
>thinks but doesn't feel.

{I think I'll respond to the original post, since we seem to have gotten
off on a tangent about emotional programmers. Geoff will have to forgive me
if I steal his words. And since this is related to that thread on Proton
Pseudo-Translation, here I'll argue against the point I made in that thread.}

  Just because emotions were evolved earliest doesn't mean they will be
easier to replicate in human-designed computers. Much of the impetus for
the 80's work on connectionism was the interesting fact that we had made
computers that could deal with all the very difficult "thinking" of
academics and scientist but that couldn't do what a 18month-old could (like
walk, see, recognize Mommy). With all those various brain structures
devoted to who-knows-what, it could be that feeling requires far more
computational power than thinking or seeing. And integrating feeling with
the rest of cognition may be pretty tricky or require a whole new
computational model.

Mind you, we've gotten pretty far since then, using programs modeled after
neural systems, in replicating vision and motor functions. Just because it
isn't easier than replicating the higher functions doesn't mean it's
actually hard...

Drea



This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:44:20 MST