Dragaera

OT: bois (was: Sethra Lavode vs. Enchantress of Dzur Mountain)

Frank Mayhar frank at exit.com
Sat Aug 24 12:59:04 PDT 2002

> #Mark A Mandel <mam at theworld.com> writes:
> #> And let's not forget that despite their name, the software constructs
> #> that their creators optimistically named "neural nets" probably have
> #> very little in common with wetware.

This is not true.  The reason they are called "neural networks" is that
the are modelled after the biological structures for which they are named.
In their implementation, of course, they are quite different, but in what
they do and how they do it they are very similar.

> On 24 Aug 2002, David Dyer-Bennet wrote:
> #I'm not so sure.  They certainly exhibit startlingly wetware-like
> #properties.  I'm thinking especially of some of the artificial insects
> #that have been built.
> #
> #And after all, we *do* understand the neuronal biochemistry pretty
> #decently.

What we understand of it, we understand pretty well.  There's still a lot
of stuff (mostly relating to how the connections and feedback systems work)
that we still don't understand.  But our understanding is getting better
all the time.

Mark A Mandel wrote:
> That's simulation of behavior at an insect level. ELIZA can simulate
> conversation convincingly at a certain level, too. How similar
> (homeomorphic?) are the innards? How far does it scale toward human
> cognition?

Considering that the most complex software neural network ever developed has
at most the complexity of an insect, comparing it to the complexity underlying
human cognition doesn't make much sense.  On the other hand, though, I am
becoming more and more convinced that the difference between the complexity
of an insect and that of a human is a matter of amount, not kind.  It's the
same set of processes, with some modifications and much, much increased
complexity.

ELIZA was a keyword-based recogizer and can't be thought of in the same
context as a neural network.  For a brief description of artificial
neural networks, I suggest
	http://www.emsl.pnl.gov:2080/proj/neuron/neural/what.html

Particularly: "Artificial neural networks are collections of mathematical
models that emulate some of the observed properties of biological nervous
systems and draw on the analogies of adaptive biological learning."  (Ibid)


(I'm somewhat pleased at the vehemence of some of the reactions I've seen
to this discussion.  It can be difficult to consider that we are not what
we think we are; the reactions suggest that some preconceptions may be
being challenged.)
-- 
Frank Mayhar frank at exit.com	http://www.exit.com/
Exit Consulting                 http://www.gpsclock.com/