From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!unix.cis.pitt.edu!pitt!speedy.cs.pitt.edu!geb Wed Dec 18 16:02:14 EST 1991
Article 2195 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!unix.cis.pitt.edu!pitt!speedy.cs.pitt.edu!geb
>From: geb@speedy.cs.pitt.edu (Gordon Banks)
Newsgroups: comp.ai.philosophy
Subject: Re: Scaled up slug brains
Message-ID: <12743@pitt.UUCP>
Date: 17 Dec 91 14:29:36 GMT
References: <40677@dime.cs.umass.edu> <12723@pitt.UUCP> <40705@dime.cs.umass.edu>
Sender: news@cs.pitt.edu
Reply-To: geb@cs.pitt.edu (Gordon Banks)
Organization: Computer Science Dept., University of Pittsburgh
Lines: 51

In article <40705@dime.cs.umass.edu> yodaiken@chelm.cs.umass.edu (victor yodaiken) writes:
>
>Generally, "computation" has to do with Turing machines or some variant. For

Well, I've heard people argue that neural nets beyond the perceptron level
are "Turing equivalents" but I'm not enough of a computer scientist to
tell you exactly how that is determined.  Certainly they do not seem to
perform their computations anything like a Turing machine does.  That
paradigm is more appropriate to von Neumann machines, it would seem.
Maybe someone else like Drew can tell us more about that.  But I would
not like to tie the idea of computation to a Turing machine.  Anything
that takes an input and has an output that can vary according to the input
in some regular and orderly way would seem to be a computer.  A neuron
certainly is, and so are networks of them.  Demonstrating that a
neural network can simulate a Turing machine is not the equivalent
of showing that a Turing machine can simulate everything a neural network
does, although we know that our current models of neural nets are for
the most part implemented on von Neumann type computers.

>a product of "computation" in the usual sense.  Note, that I do not have
>to posit a supernatural force in either case: one can easily suppose that
>there are mechanisms involving chemical transport, proteins, quantum
>effects, radio waves, spinal cords, vocal chords, endocrine systems,
>chaotic or statistical phenomena .... whatever, which are 
>intimately involved in the production of self-consciousness and language. 

Well, if you believe these physical processes are discoverable and
can be either simulated or duplicated artificially, then we aren't
too far apart, philosophically.  I'm not claiming we have the present
technology to create a brain, just that we are on the right track pursuing
it technologically.

>
>We have abilities which are not present at all in worms: as mentioned
>before, language seems to represent a radical step.

Certainly, but when you look at our brain and an ape brain, the
circuitry is very similar.  Our brain has additional cortex and
thus additional circuits, but that seems to be the extent of the
difference.  There are no radical new types of cells or new neurotransmitters,
or other geegaws.  It would seem to me that Occam's razor would say
that it is most likely that the language abilities are in the new
neural circuitry, especially since the regions of the brain responsible
for the production and understanding of language are pretty well defined
through what happens when they are injured.

--
----------------------------------------------------------------------------
Gordon Banks  N3JXP        | "When in danger, or in doubt
geb@cadre.dsl.pitt.edu     |  Run in circles, scream and shout" --Heinlein
----------------------------------------------------------------------------


