From newshub.ccs.yorku.ca!ists!torn.onet.on.ca!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!mips!darwin.sura.net!cs.ucf.edu!news Mon Jun 15 16:04:40 EDT 1992
Article 6206 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!ists!torn.onet.on.ca!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!mips!darwin.sura.net!cs.ucf.edu!news
>From: clarke@acme.ucf.edu (Thomas Clarke)
Subject: Physical versus Computaional (was Re: Transducers)
Message-ID: <1992Jun11.132823.7139@cs.ucf.edu>
Sender: news@cs.ucf.edu (News system)
Organization: University of Central Florida
References: <4138.708217481@mp.cs.niu.edu>
Date: Thu, 11 Jun 1992 13:28:23 GMT

In article <4138.708217481@mp.cs.niu.edu> rickert@mp.cs.niu.edu (Neil Rickert)  
writes:
>   I agree with Harnad's assertion that thought (or at least human
> thought) is physical.  I can find considerable evidence to support
> that view.  I don't see the evidence that pattern recognition is
> physical and non-computational.  Indeed, pattern recognition seems to
> be too rapid to have much of a non-computational component.  The
> learning, or pattern detection, may be a different matter.  In the human
> brain it is plausibly the result of physical growth, and if so could
> be considered physical.  But growth may well be a biological way of
> performing statistical procedures.  Only by either better biological
> knowedge about learning, or by the production of adequate computational
> models of learning, will this finally be settled.

You are using "physical" in a much different way than I would.
The phrase "physical growth" seems to imply what I would 
call something like "neuromorphological" (?)

My use of physical versus computational in the context of the brain
would be to distinguish between the analog of neural net implementation 
of x*y versus a digital implemenation of x*y.

Idealized neurons with continuous sigmoidal functions can compute 
x*y by "biasing" the neurons into a region where the activation is
well-approximated by the square of the input, a=s^2.  Three neurons 
then compute x^2, y^2, and (x+y)^2=x^2+2*x*y+y^2.  A fourth neuron
subtracts off x^2 and y^2, leaving (a constant times) x*y.  
Four neurons and two delay times, but analog accuracy.

A computational implementation would involve converting x and y into
some sort of coded representation (e.g. binary) and using the 
computational resources (e.g. neurons) to implement the necessary
combinatorial logic to compute the coded representation of x*y.
Many neurons, larger delay times, but digital accuracy.

My intution is that the brain uses something much closer to a
physical implementation of necessary functions.

--
Thomas Clarke
Institute for Simulation and Training, University of Central FL
12424 Research Parkway, Suite 300, Orlando, FL 32826
(407)658-5030, FAX: (407)658-5059, clarke@acme.ucf.edu


