X-Message-Number: 3955
From:  (Thomas Donaldson)
Subject: Re: CryoNet #3903 - #3907
Date: Sun, 5 Mar 1995 22:45:44 -0800 (PST)


Hi again!

This will probably be my last message tonight (I have other things to get 
done). Some will cheer and others will weep, no doubt.

But about symbol manipulation, computing, and uploading: unlike what 
Ettinger says, even when we think we are NOT manipulating symbols. I say
this because if you think about the issue even a little bit, it becomes
clear that at some level you must leave off dealing with symbols. It's
what the symbols MEAN that is important, and working with that is what
we are trying to do. Take it to a very basic level: flatworms can show
very primitive learning. When they learn to turn left or right in a 
T-maze, are they manipulating symbols?

I'm not talking about feeling at all, but about how symbols work. We build
our computers to manipulate symbols for us. Sometimes they can do it much
better than we. But the MEANING of their results and their inputs comes
entirely from us. Even if we design the computer to take actions depending
on its results, those actions ultimately come from us. 

A bit of thought will show that if we did nothing but manipulate symbols
when we thought then we'd quickly find ourselves in an infinite regress or
(perhaps) chasing our tail around and around. We can't find out the meaning
of words from a dictionary alone, whether or not it is encoded in our brain.
If we find one word, it will be defined only in terms of others (all symbols).
When we as human beings learn to speak and think, we learn words by associating
them with incidents and things in the real world. Ultimately that association
cannot be a matter of symbols but of real, concrete acts.

Moreover its very doubtful by now that we have memories in the sense that 
computers do. We are shaped by our learning and our environment, plus our
basic physical type (including our brain) to respond to incidents and events

in the world, not responding to symbols but to reality. That shaping is 
dynamic;but if we were to look at it at any instant, it would be much more as if
our responses came out of our design. A can opener, for instance, has been
designed to open cans, and no amount of symbolic manipulation will open 
a can of soup. We are far more complex, and can change much more, but we
work like the can opener and not like a computer (in the contemporary sense
of computer). Does the can opener have a memory? 

It's noteworthy here that neural nets also show some of these properties. Their
responses depend on how they are connected up, not on an ordinary computer
memory. Certainly we might use a computer to simulate them, but a real
neural net (say, using Intel neural net chips (yes, Intel makes chips for
that purpose) won't be a simulation. It will, in its own way, be more like
the can opener (or our brains) than like a computer. The difference between
our brains and the physical neural nets various scientists and engineers have
constructed (and which are now finding more and more use, in a way just like
Robin mentioned!) is that our brains also have special properties that make
the wiring far more dynamic. Such processes, among other things, involve 
the use of short term chemical changes to preserve an association while 
behind those changes other physical rewiring (involving protein synthesis)
can go on.

			Long long life,

				Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=3955