X-Message-Number: 26034
Date: Sat, 16 Apr 2005 09:20:46 -0400
From: Thomas Donaldson <>
Subject: Once more for Yvan

For Yvan:

It would be nice if you would listen. All the theorems etc which you
invoke do not normally deal with the problem of TIME. Turing did not
deal with the problem of time. Computers weren't originally invented
to deal with the problem of time (in their work, not in physics 
equations they might be used to solve). This remains a factor in their
design.

And certainly even with the fast sequential computers we have now, 
they work quite well to solve problems with do not require solution 
within the next nanosecond. And if they work fast enough on ordinary
jobs, time sharing remains quite useful.

Brains and nervous systems, however, did not evolve in a quiet world
in which they had plenty of time to solve the problems they were
presented with. They had to solve those problems fast or they would
turn into nutriment for some faster animal. Time sharing neurons
simply weren't an option --- except possibly for some animals now
long extinct and hardly noticed. This is why our brains operate not
only as parallel computers, but (as is usually not done with even our
parallel computers) with all those neurons working in parallel on
different but related jobs. The world is full of simultaneous 
events, and we could not deal with them otherwise.

You may argue, with a bit of reason, that if the processors involved
are much faster than neurons then time sharing once more becomes
an option. The problem is that simultaneous means SIMULTANEOUS. PERHAPS
a computer using time-sharing might be fast enough to deal with 
some simultaneous events, but sooner or later it will not deal
with all. As a simple example, it may be faced with a need to 
physically balance itself for an instant. If its limbs and muscles
don't work simultaneously it falls over and will eventually be
consumed (perhaps by other better designed devices). If the balance
test is hard, then even a few nanoseconds delay leads to falling.

Second, you claim that growth is quite possible IN SOFTWARE. This
statement verges on the ridiculous. Yes, you can simulate growth
in software. Yet suppose that the growth of new synapses and even
new neurons, in our dentate gyrus and elsewhere, turns out to be
essential for long term memory. (Many but not all neuroscientists
think that). Exactly how does your computer version of a brain
do that growth in software if it needs new neurons even to 
remember long term? The brain changes required are changes in
the physical anatomy of the brain. If you ran out of fuel in your
car, and told me that you could solve this problem with software
in your car's computer, I would suspect you of insanity. 

As for your reading in neuroscience, please say just what books
and articles you've read. The book by Byrne and Roberts is good
(FROM MOLECULES TO NETWORKS) but hardly more than a start. I am
glad that you've at least read some biochemistry.

          Best wishes and long long life for all,

               Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=26034