X-Message-Number: 25967
Date: Sun, 3 Apr 2005 09:30:44 -0400
From: Thomas Donaldson <>
Subject: reply to Yvan Bozzonetti

To Yvan!

I will repeat what I said before. You haven't answered my questions. 

As for the scanty comments you made in your message, I will point out
that parallel computers can do things impossible for any single 
computer. Not only that, but the NUMBER of processors makes a 
difference. There's no mystery about how this works: basically
parallel computers can do things objectively simultaneously, not
with an illusory simultaneity.

You suggest that neurons may be imitated (?) by a computer program,
and in that case their connections and number can be modified
arbitrarily. First of all, all of our neurons work simultaneously:
in short, our brains are parallel computers. This means that you 
can't just make another neuron in software. You need to add an
actual processor to the large collection of neurons which make
up our brain. As for connections between neurons (processors),
our neurons GROW new ones. Growth here means that there is no
simple limit on the number they can grow. (In practice, of course,
they also eliminate connections, but growing a connection and
eliminating another are separate processes, and don't always
occur at the same time --- in fact, they rarely occur at the 
same time).

Moreover the strongest theory about how our different kinds of
memories work is presently the idea that they are implemented
by connections between neurons. If you wish to make a neuron
based on some kind of computer chip, even if you give that 
chip a large number of different connections (so that there
are more connections than any other feature of the chip) you'll
still have a finite number which will fill up. (How to make
them go to the proper other neuron would be a problem too ---
and remember that you don't want more than one neuron per
chip). The ability  to grow new connections eliminates this
problem entirely.

As for all the work that has gone into finding out how our
memory actually works, the growth of new connections and new
neurons, I try to put all the stuff I can find into PERIASTRON.
If you want to subscribe, you can do so for a year for US $24.
If you want to subscribe using some other currency, I'm
happy to discuss that too, but want to work that out in 
advance.

Fundamentally, because neurons work independently, putting
more than one on one processor fails, and probably would do
so quickly. It's simply not valid to believe that ANY parallel
computer can be squeezed into a smaller number of processors
than the number it uses.

         Best wishes and long long life for all,

             Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=25967