X-Message-Number: 9171
From: Thomas Donaldson <>
Subject: Re: CryoNet #9168
Date: Mon, 16 Feb 1998 22:59:17 -0800 (PST)

Hi again!

I am discussing it because you raised the subject yourself.

Moreover connections between neurons generally aren't single: there may be
more than one synapse, and their location may play a role. Furthermore
if you study what is known about memory you will see that the idea that it
may consist of the formation of a new connection has wide acceptance, though
not as a proven fact, but as a very likely result of current research.

Your proposed method solves one problem but raises lots of others. Given 
that we want to create a brain rather than a computer, those connections will
have to come into being, rather than exist constantly, and be multiple
rather than single. I'm sure you've seen micrographs of all the synapses
along some stretches of axon. And as a relatively trivial problem, though
still a problem, your proposed system, being 2 dimensional, can hardly be
as compact as a 3-dimensional brain.

Many of these ideas are very easy to say but clearly very hard to implement.

As for neurons connecting only to their neighbors, that is true, but axons
of course allow their neighbors to be quite far away. In your plan, these
3-dimensional neighbors would of course somehow have to be mapped onto 
a sphere --- and their neighbors, too, and so on.

I may have more to say on this later. 

			Best and long long life,

				Thomas Donaldson

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=9171