X-Message-Number: 3805
From:  (Anders Sandberg)
Newsgroups: sci.cryonics
Subject: Re: brain as net
Date: 06 Feb 1995 11:12:35 GMT
Message-ID: <>
References: <3gsurq$>

Thomas Donaldson wrote:
>One fact about our nervous systems which may be growing more and more clear
>is that the number of our synapses increases when we learn. There is another
>article suggesting this in the latest PERIASTRON. 

This may be true. In fact, it is thought among neuroscientists that synapses
can both grow and change properties in response to learning. It is a very
vital area in neuroscience right now.

>In terms of simulating or emulating a brain on a computer, this would cause   
>a problem. Our computer-brain would need to have some way to increase the
>number of its connections. 

This sounds reasonable. However, I don't see the problem. The neural simulator
has to deal with the ways these synapses grow and change properties, something
we don't know how to do yet, but doesn't appear to be impossible. Then the
model in the computer is changed accordingly.

>The ability to grow new
>connections is clearly something hard to simulate with one of our present 
>computers. 

Not at all. Its not done in most neural network simulations because it
complicates things and is not very understood, but it is in principle
easy to add. A simulator could for example be programmed to add a new
node with synapses if certain conditions are met on "neighboring"
nodes. Its important to remember that we are doing everything in software.

--

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=3805