X-Message-Number: 13708
From: 
Date: Thu, 11 May 2000 06:54:50 EDT
Subject: Re: Msg #13693 brain simulation on computers

> Message #13693
> Date: Tue, 9 May 2000 09:44:55 -0400
> From: Thomas Donaldson <>
> Subject: CryoNet #13690 - #13691
> 
> About computer uploading and copies of ourselves (computers or not):
> 
> The fundamental problem every discussion of this possibility (that I've
> seen) seems to ignore is that we first must understand how our brain
> works in order to copy it... and we haven't attained that understanding,
> yet. Nor is doing so a simple problem. I would accept the notion that
>our brain consists of lots of neural nets with another single computer
> producing our awareness, but that is hardly enough information to
> actually do a copy. 
> 
> Problems: the neural net is not one of those used in computing, and
> operates quite differently. Not only do neurons have multiple 
> connections rather than just one, but our brain produces new neurons,
> and the connections change. >>
>
A first generation "brain on computer" will simulate the neuron activities, 
that
is what we see at small scale today. If we want a true simulation, we need 
indeed to understand everything about the brain. Some "secondary" chemical 
effects may be hard to include in that scheme; So, I think a real brain on 
machine will use another  system.

I think a second generation simulator, will in fact simulates a space domain 
with a 3 D grid. At each crossing of the grid, there will be a possibility to 
simulate a small molecule, that molecule will be allowed to interact with 
nearby crossing positions or move to such a position. If we know how such 
molecules behave and we can get a list of them in the brain ( or any other 
organ) with their position, then we can run a brain simulation even if we 
don't understant how it works. To have a perfect simulation, the grid unit 
length must simply be half the mean diameter of a molecule (from Shanon's 
theorem on sampling data).

Assume the grid mesh is one nanometer, a 10 cm cube will contains 10^24 
crossings. This is a big number, but optical or quantum computers may indeed 
be up to the job 20 years or so from now.

Yvan Bozzonetti.

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=13708