From newshub.ccs.yorku.ca!torn!utzoo!helios.physics.utoronto.ca!utcsri!rpi!usenet.coe.montana.edu!news.u.washington.edu!carson.u.washington.edu!forbis Thu Oct  8 10:11:10 EDT 1992
Article 7109 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!torn!utzoo!helios.physics.utoronto.ca!utcsri!rpi!usenet.coe.montana.edu!news.u.washington.edu!carson.u.washington.edu!forbis
>From: forbis@carson.u.washington.edu (Gary Forbis)
Subject: Re: Simulated Brain
Message-ID: <1992Oct3.201019.10386@u.washington.edu>
Sender: news@u.washington.edu (USENET News System)
Organization: University of Washington, Seattle
References: <1992Sep29.151801.8240@Informatik.TU-Muenchen.DE> <1992Sep29.225005.4267@usl.edu> <60864@mimsy.umd.edu>
Date: Sat, 3 Oct 1992 20:10:19 GMT

I know this is quite derivative from the original subject but here goes anyway.

In article <60864@mimsy.umd.edu> waander@cs.umd.edu (Bill Andersen) writes:
>   Just because the brain doesn't have
>a "single" processor doesn't mean that, assuming it is possible at all
>to simulate a brain on a digital computer, that you can't do it on a
>sufficiently powerful von Neumann machine with enough memory.  The 
>reason for this is simple: one can simulate any parallel digital 
>computer on any scalar digital computer; parallelism buys nothing
>additional in this sense.

While may ideally simulate a parallel digital computer on any scalar digital
computer, it is not clear to me that one will necessarily get the same 
results on a parallel digital computer as on a scalar digital computer.
The reason for this is that the scalar digital computer makes some 
idealizations about the environment.  As with any chaotic system, small
changes in input may make very large changes in output.  Things such as
transmissions delays and variations in clock speeds lead to problems on
parallel computers which must be overcome by synchronization messages.
These problems may never arrive in the simulation.  If the simulation of
a brain succeeds because of the non-determinism of the parallel environment
it may not succeed on the scalar machine where the non-determinism has been
removed.


