F. Frank LeFever wrote in message <72ljqc$ki at sjx-ixn8.ix.netcom.com>...
>>I must confess I have never heard of this constraint, but my first
>question would have to be, how is "computation" defined?
Let this be the devil's question. Frank has just killed the whole debate.
I have to marvel at the conceptual confidence of those who can make such
quanitative analyses of that grey stuff upstairs. Many years ago I read a
wonderful book, The Limits of Analysis, Stanley Rosen, in which he wrote,
"After all, mathematics is itself an extraordinary way of treating ordinary
experience. Perhaps there are other extraordinary manifestations of
rationality and the devotion to measure."
Computation is a convenient word but all too often it seems the word allows
people to slip down the path of thinking that all human thinking is
computational. I don't buy this (never liked Plato), the polymodal
processing capabilities alone suggest the brain probably uses a number of
differing strategies to achieve its goals. As such, trying to find the "key"
mode may be a furphy and for the life of me I can't think of how one would
define a computation at the neurological level. A big bump on the head could
create wholly new computations.
The amount of computation required for any given action may also be much
less than most suppose. I suspect that training works because it creates
mulititudes of topographical maps in the relevant cortical areas allowing
for a more discriminating choice of 'action plan' to be initiated and
possibly modified during execution by way of cerebellum feedback (ok I'm
guessing bigtime but that ain't so unusual 'round here). If this were the
case the important processing would be at the primary sensory level, making
the correct judgements so that the correct motor sequences are called.
William Calvin's throwing Madonna hypothesis might throw some light on this
ramble.
An important
>consideration is that there are many lines of evidence supporting the
>idea of the brain as made up of parallel processers, so there may be
>much more "throughput" than would be possible with a single channel
>serial processer, howsoever many "computations" might be possible in a
>given period.
This reminds me of a tale told by Daniel Dennett in Consciousness Explained.
Dennett was umpiring a baseball game when he had to make a very close call.
Uncertain he signalled "out" with his right hand but called "Safe" and felt
somewhat embarrassed. Dennett explains this by reference to his modular
model, where the brain employed multiple means of determination but didn't
have enough time to reach agreement amongst the competing processes.
john
johnhkm At logicworld.com.au
