Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!mit-eddie!uw-beaver!cornell!rochester!rit!txo6870
From: txo6870@cs.rit.edu (Timothy X Oertel)
Newsgroups: comp.ai.philosophy
Subject: Re: My view of intelligence...
Keywords: I and intelligence
Message-ID: <2209@cs.rit.edu>
Date: 23 Mar 91 17:45:22 GMT
References: <13577@helios.TAMU.EDU> <69649@brunix.UUCP>
Sender: news@cs.rit.edu
Organization: Rochester Institute of Technology, Rochester, NY
Lines: 61


In article <69649@brunix.UUCP> cs196006@cs.brown.edu (Josh Hendrix) writes:
<In article <13577@helios.TAMU.EDU>, rpb0804@venus.tamu.edu (BATES, ROBERT PATRICK) writes:
<|> I dunno about this philosophy about intelligence being inbred or evolving.  I
<|> personally feel that the brain is a large, organic macrocomputer.  After all,

[stuff deleted]

<Here's where I disagree:
<1. The brain is indeed large (considering the size of its constituent parts, and compared to the brains of other animals), and it is most definitely organic. I have a problem, however, with any sentence, phrase, or other semantically meaningful arrangemen
<t that contains the words 'brain' and 'computer' within 10 words of each other, unless the words 'is not a' fall in between :-).
<
<First, nerve cells are actually slow, firing on the order of 10,000 times/second, while the cycles per second of your average Sun workstation is measured in the millions. Second, this means that, to do all the things that we effortlessly do, (walk around,
< avoid things, etc.) the 'computer' in our skulls have to be massively parallel. Third, this massive parallelism means that the neurons, the 'computing elements', are hooked together very differently from the way CPU's, busses, and chips are. Everyth

[stuff eaten due to lines too large]

While, as far as I know, the above is mostly true, just because there is no
one-to-one relationship between processors and neurons doesn't mean that a
group of neurons couldn't be modeled by a single processor and a brain modeled
by a large group of processors.  Although, this comes together again in that
the size of the memory and processing would be mighty monsterous.  I've seen
figures that the number of neurons in the brain is ~1-4 billion and that each,
on average (presumably), has ~1 million connections.  No kidding massively
parallel... :)  
  Also, in modeling neural networks, people normally pay attention to a 100
step limit, where no computation, to be "neurologically correct", can exceed
this.  So, by what we've said, apparently we might need an order of magnitude
or so increase in processor speed AND THEN, to deal with all the computation,
we would need one processor per neuron.... goodness.  Seems almost a tad
hopeless for the near future.

<at happens in your workstation happens one 'operation' at a time, whereas all of the 'computation' that is going on to allow you to see the screen right now is happening at the same time.
<
<Of course, there's a lot of debate about that last sentence, so I do not claim here to be the final word. But that's what seems to me to be going on.
<
<|> it stores data in "chips" (cells) using electrical currents.  It constantly 
<
<Again, I disagree. Cells may contain information of their own (DNA), but the information that makes up a thought is stored in the connections of many cells.
<

Although the thought is distributed, each cell DOES hold a certain amount of
nonspecific information in electric potentials in each cell and in their 
connection strength between neurons.

[stuff deleted]

<|> Roberto
<|> RPB0804@TAMVENUS
<
<Josh Hendrix
<cs196006@brownvm.brown.edu

I admit that the total of my NN knowledge is by no means state of the
art, but I thought I'd toss in my two cents.  I won't say more since
this is philosophy, not quite NN's.


---> Tim Oertel
txo6870@ultb.isc.rit.edu