NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/talk/abortion/57107

Text File | 1993-01-21 | 6.8 KB | 139 lines

Newsgroups: talk.abortion
Path: sparky!uunet!news.encore.com!jbates
From: [e-mail] (John W. Bates)
Subject: Re: Myelin (Was Re: Spoken Like a True ProLifer)
Organization: Encore Computer Corporation
Date: Thu, 21 Jan 1993 08:54:38 GMT
Message-ID: <[e-mail]>
In-Reply-To: [e-mail]'s message of 21 Jan 93 03:44:31 GMT
References: <[e-mail]>
    <[e-mail]>
Sender: [e-mail] (Usenet readnews user id)
Nntp-Posting-Host: pinocchio.encore.com
Lines: 124


By the way, I just started catching up on this thread, and noticed
that earlier Steve provided a reference to the same research I did,
except he did it much earlier than I did. My apologies for posting
it again. I could cancel the article, but I don't feel like it. Nyah.

In article <[e-mail]> [e-mail] (Stephen Matheson) writes:
> From article <[e-mail]>, by
> [e-mail] (Mark A. Cochran):

[much deletion.]

>>>>>>>My contention remains: large segments of the nervous system
>>>>>>>function beautifully without myelin.  Myelin has a very
>>>>>>>specific purpose: it allows for fast conduction of impulses.
>>>>>>>It must not be indispensible for function, because it is
>>>>>>>anything but ubiquitous.  While it is reasonable to guess that
>>>>>>>speedy conduction would be advantageous to a complex network,
>>>>>>>it seems reasonable to assume that one can design a network
>>>>>>>without it.

>>>>>> Kind of like trying to build a hypercube out of a bunch of
>>>>>> C=64's? I'm sure it can be done, but would it actually be
>>>>>> capable of performing any resonable work?

>>>>>Huh?  Can you provide some hypercube and C=64 references? :-) Are
>>>>>they available on request?

>>>> A hypercube (as I understand it, but it's not my field at all) is
>>>> a bunch of cross-linked multi-processor computers that tries to
>>>> simulate the neural system by multiple cross-links. The C=64 is
>>>> the ancient Commodore 64 that people have always laughed about.
>>>> It would be interesting (funny even) to see what would happen if
>>>> Commodore got hte idea to try this for real. :)

>>>Wow.  Mark is a mrpmsysp UGR futrvits (Mark's occupation has been
>>>encrypted at his request; decoder available from T.S.A.K.C. BBS)
>>>*and* a neural network guru.  That's amazing.

>> Huh UH! No Way! See right up there where I specifically deny any
>> expertise in this area? I've read a litle bit, and talked to
>> people, but no *way* are you goin to get me to play neural network
>> sysadmin here. ;)

> Very well.  You're the one who brought it up.  I wonder if John
> Bates can get enough free time to comment on the applicability of
> your analogy.

Sorry, Mark, but you've got things a little mixed up. In networking
lingo, hypercubes are a network designed for supercomputing, with
a node at each vertex connecting to each neighboring vertex. It's not
related to neural networks at all. The closest neural network design
I can think of is James Anderson's "brain state in a box", in which
each output pattern is a vertex of an n-dimensional box. Nice model
for associative memory, but n tends to have to be very large (in
computational terms) for it to be useful.

I've been leary of bringing models into this discussion, since it is
often hard to relate models of neural networks to actual neural
networks. But now, let me refer to a model by Stanislas Dehaene and
Jean-Pierre Changeux, which simulated the performance of human
infants in Piaget's A not B task. Their results approximated the
performance of human infants.

The interesting part of the experiment, though, was that they varied
the amount of "noise" that the network received. At high levels
of noise, the network performed at the level of a 7-month old infant,
but at low levels, it performed at the level of a 12-month old infant.
Noise levels seemed to correspond to the development of myelin in
the frontal lobes.
(from the _Journal of Cognitive Neuroscience_ 1:3, S. Dehaene and J.P.
Changeux, A simple model of prefontal cortex function: delayed
response tasks)

>>>>>I don't see why an unmyelinated network would be incapable of
>>>>>performing any reasonable work.  I can imagine that it wouldn't
>>>>>perform as well as the faster network.  I can also imagine that
>>>>>the growing size might make the myelination important (or even
>>>>>necessary) and that the very small beginning network isn't
>>>>>myelinated because it doesn't need to be.

I noted in another article that you talked about myelination occurring
once the axon reached a ceratin diameter, and I wanted to respond to
that. It seems that two schemes have been developed by nature to deal
with signal loss and transmission speed in nervous systems, one being
myelin and the other being increased diameter. It only makes sense that
once an axon reaches a certain diameter, it should be myelinated (in
species which use myelin, of course). The squid, as an example of
an unmyelinated species, has axons that are as large as one millimeter.
Cross sections are visible to the naked eye.

As far as your statements above go, (just so you don't think I'm
rambling insanely) you're absolutely right. A small network doesn't
need myelin to function. Crabs, for example, have axons that are just
a little bit thicker than humans', and they seem to do well enough.

>>>> The resonable work in question, though, is thought. Just as you
>>>> can't use a 4 bit 16K RAM computer as an effective file server, I
>>>> don't see how you cna use the similarly limited abilities ofthe
>>>> pre-myelinated neural system as a 'thought server'.

>>>I'm reserving my judgment on the matter for a time when we know
>>>more about all the issues involved.  The hypomyelinated mice
>>>discussed later in the post may be our best window into this issue.
>>>In the meantime, it must be obvious to everyone reading this thread
>>>(all 3 of us :-) that neither of us has any clue about whether
>>>myelin is necessary or not.  I think that there is at least a fair
>>>amount of information on the biological side that suggests that
>>>myelin is not as central as some claim.  On the other hand, your
>>>arguments about the presumed complexity of the network are
>>>certainly thought-provoking (there's that smell again...myelin
>>>burning or something).

Yes. The major problem that we have in modelling brain processes is
the complexity of the whole thing. I mean, our supercomputers have
problems with 2-3000 neuron models. Massively parallel systems
reach the 16-32000 neuron level. How much of the brain is actually
dedicated to thought? Maybe what, 10^10 neurons?

[the rest is deleted 'cause I'm really tired and I've got to get
back to work. I thought I had something relevant to say, but I've
forgotten what it was.]

John