Neurologists and connectionism

Sydney M Lamb lamb at OWLNET.RICE.EDU
Fri Jan 31 21:30:46 UTC 1997


Tom --- In belated reaction to your entertaining "Get real, George"
message of the 6th, i want to touch on an incidental point at least
before we leave January forever.  You wrote:

>
>  in search of the latest fad. When I ask the real neurologists
> I know what they think of connectionism, I get an incomprehension response.
> Never heard of it. Con what?
>

Your statement has two implications: (1) that neurologists
haven't heard of connectionism, (2) that neurologists have an expert
knowledge of how the brain works that would enable them to pass judgement
on the merits of connectionism.  From my viewing platform it appears
clear that both of these implications are false.

  1.  At least some neurologists, in fact some very highly regarded ones,
have indeed heard of connectionism (1987 Rumelhart and McClelland
variety).  Two examples:  It is described in Kandel, Schwarz, and Jessel,
Principles of Neural Science (1991:836-7), which neurologists in medical
schools recommend to medical students as "the bible" on neural structures and
their operation; also in their "Essentials of Neural Science and Behavior"
(1995).  Likewise, Dr. Harold Goodglass, a very highly regarded
neurologist and aphasiologist, mentions this model in his "Understanding
Aphasia" (1993:37-8).

  2.  The above is not surprising.  What is surprising is that this
model, despite its being egregiously out of accord with known facts of
neural structures and their operation,  is mentioned favorably,
as smthg to be taken seriously, by the above
writers and others (esp. Kandel et al --- Goodglass gives it fainter
praise; possibly, one hopes, he is just being diplomatic).  Such
favorable description by Kandel et al. and others is what demonstrates
the falsity of your second implication.  If the neurologists can so
easily give credence to such an unrealistic model, they are not experts in
this area after all.  Yes, they are experts in other areas --- they know
a lot about anatomy, synapses, neurotransmitters, ALS, how to diagnose
neurological problems, etc. etc., but they have demonstrated that they don't
have much of a clue about how information is learned, remembered, and used by
real neural networks.

Warmest regards,
                Syd .




More information about the Funknet mailing list