Query: Dendritic Networks

Alan J. Robinson robin073 at maroon.tc.umn.edu
Thu Feb 15 11:46:19 EST 1996


On 13 Feb 1996 19:48:55 GMT, 
Paul Bush  <paul at phy.ucsf.edu > wrote:

>In article <3117860D.2781 at psy.ox.ac.uk>, "Simon R. Schultz" <simon.schultz at psy.ox.ac.uk> writes:
>
>|> There is room for specialisation in neuroscience, and there's plenty of room
>|> for those who specialise in theoretical analysis -- if these people (such as
>|> myself) spent 2/3 of every day running recordings, they simply wouldn't have
>|> time to develop their theoretical skills sufficiently.
>
>|> On the other hand, the physiology (and anatomy) *is* fundamental. Lets have no
>
>|> On the other hand, there isn't always an advantage to detailed (e.g.
>
>With 3 hands, you should be able to work 50% faster than anyone else...
>

Paul:

The ever increasing specialization of science has ran slap bang up 
against the strong desirability for scientists in the behavioral 
and brain sciences to have a broad basis of understanding of recent 
results throughout all those disciplines.  Without this understanding, 
the science is reduced to gathering large volumes of experimental 
data without any overall theoretical or SYSTEMS framework.

The rules of scientific conduct make it close to impossible for an 
individual scientist to achieve and sustain expertise in more than one 
discipline or one approach at a time, but many scientists from both 
inside and outside the behavioral and brain sciences appear to lack 
even the most basic credentials in the core disciplines such as 
neuroscience and psychology.

This relates to the problems associated with theoretical 
physical-mathematical model building, which has been one of the most 
important, and in some ways the most disastrous development in 20th 
century science.  

This type of model building, especially using a computer, is most 
closely associated with John von Neumann, a strong contender 
for the most powerful intellect of all time.  It is therefore 
surprising that von Neumann did not realize that almost all these 
models exhibit chaotic behavior through numerical instability and 
sensititivity to initial conditions, quite apart from the 
approximations that have to be used (contrary to popular opinion, 
physics is NOT an exact science.)

Can such models still be useful?  yes!  BUT YOU HAVE TO KNOW WHAT 
TYPE OF RESULTS YOU ARE LOOKING FOR BEFORE YOU BUILD THE MODEL.  This 
is why it is so important that such models ONLY be built by those 
who have some formal and practical background in the subject 
discipline.  A general scientific education coupled with a brief 
literature survey is NOT a sufficient qualification for model 
building or even making public statements about the brain and 
human behavior.

What I find really surprising is that this rule is repeatedly violated 
by some scientists at the world's most prestigious universities such 
as Oxford and Harvard, who should know better and set an example to 
the rest of the world.

AJR




More information about the Neur-sci mailing list