[Comp-neuro] Re: Education

Dave Beeman dbeeman at dogstar.colorado.edu
Thu Aug 14 07:37:51 CEST 2008


I've followed the discussions provoked by Jim with interest.  Here are
some of my own opinions:

With regard to training biologists to think quantitatively, I think that
it is imperative to change the education of biology majors.  In the
various modeling courses that I've taught, it was far easier to teach the
biology to physcists and computer scientists than to teach quantitative and
algorithmic skills to biologists.  This is not as it should be, but the
cure must start early.

With regard to bottom-up "biologically realistic" vs. phenomenological
models, I believe that hard problems require attacks on many fronts.
Fortunately, advances in computer hardware are on the side of realistic
modeling.  The usual excuse for avoiding detailed modeling of neurons and
their networks is that computers are too slow to model the required level
of detail.  Yet, the 1989 GENESIS olfactory cortex simulation had 4500
neurons and ran on a 0.02 GHz Sun 3. Now there is a new generation of fast
cheap quad core "gaming PCs" for less than USD $1000 that can run parallel
versions of large network models at speeds formerly associated with
parallel supercomputers.  Why not see if details matter in your conceptual
model?



More information about the Comp-neuro mailing list