[Comp-neuro] Discussion - science follows technology - what is the program

Igor Carron igorcarron at gmail.com
Wed Aug 27 14:41:34 CEST 2008


I agree with you but as a nuclear engineer, I beg to differ wrt
nuclear technology and bombs in your statement " nuclear technology
and bombs were utterly dependent on the groundwork of theory". Either
technology were highly dependent on fission. Until fission was proven
to exist physically
(http://nobelprize.org/nobel_prizes/chemistry/laureates/1944/) it did
not exist in theory nor was there any model used to discover it. Even
after that discovery, people realized you could induce chain reactions
only after it was experimentally proven that you could, on average,
get two to three neutrons out of one fission reaction with Uranium.
There was no model of an atom that led people to think that a
particular element would yield enough neutrons. Nuclear reactor
technology was not even considered remotely feasible until we
understood that there were very specific materials which could act as
moderators (the graphite used for instance by the German during WWII
was not pure enough for instance and acted as an absorber). And while
models were devised for the technology used in nuclear reactors and
for the bomb, they remain highly grounded in experimental findings
(cross sections, macro-tests...). A major issue when one comes up with
"better" models (at least in civilian nuclear technology) is the
requirement that it is entirely consistent with all the previous
experimental findings. The groundwork of theory as you put it has, in
nuclear technology, always been a way to acquire and use experimental

In nuclear physics, models may have led to some new findings like the
Higgs boson but I am sure that had it not been found, they were other
models ready to jump in to explain this non-occurence. If we look at
string theory, it looks as though they have plenty of models and/or
parameters which conveniently cannot be separated from each other
using existing experimental facilities.

What are the lessons for computational neuroscience I can think of at
least three:

- When safety is a major factor, models always must be proven to be
consistent with all the previous findings

- When several models have several hundreds or thousands of
parameters, benchmark exercises may be a way to differentiate between
models but in most cases, they don't do such a good job.

- Over time most very complex models give about the same result in a
known phase space of parameters yet there is no good mechanism for
comparing the phase spaces.



Igor Carron, Ph.D.

On Tue, Aug 26, 2008 at 3:07 PM, James Schwaber
<schwaber at mail.dbi.tju.edu> wrote:
> A lot of the discussion about the 'right way to model' or what to model may
> be a version of what my friend Mike Gruber has termed  a version of the
> science-technology fallacy, the idea that you have to understand a process
> analytically before you can use it, and he always quotes Carnot
> here--thermodynamics owes more to the steam engine than ever the steam
> engine owes to thermodynamics.  Obviously, humans used and controlled fire
> for 100,000 years before Lavoisier explained what fire was, and planes flew
> for decades before there was a theory to explain how they did it.  In fact
> theory 'demonstrated' that heavier-than-air flight was impossible. The
> reason we buy the fallacy is because of the outlier of nuclear physics--yes,
> in that unusual case, nuclear technology and bombs were utterly dependent on
> the groundwork of theory, and lasers arose from quantum research.   But this
> is not the common case in the history of technology.
> For neuroscience the question is can we predict, control, and modify brain
> function even though we never be able to 'understand' it analytically?  As
> of now the answer seems to be no.  Will this ever improve?   Drilling down
> won't cut it.  What would?   _______________________________________________
> Comp-neuro mailing list
> Comp-neuro at neuroinf.org
> http://www.neuroinf.org/mailman/listinfo/comp-neuro

More information about the Comp-neuro mailing list