possible new way to model neural nets?

Eli Meir meir at zoology.washington.edu
Fri Mar 31 18:35:39 EST 1995


I don't know about these particular examples, but in general, I think
there are very few developmental events that use reaction-diffusion
methods such as the ones you're talking about.  Basically, if nature
wants a certain substance in a certain place, it puts it there
directly, not through some sort propagating waves.  One proof of this
is the range of temperatures at which embryos can develop - you can
take fly embryos and develop them at 17 degrees, or at 30 degrees and
they come out just fine.  This changes diffusive properties only
negligably, but changes reaction processes a lot, so the conclusion is
that there's not much reaction-diffusion involved in development.

I don't know how this applies to neural nets, where the reaction speeds
don't change as much with temperature I think.  But its doubtful
whether the neurons get set up (develop) using propagating waves (once
they're set up, it may be a different story - in fact, you may want to
check out Art Winfree's work on heart attacks).


Eli Meir
Dept of Zoology, Univ of Washington
meir at zoology.washington.edu



More information about the Neur-sci mailing list