How can an engineer learn from neuroscientists
Mitchell Gil Maltenfort
mitchm at netzone.com
Mon Mar 3 12:10:15 EST 1997
In article <5fdbmu$189m$1 at news-s01.ny.us.ibm.net>, junior1 at ibm.net [Bernie
>In <mitchm-0103970153200001 at scottsdale-ts3-18.goodnet.com>,
mitchm at netzone.com (Mitchell Gil Maltenfort) writes:
>>I find that a lot of the 'computational neuroscience' work out there is
>>very elegant, very ambitious mathematical and computational work which is
>>clearly inspired by neurobiological fact but has too many simplifications
>>or assumptions to be considered a definitive explanation for the
>Yes...and that is the problem. Most (?) of the assumptions have not been
You mean "substantiated" or "accepted" right? "Blessed" sounds like it's
from the list of "fightin' words" :^>
>by neuroscientists. As a result, brain scientist don't trust the results
>more often than not the results obtained are ignored instead of being
>and discussed (good old feedback).
Sorry, look at the Neural Information Processing Systems (NIPS) conference
for starters (I *knew* there was a conference I'd forgotten when I
responded to the initial post!) This is specifically a place for
engineers and biologists to meet and mingle. Journal of Neurophysiology
runs papers based on modeling of individual neurons or systems thereof.
But an experiemental paper is not going to be well-received if it ignores
relevant questions or substantiated facts about the system that is the
focus of the paper. A computational paper has to meet the same criteria.
I admit, on the surface it seems unfair - the engineer has to meet the
scientist but the neuroscientist won't meet the engineer - but that is the
system in place.
and I'd like to refer anyone who's read this far to Kevin Spencer's
discussion of other points raised by the author of the previous post.
More information about the Neur-sci