Basic Neuron Questions

r norman rsnorman_ at _comcast.net
Thu Apr 10 13:19:31 EST 2003


On Thu, 10 Apr 2003 15:44:25 +0000 (UTC), "Didier A. Depireux"
<didier at rai.isr.umd.edu> wrote:

>John H. <johnh at faraway.xxx> wrote:
>> Just for once, could someone say, "it's really quite straightforward." I
>> wonder how often lecturers in neuro related jazz feel threatened by their
>> students ... .
>
>So in a sense "it's really quite straightforward." The way you phrased your
>question prevented me from saying that!
>
>What I mean is that, people have been arguing about whether it's a time code
>or a rate code. And the truth seems to be a lot simpler, and neither rate
>nor time codes. It's in between, one spike per relevant time-scale. 
>
>If an engineer had been asked to design a system using spikes to communicate,
>he would have chosen either a rate or a timing code of sorts. The system
>designed by evolution is both more complex (because the relevant time-scale
>will depend on what is being encoded) and a lot simpler (you can use the
>same coding scheme for any sensory input). It's economical (precise timing
>requires a high metabolic rate, population code requires a lot of neurons)
>and quite flexible.
>
>I don't know your background, John. But maybe you don't appreciate how very
>very little we actually know about the brain. Our current brain research is
>to a large extent like butterfly collecting. We arrange the facts we observe
>according to their colors, size, shape, and put them in different boxes. But
>at a fundamental level we have no idea. 
>
>						Didier

Sorry to jump in so late on this thread, but there is a certain noise
level here that makes it very difficult to follow the few posts that
do actually  make sense.

Didier's comments are quite sensible, but I disasgree that 'it is
really quite straightforward."  The problem is, as Didier says, that
the nervous system was not actually designed -- it sort of just came
out that way by evolution (the "intelligent-design folks
notwithstanding). 

It is easy for an engineer to take things like look sort of like nerve
cells and create at least conceptual schemes where binary data can be
transmitted by the details of timing in a spike train.  The problem is
in showing that anything remotely like that actually happens in a
flesh and blood nervous system.  Even putting aside the conceptual
difficulties of how to do it, the experimental difficulties of finding
a preparation on which to test any hypothesis and then actually carry
out the work is currently out of the question.

Certainly there are specific cases where timing of nerve activity,
even to the sub-millisecond level, is critical.  And certainly there
are cases where a few interpolated spikes in a relatively steady train
of a constant frequency can cause large changes in response through
short term facilitation. But as a general rule, does critical timing
really count or is only a rough running average of frequency count?
For now, the latter seems like the usual way of coding.  Anyone who
suggests otherwise has a heavy burden of proof to show that the timing
mechanism is actually a general phenomenon in real, live brains, not
in conceptual models.





More information about the Neur-sci mailing list