The neuron as an analog NOT digital system

alex taylor ataylor at
Thu Aug 10 21:43:38 EST 1995

In article <jimnash-1008951543270001 at>,
Jim Nash <jimnash at> wrote:
>In article <40787d$q6n at>, ma2 at (Marc
>Anton) wrote:
>The current view of neurons is more of a frequency coded device that has
>summed or multiplicative inputs that are either analog and slow (non
>junction) or frequency encoded and fast (junctions).  Frequence encoding
>can be represented by an analog number and hence the current view of
>nerves is essentially analog.
>Reducing nerves to a digital system is a little like taking means and
>variance as a measure of electrical activity where these are gross
>simplifications of a complex ionic channel phenomena.  One has to reduce
>to be able to model.  One has to model to gain insight.  One can only
>attempt not to extrapolate far beyond the assumptions.  But now we are
>talking about the scientific method not nerves.  Reduction is the weakness
>and power of science.  It can't be helped.

	Actualy there a number of credible top-down strategies as opposed
to strictly bottom up strategies. Even the Churchlands, with their
hard-line stance against computational psychology, concede that at the
very least some combination of top-down and bottom-up (i.e reductionistic)
strategies is required -- they call it the quaquaversal approach. Pylyshyn
wrote a turgid but rich article on the justification for a computational
approach to the brain in BBS, vol. 3.

More information about the Neur-sci mailing list