Capacity of the brain

Bill Skaggs skaggs at bns.pitt.edu
Tue Aug 31 14:39:55 EST 1999


Ralph Leonhardt <leo at neuroinformatik.ruhr-uni-bochum.de> writes:

> Hey,
> 
> you counted the memory elements of a (the) brain? Then you can probably
> tell us poor neurobiologists what they are and how they work ;-).
> 
> Greetings, Leo

Well, I *am* a neurobiologist (Asst. Professor of Neuroscience, U of
Pittsburgh), but it wasn't me who counted the memory elements.  Most
people, I believe, think that the synaptic connections between
cortical pyramidal cells are modifiable memory elements, and there are
about 10^14 of them in the human brain.  No other system in the brain
contains even a tenth as many synapses, so it is reasonable to ignore
the others for a first approximation.  It is possible, of course, that
only a fraction of cortical connections are memory elements---there is
little data to address this question.

Anyway, if we assume that all synapses between cortical pyramidal
cells are memory elements, then the remaining question is, how much
memory can each synapse hold?  This is a VERY difficult question; my
guess is that the answer is somewhere around 0.01 -- 0.1 bits per
synapse.  Why such a low number?  Because the important question is
not how much information can be stored in a synapse but rather how
much information can be extracted from it.  The way to extract
information from a synapse is to fire the presynaptic cell and look at
what happens to the postsynaptic cell.  Because cortico-cortical
synapses are individually weak and subsist in a very noisy
environment, the information that can be extracted in this way is
limited.  There is no way to calculate it precisely without making
assumptions that are very questionable---the numbers I have given are
what I consider a very crude but reasonable estimate.

	-- Bill



More information about the Neur-sci mailing list