In article <1mlikj$5lp at agate.berkeley.edu> lmk2 at garnet.berkeley.edu (Leslie Kay) writes:
>From: lmk2 at garnet.berkeley.edu (Leslie Kay)
>Subject: Re: How many inputs are needed to excite/saturate a neuron
>Date: 26 Feb 1993 17:05:23 GMT
>In article <TAL.93Feb25224725 at copley.bu.edu> tal at copley.bu.edu (Doron Tal) writes:
>>>>On the one hand, anatomical evidence suggests that a neuron has, on
>>average, on the order of THOUSANDS of neurons impinging on it. On the
>>other hand, physiological evidence shows that only a few epsps are
>>needed to generate an action potential - which would imply that ONE
>>cell is sufficient for exciting the postsynaptic neuron.
>>>>Can anyone explain this disparity between anatomy and physiology?
>>>>1. most of the thousands of synapses on a neuron are inactive
>>2. a neuron that has thousands of inputs has an extremely large and
>>>>If we knew what the range of a neuron is, on average, then we'd be
>>able to tell between the two above hypotheses.
>In addition to the excitatory connections which a neuron can
>receive on its dendrites, it can also (and often does) receive
>inhibitory synapses. Inhibition has not only a net effect when
>looking at the dynamics of neural masses, but it also has the
>effect of modulating the efficacy of excitatory input. There
>is an interesting review article in a recent issue of TINS by
>Roland Jones (TINS, Vol 16, No 2, p58). In it he discusses
>the regulation of EC input to the Hippocampus (EC= Entorhinal
>Cortex) by large networks of inhibitory neurons in both layer
>II of the EC and the Dentate Gyrus of the Hippocampus. As
>a result, he reports that it can take up to 400 axons in
>the perforant path to excite one granule cell in the Dentate
>Gyrus. (This, I believe, is because of the inhibition in
>the DG). In addition it seems to be very hard to excite the
>neurons in the EC (spiny stellate cells) which project through
>the perforant path, because of the inhibitory network in the EC.
>In answer to your question, I'd say "it depends".
Simulations of neural networks show, that memory aspects do not reside
within a single neuron, but are distributed over a lot (probably thousands)
of synaptic connections. Thus, recall from memory does not at all need
the activation of all the incoming axons.
S.Kolodzie, Duesseldorf, Germany