[Neuroscience] A general question about neural curcuit
(by r_s_norman from _comcast.net)
Tue Apr 17 10:12:00 EST 2007
On Tue, 17 Apr 2007 22:18:45 +0800, "¶Ðñ" <hslinhc from gmail.com> wrote:
>hi everyone, i'm new in computational neuroscience and I've got a question.
>If three neurons were connected to form a closed circuit. Would the spiking
>signal transfer in the circuit permanently or would it decrease and finally
>the question is derived from the modeling of spiking neuron networks.
>According to which, if one neuron spikes, and the electricity voltage is
>above the next neuron's threshold, it would spike the next one. So in a
>model neural network, the signal would transfer persistently.
A closed circle of connections like you describe is not really
functional, physiologically. Theoretically, such a circuit could
continue to fire indefinitely. The problem is that a small circuit,
like you describe, would transmit the activity around the circle too
quickly and you run into problems with the refractory period. Also,
neurons usually do fatigue or adapt by a variety of mechanisms (as in
failing to completely remove sodium inactivation with each successive
action potential) so the process will generally last only a short
time. It will fail suddenly and completely, not gradually decrease
That being said, there are circumstances where persistent cycles of
activity do occur, at least for a time. Fibrillation in the heart is
one example. Short term memory, at least some portions of it, may
involve "reverberating" circuits.
Note: I am talking about real biological neural circuits. You can
make computational "neurons" do anything you want.
More information about the Neur-sci