PHYSICAL MATHEMATICS SEMINAR



TITLE:			DYNAMICS OF NEURONAL NETWORKS 
			WITH HIDDEN NEURONS

SPEAKER:		DAVID CAI
			NEW YORK UNIVERSITY



ABSTRACT:

We study networks of all-to-all coupled, integrate-and-fire, excitatory neurons, with a portion
of the population receiving feedforward drive and the remaining, "hidden", portion receiving no
feedforward input. We demonstrate that this system undergoes a subcritical bifurcation as either 
the input or recurrent coupling is varied. Because of this transition, the network gates feedforward 
inputs and exhibits bistability and hysteresis. The hidden population allows this network response to 
be continuously tunable in the forcing and coupling strengths. Furthermore, these features persist in
the presence of synaptic failure and for small network sizes, suggesting that the computations discussed 
here could be implemented in biological networks.  Finally, we demonstrate that a long network correlation, 
orders of magnitude longer than the synaptic time, emerges from the recurrent coupling.  This correlation
time scales with network size and with synaptic connection probability. 
	

TUESDAY NOVEMBER 19, 2002
2:30 pm
Building 2, Room 338

Refreshments will be served at 3:30 PM in Room 2-349




                        

Massachusetts Institute of Technology
Department of Mathematics
Cambridge, MA  02139