Fig. 1.

Network behavior for recurrently connected leaky integrate and firing neurons (a) The mean firing rate of the network for increasing synaptic weights. The peak value of the firing rate increases for stronger couplings (green lines), and the network decays at a slower rate. At some critical value of recurrent weights, the network becomes bi-stable (black line) as the weights increase further (red lines) the firing rate of the ‘UP’ state increases. The mean firing rate is averaged over all neurons in the network and convolves with an exponential smoothing kernel as explained in the methods section. (b) Firing rate of the UP state just above the critical weight, for different synaptic time constants from 20ms to 100ms. (c) Decay time (T) increases exponentially for gradually increased synaptic weights (W). The shape of the curve depends on the synaptic time constant. Curves for 100ms synaptic time constant (dashed line) reduces the steepness of the curve slightly compared to 25ms time constant (solid line). The value of W is normalized by Wc; the critical value of the weight parameter at which the network becomes bi-stable