Figure 6: Bursting introduces multi-stability and hysteresis in the network dynamics.

(A) The increase in firing rate due to increase in external input and change in the burstiness of the neuron (dashed grey lines) is shown. The simulation protocol to generate this neuronal network hysteresis is described in Methods. It is seen that the onward (blue line) and return (brown line) curves do not trace the same path indicating the state dependence of the effect of the single neuron firing pattern on the network. The grey dots show a similar hysteresis loop for a network in which the burstiness of only 20% of the inhibitory neurons is changed. The inset plot shows the change in the firing rate of the network and therefore the burstiness of the modified-SSBN after given an initial perturbation of additional external input of 200 spikes/sec. The burstiness of the inhibitory neurons (as defined by the state variable (see Methods)) increases with the excitatory population firing rate. The increase in bursting in turn increases the population firing rate. This self-propelling mechanism continues till the single neurons produce the maximum number of spikes per burst (B = 5). (B) This panel is similar to (A), but the firing rate estimate of the excitatory population is estimated over a time window of 200 ms. The number of spikes per burst increases by 2 for every crossing of the firing rate threshold.