Neurobiologically Realistic Determinants of Self-Organized Criticality in Networks of Spiking Neurons
Figure 9
Robustness of self-organized critical network dynamics.
Statistical significance of power-law distributions of avalanche sizes (black) and durations (red) as a function of network randomization for the optimal power-law () density scaling function associated with (a) changes in external current (default
), (b) changes in conduction delays (default delays are uniformly distributed between
and
), (c) weaker postsynaptic response (default
) and slower STDP learning rate (default
) (d) doubling of module size to
neurons and network size to
neurons and reductions in postsynaptic response, STDP learning rate and external current (e) changes from soft to hard STDP weight bounds, and (f) removal of inhibitory synapses. Gray lines show the
threshold for power-law scaling. Error bars represent the standard error of the mean from
simulations.