Balanced state of networks of winner-take-all units
Fig 9
Properties of network producing heterogeneous fluctuation-driven firing rate dynamics.
(a) Sample firing rate autocovariance function from the infinite mean-field simulations in Fig 8c, averaged across all co-tuned populations (for 8 example initializations of the same network). (b) As in (a) but computed from the full spiking network simulation (N = 2000) (c). ISI CV2 distributions for several example co-tuned populations from Fig 8e. (d). Firing rate vs ISI CV2 for all neurons in the network. (e) Spike-train autocovariance function, averaged across all neurons (each of the 8 traces shows a different initialization, although the traces are very similar).