Figure 12.

Varying the simulated network size. We explore how the network output varies when we change the total number of simulated neurons. We again use the homogeneous network with Γ_k = a for k = 1, 2,…, N/2 and Γ_k = −a for k = N/2 + 1,…, N, σ_V = 0.08 μA/cm² · $\sqrt{s}$, and the box function input. As derived in the Material and Methods, Scaling when varying the simulated network size, we use the scaling a = 40/N, g = c₀(400/N), and the different colored lines correspond to different values for c₀. Error bars represent SDs over 900 repeated trials. a, We plot the inverse of the population synchrony index as a function of the simulated network size for different values of the parameter c₀ which scales the gain of the synaptic input. b, Relative error between the estimate and the actual signal as a function of N. c, The inverse of the time and population averaged firing rate during the period of zero stimulus input as a function of N. d, Square root of the mean integrated error as a function of N on a log–log plot. The two dashed black lines plot 1/$\sqrt{N}$ starting from the first cyan data point and the first magenta data point.