Figure 2.

Structural Bounds on the Balanced State

Heterogeneous networks are generated with each presynaptic population having identical CV_K and correlation coefficient between in-degrees from pairs of populations given by $c$.

(A and B) Color maps of simulation results scanning the 2D structural parameter space of heterogeneous networks firing at 10 Hz for a 60 s trial. Dotted black line shows the theoretical bound (Δ ⋅ K ≈ 1) below which we expect balance to hold. (A) Fraction of neurons quiescent throughout the 60 s trial. (B) CV_ISI over the same parameter space. Similar qualitative results are obtained at different firing rates.

(C) Fraction quiescent (blue) and CV_ISI (red) (from A and B) plotted against the deviation from fully correlated in-degrees (given by Δ ⋅ K) shows that this deviation determines the dynamics of networks independent of respective values of CV_K and c.