Fig. 5

Using a single number to predict \(d\hat{S}/dF\). (A) Paths through parameter space for the asynchronous network: white shows the part of the x-plane where \(\frac{d\hat{S}}{dF} = \frac{\nabla \hat{S} \cdot d\mathbf {x}}{\nabla F \cdot d\mathbf {x}} > 0\), black where \(\frac{d\hat{S}}{dF}<0\), where ∇Ŝ and ∇F are computed at the center of the diagram. (B) Correlation susceptibility vs. firing rate, for each path illustrated in (A). (C)–(D) As in (A)–(B), but for the strong asynchronous network