Figure 4.

Probing other ORN input noise values

Testing other ORN input noise values to show robustness of prior results for matching time-varying data statistics, but using a steady-state approximation for computational feasibility. Each triplet of input noise values has a small fixed σ (black), time-varying with a step (tan), and large fixed σ (blue); as before, the time-varying values are set to the small values in the spontaneous state and large values in the evoked state. Results for Total Error (A), Error of Variance (B), and Error of Covariance (C) are all consistent with what we observed before (Figure 3C), with the differences converging as the noise values approach each other. We tested 10 total triplets of input noise sets, using a random sample of coupling strengths (Figure S1C) that cover the best parameter values (Figure 3D), and a random set of scaling factors (Figure S1C) to approximate spike statistics in 100 ms windows. The whiskers and inter-quartiles in the box plots capture the variation of errors over the random samples of coupling strengths and scaling factors.