The MLE algorithm can be adapted to different spatial distributions (exponential, Gaussian, lognormal, linear, etc.). Our results suggest that within local circuits of the visual cortex, distance-dependent interactions are best characterized by an exponential decay.

We compared the MLE performance to related approaches: generalized linear models (GLM), transfer entropy, Granger causality and spike-count correlations, on both Poisson and Izhikevich models. Further, we applied MLE to calcium imaging data from Stringer et al. (2019).