Abstract

Critical branching processes and Poisson dynamics are the two prevailing models for the stochastic processes observed in neural circuits. However, they cannot be compared directly. The spiking process is intensely processed to generate a branching-like time series which requires various thresholds and binning. It has been shown that some results depend on the choices made in the processing of the original spike data and spurious effects might alter outcomes.

In my talk, I will provide a solution to this uncertainty due to the spike processing by circumventing it altogether. I will introduce and characterize the pumped branching process which can be tuned at will between critical branching and Poisson dynamics and present an analytical derivation of the corresponding spike statistics. Furthermore, I will compare results to experimental spike data and identify the parameters of the process, confirming that the spiking activity is in the reverberating regime