Bayesian inference of neuronal ensembles

Abstract

In many areas of the brain, both spontaneous and stimulus-evoked activity can manifest as synchronous activation of neuronal ensembles. The characterization of ensemble structure and dynamics provides important insights into how brain computations are distributed across neural networks. The proliferation of experimental techniques for recording the activity of neuronal ensembles calls for a comprehensive statistical method to describe, analyze and characterize these high dimensional datasets. Here we introduce a generative model of synchronous activity to describe spontaneously active neural ensembles. Unlike existing methods, our analysis provides a simultaneous estimation of ensemble composition, dynamics and statistical features of these neural populations, including ensemble noise and activity rate. We also introduce ensemble \"coherence\" as a measure of within-ensemble synchrony. We have used our method to characterize population activity throughout the tectum of larval zebrafish, allowing us to make statistical inference on the spatiotemporal organization of tectal ensembles, their composition and the logic of their interactions. We have also applied our method to functional imaging and neuropixels recordings from the mouse, allowing us to relate the activity of identified ensembles to specific behaviours such as running or changes in pupil diameter.