MC 6460
Speaker
Wilten Nicola, Hotchkiss Brain Institute, University of Calgary
Title
Mesoscale Imaging Reveals the Markovian Dynamics of the Brain
Abstract
Mesoscale cortical dynamics, which are recordings from the entire cortical surfaces of mice, consist of stereotyped, spatio-temporal patterns of neural activity. Whether the sequencing of these motifs follows an intrinsic ‘grammar’ is not known. Here, we use mesoscale cortical imaging and computational modelling of neocortical activity motifs as a novel lens to characterize cortical dynamics. A first-order Continuous Time Markov Chain model probabilistically describes the temporal sequences of activity motifs using Markov Elements, where the probability of transition depends only on the current element occupied. This approach creates a ‘Markovian neural barcode’ describing both the probability of Markov element transitions, in addition to the sum total of element occurrence as an occupancy distribution. The Markovian neural barcode describes an underlying grammar governing large scale cortical dynamics, and further reveals individual animal specific signatures and deviations from this overarching grammar, akin to style or dialect. Across experimental and pharmacological interventions, we show that mesoscale activity motif sequencing is highly structured and sensitive to neural perturbations using the Markovian neural barcode. This approach provides a powerful lens from which to characterize normative and pathological cortical function.