Statistical Assessment of Interaction Among Brain Regions from Multielectrode Recordings

Contemporary technologies for recording neural activity from an alert animal have created great scientific opportunities and substantial data analytic challenges. One of the big opportunities is to examine how multiple brain areas work together to achieve perception or behavior. There are many ways to characterize such coordinated neural activity, and many groups have been actively developing new methods of analysis. My statistical perspective has made me especially curious about information contained in the timing of responses, which I have felt to be inadequately understood. In this talk I will explain, very briefly, why timing relationships-which may involve oscillatory activity-seem important. I will then summarize several research projects recently completed by trainees working with me. One is based on a hierarchical Bayesian point process model for multiple population spike trains (as yet unpublished). A second produces a multivariate generalization of phase coupling (statistically, a measure of correlation for circular data) by defining an exponential family on the multidimensional torus in order to analyze oscillating local field potentials recorded simultaneously from many regions (Klein, N., Orellana, J., Brincat, S., Miller, E.K., and Kass, R.E., Torus graphs for multivariate phase coupling analysis, Annals of Applied Statistics). Local field potentials are noisy representations of population activity emanating from many local sources and can be difficult to analyze. The third piece of work deconvolves local field potentials by building a statistical model around a biophysical forward model, thereby producing a new method of estimating current sources (Klein, N., Siegle, J.H., Teichert, T. and Kass, R.E. (2021). Cross-population coupling of neural activity based on Gaussian process current source densities, to appear in PLoS Computational Biology). I will try to indicate how I think this kind of research can advance neurophysiology, and what remains to be done in order to make it fully effective.

Seminars will be held weekly on Fridays 3:30 - 4:30 pm on Zoom. After the seminar, there will be a (virtual) meet-and-greet session to interact with the speaker. Please use the chat on Zoom to ask questions to the speaker. A moderator will collect questions throughout the talk and ask the speaker at appropriate times.

Zoom Meeting ID: 923 9738 2385
Passcode: 425966