平特五不中

Growing evidence supports the hypothesis that perception and behavioral responses to relevant sensory input are determined by integrating the activities of large heterogeneous (i.e., with different response properties) neural populations in the brain. Understanding such 鈥減opulation coding鈥� has however been complicated by the fact that neural activities are not independent of each other (i.e., are correlated) and that such correlations are highly plastic (i.e., depend on behavioral context). In this talk, I will present recent results showing how heterogeneities and spatially-dependent correlations affect the ability of sensory neural populations to accurately provide information as to object location relative to the organism. Specifically, we used high-density electrode arrays (i.e., Neuropixels probes) to record from sensory neural populations in the electrosensory system of Apteronotus leptorhynchus. Correlation magnitude varied with location and was highest at the receptive field edges and strongly decreased towards the center. Further analysis revealed that such correlation plasticity increases information about object location by making neural activities less redundant. Moreover, through mathematical modeling, I will also demonstrate that there is an optimal non-zero level of heterogeneity for which object location information reaches a maximum. Overall, the many similarities between the electrosensory and mammalian systems (e.g., visual) make it likely that results obtained here will be applicable elsewhere.

This seminar will take place in-person and online (details in attached poster below)