Mathematical Biology Seminar

Brent Doiron, University of Pittsburgh
Thursday Feb. 25, 2010
4:15pm in LCB 215
Correlation shaping in the nervous system

Abstract: Correlated activity between neural outputs is widespread across the brain, well documented in sensory, cognitive, and motor areas. Neural correlations impact the coding performed by a population of neurons, as well as mediate interactions between distinct pools of neurons, and have thus received significant experimental and theoretical and attention. More recently, it has been shown that the correlation between neurons can be shaped by both stimuli and the cognitive state of a subject, and a variety of correlation shifts have been reported in diverse neural areas. While the function impact of such a shaping is clear, the underlying mechanisms responsible are poorly understood. In this talk I will present a basic framework from which to address how the shaping of neural correlation occurs and relates to simultaneous shifts in overall neural activity. I will draw from techniques in point process theory as well as non-equilibrium statistical mechanics, and discuss an emerging concept of correlation susceptibility extending the more common input-output single neuron model to encompass pairwise neural statistics. Theoretical results will match experimental findings from cortical slices, and in vivo recordings from somatosensory cortex as well as brainstem areas in weakly electric fish.