CRCNS: Linking Synaptic Populations and Computation Using Statistical Mechanics

Computations performed by single neurons result from integration of a large myriad of synaptic inputs distributed throughout complex dendritic topology. Synaptic inputs vary in substantial ways; sensory-driven activity patterns, probability of activation, synapse location within the dendritic topology, local dendritic organization, and ultrastructural characteristics are all thought to be critical for determining how synaptic inputs influence the spiking output of a neuron. Populations of synaptic inputs, ultimately, determine the coding capacity and computations single neurons can perform. Despite this fact, studies largely overlook the synaptic input population within a single neuron, instead focusing on the activity of cellular populations, using biophysical models or constructing models that create hypothetical weights or synapses (e.g. deep-neural networks). Thus, a critical question in neuroscience remains how ensemble synaptic activity is integrated in vivo and what are the fundamental principles of synaptic organization which describe neural computation. To address this issue, we present a tightly integrated experiment-theory approach. We propose to (1) measure the sensory-driven activity patterns of large populations of dendritic spines on layer 2/3 visual cortical neurons in ferret visual cortex in vivo, and (2) use a statistical physics approach to characterize the structure and computing of synaptic populations in multiple contexts. Thus, this project will provide fundamental knowledge about the synaptic architecture of neurons in the brain. The