By Arvind Kumar (Stockholm, Sweden).
Across-trial variability is an ubiquitous feature of cortical activity. Across-trial variability is not fixed and depends on the stimulus and attentional shifts. Recently it has also become apparent that across-trial variability not only affects the stimulus response but also affects behavioral performance. In my talk I will discuss mechanisms that may underlie the control of across-trial variability. In particular, I will discuss the role of input statistics in shaping of across-trial variability in a model of cortical microcircuit. We found that across-trial variability is largely determined by (1) the excitation and inhibition (EI) balance and (2) the correlation between different types of inhibitory neurons (PV, SST) and excitatory cells. In fact, the contribution of input EI balance in the control of across-trial variability is contingent on the EI correlations. This analysis also allowed us to identify conditions under which the response of the neuron population within a single trial may exhibit a heavy-tailed distribution. Overall this work helps us understand how input statistics affects the network output and isolates the role of specific components of the network.