COSYNE

Log in

Neurons exhibit stochastic variability in their spike responses to repeated stimuli. However, single-neuron measurements of variability neglect the fact that neurons in a local population exhibit statistical dependencies between their responses, which cause variability to be shared across multiple neurons. Traditional raster plots therefore reflect only a lower-bound on the precision and reliability of single-neuron responses. In this talk, I will present a joint encoding model of spiking activity in a population of macaque retinal ganglion cells, and show how it can be used probe the relative contributions of intrinsic and network stochasticity to the responses of a single cell. We find that a substantial fraction of a single neuron's variability can be explained by shared variability in the population response, and that single-neuron spike responses can be predicted more accurately when correlated population activity is taken into account.