Bayesian analysis of identification performance in monkey visual cortex: Nonlinear mechanisms and stimulus certainty

AbstractThe identification performance of single neurons in the primary visual cortex was quantified by measuring how accurately one could know the stimulus based upon the neuron's response. We found that for a typical neuron a response of 10 action potentials, following one brief stimulus presentation, was sufficient to classify the stimulus as belonging to a relatively small region in stimulus space, with a high degree of confidence. The performance was better than that which could be attained through linear summation of excitation and inhibition alone. The results suggest that the enhanced performance is a consequence of two nonlinear mechanisms: contrast gain control and expansive response exponent