Using response models to study coding strategies in monkey visual cortex

Usually the conditional probabilities needed to calculate transmitted infor mation are estimated directly from empirically measured distributions. Here we show that an explicit model of the relation between response strength ( here, spike count) and its variability allows accurate estimates of transmi tted information. This method of estimating information is reliable for dat a sets with nine or more trials per stimulus. We assume that the model char acterizes all response distributions, whether observed in a given experimen t or not. All stimuli eliciting the same response are considered equivalent . This allows us to calculate the channel capacity, the maximum information that a neuron can transmit given the variability with which it sends signa ls. Channel capacity is uniquely defined, thus avoiding the difficulty of k nowing whether the 'right' stimulus set has been chosen in a particular exp eriment. Channel capacity increases with increasing dynamic range and decre ases as the variance of the signal (noise) increases. Neurons in V1 send mo re variable signals in a wide dynamic range of spike counts, while neurons in IT send less variable signals in a narrower dynamic range. Nonetheless, neurons in the two areas have similar channel capacities. This suggests tha t variance is being traded off against dynamic range in coding. (C) 1998 El sevier Science Ireland Ltd. All rights reserved.