Schematic representation of the computation of the mutual information carried by spikes about which part of a dynamic stimulus was shown. The figure illustrates how we obtain different probabilities needed to compute (through (1)) the information about a movie carried by the spike responses (spike counts in this illustration). (a) First the stimulus presentation time is portioned into nonoverlapping windows, each considered a different stimulus (a “scene”). The set of stimuli is the set of different scenes, each of which is presented once every trial, so the probability of each scene is the inverse of the number of the scenes presented. (b) The raster plot shows a cartoon with simulated spike times across all trials and movie scenes. From these data we compute (c) probability distribution of the spike counts across all trials and scenes, (d) and (e) probability distribution of the spike counts across trials given the presented scenes and , respectively. Computing probability distributions of the spike counts across trials for all scenes, using them to calculate the joint probability of spike counts and scenes, and then applying (1), we obtain the actual value of the mutual information.