Neurontropy, an entropy-like measure of neural correlation, in binocular fusion and rivalry

Dynamic random dot stereograms were generated for which the left and right arrays were either identical (100% correlation), or uncorrelated (0% correlation), or the complements of each other (-100% correlation). Any two of these three states of correlation were presented in succession and duration thresholds for detecting the transitions were measured. These thresholds were much longer when the transition went from the uncorrelated state to the correlated state than vice versa. In order to explain the detection thresholds for the various transitions a model based on the notion of an entropy-like measure (to be called neurontropy) has been proposed. It was assumed that in binocular vision both a fusional and a rivalry process operate simultaneously, but in a dual fashion. Thus the correlated state would be regarded the same way by the fusional process as the complemented state by the rivalry process. Transitions from the uncorrelated to the complemented state (and vice versa) were the most difficult to detect, a task which only the rivalry process could accomplish. The long detection thresholds indicate that the rivalry process is less efficient than the fusional process.