Abstract
Recent evidence suggests that observers encode information about the central tendency of a visual array independent of details about individual features. For example, thresholds for judging the mean difference between two arrays of circles are equivalent to or lower than those for comparing two isolated circles. However, human performance on perceptual averaging tasks is sensitive to the variability (or heterogeneity) of visual information independent of the mean. Here, to ask whether observers automatically encode summary information about the variability of visual information in a scene, measured reaction times (RTs) on a task in which observers judged the average shape or colour of a target visual array that was preceded by an irrelevant prime array. Manipulating the mean and variability of the feature information on prime and target arrays orthogonally, we observed an interaction whereby a highly variable prime array led to faster RTs for a subsequent high variance target, and a less variable prime facilitated RTs to low-variance targets. This facilitatory variance-adaptation effect occurred with prime-target intervals as short as 100ms. A control experiment introduced variance on a task-irrelevant dimension and showed that the results were considerably stronger in the decision-relevant dimension. This variance-adaptation effect, which resembles previously reported adaptation to response conflict between sequential trials, suggests the existence of a mechanism by which the range or dispersion of visual information is rapidly extracted. This information may in turn help to set the gain of neuronal processing during perceptual choice.
Meeting abstract presented at VSS 2013