Abstract
Ensemble perception provides a visual summary of a scene, such as the mean size of a group of objects. The precise mechanisms underlying ensemble perception are still unknown. We tested whether different types of ensemble tasks rely on or compete for shared processes. Perception of the mean requires extracting the similarities across the display, whereas perception of variability requires extracting the differences. We tested whether judging both mean and variability would exert a cost on perceptual sensitivity. An array of circles of different sizes briefly flashed on the screen, and participants (N = 315) judged the mean, variability, and/or numerosity of the display. Participants were less sensitive to mean judgments when also judging the variability. Sensitivity to variability was similar regardless of whether participants also judged the mean. These asymmetrical results suggest that the mean is automatically perceived when judging variability, and thus there is no additional cost to also reporting the mean. However, variability is not automatically perceived when judging the mean, suggesting a cost to also reporting variability. In statistics, calculations of both the mean and the variability require a measure of the number of samples. We explored whether the perception of variability also required perception of the number of samples, in which case, there should be no cost to also judging the number of items. Sensitivity to variability was unaffected when also making a judgment of the number of items. When perceiving variability, judgments of other ensemble properties such as mean and numerosity do not impair variability judgments. In contrast, sensitivity to the mean circle size and numerosity were both impaired when also having to judge variability. Overall, these results show an asymmetrical process underlying ensemble perception, suggesting that ensemble perception could be driven by domain-specific mechanisms dependent on the type of summary statistic computed.