Overestimation of variability in ensembles of color value and size.

Studies have shown that people can derive summary statistics - such as the mean - from sets of similar objects for low-level (orientation, color value), mid-level (size), and high-level visual features (emotional expression) through the phenomena of ensemble perception. Recent research has identified a bias to overestimate variability in both static and dynamic arrays of lines at various orientations - referred to as the variability overestimation effect. Here, we explored whether the variability overestimation effect generalizes to other visual features, namely color value, and size and whether it generalizes to different response types. Such generalization would be consistent with the idea that this overestimation is inherent in ensemble perception processes. In the current experiments, participants saw a set of nine circles that varied in either size or color value and estimated the variability of the set. Overall, participants overestimated variability in color value and size. This overestimation was more pronounced when the set had lesser variability. The fact that the visual system overestimates variability across different features raises the possibility that this bias is encompassed within ensemble perception. The exaggerated bias when variability is low for orientation, size, and color value is consistent with a common mechanism underlying these biases. Understanding the perception of variability in ensembles has theoretical implications for ensemble perception processes and has applied implications such as how to design visualizations that require making judgments about critical but uncertain information such as the possible trajectory of the path of a hurricane.