Effects of ensemble complexity and perceptual similarity on rapid averaging of hue

The ability to extract the mean of features from a rapidly viewed, heterogeneous array of objects has been demonstrated for a number of different visual properties. Few studies have previously investigated the rapid averaging of color; those that did had insufficient stimulus control or inappropriate methods. This study reports three experiments that directly test observers' ability to extract the mean hue from a rapidly presented, multielement color ensemble. In Experiment 1, ensembles varied in number of elements and number of colors. It was found that averaging was harder for ensembles with more colors but that changing the number of elements had no effect on accuracy, supportive of a distributed-attention account of rapid color averaging. Experiment 2a manipulated the hue range present in any single ensemble (varying the perceptual difference between ensemble elements) while still varying the number of colors. Range had a strong effect on ability to pick the mean hue. Experiment 2b found no effect of color categories on the accuracy or speed of mean selection. The results indicate that perceptual difference of elements is the dominant factor affecting ability to average rapidly seen color ensembles. Findings are discussed both in the context of perception and memory of multiple colors and ensemble perception generally.