Abstract
When processing multiple visual items, the representation of individual items can be constrained by properties of the group. Prior studies have reported such ensemble bias, where memory for the size of an individual item is biased toward the average size of the group (Brady & Alvarez, 2011). However, it is unknown whether ensemble statistics influence individual representations during perception or over the course of memory retention. The current study aimed to dissociate perceptual-based and memory-based bias using a size comparison task. In the memory-bias condition (similar to prior designs), an array of 4 differently sized circles was presented briefly, and participants were instructed to remember the size of the one white-outlined target circle and ignore the three black-outlined nontarget circles. After a 2 second delay, another single white-outlined probe circle was presented at the center of the display. Participants judged whether the first (target) or second (probe) circle was larger. We expected to see the ensemble bias, where the size of the initial target circle was biased toward the average size of the array, resulting in a shift in the psychometric curve. In the perceptual-bias condition, the stimuli presentation order was reversed. The single probe circle was presented first, and the 4-circle array containing the target circle was presented after the delay. Because participants had to respond immediately after the target array presentation, any bias found in the perceptual-bias condition should indicate interaction at perception rather than bias developing over the WM retention period. In subsequent experiments, we further manipulated the length of the WM interval in the memory-bias condition. We found comparable bias toward the average ensemble size in both the memory-bias and perceptual-bias conditions. This result provides novel evidence of interaction between individual and group-level ensemble representations during the perceptual encoding stage.