Maharjan, Dubé, and Sekuler (2020 draft)

The briefest glimpse of some scene can suffice to allow the extraction of a scene's overall character. That brief glimpse produces a compressed, prototype representation of the scene, that is, its gist. Many studies have provided evidence for the remarkable accuracy of such compressed/prototype representations, but few have explored factors that control the emergence of such representations, or possibly distort the products. We addressed these questions by using small arrays of luminances. We focused on how clustering or repeating items impact perception of an array. The results showed that repetitions lowered estimation errors, though the mean of the array continued to strongly predict responses. The data suggest that subjects may capitalize on informational redundancy and emergent structure in computing summary representations. The implications for theoretical accounts of ensemble coding are discussed.