Unsupervised statistical learning of higher-order spatial structures from visual scenes

Three experiments investigated the ability of human observers to extract th e joint and conditional probabilities of shape cooccurrences during passive viewing of complex visual scenes. Results indicated that statistical learn ing of shape conjunctions was both rapid and automatic, as subjects were no t instructed to attend to any particular features of the displays. Moreover , in addition to single-shape frequency, subjects acquired in parallel seve ral different higher-order aspects of the statistical structure of the disp lays, including absolute shape-position relations in an array, shape-pair a rrangements independent of position, and conditional probabilities of shape co-occurrences. Unsupervised learning of these higher-order statistics pro vides support for Barlow's theory of visual recognition, which posits that detecting "suspicious coincidences" of elements during recognition is a nec essary prerequisite for efficient learning of new visual features.