Abstract
We evaluated the relationship between two indices of visuospatial integration: the susceptibility to visual crowding (Pelli, Palomares, & Majaj, 2004) and the encoding of summary statistics of visual arrays (Ariely, 2001). While visual crowding is the disruption of an element's identifiability when embedded in an array of multiple elements, statistical averaging of visual arrays is the accurate identification of the global statistical average relative to the identification of individual members. It has been proposed that crowding of a target occurs due to compulsory averaging of the target with its neighbors (Balas, Nakano, & Rosenholtz, 2009; Dakin, Bess, Cass, & Watt, 2009; Whitney, 2009). In the current study, we presented three squares along an imaginary circle (5 deg viewing eccentricity) and asked observers to perform a crowding and an averaging task as a function of inter-element distance. In the crowding task, we asked observers to identify the size of the central square (i.e. target) in between the two flanking squares (i.e. distractors). In the statistical averaging task, we asked observers to identify the average size of all the squares. Results show that accuracy for identifying the target improves with increasing inter-element distance in the crowding task, but accuracies for identifying the average size were unaffected by inter-element distance. However, additional analysis shows a reliable effect of choice type (i.e. target vs. average size; target vs. distractor), suggesting that “implicit” averaging contributes to the loss of identification accuracy in visual crowding. These results suggest that explicit statistical averaging of size does not share distance-dependent mechanisms that mediate visual crowding.