Abstract
Crowding refers to the deleterious influence of nearby contours on visual discrimination. A popular theory for crowding is that it arises as the consequence of inappropriate feature integration. This theory predicts that the effect of crowding increases with the number of features in close proximity to the target. We tested this prediction by examining how letter crowding depends on the perimetric complexity of flanking letters, a measurement that correlates with the number of features. We analyzed a total of 96000 trials in which 16 observers (6000 trials per observer) identified the middle (target) letter of sequences of three crowded lowercase letters (center-to-center separation = 0.8x the x-height) presented at 10° in the inferior visual field. Each letter was randomly drawn from the 26 letters of the Roman alphabet. Eight observers were tested with the Times-Roman font and the other eight with the Courier font. The perimetric complexity (perimeter squared/“ink” area) was determined for each letter and the sum was calculated for the pair of flankers on each trial. We binned the perimetric complexity of the flankers into 10 groups. In general, the error rate of identifying the target letter increased linearly with the perimetric complexity of the flankers, for both Times-Roman (r = 0.98) and Courier (r = 0.99) fonts. However, the increase of the error rate of identifying the target letter with flanker complexity depends on the target letter complexity such that target letters of low complexity are more susceptible to the flanker complexity, while target letters of high complexity are less susceptible to the flanker complexity. These findings are consistent with the prediction based on the inappropriate feature integration account of crowding, and strongly support the speculation that feature integration is a competitive process that depends on the relative proportion of features between the target and flankers.
Supported by NIH grant R01-EY012810.