Visual crowding reduces the effective contrast of target letters

Abstract

Background: When the target and the flanks have the same contrast, flanked letters (xax) require more contrast to identify than do letters presented alone (Fine, 2003; VSS). This suggests that visual crowding results, at least in part, from an increase in stimulus noise. It is well known that visual crowding is reduced when the flanking letters are spatially separated from the target letter. Here we test the hypothesis that factors that modulate visual crowding (i.e. flank-to-target spacing) will also modulate the stimulus contrast necessary for target identification. Methods: Twelve observers with normal visual acuity and contrast sensitivity identified single letters and letters flanked by an ‘x’ on each side. Target letters were presented at &+-0.34, 1.04, 1.70, and 2.38 deg for 100 msec. Inter-letter spacings for the flanked stimuli were 0 (normal spacing), 1, 2, and 3 letter-spaces between the target and the flanks. Contrast thresholds were determined for each stimulus type at each location (using a staircase procedure), as was percentage correct for full contrast stimuli. Results: Contrast thresholds were significantly increased relative to single letters for the 0- and 1-space conditions (p < 0.009). Similarly, performance for the full contrast stimuli was significantly reduced for the 0-space stimuli (p < 0.001) and marginally reduced for the 1-space stimuli (p = 0.090). Across the five stimulus types, regression analyses showed that contrast thresholds accounted for 57% of the variance in performance for the full contrast stimuli (p < 0.001). Across observers, the median variance accounted for was 64%. Conclusions: Manipulations that decrease visual crowding also decrease the stimulus contrast necessary to identify flanked letters. This finding supports our hypothesis that flanking letters decrease the effective contrast of the target letter by increasing target noise.