Abstract
Introduction. Crowding severely limits the identification of a flanked target in peripheral vision. Previous research had shown weaker crowding with dissimilar flankers, i.e., the similarity effects on crowding. Here we manipulated configural legitimacy (Experiment 1) and contour integrity (Experiment 2) of flankers to investigate the effects of similarity on crowding in Chinese character recognition. Method. Ten normally-sighted observers participated in Experiment 1 and five in Experiment 2. Target stimulus in each trial was always a real, frequently-used and randomly-drawn Chinese character from a set of 176 (Experiment 1) and 190 (Experiment 2) characters. The target (size=1.5°) was presented for 100 ms at 5° eccentricity in the right visual field. Two horizontal flankers at 2° center-to-center distance were also presented in the flanked conditions. Contrast thresholds in a target identification task (Experiment 1: 25-alternative-forced-choice [25-AFC]; Experiment 2: 20-AFC) were measured using QUEST in a target-only and different flanked conditions. Flankers in Experiment 1 were (a) real, (b) pseudo- (non-existent but configurally legitimate), (c) non- (configurally illegitimate), or (d) scrambled characters. Flankers in Experiment 2 were (a) phase-intact (i.e. real) or (b) phase-scrambled characters. Results. No statistically significant variations were found among mean threshold elevations (TEs) in the four flanked conditions of Experiment 1 (real: 3.59±0.46 [SE]; pseudo: 3.08±0.45; non: 3.47±0.44; scrambled: 2.68±0.36). In Experiment 2, TE was significantly higher with phase-intact flankers (4.97±0.65) than with phase-scrambled flankers (2.34±0.49). Conclusions. The null effect of configural legitimacy suggests that the well-documented similarity effects may be restricted to low-level visual features. We found weaker crowding with phase-scrambled flankers, inconsistent with findings from previous research (e.g., Shin, Wallace & Tjan, 2010). It is possible that visible contours in flankers determine crowding strength only when the target is of high visual complexity, as in the case of Chinese characters.
Meeting abstract presented at VSS 2015