Abstract
A letter or object target when flanked by additional letters or objects becomes difficult to recognize in peripheral vision. Recent studies show that perceptual learning can reduce this crowding effect. In crowding, the target errors may be contributed to identity errors as well as position errors in that the central target is frequently reported to a flanker position (Zhang et al., 2012). In addition, flankers are often reported to the target position (Strasburger et al., 2005), although it is unclear whether these flanker reporting errors are the cause or result of crowding. In this study we investigated how the target identity and position errors, and the flanker reporting errors, are associated with crowding perceptual learning. Observers were trained to report the central target presented at 8o retinal eccentricity (partial report). Before and after training they also reported all three letters (whole report). The results show: (1) Five sessions of training reduced target recognition threshold by 33.6% in partial report tasks, suggesting reduced crowding via perceptual learning. (2) The target identity errors revealed by whole report tasks were reduced significantly except with smallest letters where crowding was the strongest. (3) However, the target position errors, defined by the ratios between actual and predicted whole report rates, were unchanged after normalized by target identity rate changes. (4) In addition, in partial report the flanker reporting errors when normalized by target reporting rates were unchanged after training. Perceptual learning reduces visual crowding by lowering target identity errors but not target position errors. The finding that flanker report errors were unchanged when training reduced crowding suggests that flanker reporting errors are more likely a result, not a cause, of crowding, in that the observers may be forced to report a more visible flanker when they fail to recognize the central target.
Meeting abstract presented at VSS 2014