Abstract
Visual crowding describes impaired object recognition in peripheral vision due to the presence of other nearby objects. Previous studies of crowding have largely utilized letters, numbers, or Gabors as stimuli on blank backgrounds (Strasburger, Harvey & Rentschler, 1991; Greenwood, Bex, & Dakin, 2010). Additionally, crowding has more recently been demonstrated using objects as stimuli (Wallace & Tjan, 2011). The present study investigated crowding using objects in real world scenes. Virtual living-room scenes were created. Target objects were placed at 12o retinal eccentricity from fixation. In the crowded condition, four distractor objects were placed at a distance less than the critical spacing from the target, where crowding would be induced as estimated by Bouma's constant. In the uncrowded condition, the same four distractor objects were placed farther than the critical spacing from the target, where crowding would not be expected to occur. In Experiment 1, the location of the target object was exogenously cued and the image was flashed for 80 ms followed by a neutral gray screen. Participants responded to the target object's category with a 12-AFC response (including all objects shown in all versions of the scene). Experiment 2 used eyetracking to ensure participants were centrally fixated at the beginning of each trial. Procedures were similar to the first experiment, but with the image remaining on the screen for one eye fixation. In both experiments, object recognition performance was significantly reduced for crowded targets as compared to uncrowded targets. Participants who experienced a stronger crowding effect were more likely to incorrectly select a distractor object that was present in the scene versus a random object not shown, consistent with the idea that crowding involves confusion of the target and distractors. This study is unique in that it rigorously shows crowding of multiple objects in multiple real world scenes.
Meeting abstract presented at VSS 2014