Abstract
INTRODUCTION: Are global forms easier to discriminate when they are physically large?Two experiments examined whether display size/distance affects contour integration, the process that allows discrete visual elements to be perceived as a whole contour. METHODS: Displays ranged from 20 cm width/58 cm distance to 130 cm width/370 cm distance and subtended identical visual angles within experiments. All displays used the same DLP projector and custom screen. Participants discriminated (2AFC) the planar orientations of lines made from Gabor patches with either collinear (“snakes”) or parallel (“ladders”) orientations that were embedded at varying locations within a grid of randomly oriented Gabors, and which were masked after 150ms. Experiment 1 used two tasks that alternated across blocks of trials: global contour orientation judgment and local contour element orientation judgment (which yields opposite responses for snakes versus ladders). Experiment 2 used only the global task. Displays appeared at different distances during different blocks of trials. Snake and ladder stimuli were randomly interleaved within blocks of trials. RESULTS: Both experiments showed significant interactions between display size and stimulus type (snake/ladder). Increasing display size/distance significantly improved contour discrimination for snakes but not ladders, although there was a small trend for improved ladder discrimination at the largest size in experiment 2. Practice effects were accounted for by including a block-repetition factor in ANOVAs, which showed that display size effects were not byproducts of counterbalancing. There were no effects of task (local/global). CONCLUSION: Physical size and/or distance of retinotopically equivalent displays enhanced contour integration (snakes) but not texture segmentation (ladders). Ocular accommodation remains a possible cause of these effects. These results add to growing evidence that mid-level visual processes can be remarkably malleable in the contexts of different task demands.