Abstract
Can human observers adapt their saccadic strategies to a new retinal sensitivity map? Performance in a visual search task depends critically on both the observer's pattern of retinal sensitivity and the observer's saccadic strategy. Certain retinal diseases, such as macular degeneration, can dramatically alter an observer's pattern of retinal sensitivity, leading to impaired performance. It is unclear, however, whether these changes in performance result from inefficient saccadic strategies or simply reflect the loss of visual information from the damaged retinas. Here, we report the results of a study using gaze-contingent displays and an ideal observer model of visual search (Najemnik & Geisler, 2005) to determine whether human observers can appropriately modify their saccadic strategies following simulated changes to their retinal sensitivity maps. We used two experimental conditions: a ‘central scotoma’ condition that simulated a loss of vision in the central 4 degrees of the visual field, and a ‘shifted fovea’ condition that shifted the observer's sensitivity pattern upward by 2.5 degrees. In both experimental conditions, we compared the performance of our human observers with that of three types of simulated observers: 1) an ideal searcher that plans optimal fixations according to the transformed sensitivity map, 2) a naïve searcher that plans optimal fixations under the original (unaltered) sensitivity map, and 3) a random searcher that chose fixations randomly. All three simulated observers correctly updated the posterior probabilities over target locations following each saccade. In the ‘central scotoma’ condition, humans outperformed the naïve and random searchers and demonstrated near optimal performance. In the ‘shifted fovea’ condition, however, humans performed suboptimally, with performance near that of the naïve searcher. Our results suggest that human observers can rapidly adapt their saccadic strategies for visual search following simulated macular damage, but that this learning is limited and depends on the spatial pattern of vision loss.