Abstract
In everyday settings, objects frequently partially occlude each other. Thus, amodal completion is critical for object recognition in naturalistic contexts. However, the effects of aging on amodal completion remain unexplored. Therefore, we measured the extent and time course of amodal completion for younger and older adults using a shape discrimination task in which subjects judged the orientation of a moving rectangle as horizontal or vertical (Murray, Sekuler & Bennett, 2001). Observers completed three conditions: 1) complete, with the entire outline of the rectangle visible; 2) fragmented, with corners of the rectangle deleted; and 3) occluded, resembling the fragmented condition, except with corners occluded by opaque squares. In younger observers, this task is easiest with complete stimuli and most difficult with fragmented. Performance with occluded stimuli resembles that of complete or fragmented, depending on the extent to which the observer amodally completes the rectangle behind the occluders. Specifically, Murray et al. found that the younger adults performed the complete and occluded tasks similarly at stimulus durations longer than ~60 ms. We measured thresholds for each condition at stimulus durations ranging from 15-210 ms, determining the aspect ratio that led to 70% correct. Based on initial results from 10 participants, shape discrimination is impaired for older (M=69 years) vs younger (M=23 years) adults. At the shortest durations, older observers require larger aspect ratios than younger subjects to discriminate even complete shapes accurately, and fail to reach 100% accuracy even with large aspect ratios. Performance is similar across age groups for complete shapes at longer durations. Critically, our data suggest that older adults require more time than younger adults to amodally complete occluded objects (>100 ms vs ~60 ms). We currently are investigating the result further with a larger group of subjects, to determine the time for completion in older observers more precisely.
Meeting abstract presented at VSS 2017