Visual working memory for multiple moving objects in occlusion
33.4112, Sunday, 17-May, 8:30 am - 12:30 pm, Pavilion
Melissa Kibbe1; 1Psychological & Brain Sciences, Boston University
Visual working memory (VWM) capacity is typically assessed using tasks in which participants must remember two-dimensional arrays of items that blink in and out of existence. These studies have found VWM limits of around 3-4 items. In the physical world, however, VWM may be required to maintain representations of multiple three-dimensional objects as they move in and out of occlusion. This may put different demands on VWM, requiring more attentional resources than stationary 2D items. In the current experiment, we explored participants’ VWM for objects that moved into occlusion. Twenty-four participants viewed 3-D rendered displays in which sets of differently oriented bars moved sequentially behind separate rectangular occluders. Two of the occluders then dissolved, revealing the objects behind them. On half of the trials, one of the objects changed orientation. Participants reported whether they detected a change. We manipulated both set size (2-6 objects) and which object in the hiding sequence was probed (first-hidden, second-hidden, etc.) Change detection performance was best at set size 2 (88% correct), but declined significantly thereafter (set size 3: 68%; set size 4: 68% set size 5: 59%; set size 6: 60%, p< 0.001). Mean k across set sizes was 1.38 objects, fewer than typical limits observed for stationary 2D items. For all set sizes, change detection performance was best for the last-hidden object in the set (mean=79% correct). Performance then dropped significantly for the second-to-last hidden object, but only at set sizes greater than 3 (all ps< 0.01), suggesting that, as the number of hidden objects substantially exceeded participants’ limits, VWM resources were preferentially allocated to the last object participants saw hidden. Our results suggest that tracking and representing multiple occluded objects may make sizable demands on attention and VWM, limiting the number of representations that may be concurrently maintained.