Abstract
Much research has centered on understanding the nature of capacity limits in visual working memory. A largely overlooked possibility is that the nature of the stimulus itself might grossly affect estimates of capacity. Here, we investigated whether the precision or capacity of visual working memory for orientation depends critically on stimulus form, encouraged by earlier reports of better change detection performance for bars than for Gabor gratings (Alvarez & Cavanagh, 2008). Participants viewed a circular array of 2 or 6 items of either stimulus type, randomly placed at an eccentricity of 4°. After a short delay, they reported the orientation of a randomly probed item. A mixture-model analysis (Zhang & Luck, 2008) revealed that memory precision was very comparable for the two stimulus types, but that estimated capacity was about twice as large for bars (5.5 items) as for Gabors (2.6 items). Similar results were obtained when stimuli were presented for 200ms or 1000ms and followed by a patterned mask, ruling out the possibility that gross differences in capacity might be driven by faster encoding or more persistent iconic traces for the bars. Finally, we evaluated the hypothesis that working memory capacity may be superior for bars because their random arrangements can be more readily organized into perceptual groups. To disrupt perceptual grouping processes, we sequentially presented six bars or Gabors at various locations. With sequential presentation, memory capacity for bars dropped from 5.5 to 3.5 items, whereas the capacity for Gabors remained unchanged. These results suggest that higher-order patterns formed by simultaneously presented bars can be more efficiently maintained in working memory. Our findings demonstrate that working memory for orientation is highly dependent on stimulus form. Moreover, any formulations of visual working memory capacity should explicitly consider perceptual factors that allow for more efficient usage of its limited capacity.
Meeting abstract presented at VSS 2016