Remembering stimuli in different depth planes increases visual working memory precision and reduces swap errors.

Abstract

If visual working memory performance depends on competition between similarly tuned neurons in visual cortex, two items with very dissimilar features should be easier to remember than two very similar items. Thus, remembering two items with different retinal disparities, as opposed to two items with the same disparity, should reduce competition between the items because each recruits a different set of binocularly-tuned neurons. Also, introducing depth information to mnemonic displays should reduce uncertainty about the location of a remembered item in depth and reduce confusion, or 'swap' errors. To test this general prediction, we conducted a visual working memory study using a stereo display. Subjects remembered the colors of two briefly presented (150ms) stimuli that were on the same or different depth planes and reported the color of a cued item after a 750ms delay by clicking a color wheel. A mixture model analysis revealed that different depth planes had a higher precision than those on the same depth planes (p = 0.025) than stimuli presented on the same depth plane. We also found significantly fewer swap errors when the remembered items were on different depth planes (p=0.0147), indicating less confusion between target and distractor occurred for stimuli that were more separable in depth. This supports the notion that working memory capacity is at least partly limited by competition between similarly tuned neurons. Furthermore, previously-observed limits in working memory performance can be mitigated by separating stimuli in the third dimension

Meeting abstract presented at VSS 2017