Abstract
Studies have shown that visual working memory interacts with attention, but there are many varieties of attention, and how these varieties of attention modulate the contents of visual working memory is not well understood. Here, we investigated how a simple foveal discrimination task affects the contents of visual working memory using a dual-task paradigm. In the working memory portion of the paradigm, participants attempted to remember the orientation of a single object and then report its orientation after a short retention interval. On a subset of trials, we presented a letter during the retention interval. In the discrimination condition, participants were asked to report which of two letters (e.g., C or D) was presented by means of an immediate button-press response. In the no-discrimination condition, the letter was task-irrelevant and required no response. In the no-letter condition, no letter was presented. The memory data were fit by a mixture model for each target orientation to estimate response dispersion (i.e., precision) and central tendency (i.e., bias) for each orientation value. In all three conditions, cardinal orientations produced less precise response distributions than oblique orientations. In addition, responses for oblique orientations were biased away from the nearest cardinal orientation. These effects were largest when subjects performed a discrimination during the retention interval (the discrimination condition), smaller when they expected to perform a discrimination but no discrimination target was presented (the no-letter condition), and smallest when a stimulus was presented during the retention interval but could be ignored (the no-discrimination condition). In a follow-up experiment, we found the same pattern of results with a simple auditory discrimination (e.g. high/low pitch tones). These results indicate that the processes involved in making a simple visual or auditory discrimination cause visual working memory representations to become less precise and more influenced by category boundaries.
Meeting abstract presented at VSS 2015