Abstract
Aim: Given visual working memory's limited capacity, there must be rules which govern when information is thrown out of memory. This series of experiments examines the roles of decay and interference in that process. Method: Participants viewed two arrays of 8 colors separated by a lag, and compared the color at a location indicated by a central arrow cue presented 400 ms before the test array. Decay was measured by varying the lag between the first display and the cue from 400 – 1600 ms. Interference was examined by comparing three conditions: (1) the cue was presented alone, (2) the central cue was accompanied by a peripheral cue consisting of a bicolor square at the cued location, (3) the cue was accompanied by a circular array of bicolor squares with a blank space at the cued location. Results: Accuracy declined only slightly over the 1600 ms lag. However, accuracy was impaired by about 15% in conditions 2 and 3 (interference) relative to condition 1 (no interference). Since this interference occurs up to 1600 ms after the initial display, it must target memory retrieval rather than memory encoding processes. Moreover, since the interference is present even when no stimulus is shown at the cued location (condition 3), it must not be spatially specific. Conclusion: Interference with visual working memory is distinguished from more familiar perceptual masking processes by its action over long time lags and its apparent insensitivity to spatial location. Given new visual input, VWM seems obligated to update its content.
Research supported by NSF 0345525