Abstract
According to theories of visual working memory that are based on synchronous neural firing (e.g., Luck & Vogel, 1998; Raffone & Wolters, 2001), capacity limits in visual working memory arise because of inhibitory interactions that serve to isolate the representations of the objects being stored. These inhibitory interactions do not depend on the nature of the information being stored in each representation; consequently, performance in visual working memory tasks should not depend on the similarity between the representations. To test this hypothesis, we manipulated the similarity of the to-be-remembered items in a change-detection task. On each trial, subjects viewed a 100-ms sample array consisting of 3 or 4 colored squares, followed by a retention interval of 900 ms and then a test array. The test array was either identical to the sample array or differed in the color of one item, and subjects indicated whether a change was present. The colors of the items were selected from a set of reddish colors, a set of greenish colors, and a set of bluish colors. On some trials, all the colors in the sample and test array were drawn from a single set and were therefore highly similar to each other. On other trials, the colors were drawn from all three sets and were therefore highly dissimilar from each other. We found that change-detection accuracy was virtually identical for these two conditions. Thus, similar representations can be concurrently formed and maintained in visual working memory just as accurately as dissimilar representations.
This research was made possible by grants R01 MH63001 and R01 MH65034 from the National Institute of Mental Health.