Change detection and repetition detection reflect functionally distinct forms of visual working memory

To examine the roles of change detection and repetition detection in visual working memory, we analyzed three working memory tests expected to rely differentially on these processes. Subjects studied an array of colored squares and then completed three tests. In the complex-probe test, subjects indicated whether a test array matched the study array or if an item's color changed. In the single-probe test, they judged whether a single item's color matched the study color, and in the item recognition test, they identified whether a centrally presented color was studied. We collected same/different confidence responses and analyzed receiver operating characteristics (ROCs) to evaluate memory strength distributions for changed and repeated trials, using a mixture signal detection model to estimate each process. As expected, the complex-probe test showed more high-confidence memory for changed trials, while the item recognition test showed more high-confidence memory for repetitions. The single-probe test showed similar or lower-confidence memory for both trials. Moreover, model estimates indicated that the probability of recollecting a change was higher in the complex-probe than in the item recognition tests, and the probability of recollecting a repetition was higher in the item recognition than the complex-probe tests. The single-probe test showed moderate recollection for both. These results show that change detection and repetition detection are functionally dissociable, with test-type affecting their contributions to working memory. These findings have implications for studying populations, such as aging, that may exhibit impairments in one or the other and raise the question of whether different neural systems underlie these processes.