A test of interference versus decay in working memory: Varying distraction within lists in a complex span task

We tested two competing explanations of the effect of processing on working memory. According to decay models, memory representations decay during processing and can be rehearsed or refreshed in the free time between processing steps. Alternatively, one interference-based model assumes that processing involves encoding of distractor representations in working memory, and free time is used to remove distractors. In several experiments the demand from distractor processing was varied within lists, such that one burst of processing following an item on the list was either particularly demanding or particularly undemanding. The exceptional distractor burst had its greatest effect on the list item that immediately preceded it (a local effect), and it affected items that had not yet been presented as well as preceding items. Both findings are predicted by a computational interference model of working memory, and together are problematic for the viewpoint that refreshing offsets decay.