Strategic control finely tunes working memory performance without the benefit of selection history.

People can predict the probability of an event occurring and allocate mental resources accordingly. For example, the likelihood of recalling an item from working memory varies as a function of the importance of recalling that item. Strategic control is thought to finely tune working memory performance using probabilistic information, but a possible contribution of automatic processes based on selection history has not been investigated. To determine if graded working memory performance is due to strategic control or selection history, probabilistic information was either blocked (Experiment 1) or changed from trial to trial (Experiment 2). A spatial cue indicated the likelihood (i.e., 50%, 60%, 70%, 80%) that one of two colors would have to be recalled. Attention was directed externally by a pre-cue presented before the memory array or internally by a retro-cue after the memory array. As in previous studies, cue effects increased linearly with greater predictability of the cue in the blocked conditions of Experiment 1. In Experiment 2, this finding was replicated despite a lower contribution from selection history when probability changed from trial to trial. In both experiments, cue effects were similar for the pre-cue and retro-cue conditions. Taken together, these results demonstrate that probability effects on WM performance are due to finely tuned strategic control without a strong contribution from selection history.