No evidence for attentional prioritization in visual working memory through statistical learning

It is well-established that learning grounded in statistical regularities in the world can implicitly bias selective attention when oriented externally to the contents of perception. However, it remains an open question whether statistical learning can similarly bias internally oriented selective attention within working memory. To answer this question, across multiple experiments, we tested participants’ performance on established visual working memory tasks in which the relevance of two items for an upcoming behavioral task was implicitly predicted by the combination of the two items held in working memory (Exp 1 and 2a-b), an implicit retro-cue (Exp 3), or implicit temporal cueing (Exp 4). Across experiments, we found no robust evidence that statistical learning can bias internal attentional prioritization of contents held in working memory. We interpret these findings from the perspective that internal and external attention involve qualitatively different information selection mechanisms, operating to solve different goals.