Abstract
Visual working memory (VWM) is severely capacity limited. As such, the ability to use it flexibly is important in ensuring that it effectively supports behavior. Cueing the relative priorities of items in the environment before they are encoded, for instance, will result in the flexible distribution of VWM resources such that the most relevant information is also best remembered. Similarly, using a cue after encoding to retrospectively indicate which representation is most relevant will result in enhanced performance on tests of the cued representation relative to the non-cued representations, known as the retro-cue benefit. How flexible is this retro-cue benefit? Across three experiments, we manipulated the predictive validity assigned to a single retro-cue (E1), and to two consecutive (E2) or simultaneous (E3) retro-cues, and assessed whether cue validity affected the precision of color responses in a subsequent memory test. Despite observing a reliable retro-cue benefit (i.e., better performance on tests of the cued, relative to the non-cued, representations), we observed no evidence that the size of this benefit varied as a function of cue validity across trials (E1). Further, we observed no evidence that participants were able to prioritize multiple cued representations relative to each other within a given trial (E2 and E3). That is, the cue(s) prompted participants to effectively prioritize the cued representation(s), but this process is all-or-none, and resources cannot be re-distributed flexibly. As such, we conclude that the benefit that a retro-cue incurs to memory performance is inflexible, highlighting an important limitation to the re-focusing of internal attention that is not observed when assessing attention at encoding.