Abstract
Visual working memory capacity is limited (Luck, 1997) so prioritizing the most behaviorally relevant items is important. Retrospective cues (retro-cues) presented during a retention interval when sensory information is no longer available have been shown to improve item recognition (Griffin 2003), with larger effects for high validity cues (Berryhill 2015, Gunseli, 2015, Gunseli, 2018). Importantly, however, studies have only used 1–2 levels of cue-validity so it is not clear if internal shifts of attention in response to retro-cues can be directed in a graded manner or if a single item is either in or out of the privileged ‘Focus of Attention” (Cowan, 2011). Recent work (Lockhart, 2018) utilizing multiple 100% valid retro-cues found that attentional benefits did not exist when >1 item was cued, supporting the notion that retrospective attention is ‘all-or-none’. To test if retrospective attention can be directed in a graded manner, we ran two delayed report experiments with variable validity retro-cues. In E1, using a between-subjects design, each subject (n=93) was assigned a cue validity (50–90%) and we found that mnemonic precision was higher on cued trials but did not increase with cue validity while precision on un-cued trials significantly decreased with validity. In E2, each subject (n=10) was presented with a colored arrow corresponding to 3 possible cue validities (50, 65, 80%). While recall precision was higher for the cued stimulus, there was no main effect or interaction between cue probability and whether the probed target was valid or invalidly cued. Taken together, these results suggest that while individuals may be able to take cues validity into account to set one level of relative priority, there is little evidence that a given individual can flexibly direct retrospective attention to memoranda in a graded manner.