Lack of attention shift cost for features in visual working memory

Attention guides flexibly allocation of limited visual working memory (VWM) resources among representations. Within the maintenance phase of VWM, previous researchers presented retro-cues orienting to a probed feature across all multifeature stimuli and found a stable feature-based retro-cue benefit: VWM performance for cued feature was better than no/neutral-cue baseline. Two accounts might explain the mechanism: the reconstruction account suggests for separation of integrated representations, weighting on focused feature representations and removing unfocused ones; nevertheless, the priority account supports reservation and rehearsal of integrated representations with a ready state for follow-up comparison on focused feature. One distinct point between these accounts is whether the unfocused feature is impaired in VWM storage. In the current study, we used coloured, oriented bars as memoranda and manipulated the number of retro-cues to investigate these accounts. Double-cue trials displayed two sequential retro-cues orienting to different features. The single cue and the second cue in the double-cue condition are valid. Experiment 1 adopted the change detection task and three cue type conditions (no-cue, single-cue, double-cue); and in the subsequent two experiments, we elevated task difficulties and decreased expectations on cue types respectively. Our results across three experiments showed a stable pattern: VWM for the refocused feature in double-cue trials was better than no/neutral-cue baseline but was not different from those for the focused feature in single-cue trials. These findings indicate the lack of attention shift cost for features in VWM, which demonstrates that the unfocused feature in VWM is not impaired, supporting for the priority account.