Abstract
Time and space together make visual inputs distinct. Recent work shows that temporal contexts are encoded into visual working memory in a highly automatic manner, and other visual features are readily bound to their temporal positions. Less is known about whether multiple visual items can be grouped according to their temporal co-occurrence. To test this idea of temporal grouping, we assessed whether a two-item spatial retrocue was more effective when the cued items had previously appeared at the same time. Participants completed a color delayed estimation task. In each trial, six colored circles were presented sequentially but two at a time. In two-thirds of the trials, a retrocue then appeared, cueing at two locations. In half of these cued trials, the cued locations were congruent with the temporal contexts (i.e., they were locations of items that appeared at the same time); Whereas in the other half they were incongruent. The retrocues were 100% valid, guaranteeing that one of the two cued items would be tested. Arrow cues were used in Experiment 1 and peripheral cues were used in Experiment 2. In the remaining one-thirds of the trials, no retrocues were presented and all items were equally likely to be tested. Mean absolute errors were analyzed with a cue type x temporal position ANOVA. In both experiments, we found a main effect of cue type, and follow-up analyses revealed that congruent cues resulted in significantly better performance than incongruent cues and no cues. The cue type effect did not interact with the temporal position of the tested item. The data suggests that multiple items can be maintained in visual working memory as groups based on their common temporal context. These temporal groups can be effectively selected and prioritized, improving memory performance for multiple items.