Abstract
Across three experiments we investigated the interplay of visual working memory (VWM) and ensemble perception. In Experiment 1, participants estimated the mean orientation of 12 simultaneously presented bars while concurrently maintaining the shape and color of an irregularly shaped item in VWM. Critically, the color of the VWM-item could match a subset of ensemble elements oriented clockwise from the overall mean, counterclockwise, or neither – constituting three unique conditions. As predicted, when the color of the VWM-item matched a subset of elements, global mean estimation errors were systematically biased towards the mean of that subset, while such errors were centered around the global mean when the VWM-item matched neither of the subsets. In Experiment 2, we tested whether this effect was due to an explicit attentional strategy to improve VWM performance. We used the same procedure as above but changed the duration of the ensemble displays from 250 ms to 150 ms or 500 ms, reasoning that strategic attentional allocation would be less viable at short durations. Again, estimation errors were biased by elements matching the VWM-item, but the magnitude of this bias was not moderated by display duration, suggesting that the effect is unlikely to be accounted for by strategic processes. Lastly, in Experiment 3 we examined whether this effect requires active maintenance of the biasing feature in VWM. We made the color of the VWM-item irrelevant to the task, allowing the shape, but not color, of the item to change from study to test. Here, we observed no difference in estimation errors across the three conditions, which suggests that the biasing feature needs to be actively maintained in VWM to influence ensemble perception. Overall, we provide evidence that individual elements are overvalued in ensemble perception when they share a common feature with information actively represented in VWM.