Abstract
Actively maintained information in working memory (WM) has been shown to automatically bias attention toward memory-matching information in the environment, even when detrimental to the task. This has been taken as evidence for WM and visual attention being closely intertwined. Here we test the alternative hypothesis that memory-driven attentional biases arise due to inefficient task switching, such that observers might initially look for the memorized content in the environment even when it is not relevant for the new task. To test this hypothesis, in two pre-registered experiments, participants memorized a color that had to be reproduced at the end of each trial. In between, they performed a visual search task including a distractor in a color related or unrelated to the memorized color. Critically, participants were explicitly cued to focus attention either on the search task or on the memory task (Experiment 1), or were presented with an interruption task (Experiment 2), in order to aid disengagement from the memory task and switching towards the search task. Across both experiments, we replicated the memory-driven attentional capture effect, with slower responses when a related distractor was present than when it was unrelated. Critically, none of the manipulations aimed to disengage attention from the memory task convincingly eliminated this bias. These findings indicate that memory-driven attentional capture cannot be fully explained by inefficient task switching, and further support the view that representations maintained in WM automatically bias attentional selection.