The ability to exert cognitive control allows us to achieve goals in the face of distraction and competing actions. However, control is costly-people generally aim to minimize its demands. Because control takes many forms, it is important to understand whether such costs apply universally. Specifically, reactive control, which is recruited in response to stimulus or contextual features, is theorized to be deployed automatically, and not depend on attentional resources. Here, we investigated whether people avoided implementing reactive control in three experiments. In all, participants performed a Stroop task in which certain items were mostly incongruent (MI), that is, associated with a high likelihood of conflict (triggering a focused control setting). Other items were mostly congruent, that is, associated with a low likelihood of conflict (triggering a relaxed control setting). Experiment 1 demonstrated that these control settings transfer to a subsequent unbiased transfer phase. In Experiments 2-3, we used a demand selection task to investigate whether people would avoid choice options that yielded items that were previously MI. In all, participants continued to retrieve focused control settings for previously MI items, but they did not avoid them in the demand selection task. Critically, we only found demand avoidance when there was an objective difference in demand between options. These findings are consistent with the idea that implementing reactive control does not register as costly.
Public Significance StatementWhile humans use attentional control to achieve their goals and to avoid distractions, it has been theorized that it is not always deployed because of its mental costs. However, not all forms of control are equal: It has been suggested that adjusting attention in response to a stimulus may be relatively costfree. We provide a new test of this idea by integrating stimuli associated with different amounts of reactive control into a demand selection paradigm. We found that participants reactively focused their attention on items previously associated with higher demand and relaxed attention on items previously associated with lower demand. However, despite these differences in implemented control, participants did not avoid choice options associated with the previously high-demand options. This result indicates that reactively focusing attention is not costly. In addition to theoretical implications about the nature of control, these results indicate that setting up demands to rely on reactive control might substantially diminish subjective motivational and implementational constraints.