Abstract
Recent studies have shown that reward contingencies affect attentional selection. Stimuli that were selected and previously rewarded continue to capture attention even if the reward contingencies are no longer in place. In the current series of experiments, we investigated whether task-irrelevant and non-salient stimuli that merely signal the magnitude of reward available on a particular trial affect attentional selection. Crucially, in these experiments, attentional selection of the reward-signaling stimuli was never necessary but instead detrimental for actual payout. Participants searched for a target (in different experiments for either a specific color or shape) presented among five distractors. The color of one of the distractors signaled the reward available for that trial (e.g. a red distractor would indicate a high reward, a green distractor a low reward). Even though this colored distractor signaling reward availability was never part of the task set, nor physically salient, it captured attention. Follow-up experiments suggested that stimuli signaling reward get prioritized for attentional selection through reward-learning. However, this learning did not occur for participants who were not informed about the stimulus-reward association or for participants who did not become aware of the association themselves. We conclude that task-irrelevant and non-salient stimuli signaling the availability of reward gain priority in attentional selection even if selecting them is not necessary but rather detrimental for reward payout. Importantly, our data suggest that reward-signaling stimuli only capture attention once they have initially been prioritized for attentional selection. Such initial prioritization can occur through knowledge or task-relevance of the reward-associated stimuli or by rendering them perceptually salient. After the initial prioritization, learned stimulus-reward associations continue to bias attentional selection toward the reward-signaling stimuli independently of top-down or bottom-up processes, whether their selection is beneficial for reward payout or not.
Meeting abstract presented at VSS 2016