Task information overrides attentional capture by reward-associated stimuli

摘要

Reward is a powerful motivator that has robust effects on behavior. Recent work has shown that rewarded locations and stimulus features capture attention, even when the reward information is no longer task-relevant (Kiss, Driver, Eimer, 2009; Andersen, Laurent, and Yantis, 2011; Theeuwes and Belopolsky, 2012). These studies suggest that reward-associated information is automatically assigned high attentional priority. However, few of these tasks used experimental paradigms in which it would be beneficial to suppress attention to the reward-associated stimulus. We used a variant of the Posner cueing paradigm in which bilateral colored circles were presented on each trial. Subjects first learned to associate each color with an amount of monetary reward. Critically, in Exp. 1, we introduced competition for attentional priority by gradually increasing the probability of the target in one location (e.g., proportion of target on the right vs. left: 0.5:0.5, 0.6:0.4, 0.7:0.3, 0.8:0.2, 0:1). The location of the high reward color was always random. The results demonstrated an interaction between probability and reward value. Initially, performance was better (i.e., shorter RTs and higher accuracy) for targets in the high-reward location, but this pattern reversed as the strength of the spatial probability increased. The identification of a point of subjective equality (PSE) between the two biases suggested that the allocation of attention to rewarded features was counterbalanced by spatial expectancies. In Exp. 2, we tested whether this effect extends to task-irrelevant bottom-up saliency defined by luminance. In contrast to Exp. 1, subjects had shorter RTs and higher accuracy to targets in the high reward color regardless of the perceptual saliency of the exogenous cue. Together these results indicate that while reward is a powerful bias on spatial attention, there is a push-pull relationship between reward and other sources of task-relevant information. In contrast, reward-associated information appears to be prioritized over task-irrelevant saliency.