The Role of Predictable and Unpredictable Reward in the Control of Attention

Abstract

Stimuli previously associated with reward involuntarily capture attention (Anderson, Laurent, & Yantis, 2011, PNAS; Anderson & Yantis, in press, AP&P). One possibility is that this value-driven increase in attentional priority reflects the learned association between prior targets and the receipt of reward. Another possibility is that motivating visual search with reward enhances perceptual learning for former targets. To adjudicate between these two accounts, we measured the degree to which previous-target-colored stimuli capture attention following differing target̵0;reward contingencies. Each of three experiments began with a training phase in which participants received monetary rewards for correctly reporting a color-defined target in visual search. In Experiment 1, the target color remained constant across trials during training, although the magnitude of reward received for a correct response varied unpredictably. In Experiment 2, target color varied unpredictably between red and green, but the reward for a correct response remained constant. In Experiment 3, target color again varied unpredictably, but one color was associated with a greater reward than the other when correctly reported. Thus, participants received rewards for locating color-defined targets in all experiments, but only in Experiment 3 did the target color accurately predict the amount of reward. Following training, participants in all experiments completed a test phase in which stimuli carrying the color of a previously rewarded target occasionally appeared as a task-irrelevant distractor. Former-target-colored stimuli captured attention only in Experiment 3, when target color predicted reward. This finding argues against a motivated perceptual learning account of value-driven attentional capture: merely rewarding the performance of visual search was not associated with a persistent involuntary orienting of attention toward the previous target feature. Instead, our data suggest that value-driven attentional capture is mediated by mechanisms of reward learning, such that attention is involuntarily drawn to stimuli that have been learned to predict available reward. Meeting abstract presented at VSS 2013