Reward guides attention to object categories in real-world scenes

Abstract

Reward is thought to motivate animal-approach behavior in part by automatically facilitating the perceptual processing of reward-associated visual stimuli. Studies have demonstrated this effect for low-level visual features such as color and orientation. However, outside of the laboratory, it is rare that low-level features uniquely characterize objects relevant for behavior. Here, we test whether reward can prime representations at the level of object category. Participants detected category exemplars (cars, trees, people) in briefly presented photographs of real-world scenes. On a subset of trials, successful target detection was rewarded and the effect of this reward was measured on the subsequent trial. Results show that rewarded selection of a category exemplar caused other members of this category to become visually salient, disrupting search when subsequently presented as distractors. It is important to note that this occurred even when there was little opportunity for the repetition of visual features between examples, with the rewarded selection of a human body increasing the salience of a subsequently presented face. Thus, selection of a category example appears to activate representations of prototypical category characteristics even when these are not present in the stimulus. In this way, reward can guide attention to categories of stimuli even when individual examples share no visual characteristics.