Institution: Duke University, Durham, North Carolina 27708; Columbia University, New York, New York 10027; NW 177th Ter, Edmond, Oklahoma 73012
How well a stimulus predicts reward is correlated with the amount of attention that stimulus receives. However, the relationship between a cue’s predictiveness and the amount of attention it draws is not yet fully understood. There are two competing theories which posit distinct effects of a cue’s predictiveness on attention. Attentional exploitation theory states that a stimulus’s predictiveness will be directly proportional to the amount of attention it garners. Attentional exploration theory posits an inverse relationship between these two factors. In this experiment, we aimed to explore this relationship between attention and reward learning. Participants completed a modified version of a multi-armed bandit task. Their eye movements were tracked and used to measure the amount of attention participants allocated to reward-predictive versus reward-nonpredictive stimuli. We additionally measured how attentional patterns changed over time. This study found evidence in support of both theories of attentional processing. Future research is thus needed to reconcile attentional exploitation and exploration theory.
Keywords: Attentional exploitation theory; Attentional exploration theory; Probabilistic reward learning; Multi-armed bandit task