Abstract
Many previous studies report that stimuli associated with reward capture attention (value-driven attentional capture; VDAC), regardless of task-relevant or task-irrelevant features. However, necessary conditions for the formation of the feature-reward association in VDAC remain unknown. Recent studies claim that VDAC is based on prediction-based associative reward-learning, but they did not directly examine the relationship between reward prediction and feature-reward association, because the temporal relationship among feature, response and reward was fixed. In the current study, we took advantage of the flexibility of the task-irrelevant reward learning and manipulated the temporal relationship among feature, response, and reward in reward learning to reveal necessary conditions for VDAC. For this purpose, we presented features associated with reward in a variety of locations in a flanker task to create color-reward association (training phase) and then examined VDAC in a subsequent visual search (test phase). First, features were presented in a task display, and we obtained a significant VDAC, which was the replication of the previous study. In contrast, VDAC disappeared when features and reward were simultaneously presented in the feedback display, suggesting that the direct association between feature and reward is not sufficient for VDAC. When reward was presented before response selection in the task display, we could not observe VDAC, suggesting that temporal precedence of response to reward is necessary for VDAC. We next examined temporal relationship between feature and response. When features were presented before the task display requiring response (fixation display), VDAC disappeared, suggesting that temporal separation of feature and response selection does not induce VDAC. Features presented after the response selection showed the same result, indicating that synchronization between features and response is necessary for VDAC. Taken together, features presented at the time of reward prediction, namely at response selection preceding reward, induce VDAC.
Meeting abstract presented at VSS 2016