Reward-driven attentional capture: Explicit knowledge versus implicit reinforcement

Attending to visual stimuli associated with high probability of gaining rewards is highly adaptive. Research in the fields of vision sciences and addictive behaviours has shown that features associated with high reward value are preferentially attended to. It is presently unclear exactly how reward interacts to guide visual attention. Recent work suggests that reward-based learning operates independently of the classical distinction between top-down and bottom-up selection mechanisms in facilitating attentional capture. Anderson, Laurent and Yantis (2011b) found that previously rewarded features continue to capture attention even when non-salient and no longer relevant to task goals. Furthermore, stimuli associated with higher reward captured attention more robustly than those associated with lower reward. That these findings are not explained by the present top-down and bottom-up accounts has given rise to a new value-driven model of attentional capture, whereby rewards influence attentional capture separately from task goals and salience. Rewards have been postulated to operate through the midbrain dopaminergic systems, which are sensitive to the co-occurrence of features and reward. This suggests that in a value-driven attentional control mechanism, the learning of rewards would be implicit. The present research aimed to test this implicit learning hypothesis by manipulating reward and knowledge of reward using a visual search paradigm. Experiment 1 assessed the extent to which reward and reward-feedback are necessary to drive reward learning by manipulating knowledge of the reward structure and reward feedback. There was no differential reward-driven capture found, therefore the relative effects of reward and knowledge were unable to be analysed. A follow up study was designed to address potential limitations and assess whether explicit knowledge of reward structure would suffice to drive learning, or if feedback of rewards is necessary as per the implicit learning claim. Again, no differential effect of reward was found. Potential limitations, implications and directions for future research are discussed