Abstract
In daily life, individuals encounter many stimuli that predict rewarding outcomes. Such stimuli can have a powerful impact on behavior. For example, value-driven attentional capture occurs when a previously reward-predictive stimulus pulls attention away from the current task (Anderson et al., 2012). Although much is known about the acquisition of stimulus-reward associations, the extinction of such associations is less clear. Here we examined how feedback facilitates the extinction of learned stimulus-reward associations. Participants first learned to associate specific color-shape combinations with reward to earn as many points as possible. Our results indicated that participants consistently chose the high reward color for each shape during the initial training phase. Subsequent to the training phase, participants completed a testing phase where no rewards were delivered and participants demonstrated strong indices of value-driven attentional capture (i.e., an N2pc ERP component to the previously rewarding stimuli and a distraction effect in RT). Next, participants performed a second training phase where the initially rewarding stimuli were remapped to new outcomes. This was done in three ways: (1) high reward to high reward, high reward to low reward and high reward to high punishment. After the remapping procedure, participants consistently attended the high reward color (i.e., N2pc) and avoided the high punishment color (i.e., eliciting the Pd ERP component indicating distractor suppression). Our results show that reward extinction is fastest when the previously rewarding stimulus is paired with punishment. However, merely removing reward from a pairing was sufficient to bring an individual's desire to choose that previously rewarding item down to baseline indicating that reward associations used in vision laboratories can be rapidly extinguished when reward contingencies return to normal in the real world.
Meeting abstract presented at VSS 2017