Explicit knowledge modulates Pavlovian attentional biases

There is plenty of evidence showing that the history of stimulus-reward pairings can cause such stimuli to capture our attention more easily, even when attending to them actually prevents the delivery of a reward. Once established, these Pavlovian attentional biases are believed to operate automatically and persist even when they no longer predict a reward. This reflects an inflexible form of attentional sign-tracking. In this study, we investigated how reversal learning of Pavlovian attentional biases is affected by explicit knowledge. Across three experiments, we demonstrate that verbal instructions alone can modulate previously learned attentional biases (Experiments 1 and 2a) and that, when not informed of the reversal, participants still exhibit strong associations between Pavlovian attentional biases and individual differences in awareness of the new, reversed contingencies (Experiment 2b). Computational modelling of a previously published dataset reveals that instructions may prompt a rapid model-based revaluation by eliciting the automatic updating of the stimuli's value representation. However, such updating can be counterproductive, as attending to the previously high-value distractor reduces the likelihood of receiving a reward when contingencies are re-evaluated. These findings emphazise the importance of explicit knowledge in the updating process.