Abstract
Attention can be controlled by visual working memory (VWM) and by long-term memory (LTM). Extant literature on learned distractor rejection indicates long-term, proactive suppression of feature values frequently associated with distractors. However, it is debated whether this type of proactive, negative template can be represented in VWM. More generally, the nature of the interaction between VWM and LTM guidance is poorly understood. Here, we examined how learned distractor rejection and trial-by-trial negative cuing interact when applied to the same object feature. Participants searched through arrays of colored disks as their gaze position was monitored. The experiment began with a learning session, in which one of four possible colors never contained the target attribute. This successfully induced learned distractor rejection, with the probability of fixating the learned distractor color falling below chance. In a subsequent experimental session, a trial-by-trial negative cue manipulation was introduced (negative color cue or neutral cue). The negative cue either specified one of the three colors that could contain a target or, critically, the learned distractor color. Given previous evidence, we predicted that a VWM negative cue would produce a capture effect, with a relative bias to attend to the matching value rather than to avoid it (white bear effect). In this view, if VWM and LTM guidance interact dynamically, they should produce opposing effects when applied to the learned distractor color, reducing, eliminating, or reversing the learned distractor effect. This was observed. A learned distractor rejection effect persisted on neutral cue trials. However, when the learned distractor color was also the trial-by-trial negative cue color, the learned distractor effect was eliminated. The results indicate that guidance from VWM and LTM combine to determine attentional priority. Moreover, they confirm that maintenance of a negatively cued value in VWM produces an ironic bias to attend toward that value.