Fractionating Selection History: Dissociable Components of Experience-Driven Attention

Abstract

The classical view of attentional control suggests that attentional selection is guided by physical salience and task goals. Selection history, which refers to attentional selection of stimuli an observer has previously interacted with, has challenged the dichotomised view and advocates attentional selection independent of physical salience and task goals. Research to date has identified different components of selection history, including (1) reward history (2) punishment history and (3) history as a repeated target, but how these components relate and influence attentional selection remains unclear. Across five experiments, we demonstrate that although these components of selection history share a similar behavioural profile, they rely on distinct learning and neural mechanisms. Specifically, reward and punishment history are shaped by a common associative learning mechanism and recruit the dopaminergic midbrain structures, supporting the motivational salience account which suggests that attention prioritises relevance-for-survival regardless of a particular emotional valence. In contrast, history as a repeated target develops via an instrumental learning mechanism and the corresponding attentional priority is afforded by enhanced representations in the visual areas. Such dissociation between reward and punishment history on one hand and history as a repeated target on the other provides compelling evidence that the three components of selection history comprise unique sources of attentional bias that independently influences the attentional system.