Neural Mechanisms of Value-Driven Attentional Capture

Abstract

Goal-driven and stimulus-driven factors interact to determine the attentional priority of stimuli. A wealth of research demonstrates that goal-driven attention is subserved by a frontal-parietal network of brain regions, and stimulus-driven attention by a temporal-parietal network. Recently, we showed that stimuli previously associated with reward capture attention independently of their physical salience and goal-relevance (Anderson, Laurent, & Yantis, 2011, PNAS). This suggests a value-driven representation of attentional priority that may be computed outside of the goal-driven and stimulus-driven attention networks; however, the neural basis of value-driven attentional capture remains unexplored. In the present study, we measured brain activity using fMRI while participants carried out a pair of visual search tasks. They first completed a training phase in which each of two color stimuli differentially predicted a monetary reward; these same color stimuli later served as irrelevant distractors in an unrewarded test phase comprising visual search for a shape singleton target. Replicating our previous behavioral findings, we found that response time in the test phase was slowed by the presence of a previously reward-associated distractor. Neuroimaging data showed that valuable distractors elicited an increased stimulus-evoked response in contralateral regions of extrastriate visual cortex and caudate tail. The caudate tail, part of the striatum, is known to play an important role in reward processing; it has not been implicated in either goal-driven or stimulus-driven attention, but is well suited to mediate the feature-based value-driven control of attention. Our results suggest that representations of visual stimuli in the caudate tail are modified by reward learning; this in turn increases their effective salience in early visual areas of extrastriate cortex. This striatal-extrastriate locus of attentional control indicates a possible neural substrate for a key value-driven attention network. Meeting abstract presented at VSS 2013