Learned Reward Association Acts as "Template for Rejection" in Visual Search Task

Recent literatures have demonstrated that previously learned reward association modulates attentional allocation to current visual input. The manner in which attention is automatically guided by reward-associated item fits well with the benefit of stored "attentional template". However, whether the reward-associated item can be used to guide attention away as "template for rejection" remains unclear. To test this possibility, we first established reward-color association, and then tested the effect with a cued visual search task (Experiment 1). Search display consisted of a ring of twelve circles, each embedded with an oriented bar. For each trial, the circles from a half ring (left or right) were set to be one color and the other half in another color. Observers’ task was to search for uniquely oriented bar. They were instructed to exclude the circles in pre-cued color during search as target never appeared in these circles. The results showed significantly faster search RT for the pre-cued color associated with high reward than those with low reward (p<0.01) or non-reward (p<0.01), indicating a faster rejection of the high reward-associated color. To eliminate the contribution of spatial attention to the observed effect, we distorted the color symmetry of the search display (Experiment 2). The new search display consisted of eight Landolt-C in two colors with equal number. Similar pattern of results was obtained by instructing observers to exclude the pre-cued non-target colors. Moreover, a memory test at the end of each trial verified that such rejection did not impair the representation of the pre-cued color in working memory. Interestingly, the effect remained when we rendered the pre-cued color as search target and the reward-associated colors were never cued or explicitly rejected (Experiment 3). Our results suggest a flexible role of reward association in facilitating visual search performance by modulating attentional weights towards rewarded items. Meeting abstract presented at VSS 2013