Abstract
Learned reward associations can capture attention even when they are non-salient, and task-irrelevant (e.g. Anderson et al., 2011; Raymond & O'Brien, 2009). This effect is generalizable (Anderson et al., 2012), persistent over long periods of time (Anderson & Yantis, 2013), interacts with salience (Anderson et al., 2011), and appears to be spatial in nature (Anderson & Yantis, 2012; Theeuwes & Bopolsky, 2012). Here we investigated how task-irrelevant learned reward associations interact with task-relevance both in terms of task set and space. In two experiments, multiple targets (letters) were briefly presented with an equal number of distracters (numbers) within task-relevant space equidistant from fixation. Another distracter ("flanker") was also presented outside of task-relevant space (Exp. 1 letter only; Exp. 2 letter, number, or symbol). A pattern mask followed the display. Participants performed a 2-AFC (probed target vs. lure). Each item was presented within a uniquely colored circle. The color of the circles was task-irrelevant. On a subset of trials the color of one circle had a previously learned reward association (reward circle). There were two critical effects. First, when the probed target was presented within the reward circle there was a benefit to target accuracy relative to when no reward circle was presented. When any other item was presented within the reward circle there was a cost. These results add to existing evidence that task-irrelevant learned reward associations result in the spatial capture of attention. Second, task-relevance, both in terms of task set and space, interacted with the spatial capture effects of irrelevant learned reward associations. Specifically, the effect of irrelevant learned reward association was larger for target items than distracters and larger for items within task-relevant space than outside of task-relevant space. Our findings suggest that the effects of task-irrelevant top-down factors (i.e. learned reward associations) on attention are mediated by task-relevant top-down factors.
Meeting abstract presented at VSS 2014