Attentional guidance by working memory overrides saliency cues in visual search

Abstract

Visual search is influenced by a variety of factors, including how much the target stands out (i.e., its salience) and whether it is currently relevant (i.e., is it in working memory?). The contents of working memory are known to influence attention in simple visual search (Soto, Hodsoll, Rotschtein, & Humphreys, 2008). Furthermore, this memory influence likely occurs early in visual processing (Soto, Humphreys, & Heinke, 2006), as it can enhance efficient ‘pop-out’ search—wherein salient targets are detected almost automatically. Here we show that memory guidance can not only enhance but also reverse the capture of attention by visual salience. In a dual-task paradigm that combined working memory and multiple-target visual search, participants were first instructed to remember the rotation and orientation of a ‘T’ for a memory test at the end of the trial. During the retention interval, participants searched for target ‘T’s amongst distractor ‘L’s. Targets were either relatively dark, which made them highly salient against a white background, or relatively light and low-salient. In Experiment 1, working memory significantly guided search such that when the memory item matched a target, participants were more likely to find that specific target first, regardless of salience. Importantly, when the memory item did not match either target, participants found the high-salient target first, in keeping with a classic saliency effect. These results showed that the saliency effect was reversed via working memory biases. In Experiment 2, we amplified the saliency effect, such that the high-salient target was even easier to find. Participants were strongly biased to find the high-salient ‘pop-out’ target first, yet this saliency effect was attenuated when the memory item matched the low-salient target. Collectively, these findings indicate that the deployment of attention in visual search is modulated by a balance between memory guidance biases and saliency cues. Meeting abstract presented at VSS 2012