Abstract
Goal-directed control provides a flexible and powerful mechanism for biasing attention towards task-relevant features. However, the extent to which control benefits performance depends critically on how an individual chooses to wield it (e.g. whether one biases a target's color vs orientation). Recently we showed that people frequently use suboptimal strategies to control attention in visual search. We attributed this poor performance to an avoidance of the effort required to sustain optimal search. However, an alternative account is that observers are simply ignorant of the best strategy. Here we examined whether promoting information gain would increase optimal performance. We used the Adaptive Choice Visual Search task (Irons & Leber, 2016), which allows observers to freely choose between two targets, red and blue, on each trial. The relative utility of searching for each target varies periodically with changes in the ratio of red to blue distractors. Typically, individuals search for the optimal target on only 60% of trials. In Experiment 1, we previewed the search display colors prior to presenting the full display, providing additional time for observers to survey the display and determine the optimal target. Nevertheless, choice performance did not change, suggesting that individuals did not take advantage of the preview to improve their strategy. In Experiment 2, we used a more direct approach and explicitly told participants the strategy for choosing optimal targets. Now, participants who received this strategy information made significantly more optimal choices than those that did not, although performance remained well below fully optimal. Together the results suggest that the suboptimal attentional control strategies are partially due to ignorance of better alternatives. However, even explicit awareness of the optimal strategy does not guarantee full compliance, suggesting that other factors, such as effort avoidance, also contribute to suboptimal choice performance.
Meeting abstract presented at VSS 2018