Abstract
A large body of evidence suggests that statistical learning and other forms of selection history can influence the allocation of attention. However, while these effects have been extensively studied in forced choice tasks, little is known about how selection history influences which features observers choose to search for. Here, we examined whether statistical learning can facilitate the optimal choice of attentional control settings. Participants searched through an array of colored squares and identified a digit located within a red or blue target square. Each trial contained both a red and a blue target, and participants were free to choose which color to search for. On each trial, one color contained more squares than the other color. Thus, the optimal strategy was to search for the color with the fewest squares. Critically, one color was the optimal color on 75% of trials while the other color was the optimal color on only 25% of trials. Participants were faster to locate the target in the mostly optimal color and made a larger proportion of optimal choices in this color. However, because the mostly optimal color was more likely to repeat from trial to trial, it is possible that these results reflect the cumulative effects of intertrial priming rather than statistical learning. To address this issue, we conducted a second experiment in which one color was the optimal color on 75% of trials in an initial training phase but only 50% of trials in a subsequent test phase. Again, participants were faster to locate the target in the mostly optimal color and made a larger proportion of optimal choices in this color. Critically, these effects occurred in both the training and test phases. Together, these findings suggest that selection history plays an important role in determining which features observers choose to search for.