Abstract
To learn from our visual environment, we must constantly monitor for statistical regularities. Indeed, people even seem to spontaneously bias attention toward task-irrelevant regularities (Zhao, Al-Aidroos & Turk-Browne, 2013). Here, we question the extent to which this biasing depends on prior attention to regularities using a visual search paradigm. In the Training Phase, displays consisted of a 12-item "target set," containing a target T among Ls, and a 12-item "distractor set," containing only Ls. These sets appeared in red and blue, with the specific color-target mapping precued on each trial, enabling selective search. However, because of the frequent color switching, we expected participants would partially attend the distractor set; critically, we expected greater attention to the distractor set on trials in which the target color switched from the n-1 trial than when it repeated. To manipulate learning, unbeknownst to participants, we placed regularities in the distractor set (but never the target set). Specifically, every trial contained one of 16 invariant spatial configurations. 8 of these were only presented on color-switch trials, while the remaining ones were only presented on color-repeat trials. This allowed us to assess learning on trials with relatively high attention (color-switch) vs. on trials with relatively low attention (color-repeat). In the Test Phase, participants were no longer cued to the target set, so they now had to search both colors. Each display contained an invariant configuration in one color and a random configuration in the other. Results on trials containing a color-switch invariant configuration showed faster RTs for targets inside of this configuration than outside, demonstrating a clear bias toward a learned regularity. However, we found no inside/outside difference on trials with color-repeat invariant configurations. Therefore, the prior degree of attention to regularities predicts whether people will be biased toward them.
Meeting abstract presented at VSS 2017