Abstract
Recent studies show that visual search tasks in which targets occur more frequently in certain spatial regions result in improved task accuracy for more probable locations. This advantage emerges rapidly and extinguishes slowly when the target's location distribution becomes random. Here, we investigate the reference frame in which the frequently attended target locations are encoded. We changed the fixation position or the location of search items relative to fixation after participants learned to prioritize a high-probability location. Two hypotheses might explain how learning transfers: (1) spatiotopically, where the same screen location relative to the other items [DA1] is preferred regardless of fixation, or (2) retinotopically, where the same location relative to fixation is preferred. Our participants viewed four items in a diamond configuration, identifying the color of a target T among 3 distractor Ls presented briefly (160ms). Training locations were in one visual quadrant (e.g., the upper-left). One of the four locations contained the target more frequently (50%) than any of the other locations (17%). Participants in an initial experiment developed an accuracy advantage for the high-probability location in training that persisted through 10 blocks of testing in which the target was distributed equally between the four locations. In Experiment 2, the fixation point shifted after training to the center of the diamond configuration. In Experiment 3, the fixation point remained in the center of the screen during testing, but the diamond configuration was shifted to center around the fixation point. We examined whether the learned attentional preference remained in the same screen location or the same retinal location as the previously high-probability target location. Results showed no persistence in either the spatiotopic or retinotopic location. The alignment of spatiotopic and retinotopic reference frames may be important for the persistence of learned attentional biases.
Meeting abstract presented at VSS 2016