Statistical learning in visual search is easier after experience with noise than overcoming previous learning

This study compared adaptation to novel statistical learning following different environmental changes. Three groups of participants completed training with a visual search task. For the Consistent-first group, targets and distractors appeared in predefined spatial locations. For the Random-first group training contained no regularity and for the Mixed-first group training consisted of both consistent and random conditions. During the test phase, all groups received identical consistent and random conditions. Contrary to previous findings we did not observe statistical learning shutdown following experience with random visual search: a contextual-cueing effect was observed for all groups. However the effect was not stable in the Mixed-first group, suggesting an ongoing adjustment process. We conclude that initiating learning after experience with noise (i.e., random search) is easier than overcoming previous learning (i.e., encountering previously learned layouts). We argue that our cognitive system is sensitive to changes in the visual input and is actively searching for regularity even when it is absent.