Super rapid learning of new attentional sets

Humans guide attention to different targets as they navigate (e.g., going from a school zone to a construction zone while driving). To understand how observers shift focus between different sets of targets when the context shifts, we had observers perform a new visual search paradigm where target shapes are presented in multiple possible colors with distinct probabilities (e.g., 33%, 26%, 19%, 12%, and 10% baseline colors), such that some target colors are more task-relevant than others. Participants learned to prioritize the target colors that appeared with the target shape most often. We then scrambled the color-probability mapping to assess how quickly people could relearn a new set of target probabilities. Participants relearned these new colors significantly faster than their first set of colors, especially when there was less conflict between the new and old attentional sets. Our findings suggest that memory-guided attention is rapidly and continuously relearning to update task-relevant probabilities.