Proceedings of the Annual Meeting of the Cognitive Science Society

Much of what we learn comes from a mix of information thatwe select (active) and information that we receive (passive).But which type of training is better for different kinds of learn-ing problems? Here, we explore this question by comparingdifferent sequences of active/passive training in an abstractconcept learning task. First, we replicate the active learningadvantage from Markant & Gureckis (2014) (Experiments 1aand 1b). Then, we provide a test of whether experiencing ac-tive learning first or passive learning first improves the effec-tiveness of concept learning (Experiment 2). Across both ex-periments, active training led to better learning of the targetconcept, but “passive-first” learners were more accurate than“active-first” learners and more efficient than “active-only”learners. These findings broaden our understanding of whendifferent sequences of active/passive learning are more effec-tive, suggesting that for certain problems active explorationcan be enhanced with prior passive experience.