When does passive learning improve the effectiveness of active learning?

Kyle MacDonald and Michael C. Frank

Welcome to the web site for the (upcoming) paper "When does passive learning improve the effectiveness of active learning?" Here you will find links to several relevant materials:


Much of what we learn comes from a mix of information that we select (active learning) and information that we receive (passive learning). But what type of learning is best for different kinds of learning problems? In the current work, we explore this question by comparing different sequences of active/passive training in an abstract concept learning task. First, we replicate the active learning advantage from Markant & Gureckis (2014) (Experiments 1a and 1b). Then, we provide a direct test of whether experiencing active learning first or passive learning first improves concept learning (Experiment 2). Across both experiments, active training led to better learning of the target concept, but “passive-first” learners performed better than “active-first” learners. These findings broaden our understanding of when different sequences of active/passive learning are most effective, showing that for certain learning problems active exploration can be enhanced by prior passive learning.