Relational Concept Learning via Guided Interactive Discovery

A key goal in both education and higher-order cognition research is to understand how relational concepts are best learned. In the current work, we present a novel approach for learning complex relational categories – a low-support, interactive discovery interface. The platform, which allows learners to make modifications to exemplars and see the corresponding effects on membership, holds the potential to augment relational learning by facilitating self-directed, alignably-different comparisons that explore what the learner does not yet understand. We compared interactive learning to an identification learning task. Participants were assessed on their ability to generalize category knowledge to novel exemplars from the same domain. Although identification learners were provided with seven times as many positive examples of the category during training, interactive learners demonstrated enhanced generalization accuracy and knowledge of specific membership constraints. Moreover, the data suggest that identification learners tended to overgeneralize category knowledge to non-members – a problem that interactive learners exhibited to a significantly lesser degree. Overall, the results show interactive training to be a powerful tool for supplementing relational category learning, with particular utility for refining category knowledge. We conclude with implications of these findings and promising future directions.