Generative processes and knowledge revision

Which learning processes best support revision of educationally-relevant misconceptions in college students? In this study, we investigated the generative process self-derivation through memory integration, the integration of multiple facts to generate novel, never-directly-taught information. Specifically, we examined whether self-deriving correct information supported revision of misconceptions above and beyond an active, yet non-generative, learning process: rephrasing directly-provided correct information. We found that misconception revision occurred across both conditions, but at higher rates in the rephrase condition. However, the two learning processes led to equal success in revising misconceptions when only correct, generative self-derivation trials were considered. Thus, correctly generating one's own information was critical for success in the self-derivation condition. These findings contribute to our understanding of generative learning processes such as self-derivation. The results also offer insight into scaffolding misconception revision, particularly highlighting the benefit of directly exposing learners to corrective information.