Neural mechanisms by which spaced learning promotes understanding

Abstract

Human learning must build understanding that can be reconstructed and extended, especially as artificial intelligence advances. Spacing improves retention, with neural mechanisms studied for memory; a few studies show benefits for understanding, yet none has examined its neural basis. Here, learners inferred unfamiliar word meanings from familiar component words, reconstructing an intrinsic explanatory relation. Schedules were equated at a criterion-verified immediate endpoint before delayed EEG retrieval. Spaced learners revised written explanations more often, in smaller steps converging on shared formulations, and later reported greater explanatory retrieval fluency than massed learners. Spacing showed greater differentiation from once-learning than massing over right frontocentral and temporal sensors (520-740 ms); a right-temporal subset was indirectly associated with the fluency advantage. A visual word-recognition model matched early-stage activity only under once-learning and massing, but word identity under all schedules. Earlier encoding-stage similarity and retrieval-stage differentiation may be successive windows on the reorganisation making understanding reconstructable.