General knowledge and detailed memory benefit from different training sequences.

Real-world decisions require understanding generalities (e.g., sorting an album collection by genre) and the ability to remember specific events (where one acquired a particular album). Discriminating between broad categories versus individual events requires contrasting features at different levels of specificity, and therefore have different representational demands. Here, we used a within-participant design to test the hypothesis that different training protocols (blocked or interleaved order) would have dissociable impacts on representation of generalities and specifics. On each trial, participants viewed a painting from one of twelve artists along with information about its unique location. Category generalization and source memory were tested immediately and after one week. Interleaving enhanced generalization, while blocking improved incidental learning of episodic details. Furthermore, category knowledge remained stable over time, whereas episodic details declined. These results indicate that interleaving and blocking optimize discrimination at different levels of specificity, with differential impacts on inferring generalities and remembering specific events.