Rohrer and Taylor (2007) showed the effect cleanly in mathematics. Students who practised problem types in blocks (“do 20 calculus problems, then 20 algebra”) outperformed interleaved students during practice. On the delayed test, the interleaved group dominated — often by wide margins.
Blocked practice allows autopilot. You know the next problem will use the same technique as the last, so you do not have to think about which technique. Interleaved practice disables autopilot. Every problem is a fresh judgement call.
Fluera’s canvas is interleaved by nature. Navigating your knowledge graph crosses subject boundaries. Spaced review schedules mix topics on purpose. The AI surfaces cross-domain connections precisely where autopilot would take hold.