Interleaving: Mixing Topics Learns Better Than Mastering One at a Time

The short answer: mixing the topics learns them better than mastering one at a time

Common sense says: pick one topic, drill it until you've got it, then move on. The lab says the opposite. Rohrer and Taylor (2007) had students study mathematics problems either blocked by problem type (all the volume problems together, then all the area problems) or interleaved (alternating types). On the immediate practice the blocked group felt confident and performed better; on the test a week later, the interleaved group scored more than twice as high (source). The effect — that mixed practice feels worse and works better — is one of the most robust findings in modern learning science, replicated across maths, art, sport, and language.

What's actually happening

Blocked practice is procedurally easy. The previous problem warms up the same machinery the next problem needs, so retrieval is cheap and the right method is already in your hand. You feel like you're absorbing the material because the in-session performance is good. But what you're building is a narrow groove for that exact context — and when the context changes (the test, the real world), you can't tell which problem is which, because you never had to choose during practice.

Interleaving forces the discrimination work. Each problem starts cold, with the previous one a different shape, so before you can solve, you have to first decide what kind of problem this is. That extra step is the actual learning. Robert Bjork's framework calls these "desirable difficulties" — interventions that make the practice session feel worse while measurably improving long-term retention and transfer.

How to do it in real life

• Mix problems of the same broad family, not unrelated subjects. Interleave volume / area / perimeter problems within geometry, not geometry with French grammar. The technique works on discrimination between similar things; mixing wildly different subjects doesn't trigger the mechanism — it just confuses both sessions.

• Trust the felt difficulty. Interleaved sessions feel objectively harder and the in-session score will be lower. Multiple studies have shown that students consistently rate blocked practice as more effective and prefer it — even after seeing their own test results from the interleaved condition. The feeling is wrong. The data is right.

• Pair it with retrieval. Interleaving + active recall is the compound. Each problem starts cold (interleave), and you try to solve before looking at notes (retrieve). The pairing is roughly multiplicative in effect size; doing only one of the two leaves a lot on the table.

• Start with three categories, not ten. Too many categories at once produces overhead — you spend the session figuring out the categorisation rather than the problems. Three to five distinguishable types per study session is the practical ceiling for most readers. Add more once you're discriminating fluently.

• Schedule it from the start, not when you feel ready. The temptation is to block until you "know each topic" then interleave. That's wasted blocked time. The faster path is to interleave from day one with the small set you've got — even three problems on day one with two types qualifies — and grow the bank.

Why this pays double for ADHD

Blocked practice for an ADHD reader runs into the same problem twice: the in-session boredom of repetition is steep, and the brain that desperately wants novelty ends up either rushing through repetitions on autopilot or shutting down entirely. Interleaving converts that wall into momentum. Each new problem is a new shape, novelty is built into the rhythm of the session, and the attention system gets to do what it likes — engage with something different. The cognitive cost is real, but it lands on the part of the brain that was going to be the bottleneck either way.

Practically: ADHD readers often discover that interleaving is the first study technique that doesn't make them want to die. The session feels harder per minute but more engaging — and the engagement is what lets the session happen at all. Combined with the spacing effect (short sessions across days), the two techniques together turn study from an exercise in willpower into a sequence of short, varied, mildly difficult passes that the attention system can actually tolerate.

Where it fails (and the repair)

• Interleaving genuinely unrelated material. Mixing chemistry equations with Spanish vocabulary doesn't create useful discrimination — it just splits a session that should have been two. Interleave within a domain; cross-domain switching is context-switching, which is a different and costlier thing.

• Quitting because the practice feels bad. The first few interleaved sessions feel discouraging by design. Students who drop the technique at week two and revert to blocked practice never see the test-day benefit — they only experience the painful part. Commit to two weeks before evaluating; the test result is the measurement, not the felt confidence.

• Too few representatives per category. If you only have two examples of each problem type, the interleaving doesn't produce enough discrimination practice. You need a bank of at least five or six per category before the technique starts to bite. Build the bank first, then mix.

FAQ

Won't I forget what I learned about topic A while doing topic B?

Some, by design. The forgetting between encounters is part of the spacing effect — it forces real retrieval the next time you see topic A. The test isn't "do I know this in the next ten minutes?" — it's "do I know this in a week?". Interleaving makes you worse at the first test and reliably better at the second. The trade is in your favour for any study purpose that's longer than the day.

What if I'm a beginner and don't have the basics yet?

Some blocked practice is fine at the very start, to build initial procedural fluency on one type. The mistake is staying blocked. As soon as you have two or three categories with basic competence in each, switch to interleaving. The earlier you make the switch, the more interleaving benefit you compound over the course of your learning.

Does this work for skills, like sport or music?

Yes. The motor-learning literature has consistently found interleaving ("random practice") superior to blocked practice for retention and transfer of motor skills, despite producing worse in-session performance. Coaches have known this for decades; the science caught up around the same time as the cognitive interleaving work in the late 1970s and 80s.

What's the simplest possible application?

Three categories, three problems per session, drawn in random order. That's the minimal viable version. As your category bank grows, increase the per-session count but keep the order randomised. The randomisation is the technique; the count and the scheduling are the dressing.

Why do students consistently prefer the worse method?

Because in-session performance feels like learning. Blocked practice produces visible smooth performance — the next problem solved a little faster, the wrong answer rarer. Interleaving produces stop-and-think and more wrong answers. The mistake is reading those signs as learning failures rather than as the learning happening. The data has been clear since the 70s; the felt experience pushes the other way.