Interleaving
Mix problem types instead of blocking them.
definition
Mixing different problem types or topics in a single study session, rather than blocking them together.
Interleaving is the practice of mixing different problem types or topics within a single study session, rather than blocking them together. The counter-intuitive finding: interleaved practice feels harder in the moment (which makes students avoid it) and produces dramatically better transfer to novel problems later.
This is one of the clearest examples of what Robert Bjork calls a 'desirable difficulty' — a training condition that reduces short-term performance but improves long-term learning. Blocked practice feels smooth because you're rehearsing the same procedure over and over; interleaved practice feels choppy because you have to identify the problem type before you can solve it.
That identification step — figuring out what KIND of problem you're looking at before you solve it — is exactly what exams test. Blocked practice skips it (you know every problem in the block is the same type). Interleaved practice trains it. Which is why students who practice blocked often score below their apparent skill level on mixed exams.
Brainback's recall queue interleaves by construction. Cards from different subjects, different topics, and different difficulty levels arrive in mixed order. You never review 20 organic-chem cards in a row unless you deliberately filter to — the default is the healthiest one. This isn't a feature we added; it's the emergent behavior of scheduling on the SM-2 algorithm across multiple subjects simultaneously.
What the research says about interleaving.
Three papers that established the finding.
Every design decision Brainback makes about interleaving flows from these papers. Cited so you can verify.
The Shuffling of Mathematics Problems Improves Learning
College students who practiced math problems in an interleaved order performed 43% better on a delayed test than students who practiced the same problems in blocked order — despite the interleaved group performing worse during practice.
Learning Concepts and Categories: Is Spacing the 'Enemy of Induction'?
Students studying paintings by different artists learned to distinguish styles better when the paintings were interleaved (Artist A, then B, then C, then A again) than when blocked. Interestingly, students self-reported that blocked practice felt more effective — the opposite of the actual result.
The Effects of Interleaving Versus Blocking on Foreign Language Vocabulary Learning
Interleaved vocabulary practice produced better long-term retention than blocked practice. The gain was largest for confusable items (words that share phonetic or semantic features), which is exactly where interleaving is theoretically predicted to help most.
How it actually works, at the cognitive and neural level.
The mechanism isn’t decorative. Understanding WHY the effect exists lets you tell when it applies and when it doesn’t.
The cognitive mechanism is discrimination learning. Blocked practice teaches you the procedure for a specific problem type; interleaved practice teaches you both the procedure AND the ability to identify which procedure applies. Real exams test the identification step first, then the procedure — blocked practice skips the harder half.
The neural correlate involves the interaction between recall of similar-but-different patterns. When you interleave, each problem forces retrieval-plus-discrimination against recently-seen problems. This co-activation of related patterns strengthens the ability to distinguish them — which is a different cognitive skill than procedural fluency.
The reason interleaving feels harder: it produces more retrieval failures during practice than blocking does. Retrieval failure is uncomfortable but productive — it forces you to notice the confusion between similar patterns. Blocking hides the confusion by only ever giving you one pattern at a time.
Same scenario. Two study strategies.
Calculus · integration techniques.
You're learning u-substitution, integration by parts, and trig substitution. You have 30 problems to practice.
You do the 10 u-substitution problems, then the 10 integration-by-parts problems, then the 10 trig-substitution problems. Each block feels smooth because you're rehearsing the same technique. On the exam, you see mixed problems and freeze — you can't quickly identify which technique applies.
You mix all 30 problems in random order. Each problem forces you to first ASK 'which technique should I use here?' before executing. It feels choppy and slow — you make more errors during practice. On the exam, you see mixed problems and quickly identify the technique for each, because you've been practicing the identification step for a month.
Rohrer & Taylor's 2007 result was measured on exactly this scenario. Interleaved group scored 43% better on the delayed test despite performing worse during practice. This is one of the most reliable findings in cognitive psychology — and one of the most under-used.
The method, made operable.
The recall queue interleaves by construction. Cards from different subjects, different topics, and different difficulty levels arrive in mixed order. You never review 20 organic-chem cards in a row unless you deliberately filter to — the default is the healthiest one.
Three subject areas where the effect is strongest.
The method works everywhere. It works especially well in these three domains — for reasons specific to each.
Where problem-type identification is a distinct skill from solution execution. Interleaving trains identification; blocking doesn't.
Ser vs. estar, imperfect vs. preterite, subjunctive triggers — all confusable. Interleaved practice trains the discrimination.
Discrimination between similar-presenting conditions is exactly what med students train for. Interleaving is native to the domain.
Three weaker approaches — and why they underperform.
Understanding what the method beats (and by how much) tells you when to reach for it.
why weaker · You know every problem in the chapter uses the same technique. Zero identification-step training. Feels productive, transfers poorly.
why weaker · Same failure. You never practice identifying the type; the problem set tells you.
why weaker · Interleaving without spacing loses the retention benefit. Both together (which is what Brainback does) produce the largest effect.
How to actually do it tomorrow morning.
Same time as blocked practice — the shape changes, not the volume. The steps below are the sustainable protocol — not a heroic one-off.
Instead of doing a chapter's problem set in order, print or list the whole chapter's problems and shuffle them.
Solve in shuffled order. Note the moment of 'wait, which technique?' — that IS the training.
After solving, note WHY each problem uses the technique it does. That's the discrimination skill you're building.
In Brainback: the recall queue interleaves cards automatically. Don't filter to a single subject unless you have a specific reason.
For a real exam simulation: create a mixed practice set drawing from multiple recent chapters.
Assuming that because interleaved practice feels harder, it's less effective. It's exactly backwards — the extra effort is the mechanism. Bjork calls this a 'desirable difficulty.'
The mechanism, in one paragraph.
Interleaving forces you to identify what kind of problem you're looking at before solving it — which is exactly what you have to do on exams. Blocked practice skips that identification step, so blocked-practice students freeze on mixed exams.
Everything students ask about interleaving.
Does this apply to problem sets I have to do in order?▾
You can do the problem set in order but interleave your practice review of past material during the same session. Or interleave across problem sets.
How mixed is mixed enough?▾
Any mix of 3+ topic categories in a session gets most of the effect. Random ordering is best; alternation is second-best.
Does it work for language learning?▾
Yes — mixing grammar drills, vocab, and comprehension in a single session beats blocking. Duolingo does this by design.
Does it apply to reading, not just problem-solving?▾
Yes — reading a physics chapter, then a chem chapter, then back to physics beats reading each cover-to-cover then moving on. The discrimination benefit applies to conceptual understanding too.
How does it interact with spaced repetition?▾
They're separate effects that stack. Spaced (right timing) + interleaved (mixed content) = both benefits. Brainback delivers both simultaneously via the SM-2 queue across multiple subjects.
Three more methods that stack with this one.
interleaving
Try the method in the product.
Spaced recall runs on every account, free tier included.