Interleaving: Blocking Feels Better, Works Worse

2026-05-20 status: finished importance: 6 confidence: likely practicetransfermotor learning

Give a batting practice session to two groups. Group A sees 15 fastballs, then 15 curveballs, then 15 changeups, all blocked by type — the standard way batting practice has been run for a hundred years. Group B sees the same 45 pitches in random order, mixed together, so they never know what's coming next. During practice, Group A hits more pitches, feels more in control, and rates the session as more effective. On a test given days later using a new, random sequence of pitches, Group B outperforms Group A by a wide margin.^[1] This is interleaving, and the batting example understates how consistently it replicates: it shows up in badminton serves, painting-style identification, and mathematics problem sets with nearly the same shape every time.^[2]

The mechanism: discrimination, not just retrieval

Interleaving is often grouped with spacing and testing as a "desirable difficulty," and it shares their signature — worse performance during practice, better performance on a delayed test — but the proposed mechanism is distinct. Spacing is mostly about retrieval strength: restudying after some forgetting exercises recall. Interleaving is mostly about discrimination: when problem types are blocked, you know in advance which strategy applies, so the actual skill you are practicing is executing a known strategy, not selecting one. Mix the types together and every trial requires first classifying what kind of problem is this before you can solve it — and that classification step is precisely the skill a real test, with its unlabeled, unsorted problems, will demand.^[3]

Rohrer and Taylor's math studies make this concrete: students given blocked practice on four types of geometry problems (each block practiced immediately after being taught) scored 89% on a same-day test of those problem types, but only 20% a month later on a mixed test. Students given the identical problems interleaved scored lower immediately — 60% — but 63% a month out, essentially unchanged.^[4] The blocked group had not really learned to solve the four problem types; they had learned to execute a pre-selected procedure once told which one to use, and blocking hid that gap until the sorting was taken away.

Why nobody does it on purpose

Textbooks, courses, and training programs are blocked by default for a defensible-sounding reason: a new skill is often unlearnable at all until it's isolated from competing skills. Nobody interleaves a beginner's first day of scales with their first day of arpeggios. But the same blocked structure tends to persist well past the point where it stops helping, because blocked practice produces steady, visible, immediate improvement, while interleaved practice produces a rockier session that both students and instructors read as a sign the method isn't working — the metacognitive illusion runs in both directions of the classroom.^[5] Surveys of how students voluntarily schedule their own practice, when given the choice, find they overwhelmingly choose blocking, and rate it as more effective immediately after a session in which interleaving actually produced better retention.^[6]

When blocking is still right

The evidence for interleaving is strongest when the different practice categories are easy to confuse with each other and the real-world test will require telling them apart on the fly — different pitch types, different geometric formulas, different painting styles. It is weaker or reversed for material with no discrimination component, and for true beginners who have not yet built enough of a base skill in any one category to benefit from contrasting it against others; some scaffolded blocking early on, followed by a shift to interleaving as competence builds, tends to outperform either pure approach.^[7]