The Science of Interleaving: Why Mixing Topics Helps You Learn Faster

The Problem With Blocked Practice

When people learn new material — a set of mathematical procedures, a vocabulary list, a collection of historical events — the intuitive approach is to work through each type systematically before moving to the next. Finish all the problems of type A before beginning type B. Study all the words in category one before category two. This approach is called blocked practice, and it feels efficient because it does: within a single study session, performance improves rapidly as the learner becomes temporarily fluent with the material at hand.

The difficulty is that this temporary fluency does not persist. The improvement achieved during a blocked practice session is partly illusory — it reflects the learner's familiarity with recently processed examples rather than deep encoding. When the same material is tested later, after a delay, blocked learners consistently underperform interleaved learners, often substantially.

What Interleaving Involves

Interleaved practice mixes different types of problems or material within a single session rather than grouping them by type. Instead of completing all type-A problems before moving to type B, an interleaved learner alternates: A, B, C, A, B, C, or some other mixed sequence. The result is that each problem requires identifying which approach applies before solving it — an additional cognitive step that blocked practice does not require.

This additional step is the source of both interleaving's difficulty and its benefit. Blocked practice allows the learner to apply the same procedure repeatedly while it is still warm in working memory. Interleaved practice forces the learner to retrieve and select the appropriate procedure each time — a form of retrieval practice that strengthens both the procedure itself and the ability to discriminate between procedures.

Why Interleaving Feels Ineffective

Research on interleaving consistently finds a dissociation between perceived effectiveness and actual effectiveness. Students and teachers alike tend to rate interleaved practice as less effective than blocked practice, even when their test scores demonstrate the opposite. This is because interleaving is harder — it produces more errors, more confusion, and less fluency during practice. These experiences are interpreted as evidence of poor learning, when they are actually evidence of effective learning.

This metacognitive miscalibration is consequential. Learners who experience difficulty during interleaved practice often abandon it in favour of the more comfortable (and less effective) blocked approach. Understanding the mechanism — that the difficulty is the point — shifts the interpretation of that difficulty from a sign of failure to a sign of productive engagement.

The Discrimination Hypothesis

One prominent explanation for interleaving's effectiveness is the discrimination hypothesis. Blocked practice does not require the learner to distinguish between different types of problems or material, because each block contains only one type. Interleaved practice requires constant discrimination — identifying what kind of problem this is before deciding how to approach it. This discrimination practice builds a capacity that is directly relevant to real-world application, where problems do not arrive pre-labelled by category.

The implication is that interleaving may be particularly valuable in domains where a key skill is problem classification: identifying which tool or procedure applies to which situation. Mathematics is the most studied domain in this context, but the principle extends to any field where practitioners must select among multiple approaches based on situational features.

Interleaving and Creativity

Some researchers have proposed that interleaving may benefit not just procedural learning but creative cognition. The requirement to shift between different types of material within a session may promote cognitive flexibility — the ability to make unexpected connections across domains. While this hypothesis is less thoroughly established than the procedural learning findings, it aligns with broader research on the cognitive benefits of varied experience.