Interleaving

“Conditions that create challenges for the learner and appear to slow the rate of learning often enhance long-term retention and transfer.” —Robert A. Bjork

Interleaving is the practice of mixing different topics, problem types, or skills within a single study session, rather than studying one thing to completion before moving to the next. Its opposite—studying in long, single-topic runs—is called blocked practice. Interleaving is the third of the three pillars of durable learning, alongside active recall and spaced repetition, and like its siblings it is counter-intuitive: it feels worse while you do it and works better when it counts.

Blocked versus Interleaved#

Suppose you must learn to solve four kinds of problem, A, B, C, D. Blocked practice drills each in turn—AAAA BBBB CCCC DDDD. Interleaved practice shuffles them—A C B D B A D C .... The total work is identical; only the order differs.

Blocked practice feels more productive. Within a block you settle into a rhythm, your answers speed up, and progress feels tangible. But that fluency is largely an illusion of the moment: you are repeatedly applying a strategy you have just been told to apply, so you never practise the hardest part of real problem-solving—working out which strategy a problem calls for. Interleaving forces that act of discrimination on every single item, which is precisely the demand of an exam, a clinic, or a code review, where problems do not arrive pre-sorted by type.

The Bean-Bag Experiment#

The most charming demonstration of the principle predates the modern interest in interleaving and concerns not facts but a motor skill. In a now-classic study recounted by Brown, Peter C. and Roediger III, Henry L. and McDaniel, Mark A. (2014), a group of eight-year-old children practised tossing bean-bags into buckets. One group practised with the bucket a fixed three feet away; the other practised only with buckets at two and four feet, never three. After the training period every child was tested at the three-foot distance.

The children who had practised at the varied distances out-threw the constant-practice group at three feet—even though they had never once practised at that distance. Drilling the exact target taught a single ballistic movement; mixing the distances taught the more general, transferable skill of calibrating a throw. Strictly, this is a study of variable practice, a close cousin of interleaving; both belong to the family of desirable difficulties Bjork, Robert A. and Bjork, Elizabeth L. (1992), conditions that slow acquisition but deepen learning and transfer.

Why Interleaving Works#

Three mechanisms, all reinforcing, explain the effect.

• Discrimination. Interleaving juxtaposes different problem types, so the learner must repeatedly notice what distinguishes them. Blocked practice hides this: when every problem on the page is the same kind, the choice of method is given away for free.
• Retrieval and spacing. Returning to a topic after others have intervened means retrieving it from a partly-decayed state—the same effortful retrieval that powers active recall, and the same gap that powers spaced repetition. Interleaving and spacing are deeply entangled: you cannot interleave without spacing.
• Reduced illusion of competence. Because performance during interleaved practice is visibly worse, the learner is denied the false confidence that blocked drilling breeds, and judges their true mastery more accurately.

A second canonical result makes the same point with mathematics. College students learning to compute the volumes of unfamiliar solids did better on a delayed test when the problem types were interleaved than when they were blocked—again, despite the blocked group performing better during practice Brown, Peter C. and Roediger III, Henry L. and McDaniel, Mark A. (2014). The blocked students learned to execute formulae; the interleaved students learned to choose them.

The Cost, and When Not To#

Interleaving is not free. Its very mechanism—forcing discrimination on every item—makes it slower and more error-prone in the moment, and for a genuine novice who has not yet grasped a single technique, a short block to establish the basic procedure first is sensible. The rule of thumb: block to acquire, then interleave to consolidate and transfer. Interleaving also presupposes related material between which discrimination is meaningful; shuffling utterly unrelated facts buys only the spacing benefit, not the discrimination one.

Interleaving and the Spaced-Repetition Tool#

The happy accident of a tool like Anki is that interleaving comes for free. Because each day’s review queue is drawn from your entire collection, a single session naturally mixes a Japanese vocabulary word, an anatomy fact, and a programming idiom—each demanding a context switch and a fresh act of retrieval. You do not have to engineer interleaving; you have to resist the urge to defeat it by studying one deck at a time.

References#

References#

Bjork, Robert A. and Bjork, Elizabeth L. (1992). A New Theory of Disuse and an Old Theory of Stimulus Fluctuation, Erlbaum.

Brown, Peter C. and Roediger III, Henry L. and McDaniel, Mark A. (2014). Make It Stick: The Science of Successful Learning, Belknap Press.