Interleaved practice: when mixing subjects helps

Interleaved practice means mixing topics or problem types in the same session, instead of finishing one topic entirely before starting the next. Instead of doing twenty identical exercises, you alternate the types. In some studies, the mix felt harder in the moment and paid off more a few days later. The effect, however, does not hold equally for every subject.

Blocked and interleaved: what is the difference

Studying in blocks means concentrating on one topic at a time, like a whole math chapter before moving to the next. Interleaving means alternating. The difference seems small, but it changes the kind of effort. In a block, you repeat the same rule. In a mix, you have to decide which rule to use before solving. One possible explanation is that choosing the right rule is a separate skill to train. That is a hypothesis, not something the studies below prove on their own.

What the research found in math

Rohrer and Taylor (2007) ran two experiments with university students, in a lab. In the first, what varied was the spacing of practice, not the mixing: practicing over several sessions paid off much more, on a test a week later, than practicing everything in a single session. In the second, students learned several types of problem and practiced either in blocks, by type, or with the types mixed at random. On the test a week later, the mixed-practice group was far superior (Rohrer & Taylor, 2007). During practice, though, the mixed group got fewer right, and that is expected: it looks worse in the moment and pays off more later.

The problems were about the volume of geometric solids, and the group was small. The result holds with caution for other subjects and other ages.

What the research found with categories

Kornell and Bjork (2008) tested category learning with paintings by different artists. When the paintings by each artist appeared in a block, people were worse at identifying the artist of new paintings than when the paintings were interleaved with those of other artists. The important detail: participants thought the block worked better, even when their own performance showed the opposite (Kornell & Bjork, 2008). The feeling of learning can mislead, which is why it is worth measuring the result, not only the comfort.

What the reviews conclude

Dunlosky and colleagues (2013) ranked interleaved practice as moderate utility. It has promising evidence, but its general effectiveness is not yet well established. That means it is worth testing in your own routine, and not treating as an absolute rule.

Examples of an interleaved routine

Math, in a one-hour session: 10 minutes of linear equations, 10 of proportions and 10 of percentages. Then solve a mix of five problems of different types and check.

German: alternate connectors with grammar. In one session, mix "weil" and "obwohl" with verb position, and write three short sentences that use all three.

German again, with other material: alternate articles ("der, die, das") with the conjugation of "sein" ("ich bin, du bist, er ist"). Then read a short sentence and identify the article and the verb form.

Mixing different subjects, like ten minutes of the French Revolution followed by ten of mitosis and then mixed questions, is a variation close to what the research tested, but the sources in this article deal more with problem types and categories within a topic. Treat mixing between subjects as a test in your routine.

Can you interleave within one subject?

Yes. That is the most common case in the studies cited: problems of different types within the same discipline. A practical suggestion, which does not come from a specific study, is to first learn the rule for each type in a short session, and only then mix. That way you do not mix what you do not yet understand.

Does interleaving work for every subject?

There is no single answer. For categories and problem types, the research cited suggests gains. For long texts or sequences you have to follow in order, like the steps of a procedure, mixing can hurt more than help.

A simple test: if you can solve the mix with fewer mistakes, without looking at the material, the strategy is on track. If mistakes rise a lot, go back to blocks for a few days and mix again later.

How to set up an interleaved session

1. Choose three to five types of exercise that you already know how to solve separately, for example the four operations with fractions, proportions and percentages.
2. Shuffle: write the exercises on slips of paper and draw the order, or ask someone to mix the list.
3. Before calculating, say which type of problem it is and why. This identifying step is the new part compared with a block.
4. Check at the end. Note which type you confused most and devote the next session to it, in a short block, before mixing again.

How much to mix

There is no proven number. As a starting point, three types per session. If you are mixing too much and getting stuck all the time, reduce. If it is too easy, add a type.

Common mistakes

• Mixing before understanding the rule for each type.
• Thinking that making mistakes in the mix is a sign the method does not work. The mistake shows what to review.
• Giving up in the first week. The effect, when it appears, tends to be noticed after a delay, not in the moment.

To organize reviews over time, see spaced repetition. For the part about trying to remember, see active recall.

Sources

1. Rohrer, D., & Taylor, K. (2007). The shuffling of mathematics problems improves learning. Instructional Science, 35(6), 481-498. https://doi.org/10.1007/s11251-007-9015-8
2. Kornell, N., & Bjork, R. A. (2008). Learning concepts and categories: Is spacing the "enemy of induction"? Psychological Science, 19(6), 585-592. https://doi.org/10.1111/j.1467-9280.2008.02127.x
3. Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4-58. https://doi.org/10.1177/1529100612453266