I have always been cautious with statements like “this works.” Readers of this blog, or of my books, will know that I usually prefer “this can work.” Especially when it comes to instructional principles that, in practice, can be easily reduced to simple recipes. Interleaved practice is a good example. Alternating different types of practice has long been presented as a powerful, desirable difficulty. And often, that is true. But anyone who has tried to apply it to genuinely complex content will also know how quickly things can derail. That is precisely why a recent field experiment by Maria Danzglock and colleagues in Learning and Instruction is worth a closer look.

Interleaved practice means that students practise different, but related, types of content in a mixed sequence. Not ten problems of type A followed by ten of type B, but A, B, C interwoven. The idea is that learners must repeatedly decide what applies in a given situation, rather than automatically repeating the same procedure.

The study involved 376 students in the final years of upper secondary education and focused on a notoriously difficult topic in physics: the motion of charged particles in electric and magnetic fields. The content is abstract, visually complex and conceptually confusing. Exactly the kind of material for which interleaving looks promising in theory, but can disappoint empirically. The study’s design is solid: a 2×2 design comparing interleaved versus blocked practice and individual versus collaborative learning. This was not a laboratory experiment, but was conducted in authentic classroom settings. There was an immediate post-test and, one of my favourite features alongside preregistration, a follow-up test eight weeks later.

The core finding is easy to summarise and immediately invites further reading. Interleaving on its own did not work here. In fact, interleaved practice led to slightly worse long-term outcomes than blocked practice. Collaboration on its own did not work either. But the combination of both did. Students who worked in pairs and practised in an interleaved way achieved the highest scores for this topic, both immediately and after eight weeks. These are not spectacular effects, but they are consistent and robust interaction effects that remain after controlling for prior knowledge, interest and self-concept.

From an instructional perspective, this is an important nuance. This is not a confirmation of the idea that interleaving always works, nor of the claim that collaboration automatically leads to learning gains. On the contrary. The study suggests that, with complex content, interleaved practice can easily place too high a demand on working memory. The desirable difficulty then turns into an undesirable overload.

In this context, collaboration acts not as a social add-on but as cognitive support. When students reason together, discuss errors and make strategies explicit, they share the cognitive load. The data back this up. Students in the interleaved–collaborative condition experienced the material as less complex, according to the cognitive load measures.

Methodologically, the study stands on solid ground. The authors preregistered their design, recruited a sample large enough to detect the intended interaction effects, and analysed the data in ways that respect the nested structure of students within classes and pairs. The knowledge tests closely align with the content and capture more than just procedural fluency. The eight-week follow-up further strengthens the design. This study does not chase short-term gains; it focuses on learning that lasts.

The limitatins, however, are real and openly acknowledged. The study cannot show what actually happens inside the collaborative interactions. There are no think-aloud protocols and no fine-grained analyses of interaction quality. Claims about shared reflection, conceptual conflict and conceptual change, therefore, remain plausible rather than directly demonstrated. The assessment instrument also makes it hard to separate conceptual understanding, situational knowledge and procedural execution. That matters, because interleaving is assumed to influence precisely the first two.

For practice, the message again warns against oversimplification. Interleaved practice is not a switch you can simply turn on. Not with complex subject matter, and not when students work on their own. Without additional support, interleaving can even undermine learning. At the same time, this study shows that collaboration can be more than “working together”. In the right conditions, it creates the cognitive space that demanding learning strategies need in order to pay off.

In education, we often discuss instructional approaches one by one, but rarely examine how they interact. Classroom teaching, however, almost never consists of isolated ingredients. Instruction works through combinations.

Image: https://universeofmemory.com/interleaved-practice-maximize-learning-pace/