Retrieval practice is one of those strategies that’s hard to argue with. The idea is simple: have students actively retrieve information from memory rather than just review it. No passive rereading. Just trying, failing, correcting, and trying again. Its effectiveness has been well-documented for years: what you retrieve sticks better.

But as is often the case, the story doesn’t end there.

A new study by Ophuis-Cox and colleagues examines a topic that has received less attention. Not whether retrieval practice works for factual knowledge—we already know that—but whether it also helps with learning procedural knowledge. Specifically: verb conjugation among fourth-grade students. So not isolated facts, but rules that you must not only know but also apply correctly.

The setup is quite familiar to classroom practice. All students first receive instruction through worked examples. Then they are divided into two groups. One group continues working with examples. The other group combines that with retrieval practice: they must actively recall the rules, receive feedback, and then see examples again.

A week later, two things are tested. First: do they still know the rules? Second: can they also apply them correctly in new assignments?

The first result is clear. Students who received retrieval practice remember the rules much better. The difference is not small, but substantial. That fits perfectly with what we already know: retrieval practice is a powerful way to store knowledge in the long term.

But then comes the second finding, which is also clear and, in my opinion, even logical—though in a different way: there is no difference in application. Both groups improve in their correct application of the rules, but the group that remembers them better does not apply them better. In this case, knowing more does not automatically lead to doing better.

That might feel counterintuitive. If you know something better, shouldn’t you also be able to use it better? But that’s precisely where an important distinction lies that we sometimes overlook too quickly: application is not a simple extension of memorisation.

To spell a verb correctly, you don’t just need to know the rule. You also need to make the right analysis: what is the tense, what is the subject, which rule applies here? And you have to combine all of that, often under time pressure. That requires not only memory, but also processing, selection, and sometimes even the conscious or unconscious suppression of intuitive – but incorrect – answers.

In other words, you must not only be able to retrieve the rule, but you must also use it at the right time and in the right way. And that is a different beast cognitively.

The authors suggest a few possible explanations:

• The combination of steps may simply be too much for working memory to handle, especially for beginners—exactly, cognitive load theory explains this.
• Or students may have memorised the rules better, but haven’t yet internalised them enough to apply them fluently.

It’s also possible that they fall back on faster strategies, such as recognition or linguistic intuition, even when they know the correct rule.

In short, what stands out here is that retrieval practice does what it’s supposed to do. It strengthens memory. But that doesn’t automatically translate to more complex tasks. That doesn’t make this an argument against retrieval practice. On the contrary. It confirms just how powerful that strategy is. But it does set a limit. Or rather: a condition. If you want students to be able to apply something more effectively, you have to explicitly work on that application.

More retrieval practice is not necessarily the solution. Perhaps you need more practice following the steps. Perhaps you need to temporarily reduce the complexity. And perhaps you must first ensure sufficient understanding and only then focus intensively on retrieval. Or the other way around, depending on the task. The study does not provide a definitive answer to that. But it does make clear that the order and combination of strategies matter.