Learning from mistakes is a topic that has come up several times on this blog. A new study in the Journal of Educational Psychology once again shows that the answer is more nuanced than a simple yes-or-no. What did you expect? But keep reading, because the practical implications are more interesting than you might think.
A quick reminder. Researchers have long distinguished between making your own mistakes while learning and learning from errors deliberately embedded in worked examples. The latter has received considerable attention in recent years. The idea is appealing: if learners first have to identify and correct an error, they should process the material more deeply and remember it better.
This new study by Beege and colleagues investigated exactly that. Over 300 university students learned the basics of Java programming. Some studied correct worked examples, while others received worked examples containing deliberately inserted errors. The researchers also varied the instructions. Some students were simply asked to understand the examples, whereas others had to actively search for errors and correct them.
The results may surprise many readers. Students who searched for errors did not outperform those who studied correct worked examples. Their scores on both retention and transfer were very similar. They did, however, spend considerably more time on the task and experienced greater cognitive load. In other words, they achieved the same learning outcomes, but at a higher cost.
In a second experiment, the researchers went one step further. They compared three approaches: searching for and correcting errors independently, correcting errors that had already been identified, and reflecting on examples in which the errors had already been corrected. Once again, independently searching for errors proved to be the least instructionally efficient approach. Reflecting on already corrected errors was at least as effective while requiring less cognitive effort.
This does not mean that previous research was wrong. Quite the opposite. The study mainly helps us understand when learning from errors is likely to be beneficial and when it is not. For novices, searching for errors is a demanding task. They must understand the material, identify the error, determine the correct solution, and explain why the original solution is incorrect—all at the same time. Unsurprisingly, this places substantial demands on working memory. From the perspective of Cognitive Load Theory, that is exactly what we would expect.
For classroom practice, this is perhaps the most important message. We sometimes assume that making a learning task more difficult automatically leads to better learning. That is not necessarily the case. A student who spends twenty minutes searching for an error they ultimately fail to understand may learn no more than a student who spends five minutes studying a correct worked example and reflecting on it.
This certainly does not mean that errors should never be used. Rather, how you use them—and with whom—matters far more than simply including errors in instructional materials.
So what are the practical lessons?
- For beginners, correct worked examples are often the best place to start. They help learners build an initial mental model without imposing unnecessary cognitive load.
- If you do want to use erroneous examples, make sure learners already have sufficient prior knowledge. Searching for errors only becomes a meaningful learning activity once students have a basic understanding of what a correct solution looks like.
- Another promising approach is to present an erroneous example, then provide the correct solution and ask learners to explain why the original reasoning was incorrect. That aligns well with the findings of this study.
Overall, the study reinforces a broader lesson from educational psychology. Active learning is important, but not every form of activity is equally productive. Good teaching is about asking learners to invest the right kind of cognitive effort at the right moment. For beginners, that often means understanding first. Only then does analysing errors become truly valuable.