Traditionally, error avoidance was unequivocally advocated by influential educational psychologists. The argument put forward by well-known researchers such as Skinner, Ausubel and Bandura was that allowing students to make errors was counterproductive because they would be practicing incorrect approaches which would then need to be “overwritten” at a later stage. Up until the late 70s then, the focus of research was on reinforcement, with students encouraged to practice correct answers.

At this point evidence began to suggest that giving the wrong answer could, in certain circumstances, lead to better memory for the correct answer. Here, the type of error made was also important. If students had simply forgotten something, the effects on memory were not as great as when they made a genuine mistake. Moreover, there was no effect if students had absolutely no idea about the answer, but were forced to guess anyway. It seems that believing one’s answer is correct plays an important role. Here, psychologists were faced with a problem. According to cognitive theory, answers given with high confidence should be the most prominent in memory and thus be the most difficult to “overwrite” and replace with a new response. The puzzling phenomenon was termed hypercorrection, but how could it be explained?

Responses we are highly confident in are nearly always correct. In contrast, we usually have low confidence in our incorrect answers. This indicates that in general we have a good idea about what we do and do not know. Cognitive psychologists think that the explanation for the hypercorrection effect lies in the element of surprise involved. People rarely make high confidence errors so when they do make them, they are much more easily remembered.

An alternative explanation from the field of machine learning and neural networks suggests that when things happen as predicted, no learning needs to take place. It is only when there is a so-called prediction error that the network needs to change to accommodate the unexpected result.

Whichever explanation we subscribe to, provision of corrective feedback is important for error correction. Unsurprisingly, it is not enough to simply tell learners whether they were right or wrong. Students get virtually no benefit unless the feedback they receive provides the correct answer.

A number of interesting educational approaches have been based on the hypercorrection finding. Of particular intertest are attempts to deal with student misconceptions (see my comments below). Clearly, focussing on errors and their correction rather than the right answers also has positive effects for teachers—we find out what our students are struggling with.

Unfortunately, research shows that learning achieved through making mistakes may not be noticed by our students. In a study by Huesler and Metcalfe (2012) students who had made a 20% learning gain in the experimental condition (where the hypercorrection effect was leveraged) still believed they had learned more in the control situation where no mistakes were made.

Comment: One of the major areas of debate in science education research is the topic of misconceptions and how to overcome them. Misconceptions are common ideas that students have that are at odds with scientific descriptions. They originate in unconscious explanations based on our everyday experiences—our innate sense of how things work. Because they are based on unconscious assumptions rather than reasoned logic, they are notoriously difficult to change. For an everyday discussion of this phenomenon, based on Andy diSessa’s (1993) notion of phenomenological primitives, see Olga Werby’s blog.

A common misconception in physics for example is the idea that in order to keep something moving we need to keep pushing. It is quite clear where this misconception originates—everything in our everyday experience tells us that left alone things don’t just keep moving—they slow down and stop! However, the accepted physics explanation is that it is the forces of friction that slow things down—take those forces away and things would just carry on moving for ever. Clearly, if students are operating with a “left alone things always slow down” misconception, it is very difficult for them to understand why the planets in our solar system keep moving around the Sun.

The concept of hypercorrection described in the article offers an intriguing method for tackling such student misconceptions. The argument is that a teacher should intentionally create situations where typical misconceptions are cued and, once mistakes are made, take time to discuss why the “obvious” answer is in fact a misconception. Here, it is the surprise value of getting something wrong that is leveraged to help students remember and actively suppress cuing of the misconception in the future.

Another popular approach is the use of so-called refutation texts—documents that students read that explain why a misconception is a misconception (see for example Van Loon et al. 2015). More recently, Zengilowski et al (2021) have provided an overview of work in this area, pointing out some potential pitfalls and areas for future research.

Text: John Airey, Department of Teaching and Learning

The study
Metcalfe, J. (2017). Learning from errors. Annual review of psychology, 68, 465-489.

Keywords: errors, error correction, hypercorrection, engagement, misconceptions, learning outcomes

Further reading