Getting feedback is crucial in learning new skills or knowledge. But how much is needed and when should it be given? PhD research carried out at Vrije Universiteit Brussel and KU Leuven now shows that feedback is especially useful at the beginning of the learning process. But when our brains have acquired sufficient knowledge, feedback is counterproductive and slows down the development of cognitive automation.

To investigate the role of feedback in complex learning processes, psychologist Dr Katleen Vandist, under the supervision of promotors Prof Tim Vantilborgh (VUB), Prof Eva Van den Bussche and Prof Gert Storms (KU Leuven), conducted an experiment in which 34 participants were asked to learn images from two different complex categories over five consecutive days.

On the first two days of the experiment, all participants were told after each image whether they had categorised them correctly or incorrectly. For the remainder of the experiment, the researchers selected the participants who had achieved a certain level of expertise. On days three and four, the remaining participants were then divided into two groups. Group 1 still received feedback after each assignment, while group 2 only received feedback after a quarter of the exercises. On the last day of the experiment, the participants were again presented with images, but this time no one received feedback and they had to categorise the images as quickly as possible. The goal was to see how well they had automated the new skill – meaning they were able to perform the skill correctly and quickly. Participants who had received little feedback on the third and fourth days were quicker and gave the right answer as often as the participants who had received constant feedback. This shows that the knowledge of this group had become more automated.

Double validation

Dr Vandist carried out a similar follow-up experiment to rule out the theory that the participants who had received feedback after each assignment reacted more slowly on the final day due to the sudden disappearance of feedback. This time, 38 participants were asked to divide a series of images into categories. Half were told whether they had answered correctly after 95% of the exercise; the others received feedback after 5% of the assignments. The experiment was arranged in such a way that these groups could be compared with the first experiment. It still turned out that the knowledge of the test subjects who occasionally received feedback on the third and fourth days was more automated.

“This research shows that feedback is necessary at the beginning of the learning process until we have mastered the new skill quite well. Once we have reached a certain level of expertise, the development of automation is facilitated when feedback is only given occasionally”, Dr Vandist says. “Continuous feedback hinders the person from automating the skill. Why this is the case needs to be investigated further. A possible hypothesis is that participants become less attentive because they get the right solution every time. Another hypothesis is that subjects who receive occasional feedback motivate themselves more to keep performing well. They feel more confident about their performance. The proportion of feedback is also important: a very small amount of feedback, like the participants who did not know whether they were doing well in the second experiment 95% of the time, is clearly too little because they made more mistakes. These research results are also interesting for education. Automation is important in a lot of school skills such as reading, learning multiplication tables, categorising animals etc.”