Threat of punishment makes us better judges of our own knowledge

There are some walks of life where trying to be right as often as possible is not enough. Just as important is having insight into the likely accuracy of your own knowledge.

Think of doctors and surgeons making diagnostic decisions. They can't be right all the time, and neither can they be completely certain over their judgments. What becomes important then, is that they have an accurate sense of the reliability of their own knowledge. Psychologists call this "metacognitive accuracy".

Michelle Arnold and her colleagues have performed the first ever test of whether people's insight into the accuracy of their own knowledge is affected by whether they are tested under conditions of high reward for correct answers, or high penalties for wrong answers.

Across two studies, dozens of undergrads attempted to score as many points as possible on a series of multiple-choice general knowledge questions. Crucially, for each question, they were compelled to state their favoured answer, and then whether they wanted to submit it or not for scoring (with points gained for correct answers, and points lost for wrong answers). This two-stage process gave the researchers insight into the students' accuracy of their own knowledge. Withholding answers that would have been incorrect and submitting correct answers would be a sign of high metacognitive accuracy.

The key finding was that the students showed more metacognitive accuracy when they were tested under high punishment conditions in which wrong answers led to a loss of four points, whereas correct answers gained just one point. This is compared to a baseline condition in which correct answers earned one point, and wrong answers lost one point; and compared to a high reward condition in which correct answers earned four points, and wrong answers lost just one point.

Under threat of high punishment, it makes sense that the students were more cautious in the answers they chose to submit. But critically, their submission of correct answers was only moderately reduced, whereas their withholding of wrong answers was significantly increased. In other words, the threat of high penalty for wrong answers had the effect of improving the apparent insight the students had into the accuracy of their own knowledge.

This is a new area of study and the effect needs to be tested in other contexts and over longer time periods. However the real-life implications are exciting. As the researchers concluded: "the positive impact of punishment on strategic regulation may be ripe for widespread application, especially in areas where training effective metacognitive discrimination is vital – such as medicine and business." It will be particularly interesting to see whether repeated testing under high penalty conditions – imagine medical students taking diagnostic exams in this format – leads to lasting, beneficial changes in people's insight into their own knowledge.

Further reading

Arnold, M., Chisholm, L., & Prike, T. (2014). No pain no gain: The positive impact of punishment on the strategic regulation of accuracy Memory, 1-8 DOI: 10.1080/09658211.2014.990982