A new study looks at overconfidence in chess players, and suggests that efforts to reduce it in everyday life via feedback can still fall flat.
06 October 2025
By Emma Young
Overconfidence has been dubbed "the most significant of the cognitive biases", with research suggesting that it contributes to poor decisions in fields as diverse as medicine, financial investing, political leadership, and management, note the authors of a recent paper in Psychological Science.
Various ideas about what causes this bias have been proposed. One theory is that if we don't receive good objective feedback, it's hard to make accurate judgements about ourselves. If your partner tells you that you're really smart — or a wonderful cook, for example, whether that's true or not, you might well believe them. Patrick R. Heck and colleagues therefore set out to investigate whether a lack of quality feedback really could be to blame.
The researchers picked a group of people who get very regular, accurate and precise feedback on their performance in one particular domain: tournament chess. If these people are overconfident about their chess-playing abilities, the cause can't be a lack of quality feedback, the team reasoned.
With help from the US Chess Federation, a chess magazine, and a chess-training app, Heck and colleagues recruited a total of 3,388 tournament players aged 5 to 88, from 22 different countries. These players had an average of almost 19 years of tournament experience, and included people with a range of abilities, up to a grandmaster who at the time was ranked as the 20th best player in the world.
All of these players had official ratings, known as Elo ratings, which are known to accurately predict the outcomes of games. These ratings are displayed next to each player's name at tournaments, updated after every tournament and regularly discussed between players, meaning that most tournament players are well aware of the rating system and how it works, the team writes.
The participants completed a questionnaire, which, among other things, asked them to indicate their current, official tournament chess rating, and then to supply the rating that they believed would accurately reflect their true current chess ability. The participants also made predictions about the outcome of a set of hypothetical 10-game matches between themselves and other players of varying hypothetical ratings.
When the team checked official records, and analysed the participants' responses, they found that on average, these players knew their own ratings, but felt that their true ability was substantially higher. They also believed that in the hypothetical matches, they would do better than would be expected based on their current rating. To put this difference in perspective, the researchers explain that the average degree of overconfidence among these players would have shifted expectations for a match that actual ratings data would predict to be a tie "to a resounding victory".
Not all the players were equally over-confident, however. About 29% believed their current rating was accurate, and 14% believed it was too high. When the researchers dug further into the data, they found that highest-rated players tended to be the most realistic. This finding is consistent with the well-documented Dunning-Kruger effect.
Perhaps these apparently overconfident players may have actually been capable of playing better than their initial official ratings suggested; to check this, the team looked at their ratings six and twelve months after they completed the survey — and found no evidence this was the case. At both time points, only around 8% of the players who had felt that their initial actual rating was too low had reached or exceeded the rating that they felt would have accurately represented their true ability.
The results of this work do show that (in this sphere at least) accurate, plentiful feedback did not eliminate overconfidence, suggesting that players' self-perceptions are biased towards optimism, the team writes. This observation might support an alternative theory for the cause of the optimism bias beyond the realms of chess — that it a fundamental, deep-seated human bias. If so, efforts to tackle overconfidence in critical areas such as medical decision-making or financial investing by improving feedback would seem unlikely to be effective. Further investigations involving different populations would be needed to confirm this idea, however.
Read the paper in full:
Heck, P. R., Benjamin, D. J., Simons, D. J., & Chabris, C. F. (2025). Overconfidence Persists Despite Years of Accurate, Precise, Public, and Continuous Feedback: Two Studies of Tournament Chess Players. Psychological Science. https://doi.org/10.1177/09567976251360747
SOURCE:
https://www.bps.org.uk/research-digest/overconfidence-can-thrive-even-detailed-continuous-feedback(accessed 24.10.25)
No comments:
Post a Comment