The Benefits of Overconfidence Have Been Exaggerated

To the extent that a fool knows his foolishness, - He may be deemed wise.

A fool who considers himself wise - Is indeed a fool.

– Dhammapada

When I recently reviewed part of Stuart Vyse’s (2022) book on The Uses of Delusion, I noted that the author is open-minded in his quest for false but useful beliefs (Krueger, 2025). When research suggests that such intriguing beliefs have been found fails to replicate, Vyse says so. We have now learned that rituals do not enhance performance on other tasks, that the belief in the freedom of the will does not make one a better person, and that a superstitious fear of being jinxed has no motivating force to make one work with greater effort or prudence.

But the quest for false but beneficial beliefs continues. At the heart of this quest lies the siren song of overconfidence, and specifically the intuition that higher-than-warranted confidence can lead to better outcomes as compared with a well-calibrated sense of confidence. In the authoritative survey of pertinent research, Moore (2020; reviewed by Krueger & Heck, 2021) concludes that at the end of the day, well-calibrated confidence serves a person’s interest best. Underconfidence leaves opportunities unexploited, while overconfidence risks failure and social rejection, especially in the long run. Overconfident individuals fail at tasks they wrongly assume are tractable with the level of ability they have. They bite off more than they can chew. Projecting overconfidence may be appealing to the casual observer, but eventually, overconfidence reveals itself as an empty and manipulative signaling device.

I suggested in my earlier essay that people, actors and observers like, nurse a desire for delusion. They want overconfidence to work; they want to collect gains that lie beyond their actual abilities. This idea that they can is seductive because it amounts to getting something for nothing (or very little). At the limit, this delusion devolves into the cliché that we can achieve anything if we only put our minds to it. Here, ability or competence at a task is screened off. Expectation, in the form of optimism, is left as the only predictor of consequence.

Moore (2020) relates a thought experiment William James proposed to illustrate his pragmatic “will to believe.” James fancies himself hiking in the Swiss Alps. Coming to a crevasse, he feels it is too wide for him to jump across. Yet, he hopes to complete the leap by first psyching himself into an overconfident state of mind. He will succeed, he figures, even though his physical ability technically does not allow it. Wisely, James leaves this be as a thought experiment. Had he actually attempted the leap, he would probably have plunged to his death, thereby refuting his bold hypothesis, or he could have made it to the other side, in which case we would be left wondering if his positive expectation was a case of overconfidence at all. Perhaps James was initially underconfident, and only after psyching myself up did he correctly calibrate expectation with ability, and lived to tell the tale.

Now suppose there are 100 Jameses. If some, but not all, of them complete the leap, we should worry about survivor bias. Having found the occasional success, a causal claim would be misleading when the failures are ignored. As Diagoras of Melos pointed out, the survivors of shipwrecks can feel fortified in their belief in the benevolence of the gods; those who drowned are not there to dispute the claim.

Of the studies Vyse (2022) describes in order to corroborate the beneficence of overconfidence, the report by Chen and Schildberg-Hörisch (2019) is most fascinating. Vyse describes the study in detail and concludes that at last there is good evidence for the idea that delusional overconfidence is good because it makes people work harder. When we take a closer look at the report, however, our enthusiasm dims. The authors set up a visual-attention task where participants can earn money with correct responses. Participants’ ability is assessed and so is their perception of their own ability. When the former is subtracted from the latter, we see that most participants are overconfident. Half of the participants are then informed about their actual ability level, whereas the other half carries on with the task without knowing that they are overconfident.

Three sets of results are relevant: First, the authors focus on the amount of effort exerted after half of the participants have been disabused of their overconfidence. These informed participants worked on 13.1 tasks on average (standard deviation s = 8.03), whereas the unaware participants worked on 15.63 (s = 6.8) tasks. The difference is significant, p = .04, by a non-parametric test, and I estimate the standardized effect size to be d = .34. The authors conclude that overconfidence is good because it gets people to work longer.

We also should ask, however, if harder and longer work pays off. The authors present, but do not statistically analyze, the number of problems solved and the money earned. We can estimate the effect size to be .19 for both these variables, a result hovering just about the significance threshold. Without independent corroboration, this is a rather tenuous finding, and we wonder if overconfidence is a positive delusion if it merely gets people to exert more effort without earning more. Wouldn’t this extra effort be a waste? Still, we should cautiously note it would be premature to close the book on the potential blessings of overconfidence.

Vyse also looks at overconfidence in the context of competition and the projection of signals it entails. Highly confident individuals prevail in games of market entry (Camerer & Lovallo, 1999; see Chwolka & Raith, 2023, for a critical review) and mate attraction (Murphy et al., 2015). In these contexts, what matters is not overconfidence per se, but being more overconfident than one’s rivals. Diagoras, however, again counsels caution. The over-overconfident survivors have a good story to tell, whereas the losers take their moderate levels of overconfidence to their early graves or social exclusion. Overconfidence, is a dangerous game to play, and we may not have the liberty to opt out — unless of course we surrender desire (see the Dhammapada for more on this).