The evolutionary logic of overconfidence

People in general are overconfident, excessively optimistic and think of themselves as superior to others.

How can I say such a thing?

Ask anyone with a license to rate their driving abilities, and most people will tell you that they are above average. However, this is not just an isolated case of cocky drivers. The same effect is found when asking people to assess their own intelligence, evaluate their attractiveness, or reflect on how kind they are.

survey of American school students, which received over 1 million responses, found that no less than 70% rated themselves as above average leaders. Conversely, only 2% of these students humbly stated that they are actually below average.

This all begs the question: why do we think more highly of ourselves than who we really are?

To explain the prevalence of overconfidence, psychologists have historically pointed to the mental health benefits of positive illusions— such as maintaining self-esteem, or helping reduce anxiety about an uncertain future. In other words, that these biases protect us from threatening information.

However, a more intriguing explanation has been proposed by the evolutionary biologist Robert Trivers.

Robert Trivers is widely known as the ‘bad ass’ of evolutionary biology, for his antics both in and outside of academia. However, this reputation has not diminished his scientific contributions. Steven Pinker has described Trivers as “one of the great thinkers in the history of Western thought”, and Time magazine has named him one of the greatest scientists of the 20th century.

In his book The Folly of Fools, Trivers argues that a glowing view of  the self makes others see us in the same light— leading to more cooperative and romantic opportunities in life. This is because self-deception requires no conscious input, which therefore eliminates any tell tale signs of lying– ultimately making one more convincing when trying to persuade others. In other words, these positive illusions evolved not to protect us from threatening information per se, but rather to help us persuade others of our superiority.

The logic of the theory is as follows: deception is a fundamental aspect of communication. Those of our ancestors who were better at detecting deception were at an advantage, causing an evolutionary arms race in deception and detection. As a result, self-deception evolved to better mask deception, “hiding the truth from yourself to hide it more deeply from others”.

Trivers argues that we fool ourselves in all realms of life— when overestimating our looks or abilities, when justifying our beliefs, or when convincing ourselves that a lie we’ve told is actually true. Apparently, it’s all part of advancing our own agendas.

Although Trivers formulated his theory back in 1970s, only recently has it begun receiving empirical support. Fundamentally, as we don’t have access to people’s private beliefs, we can’t say with certainty whether someone is actually lying or deceiving themselves (from the best of my knowledge, we’re yet to develop a mind-reading machine). This makes testing such a hypothesis particularly tricky.

But it isn’t impossible. Answering Trivers’ call to arms, economists Peter Schwardmann and Joël van der Weele have recently published a paper in the journal Nature Human Behaviour, that successfully tackled this challenge.

Put your money where your mouth is

So, how do you measure people’s beliefs about their own abilities if you’re unable to read their minds? Peter and Joël’s simple yet ingenious idea was this: get people to bet on their own performance.

To elaborate, Peter and Joël conducted experiments where contestants essentially bet on their performance on an intelligence test. This meant that people had money at stake for accurate assessments of their intellectual performance.

However, there was a twist.

To see if introducing opportunities for social gain induced self-deception, some of the participants were invited to a speed-dating style interview with a panel of ’employers’. Here, the contestants were awarded more money if they could convince the employers that they were actually a top performer (whilst still not knowing their results on the intelligence test, which were provided to them at the end of the experiment).

Of course, there’s two sides of the coin— are people actually fooled by overconfidence?

As with the contestants completing their initial assessments, the employers were asked to essentially bet on who the top performers on the intelligence test were. This meant that the employers also had money at stake when deciding who actually performed well (in other words, their bullshit detectors were well calibrated).

The research team had another trick up their sleeves.

Before the interviews, each contestant received cryptic feedback on their performance on the intelligence test, and were then asked to reevaluate their performance in light of this feedback. This allowed the researchers to see if contestant’s overconfidence influenced the employer’s judgements of them— as their feedback had a direct and measurable impact on the contestant’s confidence levels.

Once the interviews came to a close, employers were given a few seconds where they could scrutinise contestants’ body language and facial expressions. Employers were asked to write down what each contestant said, and how well they think each performed on the intelligence test. The employers were also asked to evaluate how honest, likeable, attractive and confident they thought each contestant was.

In total, 688 German students participated in these series of experiments (410 women and 278 men).

Fooling yourself the better to fool others 

So, what did they find?

Peter and Joël found that overconfident contestants received higher evaluations from the employers. In other words, those who privately bet that their performance was relatively high were seen as higher performers on the intelligence test.

The researchers were able to isolate the impact of contestants receiving a noisy yet positive signal before their interviews (that is, cryptic feedback suggesting superior performance). Contestants who received a positive signal subsequently received higher evaluations from employers– even when controlling for their actual test scores. What this means is that overconfidence caused contestants to receive higher ratings from the employers— independent of their actual performance on the intelligence test.

Finally, the contestants invited to the speed-dating style interviews were noticeably more confident than those who didn’t participate. That is, people became overconfident when they were given the opportunity to profit from persuading others.

In summary, Peter and Joël found that overconfidence pays. The contestants who were more confident received higher ratings from the employers, and therefore won more money. Likewise, they showed that opportunities to profit from persuasion induced overconfidence, and that it was overconfidence which led contestants to receive higher ratings— above and beyond their actual performance.

Overall, these results provide compelling evidence in favour of Robert Triver’s theory of self-deception.

Implications

As is always the case, there are aspects of this paper that can be critiqued.

Although overconfidence is a well established finding from the field of psychology, there is evidence suggesting some cultural differences in how overconfidence manifests.

These experiments were conducted with German university students. However, we know that these people are really WEIRD. That is, they are Western, educated, industrialised, rich and democratic, and therefore may not paint a complete picture of humanity overall… Although we can have confidence in these research findings, we don’t want to become overconfident. 

To add more weight to the study, other recently published papers and preprints point in the same direction. However, it remains the case that these experiments would benefit from more diverse samples.

These points aside, this paper arguably has far reaching implications for business.

What this strand of research suggests is that positive illusions are far more pervasive than is usually portrayed by psychologists.

Evidently, positive illusions offer individuals a wealth of benefits. Those of us who are the most optimistic and overconfident reap many benefits in life, including more career success. As noted by Daniel Kahnmenan in Thinking Fast and Slow, entrepreneurs and business leaders are the most optimistic and overconfident among us.

Positive illusions encourage us to take gambles— gambles which can come with big pay offs. But of course, risky bets can also end in bankruptcy. Overconfidence and excessive optimism can lead financial traders to lose money, CEOs to initiate value-destroying mergers, and lenders to invest in businesses that are built on sand.

The key takeaway for me is this: if positive illusions are not so much a means of protecting us from threatening information, but rather a form of self-deception to help us persuade others, then the problem has been misdiagnosed by psychologists and practitioners alike. Initiatives such as teaching people debiasing techniques as a means of overcoming overconfidence are simply not going to work.

These findings also have obvious implications for recruitment and selection. Indeed, the experiment is inadvertently a parody of the interview process itself.

In light of these findings, it’s not too surprising that interviews are a weak predictor of job performance. These series of experiments mark yet another red cross against job interviews as a recruitment method, and lend credence to the practice of blind assessments.

By coincidence, Peter and Joël concluded their paper by discussing the business implications of their research (emphasis added):

One implication of our findings is that overconfidence is likely to be more prevalent in settings in which its strategic value is highest, that is, in cases in which measures of true ability are noisy, competition is fierce and persuasion is an important part of success. It may arise in employer–employee relationships because of its strategic benefits in job interviews and wage negotiations. Arguably, confidence may be even more valuable among the self-employed, whose economic survival often depends more immediately on persuading investors and customers. We would also expect overconfidence to be rife amongst high-level professionals in finance, law and politics.

Peter and Joël’s reflections are likely to resonate with people working in politics or finance. For those of us working in such fields, vigilance is required. For example, when evaluating candidates for leadership positions, it’s important to gauge whether candidates are overconfident. If they appear to be raising the stakes by promising the world, you have to ask yourself: are you really going to win this bet?


Written by Max Beilby for Darwinian Business

4 thoughts on “The evolutionary logic of overconfidence”

  1. Fascinating piece. “overconfidence .. rife amongst high-level professionals in finance, law and politics” certainly resonates!

    In the Schwardmann/ van der Weele work I wonder what impact gender had? Organisationally we can struggle with the phenomenon of men reaching higher levels of status and pay than sometimes more able women. Structural factors and unconscious bias are offered as causes. I wonder, if “overconfidence pays.. the contestants who were more confident received higher ratings from the employers” whether there was a gender difference measured in this study in (self delusion) confidence?

    Like

  2. Surely the interchanging use of “confidence” & “overconfidence” suggests an element of inaccuracy, or at least a lack of specificity, which must invalidate any study to a degree, but certainly diminishes one (ironically) focussed on duplicity and self-deception.
    Also providing monetary incentive to perform well in interviews would (typically) encourage more effort, but that is as, if not more, likely to encourage a conscious act of subterfuge. Regardless, it does not define a correlating link between reward:self-deception:overconfidence.
    Finally, there should be a distinction made between a level of confidence in one’s abilities and the manner in one presents themselves. I.e. an introvert is reasonably as likely to engage in an internal delusion as an extrovert, though less likely to present as positively in an interpersonal dynamic.

    Like

Leave a comment