By Emma Young
There will always be some people within a group who are more confident than others. But some groups as a whole tend towards modesty — as with the !Kung hunters of the Kalahari Desert, for example, who deliberately downplay their own achievements and efforts. However, the opposite can also occur — and widespread overconfidence can of course become a problem, as with the US energy company Enron, whose “culture of arrogance” ultimately led to its downfall.
These two examples are highlighted in a new paper published in the Journal of Experimental Psychology, which reveals a route by which a bias towards overconfidence can develop. In their paper, Joey T Cheng at York University and colleagues first propose and then provide evidence for the idea that if we’re exposed to people who are overconfident, this rubs off on us. In other words, we calibrate our self-assessments based on the confidence level of those around us. Overconfidence can, then, be transmitted socially — and this could help to explain how groups, teams and organisations form their own, sometimes drastically different, confidence norms.
The team ran a series of six studies to explore all this. In the first study, involving 104 participants, pairs of previously unacquainted students were put to work, first individually and then together, on a lab task. The task involved guessing the personality traits of 10 people from photographs. After completing the task both times, each participant ranked themselves according to how well they’d thought they’d done, relative to everyone else in the group. The team found that working with someone who’d over-ranked their own performance rubbed off on the other member of the pair — after the joint task, they became more confident in their own ability. (This happened though neither member knew what the other’s self-ranking had been.)
As the researchers write, “after working together, initially non-similar strangers became more similar to each other in terms of over-placement, suggesting the convergence of over-confidence.”
The subsequent five studies explored this effect further experimentally. A study that involved asking participants to guess the weights of people from photographs revealed that overconfidence (or in this study, at least, reduced under-confidence) can be transmitted indirectly, from person to person to person, “highlighting the extensive reach of confident peers”. (In this study, participants were indirectly influenced by a fictitious former partner of their own partner in the weight-guessing task.)
Further work revealed that these confidence effects can persist, still being evident several days later. Importantly, two of the studies produced evidence that the influence of overconfident peers on a participant’s own self-estimations happened largely outside their conscious awareness. As the team writes, if you’re unaware of such a “stealthy” transmission of bias, this could make it harder to resist.
The work also reveals one important qualifier to all these effects: overconfidence transmission occurs only within in-groups. In this case, student participants were influenced by the responses of “partners” who were identified as attending the same university, but not by responses from people identified as coming from a rival university sports team, for example. “That is, individuals do not copy indiscriminately”, the team writes. “Instead they are sensitive to whose mental representations are on display and selectively acquire the over-placement of in-group but not out-group members.” This is consistent with theories about cultural learning.
However, as the researchers themselves point out, these studies focused on over-placement as one form of overconfidence. More work will be needed to investigate whether this kind of social transmission occurs for other forms, such as over-estimation, which relates strictly to your stand-alone performance — the belief that you did better than you actually did in a test, for example. Similarly, it’s not yet clear whether it applies to over-precision (being convinced that you scored at least 80 per cent, for example, when you didn’t.)
Still, the work does contribute to our understanding of how an atmosphere of overconfidence (in the form of over-placement, at least) develops.
The theory doesn’t negate the possibility that there are also cultural effects relevant to overestimation. For example, the researchers write, in the US, the most individualistic society in the world, unusually high levels of overconfidence might be triggered by cultural cues that emphasise success and self-sufficiency — and as those cultural values spread between people, so too can over-confidence.
However, the work still suggests that overconfident beliefs among a few may lead to a cascade-like spread of biased beliefs through a social group, team, or society. A recognition that this happens, will surely be important for all kinds of organisations.