Risk Experts Are Just As Afraid of the Unknown As Everyone Else
Imagine you have to bet on the color of a card drawn from a deck. In the first scenario, you know the deck contains 50 red cards and 50 black cards. In the second, you know nothing about the deck’s composition. Which one do you bet on?
Most people choose the first deck. This behavior—preferring a known risk to an unknown one under equal conditions—is called ambiguity aversion. The debate over its meaning has been ongoing since Daniel Ellsberg described it in 1961: is it a cognitive error or a rational and legitimate preference? This is not an abstract exercise: understanding why we do it—and whether it is rational to do so—changes the way climate policies are designed, health decisions are made, and financial markets are regulated.
One influential explanation argues that ambiguity aversion is, at its core, a problem of complexity: people struggle to manage complex probabilistic situations and avoid them. If this were the case, those who spend their entire careers pricing complex risks should be nearly immune to it.
This is the hypothesis that I, together with Ilke Aydogan and Loïc Berger (IESEG, Université de Lille), have put to the test directly: we went to an international actuarial conference in Berlin and had the participants sit down for an experiment.
The number that shouldn’t be the same
To answer this, we compared two groups with very different profiles: 84 professional actuaries recruited during an international conference in Berlin—average age 40, 13 years of experience, 90% holding at least a master’s degree in mathematics or actuarial science—and 125 social science students at Bocconi, used as a control group.
The result is clear: the two groups show exactly the same level of aversion to ambiguity. Actuaries avoid the unknown just as much as the students—neither more nor less.
But the paper goes further. Among the students, those who struggle with complex probabilistic situations also tend to be more averse to ambiguity: the two behaviors are correlated. Among actuaries, this link disappears: they handle complexity better, but this does not make them less sensitive to the unknown. For actuaries, in short, not knowing how to handle a complex calculation is one thing; distrusting probabilities they do not know is another. Two distinct reactions, with distinct origins.
The result suggests that aversion to ambiguity cannot be reduced to a computational error. If even those trained to reason in probabilistic terms continue to avoid the unknown, then that reaction is not merely a computational problem. It resembles a preference. We do not treat unknown probabilities as difficult probabilities. We treat them as something different.
This distinction is not merely theoretical. In many contexts—from financial decisions to climate policies—choices are made precisely under conditions of ambiguity, not well-defined risk. If models assume that people react to the unknown as if it were a problem of complexity, they risk misrepresenting actual behavior. And thus, the decisions that result from it as well.
