Ambiguity aversion
We prefer known risks to unknown ones.
What it means
Ambiguity aversion is the preference for gambles with known probabilities over equivalent gambles whose probabilities are unknown or vague, such that people will pay a premium simply to avoid not knowing the odds. It is distinct from ordinary risk aversion: what repels is uncertainty about the probabilities themselves, not the spread of outcomes, so two bets with the same expected value and risk can be ranked differently purely by how well their odds are specified. The Ellsberg paradox is its defining demonstration, and the pattern is robust enough that no single assignment of subjective probabilities can rationalize the choices, which is why it motivated decision models beyond standard expected utility, such as maxmin and smooth-ambiguity approaches. A nuance is that the attitude is not universal: people can be ambiguity-seeking for low-probability gains or in domains where they feel competent, so 'competence' and framing modulate it. It matters in insurance, investing, medical decision-making, and policy under deep uncertainty, where vague or contested probabilities can drive excessive caution or distort the price people will pay for information.
Why it happens
No single mechanism explains it, and the candidates predict different things. One is comparison: Fox and Tversky found the premium mostly appears when a clear and a vague prospect sit side by side, and largely disappears when either is judged alone, though later work found it reduced rather than eliminated, making it partly contextual rather than a standing taste. A second is competence: Heath and Tversky showed people pay to bet on their own judgment in domains they know, and shun bets that would expose ignorance, pointing at anticipated blame, not probability. A third is suspicion: an undisclosed urn might have been filled by someone who knows. A fourth is arithmetic: Halevy found that those who fail to reduce a two-stage lottery to single-stage odds are largely those who avoid ambiguity, which he reads as recasting it as a reasoning limit.
What the evidence shows
The sharpest demonstration is Ellsberg's three-colour urn: thirty red balls and sixty black or yellow in unknown proportion. Most people bet on red over black, yet also on black-or-yellow over red-or-yellow, and no assignment of odds to black survives both choices. Ellsberg reported colleagues' reactions rather than a controlled experiment, and later lab work complicates the picture. The effect is far stronger in the two-urn design than the one-urn version, so the share of people showing the pattern depends heavily on the setup. Kocher, Lahno and Trautmann replicate aversion for moderate-likelihood gains but find neutrality or ambiguity seeking once losses enter. Binmore, Stewart and Voorhoeve, hunting for indifference points, found the principle of insufficient reason fit their data better, with ambiguity aversion only secondary.
How it is measured
The workhorse is the matching probability: the known chance at which someone is indifferent between the transparent and the ambiguous bet. If it sits below their own best estimate for the vague event, the gap is the ambiguity premium, in probability units and so comparable across people and domains. Willingness-to-pay gaps are easier to collect but mix in risk attitude and scale effects. Careful work splits the attitude in two: a level effect, a general discount on ambiguous bets, and insensitivity, the compression of vague likelihoods toward the middle, so an unlikely and a near-certain ambiguous event look more alike than they are. Someone can be neutral on the first and badly distorted on the second, which is why a single number hides more than it reports.
Using it in practice
For the decider, the question is whether the vagueness carries information. Sometimes it does: a counterparty who will not state its odds may have a reason, and the discount is earned. Often it does not, and treating unmeasured as bad concedes ground to whoever bothered to quantify. The disciplined move is to ask what the unknown number would have to be to flip the decision, which often reveals a wide range still points one way. For the communicator, the comparison is the lever: placing a vague option beside a precise one manufactures much of the penalty, while ranges, base rates and named sources narrow it. The lever cuts both ways: withholding the comparison to make real uncertainty invisible is manipulation, not clarity.
Examples
Offered a bet on a 50/50 urn or an urn of unknown composition, most choose the known one.
Savers take the high-street bank's stated three percent over a lesser-known issuer's four, not because the odds look worse but because nobody will tell them what the odds are.
A hiring manager picks the candidate with a predictable, unspectacular record over the unusual one whose range of outcomes nobody can pin down, even though the unusual one might be far better.
A patient turns down a newer drug whose trials are too small to pin the odds, choosing the decades-old one with a documented complication rate, even though the new drug's midpoint estimate looks better.
An underwriter loads the premium on a flood risk whose models disagree, charging extra for vagueness in the estimate itself rather than for any wider spread of possible losses.
First described in Daniel Ellsberg (1961).
Key references
- Kocher, M. G., Lahno, A. M., & Trautmann, S. T. (2018). Ambiguity aversion is not universal. European Economic Review, 101, 268-283. doi.org/10.1016/j.euroecorev.2017.09.016
- Trautmann, S. T., & van de Kuilen, G. (2015). Ambiguity attitudes. In The Wiley Blackwell Handbook of Judgment and Decision Making (pp. 89-116). Chichester: Wiley. doi.org/10.1002/9781118468333.ch3
- Binmore, K., Stewart, L., & Voorhoeve, A. (2012). How much ambiguity aversion? Journal of Risk and Uncertainty, 45(3), 215-238. doi.org/10.1007/s11166-012-9155-3
- Halevy, Y. (2007). Ellsberg revisited: An experimental study. Econometrica, 75(2), 503-536. doi.org/10.1111/j.1468-0262.2006.00755.x
- Fox, C. R., & Tversky, A. (1995). Ambiguity aversion and comparative ignorance. The Quarterly Journal of Economics, 110(3), 585-603. doi.org/10.2307/2946693
- Heath, C., & Tversky, A. (1991). Preference and belief: Ambiguity and competence in choice under uncertainty. Journal of Risk and Uncertainty, 4(1), 5-28. doi.org/10.1007/BF00057884
- Ellsberg, D. (1961). Risk, ambiguity, and the Savage axioms. The Quarterly Journal of Economics, 75(4), 643-669. doi.org/10.2307/1884324
Where this comes up
- Inside Squid Game: A Behavioral Scientist’s PerspectiveWhat makes Squid Game so compelling? Explore how optimism bias, loss aversion, social proof, and other behavioral scien…