Cognitive Biases · CB-31
People systematically prefer a known risk over an unknown one — even when the unknown option is, by the numbers, no worse and sometimes better.
A preference for known risks over unknown ones, such that people will choose an option with clearly stated odds over an option with the same or even better expected value but ambiguous, unstated probabilities — a distinct phenomenon from ordinary risk aversion, which concerns variance in outcomes rather than uncertainty about the probabilities themselves.
Formally demonstrated by Daniel Ellsberg in his 1961 'Ellsberg paradox' experiments, where participants strongly preferred betting on a known 50/50 urn over an urn with an unknown mix of colored balls, even when the ambiguous urn offered identical or better expected odds.
The Mechanism
A known 50/50 beats an unknown mix, even at equal odds
Ellsberg's participants weren't behaving as ordinary expected-value or even risk-averse reasoning would predict — the known urn and the ambiguous urn had identical expected payoffs by construction, yet the known urn was strongly preferred, revealing that people specifically dislike not knowing the odds, distinct from disliking risk itself.
01 · IT'S DISTINCT FROM RISK AVERSION — AMBIGUITY IS ABOUT NOT KNOWING THE ODDS AT ALL
Risk aversion concerns known probabilities with uncertain outcomes; ambiguity aversion concerns unknown probabilities
Standard risk aversion is a preference for a certain smaller gain over a risky larger expected gain when the odds are known — ambiguity aversion is a separate phenomenon specifically about disliking situations where the probabilities themselves are unknown or unstated, even holding expected value constant.
02 · IT HELPS EXPLAIN WHY NOVEL OPTIONS ARE SYSTEMATICALLY UNDERVALUED
A well-understood but unproven old option can beat a promising but ambiguous new one, purely due to ambiguity
New technologies, unfamiliar markets, and untested strategies often carry genuinely comparable or better expected value than established alternatives, but face a systematic uphill battle purely because their true odds of success are less well specified — a real, empirically documented barrier to adopting genuinely good new options.
03 · IT CAN BE PARTIALLY REDUCED BY MAKING THE UNCERTAINTY EXPLICIT AND BOUNDED
Structuring an ambiguous choice with a stated range of probabilities reduces the aversion
Research finds that presenting an ambiguous option with an explicitly stated probability range (rather than leaving it entirely unstated) meaningfully reduces the aversion, suggesting the aversion is driven specifically by the total absence of stated odds, not merely by genuine underlying uncertainty in the world.
Where It Fails / Inversion
Where it fails / inversion
Avoiding ambiguous options is not always irrational — genuine additional risk (fraud potential, unknown unknowns, lack of any track record) often does accompany ambiguous situations in the real world, unlike the artificially constructed, provably-equal-odds Ellsberg urns. The bias specifically concerns cases where the true expected value is actually equal or better despite the ambiguity, not a blanket argument that ambiguous options are always just as good.
How To Use It
Worked example · evaluating a new market opportunity versus a familiar one
A company deciding between expanding into a well-understood existing market (known competitive dynamics, known customer behavior) and a new, unfamiliar market with genuinely comparable underlying opportunity should explicitly separate the question of 'how good is this opportunity' from 'how averse am I to not knowing the exact odds' — otherwise, ambiguity aversion alone, rather than a fair comparison of expected value, may be driving the decision toward the familiar option.
How to use it
When you notice yourself strongly preferring a familiar option over an unfamiliar one, explicitly check whether that preference is based on a genuine difference in expected value, or purely on disliking that the unfamiliar option's true odds aren't as clearly stated. If the latter, consider whether more research could make the ambiguous option's real odds clearer, rather than avoiding it by default.
See Also