In single evaluation, people prefer an ice cream cup that's overfilled with 7 ounces over a larger cup that's underfilled with 8 ounces. They prefer less ice cream. The moment you show both options side by side, the preference reverses to the objectively larger serving.
The Framework
Preference reversals occur when the order of preference between two options flips depending on how the options are evaluated — singly or jointly, by pricing or by choice, by category assessment or by prototype matching. The ice cream example demonstrates the pattern: in single evaluation, the prototype (a full cup is 'good') dominates. In joint evaluation, the measurable attribute (8 > 7 ounces) dominates. The preferences aren't stable internal states — they're constructed on the fly from whatever evaluation mode is active.
Kahneman's Chapter 33 presents preference reversals as the most philosophically disturbing finding in prospect theory — more disturbing even than framing effects, because framing effects can be attributed to how information is presented, while preference reversals suggest that preferences themselves are constructed rather than discovered. There is no 'true' preference between the two cups — the preference depends on the evaluation context.
Where It Comes From
Chapter 33 of Thinking, Fast and Slow presents multiple types of preference reversals: Hsee's ice cream and dinnerware studies, pricing-vs-choice reversals (people price a gamble higher than a sure thing but choose the sure thing), and the joint-vs-single evaluation reversals that the evaluability hypothesis explains.
> "What you see in joint evaluation is often more important than what you see in single evaluation." — Thinking, Fast and Slow, Ch 33
Cross-Library Connections
Hormozi's bonuses in $100M Offers exploit preference construction: each bonus is evaluated singly (does it sound valuable?) rather than jointly (is the aggregate worth the price?). The single-evaluation mode favors vivid, emotionally resonant bonuses even when their objective value is modest.
The Implementation Playbook
Product Positioning: Understand whether your product will be evaluated alone or alongside competitors, and optimize for that context. Features that dominate in comparative evaluation (measurable superiority) may be invisible in standalone evaluation (where emotional and prototype-based assessment dominates).
A/B Test Design: Recognize that user preferences revealed in A/B tests (joint evaluation — users implicitly compare) may differ from preferences in the live product (single evaluation — users experience one version). This can explain why A/B test winners sometimes underperform in production.
Negotiation: Present offers in the evaluation mode that favors your position. If your offer has measurable advantages, push for side-by-side comparison. If your offer has emotional or prototype advantages, encourage standalone evaluation.
Key Takeaway
Preference reversals prove that preferences are not stable internal states waiting to be discovered — they're constructed in the moment by whatever evaluation mode happens to be active. This means that the context of evaluation is not a neutral container but an active determinant of choice. Whoever designs the evaluation context designs the preference.
Continue Exploring
[[Evaluability Hypothesis]] — The mechanism that produces many preference reversals
[[Framing Effects]] — The broader principle: logically equivalent presentations produce different choices
📚 From Thinking, Fast and Slow by Daniel Kahneman — Get the book