Chapter 15: Linda: Less is More

Type: Chapter 15

Category: Psychology

Difficulty: Intermediate

Reading time: 6 min

Core idea

Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations. Now: which is more probable?

1. Linda is a bank teller.
2. Linda is a bank teller and is active in the feminist movement.

In surveys, 85–90% of respondents chose option 2. This is a logical impossibility. The probability that Linda is a bank teller and a feminist cannot exceed the probability that she is a bank teller — because every person who satisfies both conditions also satisfies the first alone. Adding a conjunction (and) can only reduce or equal probability; it can never increase it.

The conjunction fallacy: representativeness produces a probability judgment that violates the conjunction rule of probability theory. The description of Linda fits the prototype of a feminist activist far better than it fits the prototype of a bank teller. The specific conjunction “bank teller and feminist” is more representative of the description than “bank teller” alone — so it feels more probable. But representativeness and probability are different things, and in this case, they point in opposite directions.

Why it matters

The less-is-more effect

“Linda is a bank teller” is a less detailed, less vivid description — and for that reason it feels less likely. “Linda is a bank teller and a feminist” is a richer, more coherent story that better fits what we know about Linda — and for that reason it feels more likely. This is the less-is-more effect: adding true but confirmatory detail to a description increases its judged probability, even though adding any detail to a conjunction can only reduce or maintain the mathematical probability.

This runs directly against probability theory. A less specific description always includes the more specific one as a subset — it is necessarily more probable. System 1 does not understand this. It evaluates coherence and representativeness, not set inclusion.

The difficulty of correcting the fallacy

Kahneman and Tversky subjected the Linda problem to extensive testing to rule out alternative explanations. Even when participants were reminded of the conjunction rule, many still violated it. Even when the problem was expressed in terms of frequencies (“Of 100 people who fit Linda’s description, how many are bank tellers? How many are feminist bank tellers?”) — a framing that makes set inclusion visible — some participants still chose the conjunction.

The representativeness response is robust even against explicit instruction. This distinguishes it from a mere reasoning error that can be corrected by careful attention — it reflects a deep feature of System 1’s assessment process.

Representativeness in scenarios and forecasting

The conjunction fallacy generalizes beyond the Linda problem. In scenario planning and forecasting, detailed, coherent narratives feel more probable than abstract statements — even though every additional detail in the scenario makes it strictly less probable. A geopolitical scenario that describes a specific chain of events (election → political shift → trade dispute → currency crisis) feels more likely than the simple prediction of a currency crisis — even though the specific chain of events is a conjunction, and every conjunction reduces probability.

Author’s argument: The more detail you add to a scenario, the more coherent and compelling it becomes — and the less probable it is. The persuasiveness of a narrative works against its statistical credibility.

Key takeaways

Key takeaways

• Conjunction fallacy: adding a conjunction (and) to a description can increase its judged probability even though this violates the mathematical rule that P(A and B) ≤ P(A) for all A and B.
• The Linda problem: 85-90% of respondents judge 'feminist bank teller' more probable than 'bank teller' — because the conjunction is more representative of the description, not because it is more probable.
• Less-is-more effect: adding vivid, consistent detail to a description increases its judged probability, even though adding any detail to a conjunction must reduce mathematical probability.
• The fallacy is robust: reminding participants of the conjunction rule, using frequency formats, and other corrections reduce but do not eliminate the effect.
• Scenario planning implication: detailed, coherent scenarios feel more probable than vague ones — narrative coherence is mistaken for statistical probability.
• System 1 evaluates representativeness, not set inclusion — and representativeness and probability come apart whenever descriptions are vivid and prototypically consistent.

Mental model

click to zoom

The conjunction fallacy — how representativeness produces a probability judgment that violates set theory.

Read it as: The description of Linda activates her prototype in System 1. The conjunction “bank teller and feminist” is closer to that prototype than “bank teller” alone. System 1 reads this as greater probability. Mathematics says the conjunction is a subset of the simpler statement and must be less probable. The conflict between logical and intuitive probability is real and robust — most people resolve it in favor of representativeness.

Practical application

Scenarios and business planning

Every business planning process that generates detailed scenarios is potentially subject to the conjunction fallacy. A scenario with six specific elements — market growth, regulatory shift, competitor exit, consumer trend, technology development, and management execution — is a six-way conjunction. Each element must independently come true. The more vividly each element is described, the more coherent and probable the scenario feels. But statistically, each added element reduces the probability of the whole. Challenge vivid scenarios by counting their conjunctive requirements.

Practical defenses:

• Count conjunctions: before accepting a “likely scenario,” count the independent events or conditions that must all be true simultaneously. Each adds a multiplication factor that reduces probability.
• Generate the failure paths: for every scenario, actively generate the most plausible way each element fails. This counteracts the coherence effect that makes detailed scenarios feel inevitable.
• Use frequency formats: rephrase probability questions in terms of frequencies (“Out of 100 people fitting this description, how many…?”) — this makes set inclusion more visible and reduces the fallacy rate.

Example

An analyst presents a market entry forecast: “Company X will enter the market, adopt a premium pricing strategy, and successfully capture 15% market share within 18 months.” Each element sounds reasonable. The room agrees it is “quite likely.”

But the probability of the full scenario is the product of:

• P(Company X enters) × P(They choose premium pricing given entry) × P(They achieve 15% share given premium pricing)

If each is 50%, the conjunction is 12.5%. The more detail added to make the scenario vivid and coherent, the lower its true probability — and the higher its apparent probability. The analyst has confused narrative quality with statistical likelihood. The correct question is not “Is this a good story?” but “What is the probability of each element, and what is the product?”

Related lessons

Related

Chapter 14: Tom W's Specialty The representativeness heuristic — similarity to prototype substitutes for base rate probability, the foundation of the conjunction fallacy. Chapter 16: Causes Trump Statistics Statistical base rates are overridden by causal narratives — a broader pattern of which the conjunction fallacy is a specific instance. Representativeness The full concept — prototype matching as a substitute for statistical reasoning, and its consequences.