Skip to content

Conjunction Fallacy

Detect a probability-judgment error by watching for a more detailed scenario being rated more probable than the simpler scenario it is a strict subset of — the signature of resemblance quietly standing in for probability.

Version
v5 · 2026-09-28 · History · 2 corrections
Domain-specific #
198
Domain group
Social Sciences
Origin domain
Psychology & Behavioral Sciences
Subdomain
Probability Judgment Biases → Psychology & Behavioral Sciences

Core Idea

The conjunction fallacy is the systematic human violation of the rule that a conjunction P(A and B) can never exceed either component P(A) or P(B). Given a description that fits a stereotype, people rate a specific compound event as more probable than one of its own constituents. The mechanism is representativeness substitution: the judge silently answers "how well does this match the description?" instead of "how probable is this?", and the detailed version usually matches better.

How would you explain it like I'm…

Two Things Can't Beat One

Imagine a box of animal drawings. There can never be more drawings of "a dog wearing a hat" than drawings of "a dog," because every hat-dog is also a dog. But if someone tells you a story about a fancy dog, you might guess "dog wearing a hat" is more likely, just because it fits the story. That mistake is the conjunction fallacy.

When Longer Stories Seem Likelier

The chance that two things are both true can never be bigger than the chance that just one of them is true. The conjunction fallacy is when people get this backward. In a famous test, people read about a woman named Linda who sounded like someone who cares about women's rights. Many said "Linda is a bank teller and a feminist" was more likely than "Linda is a bank teller," even though every feminist bank teller is a bank teller. They were really judging which sentence fit the description better, without noticing that's a different question.

Resemblance Mistaken for Probability

The conjunction fallacy is a systematic error in judging probability: people rate a combination of events, A and B, as more likely than one of its parts, A, which breaks the rule that P(A and B) can never exceed P(A). Tversky and Kahneman (1983) showed it with the Linda problem, where people ranked "bank teller and active feminist" above "bank teller" after reading a description fitting the feminist stereotype. The cause is representativeness substitution: without realizing it, people answer "how well does this match the description?" instead of "how probable is this?" The combined statement matches better because the added detail fits the stereotype. The effect isn't just a misunderstanding of wording or carelessness; it appears in frequency formats (though smaller), with explicit subset instructions, with incentives, and in trained experts. It is strongest when A alone doesn't fit the description but A and B together fit it well.

 

The conjunction fallacy is the systematic violation, in human probability judgment, of the conjunction rule P(A and B) <= min(P(A), P(B)): people judge a specific compound event more probable than one of its constituents when the compound is more representative of a given description. Tversky and Kahneman (1983) demonstrated it with the Linda problem, in which subjects given a description matching a politically engaged feminist stereotype ranked "bank teller and active in the feminist movement" above "bank teller," though the former is a strict subset of the latter. The mechanism is representativeness substitution: the probability question is silently replaced by a resemblance question, and because the conjunction adds a stereotype-consistent detail, it resembles the description more, inverting the correct probability order. The substitution is not conscious; the judge experiences answering a probability question. The effect is not an artifact of linguistic ambiguity, since it survives frequency formats (at reduced magnitude) and explicit subset instructions, nor of carelessness, since it survives incentives and appears in trained researchers judging within their own domains. Its magnitude is greatest when the constituent A is low in representativeness and the conjunction A and B is high, maximizing the gap between resemblance ranking and probability ranking.

Scope of Application

The fallacy lives in one domain — human probability judgment — and the settings below are application contexts of that single substrate, not distinct systems.

  • Judgment-under-uncertainty research — the canonical Linda problem and its many replications across formats.
  • Forensic, clinical, and intelligence judgment — fleshed-out threat or diagnostic scenarios rated above their own components.
  • Behavioral economics — compound-risk evaluation entering prospect-theory and ambiguity-aversion accounts.
  • Forecasting and scenario planning — the applied corrective of penalizing vivid detail and pricing each component separately.

Clarity

Naming the fallacy pins a normative anchor — the conjunction rule, P(A and B) cannot exceed P(A) — onto judgments where intuition supplies no such constraint, turning a vague "the detailed story felt likelier" into a checkable axiom violation. More deeply, it separates two things the introspecting mind cannot tell apart: a probability judgment and a resemblance judgment. The error becomes not a lapse of logic but a competent answer to the wrong question.

Manages Complexity

Checking the conjunction rule against every multi-clause forecast, diagnosis, or threat scenario would mean recomputing joint probabilities case by case. The fallacy compresses that open audit to one content-free ordering test: does adding detail raise the rated probability of a scenario the detail makes a strict subset? Whenever it does, representativeness has displaced probability — collapsing many vivid stories to one regularity and one corrective family, with no numbers needed in the moment.

Abstract Reasoning

The concept licenses a diagnostic move — reading a more-specific-rated-above-its-superset ordering as the fingerprint of attribute substitution, needing neither stimulus content nor the numbers. It supports interventionist moves that predict what reduces the inversion: decompose the conjunction into constituents, reframe in natural frequencies, and penalize rather than reward vivid detail. Boundary-drawing separates it from base-rate neglect and mere narrative plausibility.

Knowledge Transfer

Within human probability judgment the fallacy transfers intact as mechanism: the ordering test and its corrective family (decompose, frequency-frame, penalize detail) carry without translation across forensic, clinical, intelligence, and behavioral-economic judgment, because the same resemblance-displaces-probability machinery is at work. Beyond the human judge it does not travel — a sound estimator commits no fallacy. What generalizes is the parent, probability and its conjunction axiom (the theorem any estimator obeys), plus representativeness substitution as the specifically human mechanism. dag_edges:

Relationships to Other Abstractions

Local relationship map for Conjunction FallacyParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Conjunction FallacyDOMAINDomain-specific abstraction: Representativeness Heuristic — includesRepresentativen…DOMAINPrime abstraction: Probability — presupposesProbabilityPRIMEPrime abstraction: Bias — is a kind ofBiasPRIME

Current abstraction Conjunction Fallacy Domain-specific

Parents (3) — outgoing typed relations

  • Conjunction Fallacy is a kind of Bias Prime

    Conjunction fallacy is bias specialized to a systematic probability ranking displaced toward representative detail and away from set inclusion.

  • Conjunction Fallacy includes Representativeness Heuristic Domain-specific

    Conjunction fallacy contains representativeness because the more stereotype-matching scenario is ranked by resemblance instead of probability.

  • Conjunction Fallacy presupposes Probability Prime

    Conjunction fallacy presupposes probability because its decisive error is assigning a strict subset more probability than an event that contains it.

Hierarchy paths (5) — routes to 5 parentless roots

  • Conjunction Fallacy → is a kind of Bias

Neighborhood in Abstraction Space

Conjunction Fallacy sits in a sparse region of the domain-specific corpus (63rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Rhetorical Fallacies & Persuasion Tactics (24 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25