Conjunction Fallacy

Two things happening together can end up feeling more likely than either one alone, even though that's mathematically impossible. Adding matching detail to a scenario makes it feel more probable, not less.

This fallacy assigns a higher probability to the conjunction of two events than to a single constituent event, violating basic probability theory. It reflects a heuristic assessment where representativeness overrides the normative rules of probability.

A friend describes Linda: she studied philosophy, cares deeply about social justice, and participated in anti-nuclear protests. Asked which is more likely — that Linda is a bank teller, or that she's a bank teller who's also a feminist activist — most people pick the second option, even though "bank teller AND feminist" can never be more likely than "bank teller" alone.

A junior physician presented with a vignette of a 55-year-old male smoker with mild chest tightness rates the probability of acute myocardial infarction at 0.30, but rates "AMI with concurrent anxiety-driven hyperventilation" at 0.45 — a conjunctive estimate that violates the basic rule that a conjunction can never exceed either of its parts. The elaborated scenario simply matches the physician's mental image of a stressed cardiac patient more tightly, inflating its perceived likelihood above the plain diagnosis. Asking for each probability separately, then multiplying them properly, restores a sane estimate and exposes how much the added detail was distorting the judgment.

Likelihood gets judged by how well a story matches a mental image, not by the actual rules of probability. Adding details that fit the image makes the whole scenario feel truer, which is exactly the error.

A representativeness-driven evaluation uses prototype matching as its comparator, weighting conjunctive cues more heavily than the marginal probabilities actually involved. That constraint produces a systematic overestimation of the conjunction's probability.

Asking for a numerical probability of each event separately, before combining them, is the direct fix. Reminding yourself that a combined event can never be more likely than either part alone keeps the judgment honest.

Eliciting explicit marginal probabilities, and demonstrating the normative bound a conjunction must respect, recalibrates the judgment directly. Decompositional prompts that separate the constituent events reduce the pull of representativeness.

Overrating conjunctions; Ignoring base rates; Confusing plausibility with probability

An adversarial actor can craft richly detailed, stereotype-consistent narratives to make a preferred conjunction (e.g., "a suspect is a radical AND committed the crime") seem more probable than the base-rate event alone, steering jurors, analysts, or policymakers toward conclusions that violate probability norms. In disinformation campaigns, adding emotionally resonant or identity-coherent details to a false claim weaponizes the fallacy by increasing perceived plausibility without increasing actual evidence. Intelligence or financial adversaries can embed conjunctive framing inside scenario briefings or prospectuses to inflate perceived likelihood of a desired outcome and suppress scrutiny of marginal probabilities.

Practitioners should be trained to elicit and record explicit numerical marginal probabilities for each constituent event before evaluating any conjunction, using decompositional prompts that enforce the normative bound P(A∩B) ≤ P(A). Structured analytic techniques such as probability wheels or reference-class forecasting can anchor judgments to base-rate data and counteract prototype-matching heuristics. Red-team or devil's advocate reviews that strip narrative detail from conjunctive scenarios help expose whether elevated probability estimates are driven by representativeness rather than evidence.