JH
Jonathan Haber

Test each component probability separately

Before judging a joint claim, estimate each element on its own, then check whether the conjunction is lower.

Key takeaways

  • What it is: Before judging a joint claim, estimate each element on its own, then check whether the conjunction is lower.
  • Why it works: The conjunction fallacy arises because the mind evaluates scenarios by how well they fit a mental prototype, not by the arithmetic of probability. Forcing yourself to assign a probability to each component independently imposes the logical constraint that P(A and B) ≤ P(A) — a rule our intuition routinely violates when B makes A "make more sense." Explicit decomposition activates deliberate reasoning and dislodges the representativeness anchor.
  • Evidence: Backed by observational / correlational evidence.
  • Avoid: Decomposing the conjunction into components but still letting the narrative connection between them inflate the joint estimate — the components must be assessed truly independently.

Why it works

The conjunction fallacy arises because the mind evaluates scenarios by how well they fit a mental prototype, not by the arithmetic of probability. Forcing yourself to assign a probability to each component independently imposes the logical constraint that P(A and B) ≤ P(A) — a rule our intuition routinely violates when B makes A "make more sense." Explicit decomposition activates deliberate reasoning and dislodges the representativeness anchor.

How to do it

  1. 1Identify the conjunction: write out the two or more claims being combined ("she is a teller AND an activist").
  2. 2Estimate the probability of each element independently, without letting one inform the other.
  3. 3Apply the logical constraint: the joint claim cannot be more likely than the least likely element.
  4. 4If your initial intuition violates this, treat that gap as a prompt to revisit the evidence.

What the evidence says

Observational

Tversky and Kahneman (1983) demonstrated the conjunction fallacy across many populations and scenarios, including medically trained subjects; the error is robust and not explained by misunderstanding the question.

Honest caveat: Some researchers argue that subjects interpret "probable" as "plausible" or that conversational pragmatics shift the meaning; the core finding is nonetheless replication-stable.

References
  • — Tversky & Kahneman (1983), Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment, Psychological Review

Common mistake

Decomposing the conjunction into components but still letting the narrative connection between them inflate the joint estimate — the components must be assessed truly independently.

IX Coach surfaces conjunction structures in your reasoning ("I assume X because it fits with Y and Z") and prompts you to evaluate each element before treating the package as likely.

Practice this with IX Coach →

Related practices