Loading...
In 1983, psychologists Daniel Kahneman and Amos Tversky presented 142 undergraduates at the University of British Columbia with a simple description: '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.' Students then ranked eight statements by probability. The critical pair: 'Linda is a bank teller' versus 'Linda is a bank teller and is active in the feminist movement.' A staggering 85% ranked the conjunction — feminist bank teller — as MORE probable than bank teller alone. But this is logically impossible. The set of feminist bank tellers is entirely contained within the set of bank tellers. Every feminist bank...
Popular framing: People who picked the conjunction just weren't thinking carefully; logic would have caught it.
Structural analysis: Representativeness substitutes 'how well does Linda fit the story?' for 'how probable is this category?' — the rich narrative detail acts as a base-rate trap that diverts the calculation away from the relevant subset relationship. Availability of the feminist-activist image dominates the conjunction-probability axiom even when subjects are shown the explicit logical structure; expertise in decision science does not immunize. The failure is in the substitution heuristic itself; the architecture of intuitive probability is systematically broken, not the individual reasoner's effort.
Framing the Linda Problem as proof of irrationality makes it a curiosity rather than a structural insight. If we instead see it as evidence of a dual-system architecture with predictable failure modes, it becomes actionable: the question shifts from 'how do we shame people into being rational?' to 'how do we design decision environments that engage the deliberate system before the representativeness heuristic locks in an answer?' The gap matters because debiasing efforts built on the irrationality frame consistently underperform.
