Tversky & Kahneman 1973 (Cognitive Psychology) — availability heuristic; why "once burned" is NOT just recency bias

Claim: Tversky and Kahneman's 1973 paper "Availability: A Heuristic for Judging Frequency and Probability" established the availability heuristic — and thereby what is usually called recency bias: a general overweighting of recently-encountered information when estimating frequency or probability.

Source: Tversky, A., & Kahneman, D. (1973). Cognitive Psychology 5(2): 207–232.

Confidence: Verified.

For Candid — the "once burned" pattern is more specific than recency: A GC who has been burned does not simply remember the last vendor more vividly. He performs a structural reweighting of the probability estimate of vendor failure across the entire category, combined with a shift in the decision rule from expected-value to maximin (avoid worst case). Three distinct mechanisms beyond availability:

1. Disposition effect / narrow framing over realized losses — Shefrin & Statman 1985 (JoF) — disposition effect; narrow-framing / mental accounting of each vendor as separate account
2. Asymmetric trust updating after competence vs integrity violations — Kim, Ferrin, Cooper, Dirks 2004 (JAP) — competence violations repair via apology; integrity violations repair via denial; opposite mechanisms
3. Negativity dominance over arithmetic averaging — Rozin & Royzman 2001 (PSPR) — negativity bias / negativity dominance; bad weighs more than the algebraic sum of equivalent goods

Treating once-burned as "just recency bias" leads to the wrong intervention. Recency would predict the effect fades with time; the underlying mechanisms do not predict that.