The pseudodiagnosticity trap: Should participants consider alternative hypotheses?

摘要: The pseudodiagnosticity task has been used as an example of the tendency on part participants to incorrectly assess Bayesian constraints in assessing data, and a failure consider alternative hypotheses probabilistic inference task. In task, are given one value, anchor corresponding P(D1|H) may choose other either P(D1|¬!H), P(D2|H), or P(D2|not;!H). Most select P(D2|¬!H) which have considered inappropriate (and called pseudodiagnostic) because only P(D1|¬!H) allows use Bayes' theorem. We present new analysis based probability intervals show that selection is fact pseudodiagnostic, whereas choice diagnostic. Our shows pseudodiagnostic values actually increases uncertainty regarding posterior H, supporting original interpretation experimental findings argument...