Boyer-Kassem et al.'s conjunction-fallacy refutation rejects both the non-degenerate and the general (degenerate) class of quantum-like models

claim-quantum-like-models-fail-conjunction-fallacy records the vault's existing headline finding from Boyer-Kassem, Duchêne & Guerci (2016, arXiv:1606.03430; Theory and Decision 81:479–510) — that observed behavior came out "at odds with the predictions of the quantum-like models." Reading the paper's own hypothesis structure sharpens which models that headline actually covers.

The paper tests three nested hypotheses across seven experimental tasks (varying scenario, administration method, and payment). Hypothesis #1 — the simplest "non-degenerate" model variant, the version defended by Franco (2009) — is rejected outright. The paper's more consequential result is the rejection of Hypothesis #2: the broader claim, made across the field's own leading papers (Busemeyer et al. 2011, Busemeyer & Bruza 2012, Pothos & Busemeyer 2013, Busemeyer et al. 2015), that either degenerate or non-degenerate quantum-like models can account for the conjunction fallacy. Using Bonferroni-corrected (statistically conservative) tests, the authors reject this general hypothesis in at least three of the seven tasks. So the refutation is not confined to one narrow model variant that a defender could disown as unrepresentative — it reaches the general model class the field's central theorists actually use.

This distinguishes the finding from a typical "we didn't replicate X" null result: it targets the load-bearing claim of the research program, not a strawman edge case, and does so with a statistically conservative test designed to make the rejection hard to explain away.