Metraipolis › archive › dc_7fb76ad98777

MedQA-MM: Shortcuts Behind Medical Visual Reasoning — arXiv:2609.03261

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.03261
License (as archived)
CC0-1.0
Archived
2026-10-17T01:41:10.532Z
SHA-256 of the archived text
cc27b72e71d7e3132d5ca0111a6d585a066237b1008c1232ace2125f6dc1701e
Stable URL of this page
https://metraipolis.com/documento/dc_7fb76ad98777

How to cite

MedQA-MM: Shortcuts Behind Medical Visual Reasoning — arXiv:2609.03261 — arXiv (en), https://arxiv.org/abs/2609.03261 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7fb76ad98777 (Metraipolis archive dc_7fb76ad98777, sha256 cc27b72e71d7e3132d5ca0111a6d585a066237b1008c1232ace2125f6dc1701e)
BibTeX
@misc{dc7fb76ad98777,
  title = {MedQA-MM: Shortcuts Behind Medical Visual Reasoning — arXiv:2609.03261},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.03261},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7fb76ad98777; sha256 cc27b72e71d7e3132d5ca0111a6d585a066237b1008c1232ace2125f6dc1701e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.03261. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A benchmark score credits final answers, but not the route by which an item can be answered. In medical multimodal multiple-choice questions (MCQs), this distinction matters because a correct answer can be supported by the intended image finding or by benchmark-preserved cues in the wording of answers, non-visual clinical text, visible image text, artificial annotations, or device/context artifacts. We call the resulting score-level overinterpretation reasoning inflation. Here, a route is an observable input path that can support answer selection, not a claim about the model's hidden cognition. Across six medical multimodal MCQ datasets, we separate candidate cues from behavioral evidence through prompt- and image-side audits, modality ablations, and matched repairs that preserve the medical target and answer key. In a 13-configuration open-model panel, full-input accuracy is 62.63%, while text-only and options-only settings achieve 53.96% and 29.71%, respectively. Removing length-gap, absolute/conspicuous, and spatial/prepositional cues lowers accuracy by 6.58, 3.50, and 4.77 percentage points. We also construct MedQA-MM, a 1,000-item shortcut-mitigated subset, where text-only and options-only accuracy fall to 5.21% and 12.33%. This does not imply that models never use images; it shows that medical image-reasoning claims require route-level evidence.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-04 in cs.CL).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.03261 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T01:11:43.590Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).