Metraipolis › archive › dc_780cf4a346ba
(How) Do MLLMs Report Bistable Images Like Humans? — arXiv:2609.13254
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.13254
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T22:48:18.029Z
- SHA-256 of the archived text
294bd4a39bd94217921f48146df77e58b5a4c2faab373e021188799f0f5b3655- Stable URL of this page
https://metraipolis.com/documento/dc_780cf4a346ba
How to cite
(How) Do MLLMs Report Bistable Images Like Humans? — arXiv:2609.13254 — arXiv (en), https://arxiv.org/abs/2609.13254 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_780cf4a346ba (Metraipolis archive dc_780cf4a346ba, sha256 294bd4a39bd94217921f48146df77e58b5a4c2faab373e021188799f0f5b3655)
BibTeX
@misc{dc780cf4a346ba,
title = {(How) Do MLLMs Report Bistable Images Like Humans? — arXiv:2609.13254},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.13254},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_780cf4a346ba; sha256 294bd4a39bd94217921f48146df77e58b5a4c2faab373e021188799f0f5b3655},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.13254. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Bistable images such as the duck-rabbit are classic stimuli in which one image supports multiple mutually incompatible interpretations, typically reported one at a time in humans. We ask whether multimodal large language models (MLLMs) show similar report behavior and what internal computations support it. Using the LLaVA family, we study two tractable dimensions: modulability, whether reports can be biased by bottom-up visual cues and top-down linguistic priors, and exclusivity, whether responses commit to a single interpretation. We test both on the canonical duck-rabbit and on synthetic Visual Anagrams to mitigate memorization confounds. Behaviorally, both visual and linguistic manipulations systematically shift reports in human-consistent ways, while responses remain predominantly exclusive. Mechanistically, these effects arise from competing image-token representations, distinct pathways for bottom-up and top-down modulation, and a link between exclusive reporting and object-count encoding. Code and data are available at this https URL . Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.AI). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.13254 — 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 2026-10-15T22:29:03.423Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).