Elamur Research⌕
☰
home / papers / 2026 / Bears, all bears, and some bears. Language Constraints on Language Models' Inductive Inferences

Bears, all bears, and some bears. Language Constraints on Language Models' Inductive Inferences

Paper · 2026 · Sriram Padmanabhan, Siyuan Song and Kanishka Misra

Summary

The paper asks whether vision-language models (VLMs) treat three kinds of statements differently when generalizing a new property to an individual: universal statements ("all bears"), bare-plural generics ("bears"), and indefinite plurals ("some bears"). Gelman et al. (2002) found that children aged 4 and older, and adults, follow a graded pattern: all > generic > some. The authors replicate that experiment with VLMs. First, they run two prerequisite tests that mirror the original pretests: robust object-category recognition in images, and basic understanding of all and some. For the second test they build a new benchmark in both text and image+text formats. Ten VLMs were screened, and at least two passed both prerequisites. These models showed the same ordering of inductive generalization as humans. The authors also analyze premise-only representations and find that the three statement types fall into separate regions. This separation holds when surface wording varies, for example every instead of all. The work matters because it links linguistic form to inferential function in models and could help explain how generics are learned and processed.

Key points

  • Models and measure: 10 VLMs were tested, including Idefics3-8B, LLaVA-1.5/1.6/OneVision, Molmo 7B, Qwen2.5-VL 7B, Qwen3-VL 2/4/8B and SmolVLM, all accessed via the minicons library. Answers were scored as the relative probability of Yes versus No, with case and leading-space variants pooled.
  • Category identification test: Images came from THINGS, using 1222 of its 1854 categories with 5 images each. Each correct (positive) question was paired with three kinds of wrong categories: similar, taxonomic and random. A model got credit only if it was correct on all images for a category. The best joint scores were about 66.2 for Qwen2.5-VL-7B and 65.2 for Qwen3-VL-8B. LLaVA-1.5-7B scored only 15.0.
  • New all/some benchmark: It is inspired by Smith (1980) and has a few hundred items. The images were generated with Gemini 2.5 and checked by hand, and roughly 30% of generated image pairs were discarded.
  • Representation analysis: Low-dimensional analyses of premise representations alone grouped statements by their inductive strength rather than by wording. The authors note this is correlational and not a causal finding.
  • Limitations: The study uses English only, and the all/some benchmark is small. The models were trained on far more data than humans receive. Code is released on GitHub (kanishkamisra/inductive-constraints).

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksExperimental and Cognitive Psychology · 870 works

🧭 Topics 1Research subjects and languages the work is about.Language models & pretraining · 3,320 works

Cite

Cite as Sriram Padmanabhan, Siyuan Song and Kanishka Misra. 2026. Bears, all bears, and some bears. Language Constraints on Language Models' Inductive Inferences. Findings of the Association for Computational Linguistics: ACL 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.findings-acl.1842

@inproceedings{padmanabhan2026bears,
  title = {{Bears, all bears, and some bears. Language Constraints on Language Models' Inductive Inferences}},
  author = {Sriram Padmanabhan, Siyuan Song and Kanishka Misra},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1842},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1842}
}

References

51 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Emily Allaway, Chandra Bhagavatula, Jena D Hwang, Kathleen McKeown, and Sarah-Jane Leslie. 2024. Exceptions, instantiations, and overgeneralization: Insights into how language models process generics. Computational Linguistics, 50(4):1211–1275. find
  2. Emily Allaway, Jena D. Hwang, Chandra Bhagavatula, Kathleen McKeown, Doug Downey, and Yejin Choi. 2023. Penguins don’t fly: Reasoning about generics through instantiations and exceptions. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics,… find
  3. Emily M. Bender and Alexander Koller. 2020. Climbing towards NLU: On meaning, form, and understanding in the age of data. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 5185–5198, Online. Association for Computational Linguistics.
  4. Sudeep Bhatia. 2023. Inductive reasoning in minds and machines. Psychological Review. find
  5. Rosa Cao and Daniel Yamins. 2024a. Explanatory models in neuroscience, part 1: Taking mechanistic abstraction seriously. Cognitive Systems Research, 87:101244. find
  6. Rosa Cao and Daniel Yamins. 2024b. Explanatory models in neuroscience, part 2: Functional intelligibility and the contravariance principle. Cognitive Systems Research, 85:101200. find
  7. Gustavo Cilleruelo, Emily Allaway, Barry Haddow, and Alexandra Birch. 2025. Generics are puzzling. can language models find the missing piece? In Proceedings of the 31st International Conference on Computational Linguistics, pages 6571–6588, Abu Dhabi, UAE. Association for Computational Linguistics. find
  8. Andrei Cimpian and Ellen M Markman. 2008. Preschool children’s use of cues to generic meaning. Cognition, 107(1):19–53. find
  9. Claudia Collacciani, Giulia Rambelli, and Marianna Bolognesi. 2024. Quantifying generalizations: Exploring the divide between human and LLMs’ sensitivity to quantification. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 11811… find
  10. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, and 1 others. 2025. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arX…
  11. Matt Deitke, Christopher Clark, Sangho Lee, Rohun Tripathi, Yue Yang, Jae Sung Park, Mohammadreza Salehi, Niklas Muennighoff, Kyle Lo, Luca Soldaini, and 1 others. 2025. Molmo and pixmo: Open weights and open data for state-of-the-art vision-language models. In Proceedings of the Computer Vision and…
  12. Chaz Firestone. 2020. Performance vs. competence in human–machine comparisons. Proceedings of the National Academy of Sciences, 117(43):26562– 26571. find
  13. Stella Frank and Emily Allaway. 2025. VISaGE: Understanding visual generics and exceptions. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 32537–32546, Suzhou, China. Association for Computational Linguistics. find
  14. Richard Futrell and Kyle Mahowald. 2025. How linguistics learned to stop worrying and love the language models. arXiv preprint arXiv:2501.17047. — [arXiv:2501.17047](https://arxiv.org/abs/2501.17047) find
  15. Atticus Geiger, Zhengxuan Wu, Christopher Potts, Thomas Icard, and Noah Goodman. 2024. Finding alignments between interpretable causal variables and distributed neural representations. In Causal Learning and Reasoning, pages 160–187. PMLR. find
  16. Susan A Gelman. 2004. Learning words for kinds: Generic noun phrases in acquisition. Weaving a lexicon, pages 445–484. find
  17. Susan A Gelman, Peggy J Goetz, Barbara W Sarnecka, and Jonathan Flukes. 2008. Generic language in parent-child conversations. Language Learning and Development, 4(1):1–31. 36971 10 find
  18. Susan A Gelman, Jon R Star, and Jonathan Flukes. 2002. Children’s use of generics in inductive inferences. Journal of Cognition and Development, 3(2):179– 199. find
  19. Kyle Mahowald, Anna A Ivanova, Idan A Blank, Nancy Kanwisher, Joshua B Tenenbaum, and Evelina Fedorenko. 2024. Dissociating language and thought in large language models. Trends in cognitive sciences. find
  20. Simon Jerome Han, Keith J Ransom, Andrew Perfors, and Charles Kemp. 2024. Inductive reasoning in humans and large language models. Cognitive Systems Research, 83:101155. find
  21. Andrés Marafioti, Orr Zohar, Miquel Farré, Merve Noyan, Elie Bakouch, Pedro Cuenca, Cyril Zakka, Loubna Ben Allal, Anton Lozhkov, Nouamane Tazi, Vaibhav Srivastav, Joshua Lochner, Hugo Larcher, Mathieu Morlon, Lewis Tunstall, Leandro von Werra, and Thomas Wolf. 2025. Smolvlm: Redefining small and ef… find
  22. Martin N Hebart, Oliver Contier, Lina Teichmann, Adam H Rockter, Charles Y Zheng, Alexis Kidder, Anna Corriveau, Maryam Vaziri-Pashkam, and Chris I Baker. 2023. Things-data, a multimodal collection of large-scale datasets for investigating object representations in human brain and behavior. eLife, 1… find
  23. Martin N Hebart, Adam H Dickter, Alexis Kidder, Wan Y Kwok, Anna Corriveau, Caitlin Van Wicklin, and Chris I Baker. 2019. Things: A database of 1,854 object concepts and more than 26,000 naturalistic object images. PloS one, 14(10):e0223792. find
  24. Michelle A Hollander, Susan A Gelman, and Jon Star. 2002. Children’s interpretation of generic noun phrases. Developmental psychology, 38(6):883. find
  25. Philipp Kaniuth, Florian P Mahner, Jonas Perkuhn, and Martin N Hebart. 2024. A high-throughput approach for the efficient prediction of perceived similarity of natural objects. bioRxiv, pages 2024–06. find
  26. Yair Lakretz, Dieuwke Hupkes, Alessandra Vergallito, Marco Marelli, Marco Baroni, and Stanislas Dehaene. 2021. Mechanisms for handling nested dependencies in neural-network language models and humans. Cognition, 213:104699. find
  27. Andrew Kyle Lampinen, Stephanie CY Chan, Yuxuan Li, and Katherine Hermann. 2025. Representation biases: will we achieve complete understanding by analyzing representations? arXiv preprint arXiv:2507.22216. — [arXiv:2507.22216](https://arxiv.org/abs/2507.22216) find
  28. Hugo Laurençon, Andrés Marafioti, Victor Sanh, and Léo Tronchon. 2024. Building and better understanding vision-language models: insights and future directions. Preprint, arXiv:2408.12637. — [arXiv:2408.12637](https://arxiv.org/abs/2408.12637) find
  29. Sarah-Jane Leslie. 2008. Generics: Cognition and acquisition. Philosophical review, 117(1):1–47. find
  30. Bo Li, Yuanhan Zhang, Dong Guo, Renrui Zhang, Feng Li, Hao Zhang, Kaichen Zhang, Peiyuan Zhang, Yanwei Li, Ziwei Liu, and Chunyuan Li. 2025. LLaVA-onevision: Easy visual task transfer. Transactions on Machine Learning Research.
  31. Michael McCloskey. 1991. Networks and theories: The place of connectionism in cognitive science. Psychological science, 2(6):387–395. find
  32. Kanishka Misra. 2022. minicons: Enabling flexible behavioral and representational analyses of transformer language models. arXiv:2203.13112. find
  33. Kanishka Misra, Allyson Ettinger, and Julia Rayz. 2021. Do language models learn typicality judgments from text? In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 43. find
  34. Kanishka Misra and Najoung Kim. 2024. Generating novel experimental hypotheses from language models: A case study on cross-dative generalization. arXiv preprint arXiv:2408.05086. — [arXiv:2408.05086](https://arxiv.org/abs/2408.05086) find
  35. Kanishka Misra, Julia Rayz, and Allyson Ettinger. 2022. A property induction framework for neural language models. In Proceedings of the Annual Meeting of the Cognitive Science Society, volume 44. find
  36. Kanishka Misra, Julia Rayz, and Allyson Ettinger. 2023. COMPS: Conceptual minimal pair sentences for testing robust property knowledge and its inheritance in pre-trained language models. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, p… find
  37. Gregory Murphy. 2004. The big book of concepts. MIT press. find
  38. Daniel N Osherson, Edward E Smith, Ormond Wilkie, Alejandro Lopez, and Eldar Shafir. 1990. Category-based induction. Psychological review, 97(2):185. find
  39. Sandeep Prasada. 2000. Acquiring generic knowledge. Trends in cognitive sciences, 4(2):66–72. find
  40. Yulu Qin, Dheeraj Varghese, Adam Dahlgren Lindström, Lucia Donatelli, Kanishka Misra, and Najoung Kim. 2025. Vision-and-language training helps deploy taxonomic knowledge but does not fundamentally alter it. In The Thirty-ninth Annual Conference on Neural Information Processing Systems. find
  41. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2023a. Improved baselines with visual instruction tuning.
  42. Qwen Team. 2025a. Qwen2.5-vl. find
  43. Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. 2023b. Visual instruction tuning. In NeurIPS.
  44. Qwen Team. 2025b. Qwen3 technical report. Preprint, arXiv:2505.09388. 36972 11
  45. Sello Ralethe and Jan Buys. 2022. Generic overgeneralization in pre-trained language models. In Proceedings of the 29th International Conference on Computational Linguistics, pages 3187–3196, Gyeongju, Republic of Korea. International Committee on Computational Linguistics. find
  46. Juan Diego Rodriguez, Aaron Mueller, and Kanishka Misra. 2025. Characterizing the role of similarity in the property inferences of language models. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologi… find
  47. Carol L Smith. 1980. Quantifiers and question answering in young children. Journal of experimental child psychology, 30(2):191–205. find
  48. Alvin Wei Ming Tan, Chunhua Yu, Bria Lorelle Long, Wanjing Anya Ma, Tonya Murray, Rebecca D. Silverman, Jason D Yeatman, and Michael Frank. 2024. Devbench: A multimodal developmental benchmark for language learning. In The Thirty-eight Conference on Neural Information Processing Systems Datasets and…
  49. Hugh Mee Wong, Rick Nouwen, and Albert Gatt. 2025. Vaquum: Are vague quantifiers grounded in visual data? arXiv preprint arXiv:2502.11874. find
  50. Eunice Yiu, Maan Qraitem, Anisa Noor Majhi, Charlie Wong, Yutong Bai, Shiry Ginosar, Alison Gopnik, and Kate Saenko. 2025. KiVA: Kid-inspired visual analogies for testing large multimodal models. In The Thirteenth International Conference on Learning Representations. find
  51. Charles Y. Zheng, Francisco Pereira, Chris I. Baker, and Martin N. Hebart. 2019. Revealing interpretable object representations from human behavior. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/padmanabhan-2026-bears-all-bears-some-bears/