Elamur Research⌕
☰
home / papers / 2026 / Unlocking Human-Like Visible Logic: How Logic Diagrams Boost Logic Reasoning in Large Language Models?

Unlocking Human-Like Visible Logic: How Logic Diagrams Boost Logic Reasoning in Large Language Models?

Paper · 2026 · Xin Wu, Yuqi Bu, Mengchen Zhao et al.

Summary

Large language models frequently reach fluent but logically unsound conclusions in formal reasoning. Text alone struggles to encode set boundaries, inclusion and overlap. The authors test whether giving vision-language models a logic diagram alongside the textual premises helps, much as people sketch diagrams to reason. They build a rule-based, template-driven pipeline that turns canonical syllogistic and conditional reasoning items into Venn, Euler and Linear diagrams. It works in three steps: extracting logical types, filling templates and merging sub-diagrams. They also compare color and symbol cues for linking text to diagram elements. They then evaluate three VLM families and add a human analysis of how models use the images. Diagrams raise accuracy in more than 60% of configurations, but the gains are not universal. Models favour different diagram types and alignment cues. A single static picture can mislead them on invalid arguments, which require considering counterexamples. The paper's value lies mainly in a reproducible evaluation framework and a clear account of when visual logic support helps or harms.

Key points

  • Models evaluated: Qwen2-VL-7B-Instruct, Llava-1.5-7b and LLama3.2-vision-11B, on syllogistic and conditional reasoning with Venn, Euler and Linear diagrams.
  • Diagram generation is template-based: type extraction (e.g., "All A are M"), template instantiation, and combination of premise sub-diagrams. Conditional cases are marked with a star placed relative to P/Q regions.
  • Without instructions on how to read the diagrams, Qwen2-VL-7B-Instruct loses 10.87% total accuracy, mostly on invalid items, because it treats the diagram as ground truth. The other two models do not degrade.
  • In the human evaluation, Qwen explicitly uses Venn, Euler and Linear diagrams in 96%, 96% and 100% of responses, interpreting them correctly in 54%, 66% and 72% of cases. Llava's usage rates are 28%, 8% and 40%, and Llama's are 70%, 32% and 88%. Both Llava and Llama rarely use Euler diagrams.
  • Limitations: the method depends on template-compatible inputs and scales poorly beyond a few sets. Suggested next steps include stronger text-only baselines such as CoT and self-consistency, and multi-world renderings to reduce over-reliance.

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

 Institutions

🎓Guangxi UniversityNanning · China3 works · 6 citations

Also on the byline: Shenzhen Polytechnic University.

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksArts and Humanities · 17,450 worksLanguage and Linguistics · 17,296 worksComputer Vision and Pattern Recognition · 3,486 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,894 worksReasoning · 1,685 works

Cite

Cite as Xin Wu, Yuqi Bu, Mengchen Zhao, Qingbao Huang and Yi Cai. 2026. Unlocking Human-Like Visible Logic: How Logic Diagrams Boost Logic Reasoning in Large Language Models?. 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.1602

@inproceedings{wu2026unlocking,
  title = {{Unlocking Human-Like Visible Logic: How Logic Diagrams Boost Logic Reasoning in Large Language Models?}},
  author = {Xin Wu, Yuqi Bu, Mengchen Zhao, Qingbao Huang and Yi Cai},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.1602},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.1602}
}

References

31 references, 13 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Risako Ando, Takanobu Morishita, Hirohiko Abe, Koji Mineshima, and Mitsuhiro Okada. 2023. Evaluating large language models with neubaroco: Syllogistic reasoning ability and human-like biases. In Proceedings of the 4th Natural Logic Meets Machine Learning Workshop, pages 1–11.
  2. Risako Ando, Kentaro Ozeki, Takanobu Morishita, Hirohiko Abe, Koji Mineshima, and Mitsuhiro Okada. 2024. Can euler diagrams improve syllogistic reasoning in large language models? In International Conference on Theory and Application of Diagrams, pages 232–248. Springer. find
  3. Jon Barwise and John Etchemendy. 1996. Visual information and valid reasoning. Logical reasoning with diagrams, 3:25. find
  4. Yuqi Bu, Xin Wu, Zirui Zhao, Yi Cai, David Hsu, and Qiong Liu. 2025. Walk in others’ shoes with a single glance: Human-centric visual grounding with top-view perspective transformation. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)… find
  5. Jiangjie Chen, Rui Xu, Ziquan Fu, Wei Shi, Zhongqiao Li, Xinbo Zhang, Changzhi Sun, Lei Li, Yanghua Xiao, and Hao Zhou. 2022. E-kar: A benchmark for rationalizing natural language analogical reasoning. In Findings of the Association for Computational Linguistics: ACL 2022, pages 3941–3955. find
  6. Peter Clark, Oyvind Tafjord, and Kyle Richardson. 2021. Transformers as soft reasoners over language. In Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence, pages 3882–3890.
  7. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, and 1 others. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.
  8. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, and 1 others. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783.
  9. Tiwalayo Eisape, Michael Tessler, Ishita Dasgupta, Fei Sha, Sjoerd Steenkiste, and Tal Linzen. 2024. A systematic comparison of syllogistic reasoning in humans and language models. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: H… find
  10. Mihir Parmar, Nisarg Patel, Neeraj Varshney, Mutsumi Nakamura, Man Luo, Santosh Mashetty, Arindam Mitra, and Chitta Baral. 2024. Logicbench: Towards systematic evaluation of logical reasoning ability of large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computati…
  11. Arkil Patel, Satwik Bhattamishra, and Navin Goyal. 2021. Are nlp models really able to solve simple math word problems? In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2080–2094.
  12. Yuri Sato, Sayako Masuda, Yoshiaki Someya, Takeo Tsujii, and Shigeru Watanabe. 2015. An fmri analysis of the efficacy of euler diagrams in logical reasoning. In 2015 IEEE Symposium on Visual Languages and Human-Centric Computing (VL/HCC), pages 143–151. IEEE. find
  13. Leonhard Euler. 1768. Lettres à une princesse d’allemagne sur divers sujets de physique & de philosophie. find
  14. Yuri Sato and Koji Mineshima. 2015. How diagrams can support syllogistic reasoning: an experimental study. Journal of Logic, Language and Information, 24:409–455. find
  15. Simeng Han, Hailey Schoelkopf, Yilun Zhao, Zhenting Qi, Martin Riddell, Wenfei Zhou, James Coady, David Peng, Yujie Qiao, Luke Benson, and 1 others. 2022. Folio: Natural language reasoning with first-order logic. arXiv preprint arXiv:2209.00840.
  16. Yuri Sato, Koji Mineshima, and Ryo Takemura. 2010. The efficacy of euler and venn diagrams in deductive reasoning: Empirical findings. In International Conference on Theory and Application of Diagrams, pages 6–22. Springer. find
  17. Danial Hooshyar. 2016. A Flowchart-Based Intelligent Tutoring System Model to Improve Students’ Problemsolving Skills. University of Malaya (Malaysia). find
  18. Koustuv Sinha, Shagun Sodhani, Jin Dong, Joelle Pineau, and William L Hamilton. 2019. Clutrr: A diagnostic benchmark for inductive reasoning from text. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Lang… find
  19. Stephen M Kosslyn and Zenon Pylyshyn. 1994. Image and brain: The resolution of the imagery debate. Nature, 372(6503):289–289. find
  20. Johann Heinrich Lambert and Hans Werner Arndt. 1764. Neues Organon, volume 1. Wendler Leipzig. find
  21. Andrew K Lampinen, Ishita Dasgupta, Stephanie CY Chan, Hannah R Sheahan, Antonia Creswell, Dharshan Kumaran, James L McClelland, and Felix Hill. 2024. Language models, like humans, show content effects on reasoning tasks. PNAS nexus, 3(7):pgae233. find
  22. Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. 2024. Improved baselines with visual instruction tuning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26296–26306.
  23. Jian Liu, Leyang Cui, Hanmeng Liu, Dandan Huang, Yile Wang, and Yue Zhang. 2021. Logiqa: a challenge dataset for machine reading comprehension with logical reasoning. In Proceedings of the Twenty-Ninth International Conference on International Joint Conferences on Artificial Intelligence, pages 3622…
  24. Oyvind Tafjord, Bhavana Dalvi, and Peter Clark. 2021. Proofwriter: Generating implications, proofs, and abductive statements over natural language. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, pages 3621–3634.
  25. John Venn. 1881. Symbolic logic. Nature, 24(613):284– 285. find
  26. Duo Wang, Mateja Jamnik, and Pietro Liò. 2018. Investigating diagrammatic reasoning with deep neural networks. In International Conference on Theory and Application of Diagrams, pages 390–398. Springer. find
  27. Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Yang Fan, Kai Dang, Mengfei Du, Xuancheng Ren, Rui Men, Dayiheng Liu, Chang Zhou, Jingren Zhou, and Junyang Lin. 2024. Qwen2-vl: Enhancing vision-language model’s perception of the w…
  28. Xin Wu, Yuqi Bu, Yi Cai, and Tao Wang. 2026. Triple-constrained and assumption-based zero-shot logical reasoning in reading comprehension. Knowledge-Based Systems, page 115702. find
  29. Xin Wu, Yi Cai, and Ho-fung Leung. 2024. Abstract-level deductive reasoning for pre-trained language models. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 70–76. find
  30. Zonglin Yang, Li Dong, Xinya Du, Hao Cheng, Erik Cambria, Xiaodong Liu, Jianfeng Gao, and Furu Wei. 2024. Language models as inductive reasoners. In Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 209–225.
  31. Weihao Yu, Zihang Jiang, Yanfei Dong, and Jiashi Feng. 2020. Reclor: A reading comprehension dataset requiring logical reasoning. In International Conference on Learning Representations.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wu-2026-unlocking-human-like-visible-logic/