Elamur Research⌕
☰
home / papers / 2026 / Simulating Crisis Cognition: A Computational Framework for Hypothesis Generation in Crisis Communication

Simulating Crisis Cognition: A Computational Framework for Hypothesis Generation in Crisis Communication

Paper · 2026 · Changsen Yuan, Yanghao Zhou, Chong Feng et al.

Summary

Testing crisis messages on people during real disasters is ethically impossible. The authors propose CRISIS COGNITION, which uses LLM agents as an in-silico laboratory for generating hypotheses about crisis communication strategies. They frame the interaction as a generative Structural Causal Model. The experimenter sets the agent persona and the environment directly and randomizes only the message, so the causal effect is identifiable by construction.

Agents get personalities from the PsychoAgent dataset and are synchronized with telemetry (wind speed, pressure) from historical hurricane peaks. They reason through a two-stage chain of thought (risk appraisal, then response). The resulting COPE-CF benchmark holds matched counterfactual trajectories. Effects are estimated within strata of high and low Neuroticism.

The main finding is that the order of support matters. Giving emotional reassurance first and factual instructions second greatly increases survival-oriented actions, especially for high-Neuroticism agents. Delivering both at once calms agents but leaves them passive. The authors present these results as simulation-derived hypotheses to guide later human studies, not as policy.

Key points

  • The simulation uses 1,813 agents built on DeepSeek-V3 and Llama-3.1-70B. A blinded GPT-4o judge rates anxiety on a 1–10 scale; it matched human annotations at r = 0.82. Inferred Neuroticism correlates with ground truth at r = 0.76.
  • Action is measured as Valid Action Density: a count of survival verbs mapped to the FEMA taxonomy, excluding negated verbs. Cognitive Bandwidth divides the number of chain-of-thought tokens spent on planning by the number spent on appraising the threat.
  • "Sedative Effect" (Llama-3.1-70B):
    • The simultaneous strategy gives anxiety 3.30 but action only 0.18.
    • The sequential strategy gives anxiety 3.68 and action 5.74.
    • The RAG baseline gives anxiety 6.43 and action 0.69.
  • Gains by personality stratum: Sequential Unlock raised action density by about 412% over RAG in the top-25% Neuroticism group, versus roughly 85% in the bottom-25% group.
  • Ablation against a "Rational CoT" baseline (N = 1,500): that baseline lowered anxiety but not action (2.19). The proposed method reached 3.50 (p < 0.001). DeepSeek-V3 showed a different pattern, "functional alertness," where both anxiety and action increased.

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

🎓Beijing Institute of TechnologyBeijing · China43 works · 3,652 citations🎓Beijing University of TechnologyBeijing · China16 works · 1,910 citations

 Fields, topics and concepts

📚 Shelves 5The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksArtificial Intelligence · 17,965 worksSocial Sciences · 1,954 worksSafety Research · 626 worksCommunication · 104 works

Cite

Cite as Changsen Yuan, Yanghao Zhou, Chong Feng and Ge Shi. 2026. Simulating Crisis Cognition: A Computational Framework for Hypothesis Generation in Crisis Communication. 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.692

@inproceedings{yuan2026simulating,
  title = {{Simulating Crisis Cognition: A Computational Framework for Hypothesis Generation in Crisis Communication}},
  author = {Changsen Yuan, Yanghao Zhou, Chong Feng and Ge Shi},
  booktitle = {Findings of the Association for Computational Linguistics: ACL 2026},
  year = {2026},
  doi = {10.18653/v1/2026.findings-acl.692},
  url = {https://doi.org/10.18653/v1/2026.findings-acl.692}
}

References

14 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Lisa P. Argyle, Ethan C. Busby, Nancy Fulda, Joshua Gubler, Christopher Michael Rytting, and David Wingate. 2022. Out of one, many: Using language models to simulate human samples. CoRR, abs/2209.06899.
  2. DeepSeek-AI, Aixin Liu, Bei Feng, Bing Xue, Bing-Li Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dong-Li Ji, Erhang Li, Fangyun Lin, Fucong Dai, and 179 others. 2024. Deepseek-v3 technical report. ArXiv, abs/2412.19… find
  3. Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, Anirudh Goyal, Anthony S. Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao, Aston Zhang, and 510 others.… — The Llama 3 Herd of Models
  4. James A. Easterbrook. 1959. The effect of emotion on cue utilization and the organization of behavior. Psychological review, 66 3:183–201. find
  5. Amir Feder, Katherine A. Keith, Emaad Manzoor, Reid Pryzant, Dhanya Sridhar, Zach Wood-Doughty, Jacob Eisenstein, Justin Grimmer, Roi Reichart, Margaret E. Roberts, Brandon M. Stewart, Victor Veitch, and Diyi Yang. 2022. Causal inference in natural language processing: Estimation, prediction, interp… find
  6. Chen Gao, Xiaochong Lan, Zhihong Lu, Jinzhu Mao, Jinghua Piao, Huandong Wang, Depeng Jin, and Yong Li. 2023. S3 : Social-network simulation system with large language model-empowered agents. CoR-R, abs/2307.14984. — [arXiv:2307.14984](https://arxiv.org/abs/2307.14984) find
  7. John Gkountouras, Matthias Lindemann, Phillip Lippe, Efstratios Gavves, and Ivan Titov. 2025. Language agents meet causality - bridging llms and causal world models. In The Thirteenth International Conference on Learning Representations, ICLR 2025, Singapore, April 24-28, 2025. OpenReview.net. find
  8. Soyeong Jeong, Jinheon Baek, Sukmin Cho, Sung Ju Hwang, and Jong Park. 2024. Adaptive-rag: Learning to adapt retrieval-augmented large language models through question complexity. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Hu…
  9. Richard S. Lazarus and Susan Folkman. 2020. Stress: Appraisal and coping. Encyclopedia of Behavioral Medicine. find
  10. Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive NLP tasks. In Advances in Neural Information Proc…
  11. Mengzhu Liu, Zhengqiu Zhu, Chuan Ai, Chen Gao, Xinghong Li, Lingnan He, Kaisheng Lai, Yingfeng Chen, Xin Lu, Yong Li, and Quanjun Yin. 2025. Psychology-driven LLM agents for explainable panic prediction on social media during sudden disaster events. CoRR, abs/2505.16455. find
  12. Joon Sung Park, Joseph C. O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S. Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, UIST 2023, San Francisco, CA,… find
  13. Kurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela, and Jason Weston. 2021. Retrieval augmentation reduces hallucination in conversation. In Findings of the Association for Computational Linguistics: EMNLP 2021, Virtual Event / Punta Cana, Dominican Republic, 16-20 November, 2021, pages 3784– 3803.…
  14. Daniel Zelterman. 2001. Causality: Models, reasoning, and inference. Technometrics, 43(2):239–240. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yuan-2026-simulating-crisis-cognition-computational-framework/