Elamur Research⌕
☰
home / papers / 2026 / Cognitive Policy-Driven LLM for Diagnosis and Intervention of Cognitive Distortions in Emotional Support Conversation

Cognitive Policy-Driven LLM for Diagnosis and Intervention of Cognitive Distortions in Emotional Support Conversation

Paper · 2026 · Lin Zhong, Renjin Zhu, Shujuan Ma et al.

Summary

Emotional support chatbots built on LLMs can sound empathetic, but they rarely detect the irrational thinking patterns behind a help-seeker's distress, as Cognitive Behavioral Therapy (CBT) would require. The authors attribute this to two causes. First, existing counseling corpora seldom address such distortions. Second, current strategy-selection methods are coarse and imitation-based. To address this, they build CogBiasESC by filtering dialogues from D4, CPsyCounD and PsyDTCorpus. Experts then label each sample with one of Beck's eight distortion types, a severity grade and a safety-risk grade. They also propose CoPoLLM, which has two parts. The CPRL engine uses a multi-agent simulated counseling environment and a Deep Q-Network to learn which intervention suits each diagnosis. Its reward includes a hard penalty for missing high-risk cases. The DSCO algorithm then transfers this policy knowledge into the LLM. Across 15 strong baselines, including GPT-based and specialized counseling models, CoPoLLM performs best on diagnosis accuracy, intervention quality and risk control. This makes it a step from generic comfort toward safer, therapy-informed support.

Key points

  • Dataset construction: GPT-4o pre-filtered the source corpora down to 4,231 dialogues containing distortions. A manual check of 200 samples found that 94% truly contained at least one distortion.
  • Annotation scheme: each sample has three labels:
    • one of 8 distortion types (e.g., Catastrophizing, All-or-Nothing, Mind Reading, Should Statements);
    • intensity: mild, moderate or severe;
    • risk: low, medium or high, with any self-harm signal always marked high.
  • Annotation quality: three experts with at least a master's degree labeled the data. Full-scale annotation began only after agreement reached κ ≥ 0.70. The paper reports Fleiss' κ = 0.73 for at least one label dimension (the extracted text is cut off at this point).
  • Method: CPRL learns intervention strategies through DQN in simulation, using a constraint-aware safety reward. DSCO injects the learned strategies into the LLM. Experiments show the safety reward sharply reduces missed high-risk cases.
  • Release: the synthetic subset of CogBiasESC, the CPRL code and the CoPoLLM LoRA adapters are available at github.com/Chips98/CoPoLLM-for-ACL-2026. The human-subject portion is withheld for privacy reasons.

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 4The field of science this work belongs to (OpenAlex fields).Psychology · 1,552 worksExperimental and Cognitive Psychology · 870 worksSocial Psychology · 306 worksApplied Psychology · 21 works

🧭 Topics 2Research subjects and languages the work is about.Language models & pretraining · 3,320 worksSentiment & emotion · 972 works

Cite

Cite as Lin Zhong, Renjin Zhu, Shujuan Ma, Jinhao Cui, Lingzhi Wang, Hao Chen and Qing Liao. 2026. Cognitive Policy-Driven LLM for Diagnosis and Intervention of Cognitive Distortions in Emotional Support Conversation. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.806

@inproceedings{zhong2026cognitive,
  title = {{Cognitive Policy-Driven LLM for Diagnosis and Intervention of Cognitive Distortions in Emotional Support Conversation}},
  author = {Lin Zhong, Renjin Zhu, Shujuan Ma, Jinhao Cui, Lingzhi Wang, Hao Chen and Qing Liao},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.806},
  url = {https://doi.org/10.18653/v1/2026.acl-long.806}
}

References

36 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, and 1 others. 2023. Gpt-4 technical report. arXiv preprint arXiv:2303.08774.
  2. Aaron T Beck, A John Rush, Brian F Shaw, Gary Emery, Robert J DeRubeis, and Steven D Hollon. 2024. Cognitive therapy of depression. Guilford Publications. find
  3. Emily E Bernstein, Hilary Weingarden, Emma C Wolfe, Margaret D Hall, Ivar Snorrason, and Sabine Wilhelm. 2022. Human support in app-based cognitive behavioral therapies for emotional disorders: scoping review. Journal of medical Internet research, 24(4):e33307. find
  4. Yirong Chen, Xiaofen Xing, Jingkai Lin, Huimin Zheng, Zhenyu Wang, Qi Liu, and Xiangmin Xu. 2023. Soulchat: Improving llms’ empathy, listening, and comfort abilities through fine-tuning with multi-turn empathy conversations. arXiv preprint arXiv:2311.00273. find
  5. Yuqi Chu, Lizi Liao, Zhiyuan Zhou, Chong-Wah Ngo, and Richang Hong. 2025. Towards multimodal emotional support conversation systems. IEEE Transactions on Multimedia. find
  6. 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…
  7. Chen Gao, Xiaochong Lan, Nian Li, Yuan Yuan, Jingtao Ding, Zhilun Zhou, Fengli Xu, and Yong Li. 2024. Large language models empowered agent-based modeling and simulation: A survey and perspectives. Humanities and Social Sciences Communications, 11(1):1–24. find
  8. Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V Chawla, Olaf Wiest, and Xiangliang Zhang. 2024. Large language model based multi-agents: A survey of progress and challenges. arXiv preprint arXiv:2402.01680. reduces psychological distress in younger patients with cardiac dis…
  9. Jinpeng Hu, Tengteng Dong, Luo Gang, Hui Ma, Peng Zou, Xiao Sun, Dan Guo, Xun Yang, and Meng Wang. 2024. Psycollm: Enhancing llm for psychological understanding and evaluation. IEEE Transactions on Computational Social Systems. find
  10. Binyuan Hui, Jian Yang, Zeyu Cui, Jiaxi Yang, Dayiheng Liu, Lei Zhang, Tianyu Liu, Jiajun Zhang, Bowen Yu, Keming Lu, and 1 others. 2024. Qwen2. 5-coder technical report. arXiv preprint arXiv:2409.12186. — Qwen2.5-Coder Technical Report
  11. Dongsheng Jiang, Yuchen Liu, Songlin Liu, Jin’e Zhao, Hao Zhang, Zhen Gao, Xiaopeng Zhang, Jin Li, and Hongkai Xiong. 2023. From clip to dino: Visual encoders shout in multi-modal large language models. arXiv preprint arXiv:2310.08825. — [arXiv:2310.08825](https://arxiv.org/abs/2310.08825) find
  12. Dongjin Kang, Sunghwan Mac Kim, Taeyoon Kwon, Seungjun Moon, Hyunsouk Cho, Youngjae Yu, Dongha Lee, and Jinyoung Yeo. 2024. Can large language models be good emotional supporter? mitigating preference bias on emotional support conversation. In Proceedings of the 62nd Annual Meeting of the Associatio… find
  13. Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, and Ion Stoica. 2023. Efficient memory management for large language model serving with pagedattention. In Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles.
  14. Yang Li. 2025. Policy guided tree search for enhanced llm reasoning. arXiv preprint arXiv:2502.06813. — [arXiv:2502.06813](https://arxiv.org/abs/2502.06813) find
  15. June M Liu, Donghao Li, He Cao, Tianhe Ren, Zeyi Liao, and Jiamin Wu. 2023. Chatcounselor: A large language models for mental health support. arXiv preprint arXiv:2309.15461. — [arXiv:2309.15461](https://arxiv.org/abs/2309.15461) find
  16. Hongbin Na, Yining Hua, Zimu Wang, Tao Shen, Beibei Yu, Lilin Wang, Wei Wang, John Torous, and Ling Chen. 2025. A survey of large language models in psychotherapy: Current landscape and future directions. arXiv preprint arXiv:2502.11095. find
  17. Christoph Herrmann-Lingen. 2022. Treating depression in patients with heart failure: what is (not) recommended? find
  18. Lanlan Qiu, Xiao Pu, Yeqi Feng, and Tianxing He. 2025. Llms vs. chinese anime enthusiasts: A comparative study on emotionally supportive role-playing. arXiv preprint arXiv:2508.06388. — [arXiv:2508.06388](https://arxiv.org/abs/2508.06388) find
  19. Annette Holdgaard, Christine Eckhardt-Hansen, Christina Funch Lassen, Ingunn Eklo Kjesbu, Christian Have Dall, Kristine Lund Michaelsen, Kirstine Lærum Sibilitz, Eva Prescott, and Hanne Kruuse Rasmusen. 2023. Cognitive-behavioural therapy Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D…
  20. Ruiyang Ren, Yuhao Wang, Junyi Li, Jinhao Jiang, Wayne Xin Zhao, Wenjie Wang, and Tat-Seng Chua. 2025. Llm-based search assistant with holistically guided mcts for intricate information seeking. In Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information… find
  21. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, and 1 others. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971.
  22. Leandro von Werra, Younes Belkada, Lewis Tunstall, Edward Beeching, Tristan Thrush, Nathan Lambert, Shengyi Huang, Kashif Rasul, and Quentin Gallouédec. 2020. Trl: Transformer reinforcement learning. https://github.com/huggingface/trl. ↗
  23. Ming Wang, Peidong Wang, Lin Wu, Xiaocui Yang, Daling Wang, Shi Feng, Yuxin Chen, Bixuan Wang, and Yifei Zhang. 2025. Annaagent: Dynamic evolution agent system with multi-session memory for realistic seeker simulation. arXiv preprint arXiv:2506.00551. find
  24. Haojie Xie, Yirong Chen, Xiaofen Xing, Jingkai Lin, and Xiangmin Xu. 2025. Psydt: Using llms to construct the digital twin of psychological counselor with personalized counseling style for psychological counseling. In Proceedings of the 63rd Annual Meeting of the Association for Computational Lingui…
  25. Yangyang Xu, Jinpeng Hu, Zhuoer Zhao, Zhangling Duan, Xiao Sun, and Xun Yang. 2025. Multiagen-tesc: A llm-based multi-agent collaboration framework for emotional support conversation. In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, pages 4665–4681. find
  26. DX Xin Yan and D Xue. 2023. Mindchat: Psychological large language model. GitHub repository. find
  27. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, and 1 others. 2025. Qwen3 technical report. arXiv preprint arXiv:2505.09388.
  28. Qisen Yang, Zekun Wang, Honghui Chen, Shenzhi Wang, Yifan Pu, Xin Gao, Wenhao Huang, Shiji Song, and Gao Huang. 2024a. Psychogat: A novel psychological measurement paradigm through interactive fiction games with llm agents. arXiv preprint arXiv:2402.12326. find
  29. Qu Yang, Mang Ye, and Bo Du. 2024b. Emollm: Multimodal emotional understanding meets large language models. arXiv preprint arXiv:2406.16442. — [arXiv:2406.16442](https://arxiv.org/abs/2406.16442) find
  30. Binwei Yao, Chao Shi, Likai Zou, Lingfeng Dai, Mengyue Wu, Lu Chen, Zhen Wang, and Kai Yu. 2022. D4: a chinese dialogue dataset for depression-diagnosis-oriented chat. arXiv preprint arXiv:2205.11764. find
  31. Gülin Yazici-Çelebi and Feridun Kaya. 2022. Interpersonal cognitive distortions and anxiety: The mediating role of emotional intelligence. International Journal of Psychology and Educational Studies, 9(3):741– 753. find
  32. Chenhao Zhang, Renhao Li, Minghuan Tan, Min Yang, Jingwei Zhu, Di Yang, Jiahao Zhao, Guancheng Ye, Chengming Li, and Xiping Hu. 2024. Cpsycoun: A report-based multi-turn dialogue reconstruction and evaluation framework for chinese psychological counseling. arXiv preprint arXiv:2405.16433. find
  33. Haiquan Zhao, Lingyu Li, Shisong Chen, Shuqi Kong, Jiaan Wang, Kexin Huang, Tianle Gu, Yixu Wang, Jian Wang, Liang Dandan, and 1 others. 2024. Esc-eval: Evaluating emotion support conversations in large language models. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language P… find
  34. Weixiang Zhao, Xingyu Sui, Xinyang Han, Yang Deng, Yulin Hu, Jiahe Guo, Libo Qin, Qianyun Du, Shijin Wang, Yanyan Zhao, and 1 others. 2025. Chain of strategy optimization makes large language models better emotional supporter. arXiv preprint arXiv:2503.05362. find
  35. Xi Zheng, Zhuoyang Li, Xinning Gui, and Yuhan Luo. 2025. Customizing emotional support: How do individuals construct and interact with llm-powered chatbots. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, pages 1–20. find
  36. Rong Zhu, Jingyuan Huang, Zejiang He, Menglong Lu, Zhen Huang, Jinhui Zhao, and Yan Cao. 2024. Esc-cot: Easy-to-hard self-comparative chain-of-thought for news discourse profiling. In 2024 IEEE 36th International Conference on Tools with Artificial Intelligence (ICTAI), pages 476–484. IEEE. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhong-2026-cognitive-policy-driven-llm-diagnosis/