Elamur Research⌕
☰
home / papers / 2026 / RECAP: Transparent Inference-Time Emotion Alignment for Medical Dialogue Systems

RECAP: Transparent Inference-Time Emotion Alignment for Medical Dialogue Systems

Paper · 2026 · Adarsh Srinivasan, Jacob Dineen, Muhammad Uzair Sarfraz et al.

Summary

Medical chatbots built on large language models often give answers that are clinically correct but emotionally flat. They also give clinicians no view into how the patient's feelings were interpreted. RECAP (Reflect–Extract–Calibrate–Align–Produce) addresses this with a prompting pipeline that runs at inference time and needs no retraining. Drawing on cognitive appraisal theory, it breaks a patient's message into five inspectable steps:

  • a short abstraction of the situation,
  • induced psychological factors such as control, support and uncertainty,
  • candidate emotions,
  • Likert-style intensity ratings for those emotions,
  • a final reply conditioned on this emotional profile.

The authors tested RECAP on emotion-understanding benchmarks (SECEU, EmoBench) and on HealthBench, using models from 8B to 120B parameters. RECAP brought model judgments closer to human consensus, and smaller models gained the most. The intermediate outputs revealed a consistent bias: models tend to give too little weight to interpersonal aspects like the support a patient gets from others. When oncology fellows rated responses without knowing which method produced them, they favored RECAP over the baselines. This suggests that structured, auditable prompting can make clinical dialogue systems more emotionally attuned without hiding their reasoning.

Key points

  • On SECEU, RECAP reduces the distance from human consensus by about 20%. Gains shrink as models grow: roughly 28% for the 8B model versus 5% for the 120B one.

  • Oncology fellows gave RECAP a mean score of 3.41, versus 2.93–2.95 for zero-shot, emotion-priming and persona baselines (p < .001). RECAP's win rates were 76–88% across clinical scenarios.

  • The human evaluation had two parts:

    • single-turn: 25 scenarios × 4 methods, rated on five 1–5 dimensions;
    • multi-turn: 10 scenarios × 2 methods.

    RECAP had the largest share of high-rated scenarios (44%) and no scenario scored below 2.5.

  • On HealthBench, RECAP raised Completeness and Communication Quality. It lowered Accuracy and Instruction Following slightly. Overall, O-120B improved from 0.46 to 0.54.

  • The intermediate outputs show that models reliably detect control and uncertainty but underweight social support.

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

🎓Arizona State UniversityTempe · United States50 works · 2,032 citations

 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 worksMedicine · 241 worksHealth Informatics · 184 works

🧭 Topics 2Research subjects and languages the work is about.Sentiment & emotion · 972 worksDialogue systems · 624 works

Cite

Cite as Adarsh Srinivasan, Jacob Dineen, Muhammad Uzair Sarfraz, Muhammad Umar Afzal, Irbaz Riaz and Ben Zhou. 2026. RECAP: Transparent Inference-Time Emotion Alignment for Medical Dialogue Systems. Proceedings of the 8th Workshop on Clinical Natural Language Processing (Clinical NLP) @ LREC 2026.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/5epg4xxjjygt

@inproceedings{srinivasan2026recap,
  title = {{RECAP: Transparent Inference-Time Emotion Alignment for Medical Dialogue Systems}},
  author = {Adarsh Srinivasan, Jacob Dineen, Muhammad Uzair Sarfraz, Muhammad Umar Afzal, Irbaz Riaz and Ben Zhou},
  booktitle = {Proceedings of the 8th Workshop on Clinical Natural Language Processing (Clinical NLP) @ LREC 2026},
  year = {2026},
  doi = {10.63317/5epg4xxjjygt},
  url = {https://doi.org/10.63317/5epg4xxjjygt}
}

References

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

  1. Saleema Amershi, Kori Inkpen, Jaime Teevan, Ruth Kikin-Gil, Adam Fourney, Besmira Nushi, Penny Collisson, Jina Suh, Shamsi Iqbal, Paul N. Bennett, Daniel Weld, Mihaela Vorvoreanu, and Eric Horvitz. 2019. Guidelines for human-ai interaction. Proceedings of the ACM CHI Conference on Human Factors in C… find
  2. Jonas Armbruster, Florian Bussmann, Catharina Rothhaas, Nadine Titze, Paul A. Grützner, and Holger Freischmidt. 2024. "doctor chatgpt, can you help me?" the patient’s perspective: Cross-sectional study. Journal of Medical Internet Research, 26:e58831. find
  3. Rahul K Arora, Jason Wei, Rebecca Soskin Hicks, Preston Bowman, Joaquin Quiñonero-Candela, Foivos Tsimpourlas, Michael Sharman, Meghan Shah, Andrea Vallone, Alex Beutel, et al. 2025. Healthbench: Evaluating large language models towards improved human health.
  4. John W. Ayers, Adam Poliak, Mark Dredze, Eric C. Leas, Zidian Zhu, Jon-Patrick Allem Kelley, Zoe Chu, and David M. Smith. 2023. Comparing physician and artificial intelligence chatbot responses to patient questions posted to a public social media forum. JAMA Internal Medicine, 183(6):589–596. find
  5. Emma Beede, Elizabeth Baylor, Fred Hersch, Anna Iurchenko, Lauren Wilcox, Paisan Ruamviboonsuk, and Laura M. Vardoulakis. 2020. A human-centered evaluation of a deep learning system deployed in clinics for the detection of diabetic retinopathy. Proceedings of the 2020 CHI Conference on Human Factors… find
  6. Anna Bodonhelyi, Christian Stegemann-Philipps, Alessandra Sonanini, Lea Herschbach, Marton Szep, Anne Herrmann-Werner, Teresa Festl-Wietek, Enkelejda Kasneci, and Friederike Holderried. 2025. Modeling challenging patient interactions: Llms for medical communication training. find
  7. Zana Buçinca, Maja Barbara Malaya, and Krzysztof Z. Gajos. 2021. To trust or to think: Cognitive forcing functions can reduce overreliance on ai in ai-assisted decision-making. Proceedings of the ACM on Human–Computer Interaction, 5(CSCW1):188:1–188:21. find
  8. Carrie J. Cai, Samantha Winter, David Steiner, Lauren Wilcox, and Michael Terry. 2019. "hello ai": Uncovering the onboarding needs of medical practitioners for human–ai collaborative decision-making. Proceedings of the ACM on Human–Computer Interaction, 3(CSCW):Article 104, 1–24. find
  9. David Chen, Kabir Chauhan, Rod Parsa, Zhihui Amy Liu, Fei-Fei Liu, Ernie Mak, Lawson Eng, Breffni Louise Hannon, Jennifer Croke, Andrew Hope, Nazanin Fallah-Rad, Phillip Wong, Srinivas Raman, et al. 2025. Patient perceptions of empathy in physician and artificial intelligence chatbot responses to pa… find
  10. David Chen, Rod Parsa, Andrew Hope, Breffni Hannon, Ernie Mak, Lawson Eng, Fei-Fei Liu, Nazanin Fallah-Rad, Ann M. Heesters, and Srinivas Raman. 2024. Physician and artificial intelligence chatbot responses to cancer questions from social media. JAMA Oncology, 10(7):956– 960. find
  11. 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. find
  12. Maximilian Croissant, Madeleine Frister, Guy Schofield, and Cade McCall. 2024. An appraisal-based chain-of-emotion architecture for affective language model game agents. PLOS ONE, 19(5):e0301033. find
  13. Jacob Dineen, Don Kridel, Daniel Dolk, and David Castillo. 2024. Unified explanations in machine learning models: A perturbation approach. find
  14. Jacob Dineen, Aswin RRV, Qin Liu, Zhikun Xu, Xiao Ye, Ming Shen, Zhaonan Li, Shijie Lu, Chitta Baral, Muhao Chen, et al. 2025. Qa-lign: Aligning llms through constitutionally decomposed qa. find
  15. Daniel Dolk, Donald Kridel, Jacob Dineen, and David Castillo. 2020. Model interpretation and explainability towards creating transparency in prediction models. In Proceedings of the 53rd Hawaii International Conference on System Sciences, HICSS, pages 956–965, Maui, HI. Hawaii International Conferen… find
  16. Yu Feng, Ben Zhou, Weidong Lin, and Dan Roth. 2024. Bird: A trustworthy bayesian inference framework for large language models. find
  17. Yu Feng, Ben Zhou, Haoyu Wang, Helen Jin, and Dan Roth. 2023. Generic temporal reasoning with differential analysis and explanation. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 12013–12029, Toronto, Canada. Association for… find
  18. Laura Francis and Noelle Robertson. 2023. Healthcare practitioners’ experiences of breaking bad news: A critical interpretative meta synthesis. Patient Education and Counseling, 107:107574. find
  19. Ludwig Franke Føyen, Emma Zapel, Mats Lekan-der, Erik Hedman-Lagerlöf, and Elin Lindsäter. 2025. Artificial intelligence vs. human expert: Licensed mental health clinicians’ blinded evaluation of ai-generated and expert psychological advice on quality, empathy, and perceived authorship. Internet Int… find
  20. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. 2024. The llama 3 herd of models.
  21. Alexa Hepburn and Jonathan Potter. 2023. Understanding mixed emotions in organized helping through emotionography. Frontiers in Psychology, 14:1236148. find
  22. Tiancheng Hu and Nigel Collier. 2024. Quantifying the persona effect in llm simulations.
  23. Jen-tse Huang, Man Ho Lam, Eric John Li, Shujie Ren, Wenxuan Wang, Wenxiang Jiao, Zhaopeng Tu, and Michael R. Lyu. 2024. Apathetic or empathetic? evaluating llms’ emotional alignments with humans. In Advances in Neural Information Processing Systems, volume 37, pages 97053– 97087, Vancouver, Canada.… find
  24. Clayton J. Hutto and Eric Gilbert. 2014. VADER: A parsimonious rule-based model for sentiment analysis of social media text. In Proceedings of the Eighth International Conference on Weblogs and Social Media, ICWSM ’14, pages 216–225, Ann Arbor, MI. AAAI Press. find
  25. Saurav Kadavath, Tom Conerly, Amanda Askell, Tom Henighan, Nicholas Schiefer, Eric Clark, Guy Amir, Kamal Ndousse, Tom B. Brown, Steven Larson, Roger Grosse, Jared Kaplan, Natasha McAleese, David Hernandez, Micah Carroll, Deep Ganguli, Jan Leike, Catherine Olsson, David Krueger, Evan Hubinger, Colli… find
  26. Adam Tauman Kalai, Ofir Nachum, Santosh S. Vempala, and Edwin Zhang. 2025. Why language models hallucinate. find
  27. Richard S. Lazarus. 1991. Emotion and Adaptation. Oxford University Press, New York, NY. find
  28. Keyeun Lee, Seolhee Lee, Esther Hehsun Kim, Yena Ko, Jinsu Eun, Dahee Kim, Hyewon Cho, Haiyi Zhu, Robert E. Kraut, Eunyoung Suh, Eun mee Kim, and Hajin Lim. 2025. Adaptive-vp: A framework for llm-based virtual patients that adapts to trainees’ dialogue to facilitate nurse communication training. find
  29. Yoon Kyung Lee, Inju Lee, Minjung Shin, Seoyeon Bae, and Sowon Hahn. 2024. Chain of empathy: Enhancing empathetic response of large language models based on psychotherapy models. Korean Journal of Cognitive Science, 35(1):23– 48. Also available as arXiv:2311.04915. — [arXiv:2311.04915](https://arxiv.org/abs/2311.04915) find
  30. Man Luo, Christopher J. Warren, Lu Cheng, Haidar M. Abdul-Muhsin, and Imon Banerjee. 2024. Assessing empathy in large language models with real-world physician-patient interactions. In IEEE International Conference on Big Data (BigData 2024), pages 6510–6519, Washington, DC, USA. IEEE. find
  31. Pedro Henrique Luz de Araujo and Benjamin Roth. 2025. Helpful assistant or fruitful facilitator? investigating how personas affect language model behavior. PLOS ONE, 20(6):e0325664. find
  32. OpenAI. 2025. model card. gpt-oss-120b and gpt-oss-20b Andrew Ortony, Gerald L. Clore, and Allan Collins. 1988. The Cognitive Structure of Emotions. Cambridge University Press, Cambridge, United Kingdom. find
  33. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul Christiano, Jan Leike, and Ryan Lowe. 2022. Train…
  34. Samuel J. Paech. 2023. Eq-bench: An emotional intelligence benchmark for large language models.
  35. Rafael Rafailov, Pratyusha Ramesh, Avnish Narayan Das, Sishir Syed, Sam Basu, Yao Li, James Zou, Qi Yang, Yuntao Bai, Pieter Abbeel, and Kiril Rafailov. 2023. Direct preference optimization: Your language model is secretly a reward model. In Advances in Neural Information Processing Systems, pages 5…
  36. Hannah Rashkin, Eric Michael Smith, Margaret Li, and Y-Lan Boureau. 2019. Towards empathetic open-domain conversation models: A new benchmark and dataset. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 5370–5381, Florence, Italy. Association for Com… find
  37. Aswin RRV, Jacob Dineen, Divij Handa, Md Nayem Uddin, Mihir Parmar, Chitta Baral, and Ben Zhou. 2025. Thinktuning: Instilling cognitive reflections without distillation. find
  38. Sahand Sabour, Siyang Liu, Zheyuan Zhang, June Liu, Jinfeng Zhou, Alvionna Sunaryo, Tatia Lee, Rada Mihalcea, and Minlie Huang. 2024. EmoBench: Evaluating the emotional intelligence of large language models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (… find
  39. Klaus R. Scherer and Agnes Moors. 2019. The emotion process: Event appraisal and component differentiation. Annual Review of Psychology, 70:719–745. find
  40. Ming Shen, Zhikun Xu, Xiao Ye, Jacob Dineen, and Ben Zhou. 2025. Bow: Bottlenecked next word exploration. find
  41. Vlad Sorin, Benjamin Sheffer, Nada Meirow, et al. 2024. Large language models and empathy: Systematic review. Journal of Medical Internet Research, 26:e55610. find
  42. John Sweller. 2023. The development of cognitive load theory: Replication crises and incorporation of other theories can lead to theory expansion. Educational Psychology Review, 35:95. find
  43. Oyvind Tafjord, Bhavana Dalvi Mishra, and Peter Clark. 2022. Entailer: Answering questions with faithful and truthful chains of reasoning. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 2078–2093, Abu Dhabi, United Arab Emirates. Association for Comp… find
  44. Ruiyi Wang, Stephanie Milani, Jamie C. Chiu, Jiayin Zhi, Shaun M. Eack, Travis Labrum, Samuel M. Murphy, Nev Jones, Kate Hardy, Hong Shen, Fei Fang, and Zhiyu Zoey Chen. 2024. Patient-ψ: Using large language models to simulate patients for training mental health professionals. find
  45. Xuena Wang et al. 2023. Emotional intelligence of large language models. Includes SECEU scenario-based assessments. find
  46. Jason Wei, Xuezhi Wang, Dale Schuurmans, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models.
  47. Tomer Wolfson, Mor Geva, Ankit Gupta, Matt Gardner, Yoav Goldberg, Daniel Deutch, and Jonathan Berant. 2020. Break it down: A question understanding benchmark. Transactions of the Association for Computational Linguistics, 8:183–198.
  48. Qian Yang, Aaron Steinfeld, and John Zimmerman. 2019. Unremarkable ai: Fitting intelligent decision support into critical, clinical decision-making processes. Published as proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI ’19), DOI:10.1145/3290605.3300468. — [doi:10.1145/3290605.3300468](https://doi.org/10.1145/3290605.3300468) find
  49. Xiao Ye, Jacob Dineen, Zhaonan Li, Zhikun Xu, Weiyu Chen, Shijie Lu, Yuxi Huang, Ming Shen, Phu Tran, Ji-Eun Irene Yum, Muhammad Ali Khan, Muhammad Umar Afzal, Irbaz Bin Riaz, and Ben Zhou. 2025a. Evaluating medical llms by levels of autonomy: A survey moving from benchmarks to applications. find
  50. Xiao Ye, Shaswat Shrivastava, Zhaonan Li, Jacob Dineen, Shijie Lu, Avneet Ahuja, Ming Shen, Zhikun Xu, and Ben Zhou. 2025b. Cc-learn: Cohort-based consistency learning. find
  51. Saizheng Zhang, Emily Dinan, Jack Urbanek, Arthur Szlam, Douwe Kiela, and Jason Weston. 2018. Personalizing dialogue agents: I have a dog, do you have pets too? In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2204–2213, Melbou…
  52. Mingqian Zheng, Jiaxin Pei, Lajanugen Logeswaran, Moontae Lee, and David Jurgens. 2024. When “a helpful assistant” is not really helpful: Personas in system prompts do not improve performances of large language models. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 1… find
  53. Ben Zhou, Kyle Richardson, Xiaodong Yu, and Dan Roth. 2022. Learning to decompose: Hypothetical question decomposition based on comparable texts. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 2223–2235, Abu Dhabi, United Arab Emirates. Association f… find
  54. Ben Zhou, Hongming Zhang, Sihao Chen, Dian Yu, Hongwei Wang, Baolin Peng, Dan Roth, and Dong Yu. 2024. Conceptual and unbiased reasoning in language models. find
  55. Zhikun Xu, Ming Shen, Jacob Dineen, Zhaonan Li, Xiao Ye, Shijie Lu, Aswin RRV, Chitta Baral, and Ben Zhou. 2024. Tow: Thoughts of words improve reasoning in large language models. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/srinivasan-2026-recap-transparent-inference-time-emotion/