Why It Hurts: Identifying the Drivers of Negative Thoughts in Emotional Support Conversations
cs.HC, cs.AI, cs.CL
Submitted: 2026-06-02
Updated: 2026-08-29
Comments: EMNLP 2026
Code: https://github.com/seacowx/AppraiSal-and-PRISM
License: http://creativecommons.org/licenses/by-nc-sa/4.0/
The gist: Large Language Models (LLMs) are increasingly used for emotional support tasks, such as negative thought reframing.
Terminology
Abstract
Large Language Models (LLMs) are increasingly used for emotional support tasks, such as negative thought reframing. This task relies on modifying cognitive appraisals, the subjective interpretation of events that elicit negative emotions, which is typically conceptualized along multiple discrete dimensions. Current LLM-based frameworks model cognitive appraisal by exhaustively evaluating all possible dimensions, but they fail to account for the varying saliency of these dimensions across different contexts. In this work, we investigate a vital yet overlooked question: "Can LLMs infer the salient appraisal dimensions from emotional support conversations?" To address this question, we introduce the AppraiSal benchmark, containing 996 emotional support conversations with human-annotated mental states, including salient cognitive appraisal dimensions. Furthermore, we propose PRISM, a multi-agent probabilistic framework grounded in Bayesian Inverse Planning, designed to improve LLMs' ability to identify context-specific appraisal dimensions. Experimental results show that PRISM brings improvements to LLMs across various sizes, particularly in identifying the most salient appraisal dimensions.
Sources
- Scaling Synthetic Data Creation with 1,000,000,000 Personas
- SimpleToM: Exposing the Gap between Explicit ToM Inference and Implicit ToM Application in LLMs
- Psyche-R1: Towards Reliable Psychological LLMs through Unified Empathy, Expertise, and Reasoning
- DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models
- OpenAI GPT-5 System Card
- Gemma 3 Technical Report
- gpt-oss-120b & gpt-oss-20b Model Card
- Kardia-R1: Unleashing LLMs to Reason toward Understanding and Empathy for Emotional Support via Rubric-as-Judge Reinforcement Learning
- Qwen3 Technical Report
- Towards Open-Ended Emotional Support Conversations in LLMs via Reinforcement Learning with Future-Oriented Rewards
- Echo-N1: Affective RL Frontier
- Empathy-R1: A Chain-of-Empathy and Reinforcement Learning Framework for Long-Form Mental Health Support
- Is ChatGPT Equipped with Emotional Dialogue Capabilities?
- Building Emotional Support Chatbots in the Era of LLMs
- How FaR Are Large Language Models From Agents with Theory-of-Mind?
- Social World Models
Related papers
- Calibrating LLM Explanations to Human Decisions with Reinforcement Learning
- Rethinking Opinion Measurement Interfaces for Human-in-the-Loop Optimization
- EduGage: A Multimodal Dataset and Benchmark for Sensor-Based Momentary Assessment of Engagement in Self-Guided Video Learning
- EvoDesign: Agentic Editable Diagram Creation via Design Expertise Evolution
- HAGI++: Head-Assisted Gaze Imputation and Generation
- Explanation User Interfaces: A Systematic Literature Review