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

arXiv cs.CL / 4/21/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper introduces an LLM approach for Emotional Support Conversation (ESC) that explicitly targets cognitive distortions in help-seekers’ statements, going beyond surface-level emotional comfort.
It presents the CogBiasESC dataset, adding labeled information on cognitive distortion types, intensity, and safe risk levels to extend existing ESC datasets.
The authors propose CoPoLLM (Cognitive Policy-driven Large Language Model), a framework designed to diagnose cognitive distortions and generate more effective intervention strategies.
Experiments report that CoPoLLM outperforms 15 state-of-the-art baselines on diagnostic accuracy, intervention effectiveness, and safety risk control.
A theoretical analysis is provided to argue for CoPoLLM’s safety advantages in this setting.

Abstract

Emotional Support Conversation (ESC) plays a critical role in mental health assistance by providing accessible psychological support in real-world applications. Large Language Models (LLMs) have shown strong empathetic abilities in ESC tasks. Yet, existing methods overlook the issue of cognitive distortions in help-seekers' expressions. As a result, current models can only provide basic emotional comfort, rather than helping help-seekers address their psychological distress at a deeper cognitive level. To address this challenge, we construct the CogBiasESC dataset, the first dataset that expands existing ESC datasets by adding labels for cognitive distortions, includes their type, intensity, and safe risk level. Furthermore, we propose the Cognitive Policy-driven Large Language Model framework (CoPoLLM) to enhance LLMs' ability to diagnose and intervene cognitive distortions in help-seekers. We also analyze the safety advantages of CoPoLLM from a theoretical perspective. Experimental results show that CoPoLLM significantly outperforms 15 state-of-the-art baselines in terms of distortion diagnosis accuracy, intervention strategy effectiveness, and safety risk control.