Crisp: Cognitive Restructuring of Negative Thoughts through Multi-turn Supportive Dialogues

2025-04-24

AI Summary

To address critical shortages in clinical psychological services and persistent stigma surrounding mental health, this paper proposes a multi-turn supportive dialogue system specifically designed for cognitive restructuring (CR). Methodologically, we introduce CRDial—the first staged, multi-turn CR dialogue framework—and present Crisp, the first high-quality bilingual CR dialogue dataset. We further propose a strategy-driven, iterative large language model (LLM) paradigm for psychological dialogue, integrating sentence-level supportive strategy injection, multi-channel feedback loops, and instruction distillation, followed by fine-tuning of both 7B- and 14B-parameter models. Empirical evaluation demonstrates that our model, Crispers, consistently outperforms all baselines across three human evaluation dimensions: point-wise assessment, pairwise comparison, and intervention effectiveness—thereby validating the efficacy and feasibility of clinically grounded, logic-aligned system design.

Technology Category

• Cognitive Modeling & Cognitive Systems: Computational Creativity
• Natural Language Processing: Conversational AI/Dialog Systems
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Cognitive Restructuring (CR) is a psychotherapeutic process aimed at identifying and restructuring an individual's negative thoughts, arising from mental health challenges, into more helpful and positive ones via multi-turn dialogues. Clinician shortage and stigma urge the development of human-LLM interactive psychotherapy for CR. Yet, existing efforts implement CR via simple text rewriting, fixed-pattern dialogues, or a one-shot CR workflow, failing to align with the psychotherapeutic process for effective CR. To address this gap, we propose CRDial, a novel framework for CR, which creates multi-turn dialogues with specifically designed identification and restructuring stages of negative thoughts, integrates sentence-level supportive conversation strategies, and adopts a multi-channel loop mechanism to enable iterative CR. With CRDial, we distill Crisp, a large-scale and high-quality bilingual dialogue dataset, from LLM. We then train Crispers, Crisp-based conversational LLMs for CR, at 7B and 14B scales. Extensive human studies show the superiority of Crispers in pointwise, pairwise, and intervention evaluations.

Problem

• Addressing clinician shortage and stigma with human-LLM psychotherapy for Cognitive Restructuring
• Overcoming limitations of simple text rewriting and fixed-pattern dialogues in CR
• Developing multi-turn supportive dialogues for effective identification and restructuring of negative thoughts

Innovation

• Multi-stage dialogue framework for cognitive restructuring
• Sentence-level supportive conversation strategies
• Multi-channel loop mechanism for iterative CR