CARE: Cognitive-reasoning Augmented Reinforcement for Emotional Support Conversation

2025-09-29

AI Summary

Existing research on empathic support dialogue primarily focuses on data augmentation, neglecting the deep cognitive reasoning required for effective emotional support. Method: This paper proposes a cognition-enhanced framework that requires no large-scale synthetic data. It is the first to integrate reinforcement learning with explicit cognitive modeling, guiding logical reasoning and empathy strategy learning directly on the original training set to optimize response coherence and supportive quality in an end-to-end manner. Contribution/Results: By jointly modeling cognitive processes and dialogue policy optimization, the method significantly outperforms state-of-the-art approaches across key metrics—including logical validity, empathy intensity, and support effectiveness—on standard benchmarks. It establishes a novel paradigm for building human-like empathic support systems grounded in interpretable cognitive mechanisms rather than purely statistical pattern matching.

Technology Category

• Natural Language Processing: Conversational AI/Dialog Systems
• Cognitive Modeling & Cognitive Systems: Affective Computing
• Humans and AI: Emotional Intelligence

Application Category

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

Abstract

Emotional Support Conversation (ESC) plays a vital role in alleviating psychological stress and providing emotional value through dialogue. While recent studies have largely focused on data augmentation and synthetic corpus construction, they often overlook the deeper cognitive reasoning processes that underpin effective emotional support. To address this gap, we propose extbf{CARE}, a novel framework that strengthens reasoning in ESC without relying on large-scale synthetic data. CARE leverages the original ESC training set to guide models in generating logically coherent and supportive responses, thereby explicitly enhancing cognitive reasoning. Building on this foundation, we further employ reinforcement learning to refine and reinforce the reasoning process. Experimental results demonstrate that CARE significantly improves both the logical soundness and supportive quality of responses, advancing the development of empathetic, cognitively robust, and human-like emotional support systems.

Problem

• Enhancing cognitive reasoning in emotional support conversations
• Improving logical coherence of responses without synthetic data
• Strengthening empathetic support quality through reinforcement learning

Innovation

• Augments reinforcement learning with cognitive reasoning
• Leverages original training data without synthetic corpora
• Enhances logical coherence and supportive response quality