Collaborative Rational Speech Act: Pragmatic Reasoning for Multi-Turn Dialog

2025-07-18

AI Summary

RSA struggles to model the dynamic interplay between shared goals and private beliefs in multi-turn collaborative dialogue. To address this, we propose Conditional Rational Speech Acts (CRSA), an extension of RSA that incorporates an information-theoretic gain function grounded in rate-distortion theory. CRSA enables bidirectional pragmatic inference under private information constraints and supports conditional, multi-turn dialogue generation. Its core innovation lies in explicitly modeling the coupling mechanism between collaborative goals and individual knowledge, thereby enhancing pragmatic consistency and social awareness in language behavior. Evaluated on a referential game and real-world doctor–patient dialogues, CRSA significantly outperforms existing baselines, demonstrating superior behavioral interpretability, collaborative stability, and contextual adaptability.

Technology Category

• Multiagent Systems: Agent Communication
• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Cognitive Modeling & Cognitive Systems: Adaptive Behavior

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• 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

Abstract

As AI systems take on collaborative roles, they must reason about shared goals and beliefs-not just generate fluent language. The Rational Speech Act (RSA) framework offers a principled approach to pragmatic reasoning, but existing extensions face challenges in scaling to multi-turn, collaborative scenarios. In this paper, we introduce Collaborative Rational Speech Act (CRSA), an information-theoretic (IT) extension of RSA that models multi-turn dialog by optimizing a gain function adapted from rate-distortion theory. This gain is an extension of the gain model that is maximized in the original RSA model but takes into account the scenario in which both agents in a conversation have private information and produce utterances conditioned on the dialog. We demonstrate the effectiveness of CRSA on referential games and template-based doctor-patient dialogs in the medical domain. Empirical results show that CRSA yields more consistent, interpretable, and collaborative behavior than existing baselines-paving the way for more pragmatic and socially aware language agents.

Problem

• Extend RSA framework for multi-turn collaborative dialog
• Model shared goals and beliefs in AI systems
• Optimize information gain for pragmatic reasoning

Innovation

• Extends RSA with information-theoretic multi-turn dialog modeling
• Optimizes gain function from rate-distortion theory
• Handles private information in collaborative conversations