Flout at Your Own Risk: LLMs Struggle with Pragmatic Cooperativity Under Epistemic Asymmetry

2026-07-12

AI Summary

This work addresses the challenge that large language models (LLMs) often fail to effectively adhere to Grice’s Cooperative Principle in multi-agent collaboration under information asymmetry, primarily due to insufficient pragmatic reasoning. The study proposes the first evaluation framework that explicitly links Gricean maxims to task objectives, employing controlled collaborative tasks with cognitive asymmetry to systematically assess LLMs’ pragmatic competence both as speakers and listeners. Through a combination of prompt engineering and post-training strategies, experiments demonstrate that while LLMs possess basic pragmatic capabilities amenable to improvement via targeted interventions, they frequently violate cooperative maxims unconsciously when operating with incomplete information. This research establishes a novel methodological approach and provides empirical foundations for achieving pragmatic alignment in multi-agent systems.

Abstract

Fruitful collaborations rely on cooperative communications, including of contextual cues to incorporate into reasoning. The increasing use of LLMs in collaborative and agentic pipelines raises questions about the extent to which they exhibit these pragmatic capabilities, especially in scenarios where they may not have access to the same information as their collaborators. In this paper, we perform a novel investigation into the pragmatic reasoning capabilities of LLMs in a multi-party collaborative task under partial information conditions. We formalize a notion of collaborative epistemic asymmetry that explicitly connects objective task success to Grice's cooperative principle and empirically assess various LLMs' abilities to act cooperatively as both speakers and listeners, including both prompting and post-training strategies. Our results show that while LLMs exhibit certain pragmatic capabilities in collaborative settings, and these can be elicited through prompting and post-training, they still face challenges in pragmatic communication with incomplete information, and that certain failure modes do correlate with floutings of Grice's maxims that go unrecognized.

Problem

• pragmatic reasoning
• epistemic asymmetry
• cooperative communication
• Grice's maxims
• large language models

Innovation

• pragmatic reasoning
• epistemic asymmetry
• Grice's cooperative principle
• large language models
• collaborative communication