Distributed Partial Information Puzzles: Examining Common Ground Construction Under Epistemic Asymmetry
2026-03-05 • Artificial Intelligence
Artificial IntelligenceComputation and Language
AI summaryⓘ
The authors study how people work together when they have different information, which is hard for AI systems to understand, especially when using speech, gestures, and actions together. They created a task called the Distributed Partial Information Puzzle (DPIP) to encourage this kind of communication and collected data showing how people share beliefs over time. They tested two ways to model shared understanding: advanced language models and a logic-based system called Dynamic Epistemic Logic. Their results show that these AI methods still struggle to keep track of what people believe and how the task unfolds.
common groundmultimodal communicationepistemic asymmetryDistributed Partial Information Puzzlebelief trackinglarge language modelsDynamic Epistemic Logicspeech gesture alignmentcollaborative reasoning
Authors
Yifan Zhu, Mariah Bradford, Kenneth Lai, Timothy Obiso, Videep Venkatesha, James Pustejovsky, Nikhil Krishnaswamy
Abstract
Establishing common ground, a shared set of beliefs and mutually recognized facts, is fundamental to collaboration, yet remains a challenge for current AI systems, especially in multimodal, multiparty settings, where the collaborators bring different information to the table. We introduce the Distributed Partial Information Puzzle (DPIP), a collaborative construction task that elicits rich multimodal communication under epistemic asymmetry. We present a multimodal dataset of these interactions, annotated and temporally aligned across speech, gesture, and action modalities to support reasoning over propositional content and belief dynamics. We then evaluate two paradigms for modeling common ground (CG): (1) state-of-the-art large language models (LLMs), prompted to infer shared beliefs from multimodal updates, and (2) an axiomatic pipeline grounded in Dynamic Epistemic Logic (DEL) that incrementally performs the same task. Results on the annotated DPIP data indicate that it poses a challenge to modern LLMs' abilities to track both task progression and belief state.