PDDL-Mind: Reliable State Tracking is All You Need for Theory-of-Mind Benchmarks

ACL ARR 2026 January Submission7063 Authors

Keywords: Theory-of-Mind, State Tracking, LLM, Neuro-Symbolic

Abstract:

Large language models (LLMs) perform substantially below human level on existing theory-of-mind (ToM) benchmarks, even when augmented with chain-of-thought prompting or probabilistic belief updates.
We argue that these failures primarily arise from unreliable implicit state tracking rather than limitations in high-level reasoning.
We introduce PDDL-Mind, a neuro-symbolic framework that decouples environment state evolution from belief inference.
By translating narrative descriptions into explicit states and actions expressed in Planning Domain Definition Language (PDDL), and by verifying action-induced state transitions against a predefined domain, PDDL-Mind provides LLMs with a logically consistent and explicit representation of world states for ToM tasks.
Experiments on MMToM-QA and MuMA-ToM show that PDDL-Mind achieves over 5% absolute accuracy gain over the best existing state-of-the-art method on ToM benchmark questions.

Paper Type: Short

Research Area: Discourse, Pragmatics, and Reasoning

Research Area Keywords: pragmatic inference and reasoning

Contribution Types: NLP engineering experiment, Data analysis

Languages Studied: English

Submission Number: 7063