PDDL-Mind: Large Language Models are Capable on Belief Reasoning with Reliable State Tracking

2026-04-20

AI Summary

Large language models (LLMs) exhibit suboptimal performance on theory of mind (ToM) tasks, primarily due to their inability to reliably track implicit mental states. This work proposes PDDL-Mind, a novel neuro-symbolic framework that, for the first time, applies neuro-symbolic methods to ToM reasoning. By translating narrative inputs into explicit states and actions formalized in Planning Domain Definition Language (PDDL), the approach decouples environmental state evolution from belief reasoning and enforces logical consistency in state transitions. This yields a rigorously structured world representation that supports more accurate belief attribution by LLMs. Evaluated on established benchmarks—including MMToM-QA, MuMA, and FanToM—the method achieves substantial improvements in belief reasoning accuracy, surpassing current state-of-the-art approaches by over 5% absolute points.

Technology Category

• Knowledge Representation and Reasoning: Reasoning with Beliefs
• Machine Learning: Neuro-Symbolic Learning
• Cognitive Modeling & Cognitive Systems: Symbolic Representations

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: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

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, MuMA and FanToM show that PDDL-Mind achieves over 5% absolute accuracy gain over the best existing state-of-the-art method on ToM benchmark questions.

Problem

• Theory of Mind
• belief reasoning
• state tracking
• large language models
• PDDL

Innovation

• PDDL-Mind
• neuro-symbolic framework
• state tracking
• theory of mind
• explicit state representation