Constraints-of-Thought: A Framework for Constrained Reasoning in Language-Model-Guided Search

2025-10-10

AI Summary

Large language models (LLMs) frequently deviate from users’ high-level intentions and violate symbolic constraints during multi-step planning. To address this, we propose Constraints-of-Thought (Const-o-T), the first framework to incorporate structured (intention, constraint) pairs as reasoning priors within Monte Carlo Tree Search (MCTS), enabling semantically valid and intention-aligned controlled reasoning. Const-o-T dynamically prunes invalid paths via symbolic constraint modeling, substantially reducing the search space while improving plan feasibility and formal verifiability. Evaluated on three diverse tasks—Risk game strategy, CAD code generation, and arithmetic reasoning—Const-o-T outperforms chain-of-thought (CoT), tree-of-thought (ToT), and verifier-augmented baselines in both accuracy and structural plan consistency. Its constraint-aware architecture demonstrates strong cross-domain adaptability without task-specific retraining.

Technology Category

• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories
• Planning, Routing, and Scheduling: Planning with Language Models
• Multiagent Systems: Multiagent Planning

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

While researchers have made significant progress in enabling large language models (LLMs) to perform multi-step planning, LLMs struggle to ensure that those plans align with high-level user intent and satisfy symbolic constraints, especially in complex, multi-step domains. Existing reasoning approaches such as Chain-of-Thought (CoT), Tree-of-Thought (ToT), and verifier-augmented methods, expand the search space but often yield infeasible actions or hallucinated steps. To overcome these limitations, we propose Constraints-of-Thought (Const-o-T), a framework that provides a structured prior that enables Monte Carlo Tree Search (MCTS) focus search on semantically meaningful paths. Each reasoning step is represented as an (intent, constraint) pair, which serves both to compress the search space and enforce validity. Unlike prior methods that merely generate reasoning traces or validate outputs post hoc, Const-o-T uses (intent, constraint)pairs to actively focus the search toward feasible and meaningful plans. We integrate Const-o-T into MCTS using a structured representation of intent-constraint pairs constraints prune infeasible branches and guide exploration toward semantically valid actions, improving planning efficiency and verifiable decision-making. We demonstrate across three domains Risk game, CAD code generation, and arithmetic reasoning that our approach outperforms baselines, yielding higher accuracy and stronger structural alignment. Our contribution is to demonstrate that Const-of-T offers a generalizable foundation for constraint-guided reasoning, enabling more efficient, constraint-aligned, and domain-adaptable planning with LLMs.

Problem

• LLMs struggle with multi-step planning that satisfies user constraints
• Existing methods expand search space but yield infeasible actions
• Need framework to enforce validity and guide semantic reasoning

Innovation

• Monte Carlo Tree Search guided by intent-constraint pairs
• Structured prior compresses search space for efficiency
• Active constraint enforcement improves planning accuracy