Aligning with Logic: Measuring, Evaluating and Improving Logical Preference Consistency in Large Language Models

TL;DR

This paper proposes a universal framework to measure and improve logical preference consistency in large language models, enhancing their reliability and decision-making stability.

Contribution

It introduces a novel evaluation framework based on transitivity, commutativity, and negation invariance, along with the REPAIR technique to enhance logical consistency in LLMs.

Findings

LLMs often exhibit logical inconsistency in judgments.

The proposed framework effectively measures judgment robustness.

The REPAIR method improves logical consistency without losing alignment.

Abstract

Large Language Models (LLMs) are expected to be predictable and trustworthy to support reliable decision-making systems. Yet current LLMs often show inconsistencies in their judgments. In this work, we examine logical preference consistency as a foundational requirement for building more dependable LLM systems, ensuring stable and coherent decision-making while minimizing erratic or contradictory outputs. To quantify the logical preference consistency, we propose a universal evaluation framework based on three fundamental properties: transitivity, commutativity and negation invariance. Through extensive experimentation across diverse LLMs, we demonstrate that these properties serve as strong indicators of judgment robustness. Furthermore, we introduce a data refinement and augmentation technique, REPAIR, that enhances logical consistency while maintaining alignment with human preferences. Finally, we show that improving consistency leads to better performance in LLM-driven logic-based algorithms, reinforcing stability and coherence in decision-making systems.