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

Yinhong Liu, Zhijiang Guo, Tianya Liang, Ehsan Shareghi, Ivan Vulić, Nigel Collier
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:38518-38539, 2025.

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.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-liu25u, title = {Aligning with Logic: Measuring, Evaluating and Improving Logical Preference Consistency in Large Language Models}, author = {Liu, Yinhong and Guo, Zhijiang and Liang, Tianya and Shareghi, Ehsan and Vuli\'{c}, Ivan and Collier, Nigel}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {38518--38539}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/liu25u/liu25u.pdf}, url = {https://proceedings.mlr.press/v267/liu25u.html}, 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.} }
Endnote
%0 Conference Paper %T Aligning with Logic: Measuring, Evaluating and Improving Logical Preference Consistency in Large Language Models %A Yinhong Liu %A Zhijiang Guo %A Tianya Liang %A Ehsan Shareghi %A Ivan Vulić %A Nigel Collier %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-liu25u %I PMLR %P 38518--38539 %U https://proceedings.mlr.press/v267/liu25u.html %V 267 %X 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.
APA
Liu, Y., Guo, Z., Liang, T., Shareghi, E., Vulić, I. & Collier, N.. (2025). Aligning with Logic: Measuring, Evaluating and Improving Logical Preference Consistency in Large Language Models. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:38518-38539 Available from https://proceedings.mlr.press/v267/liu25u.html.

Related Material