Inference to the Best Explanation in Large Language Models
Dhairya Dalal, Marco Valentino, André Freitas, Paul Buitelaar
Abstract
While Large Language Models (LLMs) have found success in real-world applications, their underlying explanatory process is still poorly understood.This paper proposes IBE-Eval, a framework inspired by philosophical accounts on Inference to the Best Explanation (IBE) to advance the interpretation and evaluation of LLM explanations.IBE-Eval estimates the plausibility of natural language explanations through a combination of explicit logical and linguistic features including: consistency, parsimony, coherence, and uncertainty.Extensive experiments are conducted on Causal Question Answering (CQA), where IBE-Eval is tasked to select the most plausible causal explanation amongst competing ones generated by the LLM (e.g.GPT 3.5 or LLaMA 2).The experiments reveal that IBE-Eval can successfully identify the best explanation with up to 77% accuracy ( 27% above random), improving upon a GPT 3.5-as-a-judge baseline ( +17%) while being intrinsically more efficient and interpretable.Additional analysis suggests that, despite LLM-specific variances, generated explanations tend to conform to IBE criteria and that IBE-Eval is significantly correlated with human judgment, opening up opportunities for future development of automated explanation verification tools. Inference to the Best Explanation (IBE)Selection Criteria Competing Explanations Competing HypothesesPremise 1: I blew into the baloon. Conclusion:The balloon expanded.Premise 2: I pricked the baloon.Conclusion: The balloon expanded. LLM Explanation 1 (E1)Step 1: IF someone blows into a balloon, THEN it can cause the balloon to inflate.Assumption: Blowing air into a balloon increases the amount of air inside it, leading to inflation.Step 2: IF the balloon inflates, THEN it can cause the balloon to expand.Assumption: When a balloon inflates, it stretches and expands in size.Therefore, since I blew into the balloon, it caused the balloon to inflate, which resulted in its expansion. Explanation 2 (E2)Step 1: IF a balloon is pricked, THEN the balloon may deflate.Assumption: Pricking a balloon can cause it to lose air and deflate.Step 2: IF a balloon deflates, THEN there is a decrease in air pressure inside the balloon.Assumption: When a balloon deflates, the air pressure inside the balloon decreases.Step 3: IF there is a decrease in air pressure inside the balloon, THEN the external air pressure will cause the balloon to expand.Assumption: When the air pressure inside a balloon is lower than the external air pressure, the balloon will expand.Therefore, since the balloon was pricked, it may have deflated, resulting in a decrease in air pressure inside the balloon, causing the external air pressure to make the balloon expand Causal Question The balloon expanded.What was the cause?A) I blew into it.B) I pricked it.
Ask about this paper
Your agent reads all of it.
Lune indexed this paper to the last equation, along with the top-tier papers that cite it. Ask a question and the answer quotes them.
Cited by top-tier papers4
- Improving Chain-of-Thought Reasoning via Quasi-Symbolic AbstractionsLeonardo Ranaldi, Marco Valentino, André FreitasACL 2025 · 29 citations
- Verification and Refinement of Natural Language Explanations through LLM-Symbolic Theorem ProvingXin Quan, Marco Valentino, Louise A. Dennis, André FreitasEMNLP 2024 · 9 citations
- Faithful and Robust LLM-Driven Theorem Proving for NLI ExplanationsXin Quan, Marco Valentino, Louise A. Dennis, André FreitasACL 2025 · 8 citations
- Leibniz: Theory-of-Mind Driven Neuro-Symbolic Logical Reasoning via Multi-Agent CollaborationYue Fan, Hu Zhang, Yunxiao Zhao, Guangjun Zhang et al.ACL 2026
Builds on12
- Language Models are Few-Shot LearnersTom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah et al.NeurIPS 2020 · 64,255 citations
- Chain-of-Thought Prompting Elicits Reasoning in Large Language ModelsJason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma et al.NeurIPS 2022 · 22,562 citations
- Self-Consistency Improves Chain of Thought Reasoning in Language ModelsXuezhi Wang, Jason Wei, Dale Schuurmans, Quoc V. Le et al.ICLR 2023 · 681 citations
- Adversarial NLI: A New Benchmark for Natural Language UnderstandingYixin Nie, Adina Williams, Emily Dinan, Mohit Bansal et al.ACL 2020 · 602 citations
- Autoformalization with Large Language ModelsYuhuai Wu, Albert Qiaochu Jiang, Wenda Li, Markus N. Rabe et al.NeurIPS 2022 · 364 citations
Related papers
- Can we trust LLM Self-Explanations for Entity Resolution?Tommaso Teofili, Donatella Firmani, Nick Koudas, Paolo Merialdo et al.VLDB 2026 · 2 citations
- ConSim: Measuring Concept-Based Explanations' Effectiveness with Automated SimulatabilityAntonin Poché, Alon Jacovi, Agustin Martin Picard, Victor Boutin et al.ACL 2025 · 8 citations
- HEIE: MLLM-Based Hierarchical Explainable AIGC Image Implausibility EvaluatorFan Yang, Ru Zhen, Jianing Wang, Yanhao Zhang et al.CVPR 2025
- Auto-PRE: An Automatic and Cost-Efficient Peer-Review Framework for Language Generation EvaluationJunjie Chen, Weihang Su, Zhumin Chu, Haitao Li et al.AAAI 2026
- The Abstraction Gap in Vision-Language Causal ReasoningChinh Hoang, Mohammad HasanICML 2026
