Theory-Grounded Evaluation of Human-Like Fallacy Patterns in LLM Reasoning

2025/06/10 by Andrew Richardson, Richardson, Andrew Keenan, Kearns, Ryan Othniel +5
Computer Science · #Topic Modeling #Explainable Artificial Intelligence (XAI)

paper · pdf · doi:10.48550/arxiv.2506.11128

Abstract

We study logical reasoning in language models by asking whether their errors follow established human fallacy patterns. Using the Erotetic Theory of Reasoning (ETR) and its open-source implementation, PyETR, we programmatically generate 383 formally specified reasoning problems and evaluate 38 models. For each response, we judge logical correctness and, when incorrect, whether it matches an ETR-predicted fallacy. Two results stand out: (i) as a capability proxy (Chatbot Arena Elo) increases, a larger share of a model's incorrect answers are ETR-predicted fallacies (ρ=0.360, p=0.0265), while overall correctness on this dataset shows no correlation with capability; (ii) reversing premise order significantly reduces fallacy production for many models, mirroring human order effects. Methodologically, PyETR provides an open-source pipeline for unbounded, synthetic, contamination-resistant reasoning tests linked to a cognitive theory, enabling analyses that focus on error composition rather than error rate.

Citations

• Evaluation data contamination in LLMs: how do we measure it and (when) does it matter?
• A Systematic Analysis of Large Language Models as Soft Reasoners: The Case of Syllogistic Inferences
• Observational Scaling Laws and the Predictability of Language Model Performance
• Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference
• A Survey on Knowledge Distillation of Large Language Models
• Task Contamination: Language Models May Not Be Few-Shot Anymore
• Investigating Data Contamination in Modern Benchmarks for Large Language Models
• A Systematic Comparison of Syllogistic Reasoning in Humans and Language Models
• Large Language Models
• Humans in Humans Out: On GPT Converging Toward Common Sense in both Success and Failure
• Large Language Models Can Be Easily Distracted by Irrelevant Context
• Language Models Are Greedy Reasoners: A Systematic Formal Analysis of Chain-of-Thought
• BNAI, NO-TOKEN, and MIND-UNITY: Pillars of a Systemic Revolution in Artificial Intelligence
• Measurement as governance in and for responsible AI
• Compression of Deep Learning Models for Text: A Survey
• Knowledge Distillation: A Survey
• Language Models are Few-Shot Learners
• Transformers as Soft Reasoners over Language

Related

• Reason from Fallacy: Enhancing Large Language Models' Logical Reasoning through Logical Fallacy Understanding
• RULEBREAKERS: Challenging LLMs at the Crossroads between Formal Logic and Human-like Reasoning
• Follow My Lead: Logical Fallacy Classification with Knowledge-Augmented LLMs
• Reasoning as Pattern Matching: Shared Mechanisms in Human and LLM Everyday Reasoning
• Trustworthy Reasoning: Evaluating and Enhancing Factual Accuracy in LLM Intermediate Thought Processes