Research

Paper

AI LLM February 24, 2026

Linear Reasoning vs. Proof by Cases: Obstacles for Large Language Models in FOL Problem Solving

Authors

Yuliang Ji, Fuchen Shen, Jian Wu, Qiujie Xie, Yue Zhang

Abstract

To comprehensively evaluate the mathematical reasoning capabilities of Large Language Models (LLMs), researchers have introduced abundant mathematical reasoning datasets. However, most existing datasets primarily focus on linear reasoning, neglecting other parts such as proof by contradiction and proof by cases, which are crucial for investigating LLMs' reasoning abilities. To address this limitation, we first introduce a novel first-order logic (FOL) dataset named PC-FOL, annotated by professional mathematicians, focusing on case-based reasoning problems. All instances in this dataset are equipped with a manually written natural language proof, clearly distinguishing it from conventional linear reasoning datasets. Our experimental results over leading LLMs demonstrate a substantial performance gap between linear reasoning and case-based reasoning problems. To further investigate this phenomenon, we provide a theoretical analysis grounded in graphical model, which provides an explanation for the observed disparity between the two types of reasoning problems. We hope this work can reveal the core challenges in the field of automated natural language mathematical proof generation, paving the way for future research.

Metadata

arXiv ID: 2602.20973
Provider: ARXIV
Primary Category: cs.CL
Published: 2026-02-24
Fetched: 2026-02-25 06:05

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2602.20973v1</id>\n    <title>Linear Reasoning vs. Proof by Cases: Obstacles for Large Language Models in FOL Problem Solving</title>\n    <updated>2026-02-24T14:53:34Z</updated>\n    <link href='https://arxiv.org/abs/2602.20973v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2602.20973v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>To comprehensively evaluate the mathematical reasoning capabilities of Large Language Models (LLMs), researchers have introduced abundant mathematical reasoning datasets. However, most existing datasets primarily focus on linear reasoning, neglecting other parts such as proof by contradiction and proof by cases, which are crucial for investigating LLMs' reasoning abilities. To address this limitation, we first introduce a novel first-order logic (FOL) dataset named PC-FOL, annotated by professional mathematicians, focusing on case-based reasoning problems. All instances in this dataset are equipped with a manually written natural language proof, clearly distinguishing it from conventional linear reasoning datasets. Our experimental results over leading LLMs demonstrate a substantial performance gap between linear reasoning and case-based reasoning problems. To further investigate this phenomenon, we provide a theoretical analysis grounded in graphical model, which provides an explanation for the observed disparity between the two types of reasoning problems. We hope this work can reveal the core challenges in the field of automated natural language mathematical proof generation, paving the way for future research.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-02-24T14:53:34Z</published>\n    <arxiv:primary_category term='cs.CL'/>\n    <author>\n      <name>Yuliang Ji</name>\n    </author>\n    <author>\n      <name>Fuchen Shen</name>\n    </author>\n    <author>\n      <name>Jian Wu</name>\n    </author>\n    <author>\n      <name>Qiujie Xie</name>\n    </author>\n    <author>\n      <name>Yue Zhang</name>\n    </author>\n  </entry>"
}