A Closer Look at the Self-Verification Abilities of Large Language Models in Logical Reasoning

TL;DR

This paper evaluates the self-verification abilities of large language models in logical reasoning, revealing their struggles with identifying fallacies and highlighting the need for improved self-assessment techniques.

Contribution

It introduces the FALLACIES dataset for categorizing reasoning fallacies and provides a comprehensive analysis of LLMs' ability to detect logical errors.

Findings

LLMs often fail to accurately identify reasoning fallacies.

Existing models struggle to guarantee the validity of their self-verification.

The study offers insights for enhancing self-verification methods in LLMs.

Abstract

Logical reasoning has been an ongoing pursuit in the field of AI. Despite significant advancements made by large language models (LLMs), they still struggle with complex logical reasoning problems. To enhance reasoning performance, one promising direction is scalable oversight, which requires LLMs to identify their own errors and then improve by themselves. Various self-verification methods have been proposed in pursuit of this goal. Nevertheless, whether existing models understand their own errors well is still under investigation. In this paper, we take a closer look at the self-verification abilities of LLMs in the context of logical reasoning, focusing on their ability to identify logical fallacies accurately. We introduce a dataset, FALLACIES, containing 232 types of reasoning fallacies categorized in a hierarchical taxonomy. By conducting exhaustive experiments on FALLACIES, we obtain comprehensive and detailed analyses of a series of models on their verification abilities. Our main findings suggest that existing LLMs could struggle to identify fallacious reasoning steps accurately and may fall short of guaranteeing the validity of self-verification methods. Drawing from these observations, we offer suggestions for future research and practical applications of self-verification methods.