How Likely Do LLMs with CoT Mimic Human Reasoning?

TL;DR

This paper investigates how chain-of-thought prompting affects LLM reasoning by analyzing the causal structure of their reasoning process compared to humans, revealing deviations and factors influencing reasoning quality.

Contribution

It introduces a causal analysis framework to diagnose LLM reasoning, highlighting how different training techniques impact reasoning structure and identifying limitations of model size increases.

Findings

LLMs often deviate from ideal causal reasoning chains

In-context learning strengthens causal structure

Fine-tuning and reinforcement learning weaken causal structure

Abstract

Chain-of-thought emerges as a promising technique for eliciting reasoning capabilities from Large Language Models (LLMs). However, it does not always improve task performance or accurately represent reasoning processes, leaving unresolved questions about its usage. In this paper, we diagnose the underlying mechanism by comparing the reasoning process of LLMs with humans, using causal analysis to understand the relationships between the problem instruction, reasoning, and the answer in LLMs. Our empirical study reveals that LLMs often deviate from the ideal causal chain, resulting in spurious correlations and potential consistency errors (inconsistent reasoning and answers). We also examine various factors influencing the causal structure, finding that in-context learning with examples strengthens it, while post-training techniques like supervised fine-tuning and reinforcement learning on human feedback weaken it. To our surprise, the causal structure cannot be strengthened by enlarging the model size only, urging research on new techniques. We hope that this preliminary study will shed light on understanding and improving the reasoning process in LLM.