Unveiling Factual Recall Behaviors of Large Language Models through Knowledge Neurons

TL;DR

This paper explores how large language models recall factual knowledge during reasoning, revealing their reliance on shortcut pathways, and demonstrates that improving factual recall enhances reasoning performance, especially with Chain-of-Thought prompting.

Contribution

The study introduces a method to analyze internal factual recall in LLMs, showing how manipulating recall affects reasoning and how Chain-of-Thought prompts influence factual engagement.

Findings

LLMs often use shortcut pathways instead of factual recall.

Enhancing factual recall improves reasoning accuracy.

Chain-of-Thought prompting increases factual engagement during reasoning.

Abstract

In this paper, we investigate whether Large Language Models (LLMs) actively recall or retrieve their internal repositories of factual knowledge when faced with reasoning tasks. Through an analysis of LLMs' internal factual recall at each reasoning step via Knowledge Neurons, we reveal that LLMs fail to harness the critical factual associations under certain circumstances. Instead, they tend to opt for alternative, shortcut-like pathways to answer reasoning questions. By manually manipulating the recall process of parametric knowledge in LLMs, we demonstrate that enhancing this recall process directly improves reasoning performance whereas suppressing it leads to notable degradation. Furthermore, we assess the effect of Chain-of-Thought (CoT) prompting, a powerful technique for addressing complex reasoning tasks. Our findings indicate that CoT can intensify the recall of factual knowledge by encouraging LLMs to engage in orderly and reliable reasoning. Furthermore, we explored how contextual conflicts affect the retrieval of facts during the reasoning process to gain a comprehensive understanding of the factual recall behaviors of LLMs. Code and data will be available soon.