Is Moral Self-correction An Innate Capability of Large Language Models? A Mechanistic Analysis to Self-correction

Zimo Qi, Guangliang Liu, Kristen Marie Johnson, Lu Chen

Abstract: Though there has been intensive attention to the self-correction capability of Large Language Models (LLMs), conclusions regarding its effectiveness remain varied. In this paper, we investigate a fundamental question: is moral self-correction an innate capability in LLMs? To explore this, we conduct (1) a mechanistic analysis of how key components of self-correction, such as Chain-of-Thought (CoT) reasoning and external feedback, interact to enable moral self-correction; and (2) a behavioral analysis of LLMs' ability to distinguish between desired and undesired outputs, introducing a self-distinguish framework. Our mechanistic analysis reveals that LLMs struggle to effectively leverage helpful feedback, and conflicts can arise between feedback and CoT reasoning. These limitations suggest that LLMs fail to identify useful contextual information, instead prioritizing their own internal knowledge. Additionally, our behavioral analysis indicates that LLMs struggle to differentiate among their own outputs. Based on these empirical findings across two analytical dimensions, mechanism and behavior, we argue that moral self-correction is not an innate capability of LLMs.