Self-Critique and Refinement for Faithful Natural Language Explanations

TL;DR

This paper introduces SR-NLE, a framework enabling large language models to iteratively critique and refine their natural language explanations, significantly improving their faithfulness without external supervision.

Contribution

We propose a novel self-critique and refinement framework for natural language explanations that enhances faithfulness through iterative feedback mechanisms, including a new feature attribution-based approach.

Findings

Reduces unfaithfulness rate from 54.81% to 36.02% on average

Effective across multiple datasets and models

No additional training or fine-tuning required

Abstract

With the rapid development of Large Language Models (LLMs), Natural Language Explanations (NLEs) have become increasingly important for understanding model predictions. However, these explanations often fail to faithfully represent the model's actual reasoning process. While existing work has demonstrated that LLMs can self-critique and refine their initial outputs for various tasks, this capability remains unexplored for improving explanation faithfulness. To address this gap, we introduce Self-critique and Refinement for Natural Language Explanations (SR-NLE), a framework that enables models to improve the faithfulness of their own explanations -- specifically, post-hoc NLEs -- through an iterative critique and refinement process without external supervision. Our framework leverages different feedback mechanisms to guide the refinement process, including natural language self-feedback and, notably, a novel feedback approach based on feature attribution that highlights important input words. Our experiments across three datasets and four state-of-the-art LLMs demonstrate that SR-NLE significantly reduces unfaithfulness rates, with our best method achieving an average unfaithfulness rate of 36.02%, compared to 54.81% for baseline -- an absolute reduction of 18.79%. These findings reveal that the investigated LLMs can indeed refine their explanations to better reflect their actual reasoning process, requiring only appropriate guidance through feedback without additional training or fine-tuning.