Support-Contra Asymmetry in LLM Explanations

TL;DR

This study investigates how LLM explanations align with predictive lexical cues, revealing a pattern where correct predictions reference supporting evidence and incorrect ones reference contradicting cues.

Contribution

It introduces an empirical analysis of support-contra asymmetry in LLM explanations using external interpretable feature importance signals across multiple datasets.

Findings

Explanations for correct predictions reference more supporting cues.

Explanations for incorrect predictions reference more contradicting cues.

The support-contra asymmetry pattern is consistent across datasets and models.

Abstract

Large Language Models (LLMs) increasingly produce natural language explanations alongside their predictions, yet it remains unclear whether these explanations reference predictive cues present in the input text. In this work, we present an empirical study of how LLM-generated explanations align with predictive lexical evidence from an external model in text classification tasks. To analyze this relationship, we compare explanation content against interpretable feature importance signals extracted from transparent linear classifiers. These reference models allow us to partition predictive lexical cues into supporting and contradicting evidence relative to the predicted label. Across three benchmark datasets-WIKIONTOLOGY, AG NEWS, and IMDB-we observe a consistent empirical pattern that we term support-contra asymmetry. Explanations accompanying correct predictions tend to reference more supporting lexical cues and fewer contradicting cues, whereas explanations associated with incorrect predictions reference substantially more contradicting evidence. This pattern appears consistently across datasets, across reference model families (logistic regression and linear SVM), and across multiple feature retrieval depths. These results suggest that LLM explanations often reflect lexical signals that are predictive for the task when predictions are correct, while incorrect predictions are more frequently associated with explanations that reference misleading cues present in the input. Our findings provide a simple empirical perspective on explanation-evidence alignment and illustrate how external sources of predictive evidence can be used to analyze the behavior of LLM-generated explanations.