Saved in:
Bibliographic Details
Main Authors: Yang, Shuai, Yang, Qi, Tang, Luoxi, Meng, Yuqiao, Guo, Nancy, Blackburn, Jeremy, Xi, Zhaohan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.11839
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914330182680576
author Yang, Shuai
Yang, Qi
Tang, Luoxi
Meng, Yuqiao
Guo, Nancy
Blackburn, Jeremy
Xi, Zhaohan
author_facet Yang, Shuai
Yang, Qi
Tang, Luoxi
Meng, Yuqiao
Guo, Nancy
Blackburn, Jeremy
Xi, Zhaohan
contents Counterfactual reasoning has emerged as a crucial technique for generalizing the reasoning capabilities of large language models (LLMs). By generating and analyzing counterfactual scenarios, researchers can assess the adaptability and reliability of model decision-making. Although prior work has shown that LLMs often struggle with counterfactual reasoning, it remains unclear which factors most significantly impede their performance across different tasks and modalities. In this paper, we propose a decompositional strategy that breaks down the counterfactual generation from causality construction to the reasoning over counterfactual interventions. To support decompositional analysis, we investigate \ntask datasets spanning diverse tasks, including natural language understanding, mathematics, programming, and vision-language tasks. Through extensive evaluations, we characterize LLM behavior across each decompositional stage and identify how modality type and intermediate reasoning influence performance. By establishing a structured framework for analyzing counterfactual reasoning, this work contributes to the development of more reliable LLM-based reasoning systems and informs future elicitation strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11839
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle On the Eligibility of LLMs for Counterfactual Reasoning: A Decompositional Study
Yang, Shuai
Yang, Qi
Tang, Luoxi
Meng, Yuqiao
Guo, Nancy
Blackburn, Jeremy
Xi, Zhaohan
Artificial Intelligence
Counterfactual reasoning has emerged as a crucial technique for generalizing the reasoning capabilities of large language models (LLMs). By generating and analyzing counterfactual scenarios, researchers can assess the adaptability and reliability of model decision-making. Although prior work has shown that LLMs often struggle with counterfactual reasoning, it remains unclear which factors most significantly impede their performance across different tasks and modalities. In this paper, we propose a decompositional strategy that breaks down the counterfactual generation from causality construction to the reasoning over counterfactual interventions. To support decompositional analysis, we investigate \ntask datasets spanning diverse tasks, including natural language understanding, mathematics, programming, and vision-language tasks. Through extensive evaluations, we characterize LLM behavior across each decompositional stage and identify how modality type and intermediate reasoning influence performance. By establishing a structured framework for analyzing counterfactual reasoning, this work contributes to the development of more reliable LLM-based reasoning systems and informs future elicitation strategies.
title On the Eligibility of LLMs for Counterfactual Reasoning: A Decompositional Study
topic Artificial Intelligence
url https://arxiv.org/abs/2505.11839