Assessing Step-by-Step Reasoning against Lexical Negation: A Case Study on Syllogism

Mengyu Ye, Tatsuki Kuribayashi, Jun Suzuki, Goro Kobayashi, Hiroaki Funayama

Submission Type: Regular Short Paper

Submission Track: Interpretability, Interactivity, and Analysis of Models for NLP

Submission Track 2: Language Modeling and Analysis of Language Models

Keywords: Negation, Prompting, Reasoning, Language model, Model Analysis, Probing, Chain-of-thought

TL;DR: This work analyzed reasoning ability of large language models(LLMs) against lexical negation using step-by-step reasoning and shows that LLMs still struggled with addressing such negation.

Abstract:

Large language models (LLMs) take advantage of step-by-step reasoning instructions, e.g., chain-of-thought (CoT) prompting.
Building on this, their ability to perform CoT-style reasoning robustly is of interest from a probing perspective.
In this study, we inspect the step-by-step reasoning ability of LLMs with a focus on negation, which is a core linguistic phenomenon that is difficult to process.
In particular, we introduce several controlled settings (e.g., reasoning in case of fictional entities) to evaluate the logical reasoning abilities of the models. We observed that dozens of modern LLMs were not robust against lexical negation (e.g., plausible$\rightarrow$implausible) when performing CoT-style reasoning, and the results highlight unique limitations in each LLM family.

Submission Number: 1489