Analysis of the Neglect-Zero Effect in Large Language Models

2026-06-04

AI Summary

This study introduces, for the first time, the human cognitive phenomenon known as the “null-model neglect effect” into research on large language models (LLMs), investigating whether LLMs overlook trivially true statements that arise from empty sets during logical reasoning. Employing a structural priming paradigm, the authors designed priming sentences to highlight null-model contexts and examined the models’ subsequent reasoning behavior in target sentences. Experimental results indicate that prevailing LLMs do not exhibit this effect, revealing a marked divergence from human cognition. This work not only expands the evaluative dimensions of LLMs’ sensitivity to logical structure but also offers a novel perspective for understanding their formal reasoning mechanisms.

Abstract

We investigate the extent to which the language processing of LLMs resembles human cognitive processes, focusing on a human cognitive bias called the $\textit{neglect-zero effect}$. This effect refers to the human tendency to ignore $\textit{zero-models}$, which are configurations that render a proposition vacuously true by virtue of an empty set. We focus on two types of inferences driven by the neglect-zero effect, and examine how LLMs process these inferences by comparing their behavior with that in an inference that does not involve the neglect-zero effect. For this purpose, we employ a paradigm based on $\textit{structural priming}$, where recent exposure to a preceding sentence (the $\textit{prime}$) facilitates the processing of a subsequent sentence (the $\textit{target}$) due to their structural similarity. We prepare primes to force LLMs to consider the zero-model, and analyze whether they also consider it in the target. The results suggest that the neglect-zero effect may not occur in the LLMs analyzed in this study. Our code is available at https://github.com/ynklab/neglect_zero

Problem

• neglect-zero effect
• large language models
• zero-models
• cognitive bias
• vacuous truth

Innovation

• neglect-zero effect
• large language models
• structural priming
• zero-models
• cognitive bias