arXiv cs.AI· Peihua Mai, Zhuoyan Shao, Xinbao Qiao, Meng Zhang, Xinyue Zhou, Yan Pang·· 10 天前AI 评分37
研究揭示 LLM 存在"错觉模式感知"，会引发虚假推理
Illusory Pattern Perception Drives Spurious Inference in Large Language Models
AI 导读
一项被 NeurIPS 2026 接收的研究首次系统考察了大语言模型中的错觉模式感知，通过改编经典心理学范式设计三项任务并与人类行为直接对比，发现 LLM 常表现出比人类更强的错觉模式感知，倾向于将高频正面属性过度关联至少数多数群体或大型组织，并从模糊事件中构建因果叙事。研究基于稀疏自编码器（SAEs）构建特征可解释性框架分析内部表征，发现整体频率感知与分析性认知取向与错觉感知的出现相关。
正文 · 原文
Abstract:Illusory pattern perception is a well-documented human cognitive tendency to infer meaningful relationships in data that is actually random. Such a tendency, often described as "connecting the dots" where none exist, can result in systematic reasoning errors. This paper investigates whether Large Language Models (LLMs) exhibit such perceptual tendencies, which can lead to systematic errors in downstream applications. To our knowledge, this work presents the first systematic study of illusory pattern perception in LLMs, adapting classic psychological paradigms to three tasks with direct empirical comparison to human behaviors. We find that LLMs frequently exhibit stronger illusory pattern perception than humans. In particular, models tend to over-associate frequent positive attributes with majority groups or large organizations, and show increased tendencies to construct causal narratives from ambiguous events. To uncover the mechanism behind these behaviors, we develop a feature interpretability framework based on Sparse Autoencoders (SAEs) to analyze internal representations. Our results reveal that holistic frequency perception and analytic cognitive orientation are linked to the emergence of illusory perceptions. These findings highlight a previously underexplored cognitive-like illusion that may affect the reliability of LLM reasoning. Code available at this https URL.
| Comments: | accepted by NeurIPS 2026 |
| Subjects: | Artificial Intelligence (cs.AI) |
| ACM classes: | I.5.5 |
| Cite as: | arXiv:2610.07791 [cs.AI] |
| (or arXiv:2610.07791v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.07791 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Peihua Mai [view email]
[v1]
Tue, 6 Oct 2026 05:37:12 UTC (5,750 KB)
来源：arXiv cs.AI · arxiv.org