How do Humans Process AI-generated Hallucination Contents: a Neuroimaging Study

2026-05-16

AI Summary

It remains unclear why humans are susceptible to being misled by hallucinated content generated by multimodal large language models. This study addresses this gap by recording electroencephalography (EEG) signals from 27 participants performing a task that required judging the veracity of AI-generated image captions. Through analysis of event-related potentials (ERPs), the work reveals— for the first time—that hallucinated statements fail to effectively engage the canonical neurocognitive pathways associated with fact verification. The findings demonstrate significant neural dynamic differences between processing hallucinated and non-hallucinated content during semantic integration, reasoning, and memory retrieval. Moreover, distinct EEG patterns emerge when participants misclassify versus correctly identify hallucinations, providing critical neuroscientific evidence for understanding the perceptual and cognitive mechanisms underlying human susceptibility to AI-generated hallucinations.

Abstract

While AI-generated hallucinations pose considerable risks, the underlying cognitive mechanisms by which humans can successfully recognize or be misled by these hallucinations remain unclear. To address this problem, this paper explores humans' neural dynamics to characterize how the brain processes hallucinated content. We record EEG signals from 27 participants while they are performing a verification task to judge the correctness of image descriptions generated by a multi-modal large language model (MLLM). Based on an averaged event-related potential (ERP) study, we reveal that multiple cognitive processes, e.g., semantic integration, inferential processing, memory retrieval, and cognitive load, exhibit distinct patterns when humans process hallucinated versus non-hallucinated content. Notably, neural responses to hallucinations that were misjudged versus correctly judged by human participants showed significant differences. This indicates that misjudged AI-generated hallucinations failed to trigger the standard neurocognitive fact verification pathway.

Problem

• AI-generated hallucination
• human cognition
• neuroimaging
• fact verification
• cognitive mechanisms

Innovation

• AI hallucination
• neuroimaging
• EEG
• event-related potential
• cognitive mechanisms