Research map / Large language models / AI and society
How do Humans Process AI-generated Hallucination Contents: a Neuroimaging Study
Shuqi Zhu, Yi Zhong, Ziyi Ye, Bangde Du, Yujia Zhou, Qingyao Ai, Yiqun LIU
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| hallucination words correctly recognized vs non-hallucination words: N100 amplitude | human EEG response · EEG ERP, 27 participants | N100 amplitude: occipital F[1,26]=8.271, p<0.05 | Improved |
| correctly recognized hallucination words vs non-hallucination: N400 and P600 | human EEG response · EEG ERP, 27 participants | N400/P600 amplitude: central P600 F[1,26]=31.558, p<0.001 | Improved |
| misjudged hallucination words (HalluWrong) vs non-hallucination words | human EEG response · EEG ERP, 27 participants | ERP amplitude | No clear effect |
| misjudged vs correctly judged hallucination words | human EEG response · EEG ERP, 27 participants | ERP amplitude | Hurt |
| use EEG features to classify hallucinated vs non-hallucinated words and sentences (correctly recognized only) | SVM, GBDT, attention model · HalluCorrect vs NoHallu, within-subject | AUC: word SVM 0.9393; sentence attention 0.9673 | Improved |
| classify EEG when humans failed to detect hallucination (HalluWrong vs NoHallu) | SVM, GBDT, attention model · HalluWrong vs NoHallu | AUC: SVM word within 0.5330 | No clear effect |
| cross-subject instead of within-subject split (ablation) | word-level EEG classifiers · HalluCorrect vs NoHallu word-level | AUC: SVM 0.9393 -> 0.8631 | Hurt |
| sentence-level instead of word-level prediction (ablation) | EEG classifiers · HalluCorrect vs NoHallu | AUC: attention within 0.9113 -> 0.9673 | Improved |
| hallucination type (relation, entity, attribute) on ERP and accuracy (ablation) | human participants · EEG study | recognition accuracy: relation 90.88% vs attribute 86.18% | No clear effect |
Closest papers by meaning
- Auditing Meta-Cognitive Hallucinations in Reasoning Large Language Models (NeurIPS 2025 · similarity 0.87)
- In-Context Sharpness as Alerts: An Inner Representation Perspective for Hallucination Mitigation (ICML 2024 · similarity 0.86)
- Devils in Middle Layers of Large Vision-Language Models: Interpreting, Detecting and Mitigating Object Hallucinations via Attention Lens (CVPR 2025 · similarity 0.86)
- Combating Multimodal LLM Hallucination via Bottom-Up Holistic Reasoning (AAAI 2025 · similarity 0.86)
- Why and How LLMs Hallucinate: Connecting the Dots with Subsequence Associations (NeurIPS 2025 · similarity 0.85)
- HalluciDoctor: Mitigating Hallucinatory Toxicity in Visual Instruction Data (CVPR 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
