Surpassing the uncanny valley

Published: 13 January 2026| Version 1 | DOI: 10.17632/kjv4dckx8m.1
Contributors:
,

Description

These data are relative to the paper "Neural Signatures of Hyper-Realistic AI-Generated Faces: Dissociating Behavioral Indistinguishability from Implicit Neural Evaluation". The present study aimed to determine whether the human brain preserves sensitivity to the artificial origin of hyper-realistic AI-generated faces even when explicit, behavioral discrimination from real faces fails. By integrating behavioral validation with high-density EEG recordings, we sought to characterize the temporal dynamics and neural substrates underlying the processing of real versus GAN-generated faces across perceptual, evaluative, and familiarity-related stages. Specifically, the study investigated whether implicit neural markers reveal systematic differences in the processing of artificial faces that are behaviorally indistinguishable from real ones, thereby dissociating overt recognition performance from covert neural evaluation.

Files are not publicly available

You can contact the author to request the files

Institutions

University of Milano-Bicocca

Departments

Department of Psychology

Categories

Artificial Intelligence, Affective Computing, Cognitive Neuroscience

Funders

  • Italian Ministry of University and Research
    Grant ID: 2023-NAZ-0206, PsyFuture – Dipartimento di Eccellenza 2023-2027

Additional Metadata for University of Milano - Bicocca

Date the data was collected2025-06-02T22:00:00.000Z
SSD ClassificationM-PSI/02 - PSICOBIOLOGIA E PSICOLOGIA FISIOLOGICA