Abstract
Electroencephalography (EEG) signals measured under fixed conditions have been exploited as biometric identifiers. However, what contributes to the uniqueness of one's brain signals remains unclear. In the present research, we conducted a multi-task and multi-week EEG study with ten pairs of monozygotic (MZ) twins to examine the nature and components of person-identifiable brain signals. Through machine-learning analyses, we uncovered a person-identifying EEG component that served as "base signals" shared across tasks and weeks. Such task invariance and temporal stability suggest that these person-identifying EEG characteristics are more of structural brainprints than functional mindprints. Moreover, while these base signals were more similar within than between MZ twins, it was still possible to distinguish twin siblings, particularly using EEG signals coming primarily from late rather than early developed areas in the brain. Besides theoretical clarifications, the discovery of the EEG base signals has practical implications for privacy protection and the application of brain-computer interfaces.