Name Familiarity Detection using EEG-based Brain Computer Interface

Abstract

Electroencephalography (EEG)-based Brain Computer Interface has been used for image and voice familiarity detection in earlier works. It is known that efficient feature extraction is the key to a good classification accuracy in BCI systems. EEG discriminative features between familiar and unfamiliar names of people, presented in text format as visual stimuli, have not been reported in literature. Detection of familiar and unfamiliar names can play a distinct role in psychology, clinical assessment of cognitive impairment, lie detection, investigations and interrogations. In this paper, the subject's familiarity to names of people was studied based on EEG signals obtained while the subjects were exposed to familiar and unfamiliar names as a visual stimulus in the form of text. The discriminative features were found to be subject-specific. Using three subject-specific features, an average classification accuracy of 84.37% for familiar vs. unfamiliar names was achieved among 8 subjects.