Covert attention as a paradigm for subject-independent brain-computer interfacing

Abstract

Humans can voluntarily deploy attention to locations in visual space without moving their eyes. It has been shown that the direction of this so-called 'covert attention' can be decoded on a single-trial basis from EEG and MEG data. In this paper we explore the possibility of using covert attention as a paradigm for subject-independent brain-computer interfacing (BCI). We find that although there is an average decrease of classification performance by moving to subject-independent learning, good subjects tend to remain good, indicating that performance of our BCI system depends on the subject individual performance level rather than whether or not the BCI system is trained using data of the same subject. This suggests that a calibration session can be omitted, which cuts down preparation time and thereby increases the practical relevance of covert attention based BCIs. © 2012 Springer-Verlag.