A P300-based brain-computer interface: Towards a simpler approach

Brain-computer interfaces (BCIs) allow new ways of communication with computers, independent of the normal neuromuscular pathways. One of the neural mechanisms which can be used in a BCI is the P300 wave. In this paper we focused on the realization of an EEG-based P300 speller application. We used the data set IIb from the BCI Competition 2003. The results achieved with simple methods - using only six electrodes and linear classification - surpass those achieved in the competition. The system achieves full accuracy with four repetitions of the stimuli sequence, and 8.4 seconds are needed to select a symbol.