A Novel Selective Stimulus Presentation for P300 Speller

The P300 speller is one of the brain-computer interfaces, allowing users to spell letters just by thoughts. Due to the low signal-to-noise ratio of the P300, however, stimuli are repeatedly presented so that EEG signals can be averaged, which improves the accuracy but degrades the speed. The authors have proposed to discontinue the stimulus presentation adaptively to the P300 response and have shown its superiority in the performance over the standard way that presents a prefixed number of stimuli. In addition to this adaptive stimulus termination, this paper proposes to select stimuli to be presented to avoid presenting redundant stimuli. Both off-line and on-line experiments show that the proposed method is more effective than our conventional method.