Development of EEG-Based Communication and Control

D.J. McFarland, T.M. Vaughan & J.R. Wolpaw

Individuals can learn to control the amplitude of the 8-12 Hz mu rhythm or the 18-25 Hz beta rhythm in the EEG recorded over sensorimotor cortex and use it to move a cursor to a target on a video screen (Electroencephalography abd Clinical Neurophysiology, 78: 252-259, 1991 and 90: 444-449, 1994). Successful operation of a brain-computer interface (BCI) requires subject control of the EEG signal and appropriate signal processing in real time. Filtering in time and space can enhance the EEG signal and reduce the influence of artifacts. We have used the EEG signal in a task where cursor movement is a linear function of the EEG amplitude. To allow individuals to move the cursor with equal ease along each axis of control, it is necessary to estimate an appropriate intercept for this linear equation. Use of a running average provides a good estimate. In addition, the slope of the linear equation determines the rate of cursor movement and can affect user performance. The optimal slope varies between individuals and must be determined empirically. Appropriate processing of the EEG signal should allow individuals to successfully use EEG based communications for practical applications.