Conversion of EEG Activity into Cursor Movement by a Brain Computer Interface (BCI) *

Georg E. Fabiani

The Wadsworth Brain Computer Interface (BCI) uses electoencephalographic (EEG) activity to control a cursor on the screen. The aim is to provide communication for severely disabled people. A trained user is able to move a cursor up and down. ‘Yes’/’No’ questions can be answered by hitting the top/bottom of the screen with the cursor. The cursor moves ten times per second based on the user’s EEG. Currently, cursor movements are calculated by an empirically derived linear equation. This presentation discusses a more theoretical approach based on Bayesian Discriminant Analysis. The method uses available data to calculate the posterior probability that the outcome of the trial is ‘Yes’ or ‘No’. It also provides cursor movements for feedback to the user. It can process multidimensional input data and thereby improve the performance of the system. This presentation shows how the theoretically derived cursor movement is related to the currently used linear one, and how they compare in simulated practice.