Discrimination of Discrete Feedback During Performance of Motor Imagery

Experiments were performed to determine if discrete feedback could be discriminated during concurrent performance of motor imagery for brain-computer interface control. An inverse solution feature extraction method captured regions of activity indicative of conflict detection and nterpretation of feedback. Separate classifiers were trained to detect positive and negative feedback based on constrained rates of precision. Tested in simulation, classification rates suggest discrimination of discrete feedback embedded within a trial can improve accuracy in less able brain-computer interface subjects, with little impact on subjects who perform well.