ABSTRACTThe current study was concerned with the discrimination of 12–15 Hz (15μV) surface cortical EEG, recorded over the dominant hemisphere. This EEG bandwidth is sometimes called the SMR (sensorimotor rhythm), and has been associated with seizure reduction. Thirty‐six normal subjects were divided into three groups and exposed to three methods of discrimination training: response feedback, production feedback, and no‐feedback control. In the initial assessment session, all subjects were asked to detect the presence of SMR by pressing a response button in the absence of feedback. Over the next 4 training sessions, the control group continued without feedback, while the response feedback group received feedback (tone) for correct discriminations, and the production group received feedback for producing the SMR signal. Discrimination performance was assessed during a 15‐min no‐feedback test period, following each feedback segment. The final session took place two weeks after training had been completed and was the same as session 1. The results showed that both experimental groups improved discrimination accuracy over baseline, and relative to the control group. Performance of the groups at baseline was not significantly different, ranging between 16% and 29% correct. Peak performance during training showed that both experimental groups improved over baseline and relative to the control group, averaging 17%, 43% and 78% correct for the control, response and production feedback groups, respectively. In the final assessment, all groups deteriorated in performance, but the production group remained significantly above baseline levels. No changes in average SMR time or frontal EMG were noted. However, time spent in the occipital alpha bandwidth 8–13 Hz (25μV) did increase in the best discriminators. This may indicate some discrimination of subjective sensations associated with SMR discrimina