  |
Back
EEG Biofeedback Training for Hyperactivity,
Attention Deficit Disorder, Specific Learning Disabilities,
and Other Disorders
Siegfried Othmer, Ph.D., and Susan F. Othmer
March, 1989
Page 2 of 5
Back : Next
 Controlled Studies of EEG Biofeedback Training for Epilepsy
In order to establish the validity of the EEG biofeedback training technique, controlled studies were needed. For the case of operant conditioning, such controlled studies consist of ABAB studies, in which the contingency for reward is periodically reversed. Nonspecific effects of the training are ascertained by the use of non-contingent reward, using "yoked controls", where the feedback signal to the patient is derived, unbeknownst to him, from another patient.
Cabral and Scott used EEG biofeedback and relaxation in a crossover design with three cases of drug resistant epilepsy. Both biofeedback and relaxation improved patients' control of their seizures, and the benefit was maintained during the followup period (Cabral, 1976). Wyler, Lockard, and Ward showed that enhancement of EEG activity above 14 Hz, and suppression of activity below 10 Hz also was effective in seizure reduction. A contingency reversal design was used. Synchronization of the EEG worsened seizure incidence; desynchronization of the EEG improved it. Two patients showed improved incidence; two showed improved severity. A fifth was a control with EMG biofeedback alone, and showed no change (Wyler, 1976). An ABAB single-blind study of the effect of EEG biofeedback training on seizure incidence was performed by Sterman and MacDonald (1978). Two frequency bands were employed for enhancement: 12-15 Hz, the SMR band; and 18-23 Hz, in the beta band, associated with EEG activation, focus and arousal. Reduction in seizure incidence was reported for 6 of 8 patients, and amounted to 44-100%, with an average reduction of 74%. When 12-15 Hz was used for reward, and the contingency subsequently reversed, seizure incidence once again increased as expected. However, with the use of 18-23 Hz, the seizure incidence did not change significantly after contingency reversal. (This is not entirely unexpected. Once EEG regulation is effected, the tolerance to epileptogenic EEG activity appears to be permanently enhanced, and not easily reversed.) The experiment was single-blind in that the subject was unaware of the sense of the contingency.
The possibility that the beneficial effect of EEG biofeedback training is nonspecific was ruled out with studies using noncontingent or random rewards. Several studies obtained the consistent result that noncontingent reward was ineffective in seizure reduction. Wyler's and Finley's studies were the first to include such pseudo-conditioning and control periods (Wyler, 1976; Finley, 1976). Such sham training was also provided for in a more exhaustive, double-blind study (neither the experimenter nor the patient was aware of the contingency of reward) undertaken by Lubar, et al. in 1981. Eight medically intractable epileptics were subjected to a lengthy experimental paradigm which included a four-month baseline for recording of seizures, a two-month period of non-contingent reward, a four-month EEG training phase, a reversal phase, a second training phase, another four-month pseudotraining phase, and followup. During the training phase, patients were given one of three contingencies: suppression of 3-8 Hz activity, enhancement of 12-15 Hz, or simultaneous suppression of 3-8 Hz and enhancement of 11-19 Hz activity, the latter to achieve normalization of the EEG.
Five of 8 patients exhibited seizure reduction with respect to baseline, the reduction being 35% for the entire group. The training to reduce 3-8 Hz amplitude was the most effective. Reversal training to enhance 3-8 Hz was likewise effective in increasing seizure incidence, so much so that the training had to be interrupted for two patients who deteriorated rapidly with that training. The fact that facilitation of abnormal EEG patterns can exacerbate seizures provides additional verification of efficacy of the training. Restoration of the appropriate contingency restored the previous gains. Those who trained for EEG normalization did not show deterioration during the reversal phase. EEG studies performed during the training as well as during sleep confirmed a decrease in abnormal low-frequency patterns, as well as increase in mid-frequency activity (12-15 Hz and 16-19 Hz), even when that was not specifically trained for (Whitsett, 1982). Hence, relaxation effects or other nonspecific effects do not appear to explain the results obtained.
These conclusions were recently again confirmed in a controlled study of 24 drug-refractory epileptics, in which impacts of the training on motor, cognitive, and psychosocial function were investigated (Lantz and Sterman, 1988). An overall 61% reduction in seizure incidence was achieved with training, with a range of 0-100%. Cognitive and motor function improved only in that population which achieved significant seizure reduction with training. Psychosocial performance improvements, on the other hand, appeared to be uncorrelated with training history according to some tests, and to be significantly correlated according to others.
The controlled studies just referred to serve to establish that specific benefits in terms of seizure management are achievable with EEG biofeedback training which incorporates the elements of 1) suppression of excessive low-frequency activity, in the 3-8 Hz band; 2) enhancement of activity in the 12-15 Hz or 11-19 Hz bands. These results are not explainable in terms of non-specific factors related to participation in these studies.
Effectiveness of EEG Biofeedback in the Treatment of Hyperactivity, Attention Deficit Disorder, and Learning Disabilities
The utilization of EEG biofeedback in the treatment of hyperactivity was initially an incidental corollary to the study of epilepsy. It was observed that symptoms of hyperactivity subsided during training for seizure reduction with epilepsy (Lubar and Bahler, 1976a). This was not entirely surprising, since hyperactivity may also be regarded in terms of insufficient motor inhibition, and since the EEG observables are similar in general to interictal epileptiform activity: a relative abundance of low-frequency activity (beyond age-appropriate norms), and a relative dearth of intermediate frequency activity (SMR and beta).
The first systematic study of EEG biofeedback effectiveness with hyperactivity in the absence of seizure history was reported by Lubar and Shouse (1976b). An ABAB study, it employed reward for 12-14 Hz, and inhibition of excessive 4-7 Hz. The contingencies were periodically reversed. The subject was able to acquire the SMR task, increasing the fraction of time that SMR was produced above threshold. A number of behaviors associated with the hyperactivity were monitored, and significant changes, in line with expectations, were observed for 8 of 13 behavior categories. The EEG training was shown to be more effective than the use of stimulant medication (methylphenidate, or Ritalin (R)) alone.
A more comprehensive study of hyperkinesis and EEG biofeedback is that of Shouse and Lubar, (1979). A test of stimulant drug withdrawal was included in this work. 3 of 4 subjects showed contingent increases in SMR which were correlated with class room motor inactivity. Combining SMR training with drug treatment resulted in substantial improvements in tested behaviors that exceeded the effects of drugs alone, "and were sustained with SMR training after medication was withdrawn". One subject failed to acquire the SMR task. There was a correlation of pre-training SMR levels with inappropriate motor behavior, as well as with the susceptibility to training.
The undesirable behaviors monitored in the study were disruptive motor activities such as self-stimulation, object play, out-of-seat, self-talk, opposition, and non-interaction. Desirable behaviors were increased attention span and cooperation. Social behaviors evaluated were self-initiated approaches to peers or teachers, and sustained interactions with them. In the three responding individuals, percent of time with SMR increased from 5-35%, 13-25%, and 10-32%, respectively. Desirable behaviors increased from 7-40 events per day, 12-40, and 12-28, respectively; whereas undesirable behaviors decreased from 50-12, 30-12, and 38-12, respectively. The study suggested that the SMR may have both diagnostic and prognostic value in hyperkinesis remediation, with particular regard to motor rather than attentional deficits.
Finally, a clinical study by Lubar and Lubar (1984) extends the technique to attentional deficits and learning disabilities. The appropriateness of doing so is based, among other considerations, on the observation that more than 60% of the cases of learning disability exhibit EEG abnormalities (Muehl, Knott, and Benton, 1965). The experimental protocol was complemented with training in the 15-18 Hz region associated with EEG activation in general, and with arousal and focus. Changes in the EEG were documented with power spectral density measurements, which were compared with those of normal subjects. The EEG biofeedback was also accompanied by academic training. Acquisition of the desired EEG characteristics was observed in all 6 subjects under study. Significant improvements in academic performance were also documented for all of the subjects. A recent review of EEG biofeedback applied to hyperkinesis and learning disabilities may be found in Lubar, (1989). The fact that hyperkinesis and attention deficit disorder is conventionally treated with stimulant medication is evidence that we are dealing with insufficient arousal. This tends to support the enhancement of 15-18 Hz EEG activity as a strategy for activating the arousal and focus mechanisms affecting the sensorimotor cortex, as well as other cortical and subcortical areas of the brain.
One remaining question with the study is that of separating out the effects of the EEG biofeedback from those of the academic augmentation. It was observed that "five of the 6 children were receiving such academic training prior to the onset of the EEG biofeedback training, with no significant improvement over several years." Hence, the beneficial effects are ascribed primarily to the EEG biofeedback. This conclusion is buttressed by the objective changes observed in the EEG, which are appropriate to the training protocol.
A large fraction of specific learning disabilities (as distinct from obvious attentional problems or hyperactivity) also appear to find their basis in minor neurological deficits. By using a large number of indicators, some 95% of learning disabled children could be correctly identified strictly on the basis of the EEG (Lubar, 1989). Using as few as 8 variables, predictability was already above 75%. The best predictor was excessive 4-8 Hz activity in the frontal-temporal locations. Enhancement of beta activity was found to be successful in most of a group of 37 children evaluated over a period of two years. The children showed a significant improvement in Metropolitan Achievement Test scores as compared to controls.
|
 |
