Introduction / Early Research   Controlled Studies / Effectiveness   Clinical Work Discussion / Summary   References

EEG Biofeedback Training for Hyperactivity,
Attention Deficit Disorder, Specific Learning Disabilities,
and Other Disorders

Discussion of Possible Mechanisms
The EEG training technique discussed above appears to be effective over quite a range of conditions which are traceable to the existence of anomalous EEGs. In the case of hyperactivity, the observable anomaly generally consists of excessive low-frequency activity with insufficient beta activity, related to arousal. In the case of epilepsy, interictal activity is likewise characterized by enhanced low frequency activity (in addition to subclinical spike-and-wave or other epileptiform phenomena). In the case of endogenous depression, the EEG shows abnormally low amplitude overall, in particular low beta activity. In all cases, the changes effected in the EEG are such as to make the EEG more normal. In general, the changes appear to be permanent, once learning has been consolidated. Severe cases may benefit from periodic booster sessions.

The normal adult EEG in the awake and focused state is characterized by relatively low amplitude activity, with the statistical characteristics of noise. The spectral density is roughly monotonically declining with frequency. The result of training is to approach this ideal characteristic. This occurs regardless of the initial EEG characteristic. For example, if the intermediate frequency amplitude is high initially, it will come down toward normal levels, even though the training is reinforcing in that frequency band. This paradoxical result is ascribed to the existence of mid-frequency components of the adverse low-frequency EEG characteristics one wishes to suppress, as well as to the existence of excess high frequency activity (>20 Hz) in many EEGs. When the excessive low frequency components are trained out, the higher frequency components are reduced also. Training also effects elimination over time of the excess high frequency activity. For these reasons, we cannot use an increase in mid-frequency EEG amplitude as a measure of treatment success in all cases. One needs to use more comprehensive criteria for normalcy of the EEG. In practice, the more appropriate observables are the behavioral ones. Behavioral change is often noted well before changes are unambiguously registered within the EEG. The converse may also occur: dramatic changes in the EEG may be noted, with significant behavioral change noted only later. A tight correlation between what is observed behaviorally and what is seen in the EEG is probably not in prospect.

The observation that EEG training effects changes in the EEG toward more normal values, regardless of the starting point, buttresses the hypothesis that the training effects improved cortical regulation. A further observation is that the training appears to have little observable effect on a person characterized by a normal EEG. This suggests that we are not producing a particular brain state (creative or otherwise), but rather are restoring conditions of normalcy when these are absent. Cortical regulation is accomplished by activation of the brain stem and thalamic activating system, and inhibitory feedback circuits involving both nonspecific and specific thalamic nuclei. Hence, we are in all likelihood effecting change subcortically. This also helps to account for the fact that the effects of training are non-local. That is, training one hemisphere may also train the other, and the effect of training on emotional factors indicates an impact on the limbic system as well. Whereas it appears to be true that training at the sensorimotor cortex impacts on the entire brain, the converse is not true. That is, what goes on in the rest of the brain is not necessarily discernible at the sensorimotor cortex. This may account for the absence of tight coupling between what is observed behaviorally and what is seen in the EEG.

The effectiveness of training in the 15-18 Hz spectral band, as well as the 12-15 Hz spectral band, suggests that a more general mechanism than motor inhibition is involved. One may associate coherent activity such as alpha spindles and sleep spindles with self-generative mechanisms within the thalamus or reticular formation, since these spindles occur preferentially when external stimuli are excluded (closed-eyes, or sleep, respectively). By contrast, a state of focused attention which is optimally receptive to sensory inputs (which are random in phase) is likely to be characterized by desynchronized EEG activity, one governed by a stochastic, random process. When a given mechanism is shifted from coherence to incoherence, the observed spectral content shifts to higher frequency, and reduces in amplitude. Conversely, if a given cortical process is entrained to function at a lower frequency (12-15 Hz), it may do so by augmenting inhibitory functions. By using the higher frequency band (15-18 Hz), one may still be training the same mechanism, but one may be training it toward more appropriate activation, rather than specifically enhancing inhibitory processes. This is consistent with the subjective experiences reported with SMR and beta training, which are distinctly different.

Summary and Conclusion
The EEG biofeedback technique appears to be quite successful in effecting remediation in hyperactivity, attention deficit disorder, specific learning disabilities, drug-resistant and other cases of epilepsy, in sleep disorders, and in cases of closed head injury. The effects of training appear to be permanent in most cases.

The unifying criterion underlying the conditions treated appears to be that they exhibit anomalous EEG properties. In a large fraction of the cases, these anomalous EEG properties are traceable to identifiable injury to the brain in the patient's life history, including fetal drug exposure and birth trauma. Detailed family histories commonly indicate a genetic vulnerability or predisposition as well. At least partial normalization of the EEG is a demonstrated consequence of the biofeedback training in most cases. The clinical studies have considerably outdistanced the research to date, and clearly justify further research in order to put these findings on a sound basis, as well as to refine the technique and permit the determination of mechanisms. The generality of the technique suggests that we are dealing with a very fundamental mechanism of cortical regulation affecting areas of the brain beyond the sensorimotor cortex where training takes place.