A Discussion of Mechanisms
In our work, we distinguish between training at higher frequencies (15-18 Hz, which we refer to as "beta") and at lower frequencies (12-15 Hz, or "SMR" training) within the overall beta range of frequencies. These have vastly different effects. In beta training we appear to be dealing with conditions of underarousal, either induced by trauma of some kind, or of genetic origin. In SMR or alpha training, we appear to be dealing with conditions of overarousal, of anxiety, of hypervigilance, of heightened stress susceptibility. Taken together, the training appears to help normalize physiological arousal, i.e. to restore normal modulation and control of arousal level. The increase in seizure threshold with training suggests that the training confers increasing stability in the face of cortical hyperexcitability. The training appears to enhance self-regulation of fundamental arousal mechanisms where these are deficient.

Physiological arousal is under the management of the brain stem, which also regulates the sleep-wake cycle and modulates the pain response. The regulation of cortical function is mediated by the thalamus which, among other things, modulates inputs to the sensorimotor cortex where our (beta) training conventionally takes place. A distribution of frequencies within the low beta range of 12-19 Hz appears to be operative in regulating a variety of cortical functions. Training those specific frequency bands can then be used to elicit certain effects. Among these are regulation of sleep cycles, improved cognitive function, modulation of attention, and increased stability of mood. As suggested above, if one modality has such diverse effects, it must be true that a very central and basic function such as arousal control is being affected by the training. If brain stem function and arousal level are being trained, certain predictions would follow. In particular, we might expect effects on pain mechanisms. We have observed a profound effect on headache syndromes and chronic pain. Another piece of corroborative evidence is the finding that the human brain is peculiarly sensitive to whiplash injury. Even a minor car accident, involving no loss of consciousness but involving whiplash, can leave lingering deficits of the type mentioned above. In whiplash it is the brain stem that is being impacted, yet cortical function suffers! Likewise in birth injury it is frequently the spinal cord and brain stem which take the brunt. A final straw in the wind is that PET studies show the thalamus and the sensorimotor cortex to be in a stage of rapid growth and organization, their most vulnerable period, right at the time of birth. A mutually consistent view is that the EEG is the "window in" to the functioning of the cortex, as it is regulated by subcortical structures such as the brain stem/midbrain, which structures are vulnerable to injury. The EEG training renormalizes this regulatory mechanism.

Many of the above findings of efficacy of EEG training are only clinical, and remain to be confirmed in systematic research. Because of the usual disdain of the neurological community for behavioral management techniques, however, these exciting possibilities find no resonance within medical research. Partly, this is because behavioral techniques really belong to the field of psychology, not medicine. And partly it is because finding a physiological basis for the above conditions has been so elusive. Objective findings supporting the major head injury symptoms listed above are often lacking. MRI and CAT scans are frequently negative. As was recently pointed out in Newsweek, however, such tests are not even able to distinguish a live brain from a dead one. Perhaps they should not be expected to discriminate functional deficits. Tests which measure function rather than morphology, such as quantitative EEG, PET scans, and evoked response measurements, do show up such deficits. Putting it crudely, we have a small, elusive "hardware error" leading to prominent and obtrusive "software errors" in the human brain. The EEG training appears to be able to reinforce the control codes in our "fuzzy- logic" brain and thus remediate functional deficits.

Two findings promise to shift attention to the claims of EEG biofeedback. The first is the report by Alan Zametkin of the NIMH that the brains of hyperactive adults show lowered glucose uptake in the sensorimotor and frontal regions. That is, there is a discernible functional distinction and it is consistent with underarousal. Secondly, we have the recent report by Lewis Baxter of UCLA that behavior therapy for obsessive-compulsive behavior results in activity level changes in the caudate nucleus similar to those elicited by medication for this condition. Finally, we have the recent confirmation by Chris Mann of what had already been well-established, namely that the EEG statistics of ADHD children are significantly different from those of normals, and in line with the underarousal hypothesis. These results may begin to draw the attention of a reluctant medical community to this promising new field.

One key reason for the lack of interest by the medical research community is that the current focus in neurophysiology is on neurotransmitter mechanisms and interactions at the molecular level. The phenomenology we are concerned with cannot yet be described in those terms. ADHD may involve observable differences in serotonin, norepinephrine, and dopamine function, but these may be effects rather than causes. To understand "disorder", we must look at how the brain maintains "order". We must look at the brain as a control and feedback mechanism. A functioning serotonin system is a necessary but insufficient condition for maintenance of "order".