Epilepsy is listed among the most difficult conditions to address with EEG training because of the variety with which epilepsy manifests, because of the ongoing structural deficits which may underlie the condition, and also because historically it has been the most intractable cases which have been referred for EEG training. This historical circumstance introduces a bias into how matters are viewed, since in fact there are many types of seizure that respond quite readily to EEG training. Since training was done at the sensorimotor strip, and was deemed to address the motor system specifically, Sterman argued initially that the training could be expected to be beneficial only for seizures with a predominantly motor symptomatology. Subsequently, however, a controlled study was successfully accomplished with primarily temporal lobe or complex-partial seizures (Lantz, 1988).
The Sterman protocol was replicated for seizures in a number of laboratories and by a number of groups (See References for Table 1.) The technique failed to be acknowledged at the time, however, because of confounding issues regarding anticipated changes in the EEG (Quy, 1979). The cat EEG had manifested a countable increase in incidence of bursts of SMR rhythmic activity with training. The human EEG does not exhibit such bursts except during Stage 2 sleep. And whereas there was in fact an increase in sleep spindles with training in epileptic subjects, the various studies which were intended to replicate Sterman's findings did not yield consistent EEG changes in the waking state (Kaplan, 1975). We now understand that this is not a contradiction. The human EEG remains more desynchronized in the waking state than the cat EEG, and observable bursts would now be considered anomalous. Some individuals did in fact show increased amplitudes of the EEG in the SMR subsequent to long-term reinforcement; others tended toward normalization of their EEG characteristics, which in many instances meant an overall decrease in EEG amplitudes, even within the training band. At the relevant time, however, during the 1970's and 1980's, the lack of consistent EEG changes accompanying the training was thought to be fatal to the hypothesis that EEG training had taken place. The behavioral benefit of EEG biofeedback training that had been replicated in all of the studies was therefore attributed instead to non-specific factors.
Subsequent developments (in our clinical setting) extended the seizure work to absence seizures as well. These generally require higher-frequency training of 15-18Hz in addition to the SMR-training. It is important to make the distinction that in the use of EEG training with seizures, no claim is made that the seizure focus is in any sense extinguished or annihilated. Rather, it is claimed that by enhancing stability conditions in the surrounding healthy brain tissue, the irritable focus will no longer as readily lead to spreading of paroxysmal activity and hence to focal or generalized seizures. The effect of enhancing stability can often be additive to the effect of anti-convulsant medication. It may also lead to the reduction or even elimination of such medication.
More than half of seizures occur at night, and most of these are closely associated with sleep transitions, particularly with falling asleep and waking. This association suggests an intimate connection of seizure susceptibility with stability of arousal. Similarly, about half of all seizures are associated with identifiable events of brain trauma. This of course suggests a connection between the seizure susceptibility with the specific organic loss suffered in the brain injury. However, an equally compelling case can be made that the association is in fact with arousal disregulation here as well. As already indicated in our discussion of brain injury, the predominant symptomatology associated with such injury relates not to the specific location of injury, but to generalized function, in particular the management of arousal. Hence, a brain with an intrinsic seizure vulnerability could simply have been pushed over the edge by a minor head injury.
The hypothesis that efficacy for epilepsy is traceable in large measure to improved regulation of arousal comes from an unusual quarter. A Swedish study has demonstrated some 60% seizure reduction in children by behavioral methods alone (Dahl, 1992). The strategies typically involve deliberate changes in arousal level when the subject anticipates a seizure. As it happens, the 60% reduction is also the average seizure reduction obtained using the Sterman protocol in the various published studies. The nexus with arousal dysfunction helps us to address the structuralist objection that the seizure focus should be impervious to such an intervention as EEG biofeedback. (Whether articulated or not, it is this structuralist objection that has resulted in neurologists dismissing this technique out of hand for thirty years.) It is in fact quite sufficient to argue that only healthy brain tissue is affected by the training in order to explain the clinical findings.
Excerpt from "Applied Neurophysiology & Brain Biofeedback"
Edited by Rob Kall, Joe Kamiya, and Gary Schwartz