EEG Biofeedback: Self-regulation Training

The approach to EEG biofeedback which is based on the medical model can probably best be defended by someone who actually uses it. Let me describe the alternative model. The perspective is one of the healthy brain, which has the versatility to modulate arousal states and attentional styles as the immediate situation requires. In the disordered brain, this ability is diminished to some degree, or brain function is compromised by discontinuities in cortical processing, or breakdowns in intra-cortical communications. The EEG reflects first and foremost the state of arousal in which the individual finds himself. EEG biofeedback training, by favoring specific frequency bands, can "move" a person to a different physiological state in the general case, provided merely that the person is willing to "try to train". This is claiming no more than that a person can be changed in physiological state by temperature and EMG training, about which no controversy remains. We are simply using the more "central" information of EEG correlates of physiological states rather than peripheral measures. Moving a person to a different physiological state may benefit him in terms of the experiences he may have in that state, or in terms of an enhanced ability to navigate among different physiological states autonomously. Also, exercising the ability to maintain a particular state tends to reinforce and stabilize the mechanisms by which various states are maintained. These abilities need have little to do with pathology. In fact, these abilities are greater, and can perhaps be enhanced even more, in the more mentally competent person.

Whether someone derives benefit from the training has little to do with any traditional diagnostic categories of mental disorders. The training accomplishes three essential tasks, in our view:

1) It enhances the ability of the individual to access and maintain different states of physiological arousal;

2) it enhances and supports the mechanisms by which the brain manages cortical hyperexcitability;

3) it reinforces equilibrium states, i.e. homeostasis.

With respect to the first, the training enables remediation of arousal disorders; promotes entry into diminished arousal states of alpha and theta for therapeutic objectives; increases the inventory of attentional states (from narrow to broad focus); and facilitates training toward peak performance. EEG training makes a unique contribution here, insofar as traditional EMG and temperature biofeedback have tended to address conditions of overarousal, such as anxiety disorders, hypervigilance, panic, and adverse stress reactions. By contrast, EEG training also addresses conditions of underarousal such as endogenous and reactive depression with equal facility. The efficacy for ADHD can also be seen in terms of remediation of an underarousal condition. And relief from chronic pain can likewise be seen through its intimate connection with depression, i.e. underarousal.

With respect to 2), enhancement of stability conditions, we can identify several "degrees" of instability for our purposes: In the most extreme case, EEG training stabilizes the brain against chaotic excursions into pathological states such as seizures, rages, and migraines. Secondarily, it stabilizes the brain against more minor excursions which manifest themselves in such phenomena as temper tantrums, night terrors, sub-clinical seizures, ordinary headaches, motor and vocal tics, obsessive- compulsive behaviors, episodic dyscontrol, panic attacks, bipolar disorder, and PMS (see Disorder page). Thirdly, it stabilizes the brain against the even lesser disruptions and discontinuities of cortical function which manifest in disturbances of attentional mechanisms, of sequential and parallel processing, of visual and auditory processing and memory, of other specific learning disabilities, and of the normal sequence of sleep stages.

With respect to 3), reinforcing homeostasis, the training achieves normalization of the pain threshold, normalization of appetite, and apparent normalization of the blood glucose level. The stabilization of arousal level already discussed in 1) above can also be viewed as a return to homeostasis. The best evidence of the power of this tool may be seen in the fact that clients are liberated from dependency on all types of medication which modulate arousal, including anti-depressants, stimulants, sleep medications, pain medications, and anxiolytics (including alcohol). This we have consistently observed.

When EEG biofeedback is tried by the mentally competent person, he augments his natural mental skills and his range of control further. What is this worth? That is not for us to say, but we can provide information to the person training, and let him make that decision. Biofeedback, at its best, is empowerment of the individual. We are simply the agency of that empowerment. When Klaus Tennstedt was first offered the opportunity to conduct the Vienna Philharmonic, he was asked by a reporter: "I guess you won't have to rehearse very much with the Philharmonic?" "On the contrary", came the reply, "with that instrument at my disposal, just think of the possibility of refinement, of nuance, that is not possible with ordinary orchestras. We will rehearse more, not less." (My memory may have failed me on the precise quotation.) Thus with EEG biofeedback.

EEG biofeedback training at its best recognizes the heterogeneity of the clinical population and adjusts the training appropriately. In the case of ADHD, we must assess who the person is who has these attentional deficits, and we must characterize the individual in terms of attentional style and patterns of physiological arousal. We must know his family and genetic history, and any history of trauma. Out of all this comes a preferred starting training protocol. This is tried during the intake session. We then teach the individual to observe himself so that he can be a witness to the impending changes. At the next training session, we ask about the results which have been achieved. In at least forty percent of cases, the person will have something to report which is clearly traceable to the training in that first session: sleep may have changed; mood or irritability may have changed; alertness and school behavior may have changed. This allows us to judge within three training sessions in most cases that we are on the right track with a person. Otherwise we change protocol. The multi-dimensional assessment we do on a session-by-session basis is the home turf of the qualified psychological or educational professional. How unfortunate that such a person should be asked to subordinate that information to the tyranny of a single parameter, such as the theta-beta ratio. Is it changing? Is it changing the right way? Who gives a damn, if the client is responding well? The client is a rich source of information about himself, and we should not defer to a single parameter.

The case for the use of this parameter has never been adequately made even by those who insist upon such rigor. Chris Mann, in his QEEG study of ADHD, found that if all the EEG data was taken collectively from baseline, reading, and drawing tasks, a false negative rate of 20%, and a false positive rate of 26%, could be achieved. With a prevalence of ADHD taken to be 10%, the positive predictive power of the EEG, making use of Bayes' Theorem, is only 0.37.

We looked over our data several years ago, and we found in reviewing 167 client files that more than half changed in the "wrong" direction in terms of theta/beta ratio over the course of training, quite irrespective of progress made in the training. Most of the clients did not change significantly one way or the other, which is perhaps the most significant finding. We don't preselect clients for high theta/beta ratio at the outset, and so these results should not be surprising. We get normalization of the EEG, in whatever direction that may lie. (The high frequency part of the spectrum may be high initially, perhaps due to anxiety conditions, and normalization means a decrease in beta amplitudes rather than the reverse.)

One reason that Barry Sterman's early work with epilepsy was not more accepted was due to the fact that the "SMR" amplitude was not found to increase with the training, even though seizure incidence improved. Expectations for such changes in the EEG were simply wrong. A peak in the spectrum at 14 Hz is not part of the normal human, waking EEG, and should not have been expected. We are training (exercising) a mechanism which is operative at 14 Hz; we are not training for an outcome of higher EEG amplitudes there (except operationally during the training session). (Cats do have a peak at 14 Hz in the spectrum, and that did increase with training.)

The biofeedback training protocol which has been used most extensively for epilepsy by Barry Sterman is SMR training (12-15 Hz) around C3 at sensorimotor cortex. This has been true irrespective of any seizure focus elsewhere on the cortex, and irrespective of the nature of the seizure; yet the training has been effective. The strongest case for using brain map data to guide the EEG training can be made for those instances in which we have localized injury, as in seizure disorder, traumatic brain injury, or stroke. Even here, the historical data demonstrates efficacy for training a mechanism, rather than training tailored specifically toward normalizing the EEG. No doubt additional and important gains are to be made by training at the location of the deficit, as determined by EEG data or other means. We use brain mapping in these cases ourselves, and have found the data very informative. Nevertheless, the point remains. In the case of ADHD, for example, the choice of whether SMR training (12-15 Hz) or higher frequency-training (15-18 Hz in our case, or 16-20 in Lubar's) cannot now be made on the basis of the EEG. It is made on the basis of symptoms.

One of the unfortunate plagues that the field of psychology visits upon itself is an envy of physics, a yearning for the provable fact, the hard number. This tyranny of numbers is unfortunate, since the faith is so misplaced. To this, one must add the deferential posture the psychologist bears before medical authority, before medical research procedures, and before the disease model of mental disorders. And finally, one must lament the deferential posture the clinical psychologist bears before the academic research professional. At its best, psychotherapy is integrating and inductive in nature, so contrary to the reductionist propensities of research. Why is the clinical psychologist defensive? The researchers are still arguing among them selves whether recovered memories are real. Rats don't help you answer that question. The truly interesting phenomena in clinical psychology are not reducible to researcher's categories.

It is ironic that a person like Elmer Green, schooled as a physicist, should not feel embarrassed about using terms like "soul", or "subtle energies", and that a hard scientist like Francis Crick should seriously concern himself with theories of consciousness. These matters embarrass psychologists steeped in behaviorism and still trying to be like physicists. Surely the perception of consciousness, or soul, is as real as the perception of pain, and equally resistant to quantification and concrete observation. Clearly the brain manages itself along with the rest of the body. Awareness of self is merely the complement of awareness of out side stimuli; at a sufficient level of complexity, the fusion of self-awareness with a perception of self-interest (self- preservation, self-propagation, etc.) becomes what we perceive as consciousness. Insofar as biofeedback increases the range of information the brain has about itself and can act upon, we are enhancing the scope of consciousness. At its best, psychotherapy nurtures the soul. Biofeedback is its physiological complement.

We have found dramatic effects of EEG biofeedback on such maladies as PMS. The protocol we have used is essentially the same, regardless of what the EEG looks like. And most of the EEGs in fact look fairly normal. There is no known EEG correlate of PMS. Should we desist until one is found upon more subtle examination? There is no known EEG correlate of Tourette Syndrome. Yet our training can be very helpful. Do we desist until EEG criteria are developed? The EEG training is very helpful for temporal lobe epilepsy (complex partial seizures), yet the EEG is not an unambiguous diagnostic. According to George B. Murray, authority on complex partial seizures at the Harvard Psychiatry Department, "The EEG manifestations of complex partial seizures do not usually appear with Cartesian clarity The scope of the EEG manifestations can be as broad as the entire field of EEG" (Murray, 1981). Do we desist until these things are sorted out?

Unfortunately, the EEG reflects only gross anomalies according to present analysis techniques, which are largely based in the frequency domain. Even here, specificity can be improved by comparing EEG data under challenge with that at baseline. With challenge tests, however, we have three pieces of data available: 1) how well the person performs the challenge task; 2) how he reacts under the challenge; and 3) what the EEG looks like under challenge. Among these three, the first two remain the most informative. And as we move away from manifest pathology, the problem gets worse. This can be illustrated with an example from PET scans. Here, the area of the cortex involved with mathematical calculations can be revealed by measuring local glucose uptake or blood flow. If this is tried with a mathematician, however, we find little effect! Mathematical calculations do not challenge his brain enough to significantly change glucose demand. Hence, PET scans do not reveal where math problems are processed as well as where math problems are processed by those who do so badly. It exposes bottlenecks better than function. Likewise with the EEG. As we work with more functional individuals, the common EEG parameters become less instructive.