Progress in the Veterans Program
by Siegfried Othmer | December 10th, 2008
We have just completed an introductory training course which included some thirty mental health professionals who are currently working with returning veterans and active duty servicemen. Represented were the United States Navy, the Marine Corps, the Veterans Administration, and the Salvation Army. The intention is to begin pilot projects at a number of facilities to demonstrate neurofeedback efficacy in realistic settings for PTSD and TBI (traumatic brain injury).
With this additional participation, we had our largest training course to date, with some 48 attendees—a logistical challenge to our team. The remaining clinicians who could not be accommodated will come to the January course. We are also prepared to extend invitations to VA personnel from other parts of the country who want to bring neurofeedback into their facilities, on a space-available basis.
The presence of so many clinicians who were not closely acquainted with neurofeedback presented an unusual challenge. Customarily nowadays attendees come with their batteries already charged because of some prior exposure to neurofeedback–either from their own experience, that of one or more of their clients, or on the recommendation of a trusted colleague. They no longer need to be convinced; they just want to learn how to do the work. Accordingly, our training program has been shaped more and more toward the tactics and the practical experience of doing neurofeedback, to the relative neglect of the rich and quirky research history of the biofeedback field and of the scientific models that underpin neurofeedback.
Additionally, these seasoned practitioners had already been humbled by the clinical intractability of PTSD, substance abuse, and TBI, and were no doubt coming with their critical faculties on high alert. Fortunately, we had just extended our course to put the research history back into the program. We also increased the hours spent in practicum during the course.
It is the encounter between the brain and its reflection in our instrument that serves as the ultimate truth test in our course. That’s the advantage of real-time feedback under optimized conditions: most trainees become aware of the induced state shifts during the first session, and all placebo models or other psychological escape hatches are suddenly unavailing. One is confronted with the unadorned reality that state shifts have been systematically induced, mediated only by gazing at one’s own EEG.
Upon seeing the technique demonstrated at the outset with one of the attendees, an observer exclaimed: “This can’t be operant conditioning!” Indeed it is not. The state shifts are happening much too fast to be the result of a training paradigm. An entrainment model takes us closer to the reality. The reinforcements induce the brain into a kind of waveform following (spectral following?), which in turn alters its state of arousal and of local activation. A subtle challenge to the brain’s regulatory mechanisms is set up, as the reinforcements continuously provoke the brain into slight shifts from prevailing states. The watchword is “continuous.” The brain simply becomes a witness to its own unfolding activity. There is no longer a discrete event that is to be rewarded. Rather, there is a continuous process in which the brain is engaged. A byproduct of this stretching exercise is the learned control of state.
In practice, it is then the clinician’s burden to find the conditions under which such training is conducted most efficiently, most productively, and most benignly, with a particular nervous system. Fortunately, we have already mapped that terrain fairly extensively. This then takes us to the second major challenge to one’s belief system, namely the incredible specificity of the training parameters.
This was well illustrated in the demonstration session, which the trainee entered with a migraine headache and in a state of muscle tension. The reward frequency optimization procedure, keying on both of these symptoms, first narrowed things down to the range of 1 Hz to 2.5 Hz. Further optimization led closer and closer to 2.5 Hz, ending up at 2.49 Hz. So, had we been mistaken earlier with regard to 2.5 Hz? Had the brain already altered the quality of its functioning in the interim? Revisiting 2.50 Hz found neither of these to be the case. Whereas reinforcement at 2.49 Hz was welcomed, reinforcement at 2.50 Hz was experienced as adverse. This has to be seen in order to be believed.
Such specificity in reward conditions may not be commonplace, but it is also not rare. There were other instances of it as the attendees all tried to locate their own individual optimal reward frequencies. One person reported after the second such session that she felt like she was seeing herself in the feedback. We do describe this process in that language with reference to the brain: The brain quite naturally engages with the proffered signal because it recognizes its own activity represented there. Rarely, however, does that translate to such a visceral felt experience as in this case. This same person also found the ideal training conditions to be quite specific in frequency, and she was aware almost immediately when the reward parameters were altered by as little as one percent.
The immediacy of the state shift is analogous to the experience of a car changing direction as the steering wheel is turned. The brain follows the reinforcements as adroitly as a car tracks its front wheels. This may not be so problematic when we’re talking about reinforcement frequencies in the higher bands. As is apparent from our recent newsletters, however, we have gradually extended our “bandwidth” in the search for optimum reward frequencies to the infra-low frequency range. Our utilization of the low-frequency range is increasing as our clinical population shifts evermore toward the severe end of the spectrum.
The promptness of state shifts with changes in reinforcement parameters appears to hold even when we dial in very low reward frequencies. This is much harder to explain. Suppose, for example, that a client is being trained at 0.01 Hz. If an adjustment is made to 0.015 Hz, it may take mere seconds, much less than one period of that low frequency (67 seconds) for the person to be aware of the shift. Of course the reinforcements always focus on what is happening “now”—they are oblivious to history even on the 50-second timescale. The momentary change is all the brain ever has to go on to figure out what is being reinforced, even when that activity is dominated by the infra-low frequencies. In that perspective, it is perhaps less surprising that the brain could quickly recognize the mid-course correction. Even so, we may not be quite finished yet with the challenges this new modality presents to rational belief.
In the experience of the course, attendees are inevitably riveted by the most dramatic happenings and the greatest challenges to one’s belief system. However, in a course with large attendance there are inevitably a number of people who do not respond strongly to the training right within the session. The tendency might be to feel left out from all of the excitement that others are experiencing. Upon reflection, however, this could very well be seen as good news. High tolerance to the training under any conditions could be demonstrating the very brain stability that we are trying to promote. On the other hand, it could also indicate a thoroughgoing insensitivity to one’s own state. Clinical skills are once again needed to discern the difference.
The Observer versus the Witness
The principal challenges that neurofeedback presents to our belief system may be traceable to the fact that we are all inclined to look at the neurofeedback process with a “man in the loop.” We are not literally in the feedback loop, of course, but we tend to put ourselves there conceptually. That is to say, we tend to think this process through as an observer, when in fact the process must be seen in the perspective of a participant. The brain does not relate to its own EEG like an external observer. Rather, it relates to it as the active agent, as a witness to its own activity. The brain has some awareness of its own activity as it is being produced, and it detects a correlation with the feedback signal. The brain engages with that signal just as it relates to its ordinary actions in the world—the swinging of a hammer, hand-writing, or driving a vehicle. When the “mirror” to its own activity is suddenly altered, why shouldn’t the brain notice immediately? The usual limits that information theory imposes on the “detection” and discrimination of a given frequency don’t really apply here.
In information theory terms, the “observer” would have to see a substantial fraction of a cycle of 0.01 Hz signal to be certain that 0.01 Hz is in fact the dominant frequency. Discrimination between 0.01 and 0.011 would require an even longer analysis window, etc. The brain does not have that problem when it comes to terms with the feedback signal. We do not, after all, have a narrow filter here that passes only the 0.01 Hz information. We have a soft filter that simply peaks at that frequency. Higher frequency components still make it through, albeit with attenuation. The brain, aware of its own activity, recognizes the whole ‘Gestalt,’ and it is even sensitive to the manner in which the information is shaped by our filters.
The outside observer is much more handicapped in comparison to the witnessing and engaged brain. If we look at an alpha spindle in an EEG, for example, it appears as if the alpha activity is the only thing going on at that moment. If we look at the signal in spectrum analysis, on the other hand, we see that all the activity in other bands is ongoing as well. We therefore know that this activity must be present in the time waveform also, even though we may be unable to discern it. It is simply masked by the large-amplitude alpha waveform. The same thing occurs when we look at a record groove on an old LP with a magnifying glass. We tend to see the dominant frequency, and we only get the full experience of what is in fact recorded there when we listen to the record. Only under these conditions can we hear the whole spectrum of frequencies. The worlds of the external observer and of the experiencing witness are clearly very different.
Now translate all this to the brain in a thought experiment. Consider the brain to be the performing orchestra, and let us also arrange for it to see the record groove being cut into a vinyl recording of its performance at the same time. The brain detects a correspondence. If we now shape the filters on the recording to emphasize particular frequencies, the brain still detects the correlation. If this is unconvincing, imagine that a performance, as recorded and shaped by the filters, is played back in real time to a real orchestra. There will be a small signal processing delay, but musicians live with reverb in large auditoriums all the time. No problem. The playback may sound muffled and appear to emanate from the back of the hall. Is there any doubt that the musicians in the orchestra would still be able to discern the correspondence between their performance and what they are hearing from the playback? There would be no question about the authenticity of what they are hearing even if the recorded signal were to be massively distorted, and they would even be exquisitely sensitive to sudden changes in the filter parameters.
The Great Divide
This very issue is the key to the major divide within the field of neurofeedback. Some people orient toward the observables in the EEG, and explain what happens in neurofeedback through an “observer-centric” model. We, on the other hand, proceed on the basis of a process model that places the brain’s experience of itself at the heart of the matter. This makes brain dynamics the paramount consideration, because the brain attends selectively to what it cares about. This leaves the stationary properties of the EEG with only marginal relevance. The brain doesn’t care very much what its average beta amplitude is. Most of the brain’s activity organizes instantaneous function. It is focused on the present moment. It does not even have the wherewithal to store information about its own past states. It remembers selectively. Up to now, this central issue has not really ever been confronted within the field. The two principal perspectives on neurofeedback are coexisting without interacting. Unsurprisingly, in the observer perspective what we are routinely doing on a daily basis is quite simply impossible! It only makes sense in the witness perspective.
Consider the following iteration on our thought experiment. What if the performance were not played back at all but instead were used to drive the lasers of a Laserium display on the cupola of planetarium? The correspondence between the display and the music would still be apparent even though no information is left to the auditory channel. All that is left is information related to relative timing. Timing is mainly relevant to the present moment. Neurofeedback likewise is a matter of giving the brain relevant information on the present moment. This then takes us very close to what actually happens in our neurofeedback loop. The music of the brain is indeed lost to us. What remains are mere indices of change, but these are sufficient to infuse the brain’s witness of itself with meaning, thus facilitating engagement. What is most relevant to the brain is information regarding the relative timing of events. Everything else is secondary.