June 1, 2003

Inhibits Again

Author: Siegfried Othmer

For the last number of years, Jonathan Cowan has been promoting broad-band down-training of frontal sites to improve attentional functioning, based on the earlier work of Dan Maust. We of course tried this also, and found that it could not be the whole story, as many people found the training too activating or even mania-inducing. It’s also an obvious hazard for those with propensities toward Tourette-like symptoms.

Also, our own approach to broad-band inhibits took into account the fact that the typical EEG looks very different below 13 Hertz than it does above. A typical compressed spectral array looks like a topographic map of Colorado or Montana: mountains to the West, and flatlands to the East. So we furnished two thresholds, one to take care of the low end of 2-13Hz, and one to take care of the upper end, 14-30 Hz. Absent such dual thresholds, it is likely that the inhibits will be dominated by activity in the lower frequency band in most cases. Despite this modification, the use of the broadband inhibits has not taken over. The question remained open: When to move from the more “standard” theta and high-beta inhibits to broadband inhibits.

We have over the years also gained an appreciation of the specificity of the effects of the inhibits. People can often tell us that they are aware of a change in the training parameters when all we have done is to shift the inhibit band cutoff frequencies. That is sobering. Most recently we have also had some particularly favorable results with broadband inhibits that make it worthwhile to revisit this issue. Given the fact that the specifics of inhibition do seem to matter, it makes sense to us to continue on the same evolutionary pathway on which we have already started. That is, we want to refine the choices available to tailor the inhibits, even within the general context of broadband inhibit training.

It is a natural development simply to divide the two existing bands into more subbands, which opens up more choices. Which bands to choose? At the last SNR Conference there was a talk on the evidence supporting the choice of the traditional standard bands. The upshot of the talk was that there is in fact considerable support for our traditional segmentation, give or take a Hertz here and there. So one approach would divide the lower band roughly into delta, theta, and alpha, and the upper band into perhaps 12-18Hz and 18-30 (or 40) Hz. The generic choices could be “tailored” in the individual case to what is actually observed in the spectral plot to be particularly worthy of attention.

With the BrainMaster it was relatively easy to set up multiple inhibits, and to display them with the Thermobars. Surprisingly, one did not feel overwhelmed by all the data provided in feedback. One picks up the general “gestalt” of what is happening perhaps more than zeroing in on all the details. By the addition of an animation, combined with the use of the autogoal feature on each of the inhibits, it was possible to approximate the conditions that Val Brown obtains with his multiple targeting and dynamic thresholding strategy. It will be interesting to see if we get similar effects in the training as is reported for NeuroCarePro. Likewise, it will be interesting to see if this offers results similar to Jon Cowan’s traditional broadband inhibit training when it is applied to our preferred Fp1-Fp2 placement. The main point of distinction with respect to both of these techniques is that the primary focus in our approach remains on the reward signal.

There is a second alternative approach made possible by BrainMaster, NeuroCarePro, and the new generation of software coming along. We can “overlay” narrowly targeted inhibit bands, as appropriate, on top of the broader bands. This is perhaps the most direct way of bringing about a merger of the traditional inhibit strategy and the broadband strategy. The percent inhibit can be adjusted to give a suitable weighting of the broadband and narrow-band components. One might want to hew somewhat closely to the values we have traditionally used (e.g. 20% inhibit on theta), while keeping a light hand on the broadband inhibits (nominally 10%).

We know of instances in which very narrow-band regions of excess activity have been identified in the high-beta band---for example 23 Hz. One could then zero in on that activity with a 2-Hz wide filter centered on 23 Hz while still utilizing the broadband 14-30 Hz or 22-30Hz filters. We would choose a 2-Hz wide filter in this case for rapidity of response.

Finally, while we are on the subject of inhibits, I would like to recall to mind the option of using the artifact inhibit threshold more actively as an inhibit on adverse EEG activity. In those instances where children go into paroxysmal activity several times a minute, activity that is easily discernible from the ambient background, one might as well use the artifact inhibit to detect these epochs and cut off feedback entirely. There is little point in trying to extract a good reward signal during such events. The other inhibits can then be deployed to detect more subtle disregulations, and will probably end up being more useful.

Without a doubt the availability of the new generation of software from a number of vendors will unleash a flood of innovation, and a lot of that will involve the refinement of inhibit strategies.