EEG Training Protocols Guide

Guide to EEG Neurofeedback Protocols in MBTT Part 1: Standard Basic Protocols

Introduction
What Is a Neurofeedback Protocol?
Working with Protocols in Mind-Body Training Tools
MBTT Basic (or Standard) vs. Advanced Protocols
Inhibit Parameters and Inhibit Thresholds
Basic Protocols Classified by Parameter Type
Preconfigured Standard (Basic) Protocols Listing

Introduction

This guide is part of a set that teaches you how to use Mind-Body Training Tools for EEG neurofeedback. It builds on the foundation presented in the EEG Neurofeedback Training App User Guide (which you are strongly recommended to read first).

MBTT EEG neurofeedback training software is built around the concept of the protocol. A protocol describes what particular EEG parameter to train, and how. In the software a protocol is an explicit entity, having a name and several other attributes. When running neurofeedback sessions, a necessary first step is to select (or create) a protocol - you can't run a session without this step.

In neurofeedback, selecting the most appropriate protocol for any given client or trainee, is key to success. How to make that choice is beyond the scope of this guide - for that you need to take a training such as my Neurofeedback Practitioner Programme. For now, suffice it to say that the decision takes account of both the trainee's goal (or conversely the issues they'd like to overcome) and the particular patterns that manifest in the trainee's EEG (as measured and revealed by an EEG assessment).

What Is A Neurofeedback Protocol?

In neurofeedback, a protocol the decision of:

what particular EEG parameter or parameters to train (e.g. amplitude of the alpha frequency band)
whether you want to enhance it or suppress it (i.e. train up or down)
where on the head you wish to train it (including how many channels of EEG to use in training)
how long to train (within a session) and how to structure this time, e.g. 6 x 2-minute periods of active training with rests between.

As already stated, in MBTT a protocol exists as a software entity that specifies most of the above, namely the parameter(s), direction of training (up or down) and the number of channels. Note that threshold levels are also stored by the software within each protocol.

As far as the software goes, the electrode sites (and including the montage) is not explicitly represented - the software runs the same regardless of where your sensors are. That is certainly not to say it doesn't matter - the practitioner needs to be systematic about this choice, and to keep track of their choice.

Also not defined within the software entity is how many training periods to run, and how long each should be. You can make this decision “on the fly”, but again it's something you should think about beforehand, though at the same time the final judgements will depend on how the trainee responds during the session.

Working with Protocols In MBTT

Again, the first step in running a neurofeedback session is to load a protocol. You can select a preconfigured protocol, or you can create your own, either by taking a copy of a preconfigured protocol, or by setting one up from scratch.

If you're a professional practitioner working with clients, it's good practice to set up your own protocols, for example one for each client. This allows you to make small edits for the individual client (if you wish).

When it comes to assessing progress (over sessions) you'll want to use the protocol report. This is based on all sessions run using one particular protocol. It wouldn't make sense to combine sessions that used different protocols.

Classes of Protocols

The preconfigured protocols in MBTT come in two classes: (i) basic (or standard) and (ii) advanced.

What distinguishes the two? Largely it comes down to feedback, or more specifically, how many independent feedbacks are available. To understand this, let's review the concept of feedback.

The Concept of Feedback

Feedback in its most general sense is information about how well you're performing some particular skill, that enables adjustments and improvements to the performance based on learning. More specifically, in neurofeedback training, feedback is what the software presents to the trainee to see or hear. Changes in this feedback convey the information that enables learning.

Feedback can be visual, e.g. a graph that goes up or down, or a video, or a game. Or feedback can be auditory – a sound with varying volume or pitch.

A key principle of feedback is that it should (ideally) help to motivate the trainee in some way, i.e. it should incorporate a reward for “success”. Having the experience of reward makes learning more efficient in some sense.

Whatever the medium of feedback, there are different classes of feedback:

Dichotomous or Stop-go – e.g. a video, which is playing (when the parameter is above threshold) or it is stopped.
“Points-based”. Feedback is a discrete event, such as the scoring of a point in a game, which is triggered by the training parameter crossing the threshold (e.g. alpha amplitude exceeds 20uV).
Proportional. For example video feedback where brightness varies over a range. Instead of a discrete feedback event, (e.g. starting or stopping) the feedback is continuous but varies smoothly over a range. The video can be bright or dark or anywhere in between. This form of feedback mirrors the training parameter which also varies over a range.

Feedback is discussed in more depth in the Guide to Feedback in MBTT Neurofeedback Training.

Working with More Than One Feedback Parameter

Suppose you're working with a protocol that involves training more than one EEG parameter, e.g. you want to train beta amplitude up, and theta amplitude down. In the software, you'd use two thresholds, one for beta amplitude and one for theta amplitude.

This is clearly more complex than just training a single parameter such as alpha amplitude. What does this mean for how we present feedback?

In principle, we have two choices:

we can combine the two parameters into one, and present a single feedback as before, or
we can independently feed back each parameter.

A Note on Terminology: Feedback Stream vs. Feedback Parameter

Now we are speaking of one feedback vs. two feedbacks, we need a new terminology to refer to this. I settled on "feedback stream". By this I mean the aspect of the actual feedback which is varying in a way directly perceivable by the senses. I want to distinguish this from "feedback parameter" which is the calculated EEG variable that gets mapped to the feedback.

Two feedback parameters could be for example theta amplitude and beta amplitude. Two feedback streams could be the brightness of a video and the volume of the video soundtrack.

Examples of Feedback

If we're using either stop-go or points based feedback:

Combining into a single feedback stream would mean the feedback event is triggered by both threshold conditions being met simultaneously. For example, our stop-go video would play when beta amplitude is above its threshold level AND theta amplitude is below its threshold level, at the same time.
Two independent feedbacks would mean that we could for example play two different contingent sounds; a high pitched piano note when beta is above threshold, and a low pitched note when theta is below threshold.

If we're using proportional feedback:

Combining into a single feedback would mean we take an average of the two parameters, and feed back this average.
Independent feedbacks would mean we could play two soundtracks simultaneously, one whose volume varied proportionally to theta amplitude, and a second whose volume varied with beta amplitude. The two soundtracks would have to complement each other: nature soundtracks work well in this respect. For example the sound of flowing water represents theta amplitude, and the sound of birds singing represents beta amplitude. It wouldn't make sense to use video feedback in this case because you can't really pay attention to both. However you could have video brightness linked to one parameter and video volume linked to the other.

Trainee Outcomes and the Arousal Model

It's perhaps worth mentioning that in this guide we're not classifying protocols according to their intended usage or outcome. I do make some comments about application of the protocols in the listings below, where I sometimes make reference to the arousal model. The arousal model is beyond the scope of this guide to cover - it's something that would be rightly covered in a neurofeedback training programme.

Here is a very brief summary of the model:

Client issues are classed as resulting from either over-arousal in the brain, or under-arousal (though not every case falls neatly within this dichotomy).
Over-arousal cases have both (i) typical presenting problems and issues and (ii) EEG markers which can be investigated in an EEG assessment. Under-arousal cases are the same.
Some protocols are seen as activating, or aimed at increasing arousal, while others are relaxation protocols, aimed at lowering arousal.

MBTT Basic vs. Advanced Protocols

The main difference between basic (standard) and advanced protocols is that basic protocols have only one channel of feedback, while advanced protocols can have two or more.

A secondary difference is that basic protocols have only one main threshold while advanced protocols can have two or more thresholds.

Having only one main threshold means there is only one EEG parameter, or one training parameter. However, there are ways of combining EEG parameters into one, so that you are implicitly training two parameters. For example, you could train the ratio of theta amplitude to beta amplitude. Down-training this ratio is equivalent to training theta down and beta up.

Inhibit Parameters & Inhibit Thresholds

Basic protocols may also have one or two inhibit thresholds, as can advanced protocols. Inhibit thresholds are described in the main Guide to Feedback in MBTT Neurofeedback Training.

In the MBTT software, inhibit parameters and inhibit thresholds work differently from main feedback parameters, and also, differently to other neurofeedback software products. It's perhaps worth summarising the differences, for the sake of clarity.

More generally in neurofeedback, an inhibit refers to something that is trained down, or at least held back from going up.

In MBTT, an inhibit parameter is a secondary type, which is not fed back directly in the same way as a main feedback parameter. Rather, it in some sense modulates the feedback of a main parameter.

In MBTT a main feedback parameter can be trained down, but it is still a main parameter and not an inhibit parameter in this special sense.

Basic Protocols Classified by Parameter Type

The class of basic protocols can be sub-divided according to how the main parameter (the parameter of the main threshold) is calculated. Parameter types include:

Amplitude of a frequency band (e.g. theta amplitude)
Ratio of two amplitudes (i.e. amplitudes of frequency bands) e.g. theta to beta ratio, or fast to slow alpha ratio.
Coherence of a frequency band (e.g. alpha coherence) (Requires at least two channels of EEG.)
Parameters based on FFT or spectral analysis e.g. alpha percentage by FFT.

The remainder of this guide lists the preconfigured protocols, grouping them by class, and offering some comments on their usage.

Preconfigured Standard Protocols Listing

Band Amplitude Protocols

The following table shows standard protocols in which the main parameter is the amplitude of some frequency band. They all derive from digital filtering, which is one of the two main DSP (Digital Signal Processing) methods (the other being spectral analysis or Fast Fourier Transform, FFT).

Protocol Name	Main Parameter	Inhibits	Comments
Focus	SMR (12-15 Hz) Up	Theta Fast Beta	Probably the most widely used and researched neurofeedback protocol, it derives from the research of Barry Sterman and Joel Lubar. To be strictly SMR, you must train in the central region (C3, Cz, C4) but you can also train the same frequency range frontally.
Alert	Beta Up	Theta Fast Beta	Considered an activating protocol.
Relax	Alpha Up	Theta Fast Beta	Probably the most widely used and researched neurofeedback protocol, it derives from the research of Barry Sterman and Joel Lubar
Squash	1-15 Hz Down	-	TO_DO
Theta Up	Theta Up	Delta Fast Beta	TO_DO
Theta Suppress	Theta Down	-	TO_DO

Band Ratio Protocols

TO_DO

Protocol Name	Main Parameter	Inhibits	Comments
Theta:Beta	Theta:Beta	Fast Beta	This protocol may be used when the practitioner judges the theta to beta ratio to be on the high side, and where the trainee has goals such as improving focus or executive function. In this case you would train the T:B ratio down. In principle the brain could achieve this by lowering theta amplitude, or increasing beta amplitude, or both. An alternative use would be to train T:B up, as a de-arousal protocol for cases where the goal might be to reduce anxiety or agitation.
Brain Efficiency	Fast Alpha : Slow Alpha	-	Alpha speed, or alpha dominant frequency, has been described as a marker of brain efficiency (slow speeds being inefficient, faster speeds better). Therefore training upwards the ratio of fast to slow alpha is one way to train increased brain efficiency (though there are others). Training faster alpha is sometimes known as brain brightening.
Alpha R:L	Alpha R:L Ratio	-	The design intent of this protocol is to train balance between the left and right sides of the brain (it is a kind of “asymmetry” protocol.) It requires two channels of EEG, which should be on opposite sides of the midline, for example F3 and F4. L-R imbalances in alpha amplitude may be a marker of mood regulation issues. Training R:L alpha upwards could improve mood regulation in cases where alpha is more dominant on the left side (F3). This is a relatively more advanced protocol, and should only be used where an assessment has uncovered an imbalance, and where the trainee experiences mood regulation issues.
TO_DO	TO_DO	TO_DO	TO_DO

FFT-Based Protocols

To the best of my knowledge, most neurofeedback software uses digital filtering to calculate band amplitudes as feedback parameters. (Band ratios are ratios of band amplitudes calculated from digital filtering.) An alternative method of computing training parameters is spectral analysis or FFT (though it could be considered as more experimental).

The results of spectral analysis are commonly displayed as a spectral chart, with frequency along the x-axis, like this example:

eeg spectral analysis or fft display

An FFT contains much more information than does one single band filter amplitude. The advantage is that we can compute more general or global measures of brain activity.

Conceptually, the FFT is a list of numbers, each number corresponding to some particular frequency, and the size of the number is analogous to the amplitude at that frequency. We can calculate EEG measures by adding these numbers together, across frequency ranges. For example, we could get a measure of alpha activity by adding all the numbers in the alpha range, say 8-12 Hz. (This would be like adding up the heights of the yellow and green bars, in the above FFT chart.)

It would be rather different from the alpha amplitude delivered by a digital filter - conceptually this would be like the height of the highest bar in the alpha range.

A more useful FFT measure would be to calculate alpha % - this would be a ratio of the sum of the alpha numbers, to the sum of all the numbers (or say the sum of the 1-40Hz bars).

Protocol Name	Comments
Squash	Generally the squash protocol down-trains a broad frequency range, typically the slow waves, from delta through theta to alpha. It is thus regarded as an activation protocol. A squash protocol based on digital filtering (DF) also exists - in effect this would pick out the single biggest frequency component. This component might at times be in the delta range and at other times it might be in the alpha range. The FFT version of squash adds together all the components in the frequency range, e.g. 1 to 12 Hz. Conceptually we can think of this sum as the area of the FFT chart, between the two frequency limits. Thus it comes closer to training the full range of slow waves, all at the same time. Personally I see this as an advantage over the DF version. There is an option to weight the sum in a way that balances the “1/f” nature of the EEG spectrum. This means that higher frequencies, which by the 1/f rule tend to have lower amplitudes, are weighted to contribute more to the sum.
Squash with Alpha Window	This protocol is similar to the FFT squash above, in that it sums up the frequency components across abroad range, but this time excludes those in the alpha range. Suppose you were training between 1 and 25Hz, with a window in alpha. Conceptually the feedback parameter is the area of the FFT chart excluding the alpha range. That would mean, with reference to the above chart, the areas of the gray, cyan, orange and red regions, but NOT the yellow and green regions which are alpha.
Alpha %	In this protocol we calculate firstly the area of the FFT across a broad frequency range, and secondly the area in the alpha frequency range - that would be the yellow and green regions in the above chart. The feedback parameter is then the second area as a percentage of the first area. This is a way of training up alpha activity, much the same as the more standard DF based protocol, but it has no need of separate inhibits - in a sense the inhibits are “built in” to the calculation, because if say theta activity rises then the alpha percentage decreases.
Beta %	This protocol is conceptually the same as alpha % above, except that it uses the beta range in place of the alpha. Again there is no need for separate inhibits. Training beta up (while in effect training other bands down) is an activation protocol.
Theta:Beta Ratio	This protocol is similar to the standard theta:beta digital filtering amplitude-based protocol, except here we use the sum of theta activity to the sum of beta activity, based on the FFT output. Conceptually it would be the area in cyan divided by the area in red, with reference to the FFT graphic above. This FFT protocol would have much the same application as the DF-based theta:beta protocol - it can be down-trained as an activation protocol, e.g. to improve focus and executive function.
Centre of Gravity or Centroid	The FFT centroid protocol is a more experimental application. The parameter is the centroid frequency. Centroid means something like centre of gravity. With reference to the FFT chart shown above, the centroid frequency is the point on the horizontal axis where half the area is to the left and half to the right.

Other Protocols (inc. Experimental)