Coherence and Synchrony

Discussion about coherence and synchrony as they relate to brain function and training.


Coherence is a measure of how effectively two sites are able to link and unlink, to share information. Whether coherence is a good thing or a bad thing depends to some extent on what frequency we are talking about and what sites and what task. Slower frequencies, alpha and lower, are generally produced by rhythm generators in the middle of the brain, below the cortex. When the cortex, the thinking part of the brain, is working, we expect to see its neurons “de-synchronize” from those rhythm generators, just as you might turn off music you were listening to when you had to focus on a task. Faster frequencies, the beta frequencies, are produced in the cortex, generally in local pools of neurons which work together to perform specific functions.

Coherence is not measured in Hertz or microvolts. It is a measure, like the correlation coefficient, which ranges from 0-1 or (in percentage terms) from 0-100. When we look at coherence in alpha or theta in a brain, especially during the eyes closed or eyes-open baseline conditions–not at task–we expect to see it run between about 40-70. All the non-working neurons who are listening to the specific “radio station” (frequency generator) should all be dancing to the same rhythm. When this coherence does not appear, it suggests that the neurons in the cortex are having difficulty letting go of their working state and shifting into a lower-energy resting-ready state. People who are anxious or stressed or depressed may show these low slow coherence levels. When the coherence stays locked in during a task, it suggests that the neurons are not able to shift into productive states: they are listening to their i-pods when the boss is telling about the new project and their tasks in it.

Fast beta frequencies come from the neurons in the cortex. There is no reason for them to be generally coherent, as there is with the slower frequencies. So fastwave coherence, especially with eyes closed or open in baseline conditions, should generally be below 40. In a productive, functional brain, working neurons in a given pool link up quickly with others to share information or perform a function, then quickly unlink and perhaps link with others. This would be like commodities traders, taking buy or sell orders by phone from clients, hanging up, calling to execute the orders, then accepting the next call. At task we should see shifting patterns of connections. When fastwave coherence is high consistently, it suggests that the neuron pools are locked together–as if the trader were talking to an old friend for an hour in the middle of the work day. High fastwave coherences can relate to brains which are less productive, sometimes rigid, locked-in, so high frontal fastwave coherence would suggest a person who easily got stuck.

I’m going to give you a trainer’s understanding as I’ve developed and learned it.

Slow frequencies should be coherent/synchronous. There are 5-6 frequencies produced by sub-cortical rhythm generator nuclei in the thalamus, the hippocampus and perhaps the brain stem. These frequencies are described as Global (Delta and Theta), Regional (Alpha and SMR) or Local (beta). Beta is produced by cortical neurons when they are performing a task. When an area of the cortex is resting, it can resonate to one of the sub-cortical rhythms. Since all sites are resonating to the same source, we would expect them to be coherent (consistently related) and in phase (firing at about the same time)–synchronous.

Fast frequencies–beta and above–are energy intensive but they only happen in specific areas at a time, pass on the job and rest. They would be expected to be very independent. A high level of coherence in these frequencies–especially in non-task recordings–would suggest that the brain is overly activated, not able to rest. It’s like a basketball team where all five guys try to dribble the ball down the court at the same time.

I really haven’t worried much about coherence values too HIGH in slow frequencies, though I have seen it from time to time in delta or alpha, but I’ve never focused training on it.

Coherence is expressed in numbers from 1-100 (or as percents). 1 being absolutely no relatedness between the signals. 100 being that if you knew what was happening in one signal you could predict what the other would be doing with complete accuracy. I use 40 as a cutoff. I can’t give you sources for that. I’ve been using it for many years but I don’t recall the source. Low frequencies should be above 40. Fast frequencies below 40.

Low coherence in slow frequencies suggests an irritated, excited brain, one which does not communicate well within itself, wastes a lot of energy, an engine that doesn’t know how to idle.

High coherence in fast frequencies sometimes appears as extreme sensitivity–sometimes as a wall. Fastwave coherence in the occipitals can be related to light sensitivity. Locking the neurons together can either dramatically increase the effect of a stimulus; or it can block the neurons from doing their processing job.

I haven’t really worked with coherences greater than 75 or 80 in slow frequencies. Nor have I worked with “low” fastwave coherences.

Coherence at Locations

Coherence is a measure of the brain’s ability to communicate through a clean, accurate channel between two sites. When coherences in slow frequencies between sites like F3 and F4, C3 and C4, P3 and P4, O1 and O2, which connect at the midline are low, that suggests that something is interfering with the connections. That something may be excessive excitement in the neurons (they won’t stop producing bursts of beta, even when there is no task requiring beta) or it may be a physical disruption caused by a head injury or some sort of lesion. In either case, it’s likely that performance will be affected.

Sites like F7 and F8, T3 and T4, T5 and T6 don’t share a boundary at the midline. They are very far apart (in brain terms), and they have to send their communications either through the corpus callosum or the anterior commissure, so the delay is greater. They tend to have very low coherences, especially the temporal sites. However, coherence of zero at any site is definitely an error.

Alpha Coherence

If you understand alpha coherence as a measure of neurons’ ability to “let go and dance” to the rhythm of the thalamus, which is a restful and “present” state, then low coherence can be considered as an indicator that the neurons are a bit “up-tight” and can’t let go very well. The client may also demonstrate some of this type of behavior as well. If alpha/theta ratios are low (below 1.0 in the front of the head or 1.5 in the back with eyes closed) and coherence is down around .4 (or 40 depending on how it is being reported) or below, then the client may be depressed or anxious, almost always stressed, probably tired, perhaps struggling with staying connected to the environment.

Coherence is a measure of how the neurons are synchronizing with the thalamus, so it is quite variable (unless you happen to have done a lot of training or meditation). What we look at is the overall “average” coherence over a period.

You can train alpha coherence wherever you wish, though I would usually do so behind the central strip. Because it is a relationship between two sites, it must be done two-channel, and it is best done (in my opinion) with eyes closed and using monopolar (ear-referenced) placements. P3/A1 and P4/A2 or O1/A1 and O2/A2 are very good sites to work with.

The alpha/theta ratio could be low because alpha is low–or because theta is very high. If the theta/beta ratio is high and the alpha/theta ratio is low, chances are that theta is “swamping” the other frequencies. You want to train it down. If theta/beta looks pretty good but alpha/theta is low, then you probably want to train up alpha. I usually try alpha coherence and try P4/A2 alpha up/theta down in these cases. Some clients prefer one, some the other.

Beta Coherence

Remember that eyes-closed coherence means that the neuron pools are either locked together or are linked to a rhythm generator. There isn’t a rhythm generator for beta, so high levels of coherence suggest that the neurons are locked into a communication loop. With eyes closed, what are they communicating about? Beta coherence, then, should generally be a relatively sporadic, short-lived phenomenon that occurs when two pools of neurons are sharing information or working on the same task. When it appears at high levels, or during eyes-closed or eyes-open baseline tasks, the implication is that those neuron pools are no longer operating independently.

In the back of the head, where we process and integrate sensory information, locked up neurons tend to result in higher levels of sensitivity of sensory inputs, reduced ability to differentiate and deal with multiple types of information. It would result in a very simplified and very LOUD view of the world.

In the front of the head, where the executive centers are supposed to be handling inputs from many different sources and making complex decisions–or planning and producing motor outputs–the result can be a kind of rigidity of thought or getting easily overwhelmed by complexity in processing requirement.

HEG and Coherence

Robert Coben, who has done a fair amount of detailed work with HEG is the source of the comment about the different approaches to HEG and their effects on coherence. I don’t know that anyone else has ever verified that, and I don’t know what his basis for the comment might have been. In any case, it’s usually fairly easy to train state coherence up or down in a frequency band once you know it is there. I’m not sure I would make that the source of deciding how to approach HEG.

High Coherences in All Sites

If you see this, my first guess would be that there was some muscle or electromagnetic artifact, either of which can cause coherence values to rise artificially (since the signal is coming from the same place–the muscle).

Phase is a timing relationship. Two waveforms are “in phase” when the peaks and troughs occur at the same time. Coherence is a measure of how stable the relationship remains over time. Synchrony is coherence in phase. Synchrony will always be coherent, but coherence need not always be synchronous.


Synchrony indicates that two waveforms are coherent (consistent relationship between their peaks and valleys) and in phase (peak and valleys happening at the same time in both waveforms.) If that is true when you add the waves together, the peaks are added to the peaks, giving the largest possible positive number, and the valleys are added to the valleys, giving the largest possible negative number. Thus the sum of the two, when we look at the absolute values, removing the plus and minus signs, will be the largest possible values for those two waveforms.

When you subtract the peaks from the peaks, the difference will be the smallest possible number or negative number given the two waveforms, so the absolute value will be smallest.

Since synchrony is distance related, the further apart two sites are, the harder it is for them to respond at the same time. Sites like T3/T4, T5/T6 and F7/F8 are not generally as synchronous as F3/F4, C3/C4, P3/P4, etc. O1/O2 are very close together and pretty easy to get synchronous.

Why would two signals be synchronous?

1. They are coming from two sites communicating with one another in a particular frequency; or

2. Both are coming from the same source.

The reason that most synchrony training is done in slower frequencies (alpha and lower) is that these frequencies are not produced cortically, but are the result of surface neurons synchronizing with sub-cortical rhythm generators like the thalamus or hippocampus. Artifact could be defined as signals that appear in the EEG which do not come from the brain. Hence, if there is artifact, it is likely to appear in both channels and come from the same source in both channels (e.g. an eyeroll, a muscle bracing). That would make it likely to be coherent or even synchronous.

The phase angle, for a synchronous signal, should be close to zero. The TQ7 (after looking at several options) shows the percent of time that the signal was between +/- 30 degrees–very close to zero. The combination of the phase data and the coherence values give us an idea of the level of synchrony between two sites in a frequency.

A Fully Synchonous Brain?

It is unlikely that we will find the entire EEG in all frequencies to be in phase and coherent. It’s tough enough to achieve that with a single band in a naturally synchronous frequency like alpha, which forms nice neat sine waves when rocking along at 10 Hz or thereabouts. But to find two waveforms where ALL the frequencies are synchronous at the same time would be very unusual. So if you take the whole raw EEG and add it together–even if there is synchronous alpha in it, you’ll be adding beta peaks to beta troughs, theta peaks to theta crossing points, delta troughs to delta peaks, etc.–thus you will obscure the phase relationship in alpha.

About Alpha Synchrony

Alpha synchrony relates to the brain’s ability to keep the channels clear and open when the neurons are not active. It’s not just about raising alpha amplitude.

The key is to understand that alpha shouldn’t be dominant in front, but that doesn’t mean that the alpha that exists there shouldn’t be synchronous. Still, in most cases, training it should be done in the parietals/occipitals. You can train in the central strip as well.

It is often true that when someone is producing synchronous alpha, most other frequencies go away, but certainly not always when one is learning to do it will that happen. If you want to pick out and reward those moments when synchrony is starting to happen or happening briefly, work with the alpha band, not the whole EEG.