Mar 082016

For a long time, I have wondered whether my brain waves change with the rhythm of my breath. Can we observe characteristics of the EEG signal that correlate with the pattern of inhalations and exhalations during zazen?

I wanted to test this during the February 2016 osesshin at the Tahoma Zen Monastery. But first I needed to make sure that data in the EEG signal were synchronized with the respiration data. Since these two signals were coming from two separate devices, I needed to devise a system (which I describe at the end of this post) to insure that the two data streams were in sync.

Let us take a look at a 25-minute recording made on Day 6 of the sesshin. Following instructions from my teacher, Shodo Harada Roshi, I have been working on extending the exhalation as long as possible. Here is a spectrogram for the left back (lb) sensor.

spectrogram for lb sensor during zazen

The most prominent feature is the strong alpha oscillation at around 8 Hz, which we can verify by looking at the PSD spectrum for this interval.

PSD spectrum for lb sensor during zazen

At the lb sensor, we see a strong oscillation at 8.3 Hz. Now compare this with the spectrum for the left front sensor, lf.


PSD spectrum for sensor lf during zazen

The Power Spectral Density graph for sensor lf shows an alpha peak at the same frequency 8.3 Hz but at much less intensity (about 2.4 dB/Hz) compared to sensor lb (about 100 dB/Hz). In addition, there is considerably more power in the higher frequencies (beta and gamma) than we saw for lb.

Now let’s take a closer look at the spectrogram for sensor lf.

spectrogram for sensor lf during zazen

The alpha band is diminished (faint yellow horizontal strip near frequency 8 Hz). The frequencies from about 15-50 Hz are more prominent– note the orange smear centered from about 20-30 Hz.

But this spectrogram shows something else that is very interesting: the vertical bands with a semi-regular spacing. What could be the cause of these bands? At first, I thought maybe they were some kind of mathematical artifact. But when I took a look at the time intervals between successive bands, another idea came to me.

annotated spectrogram for lf

There are about 11 bands in an interval of 200 seconds, which is equivalent to about 3.3 bands per minute. But 3.3 per minute is a typical zazen breathing rate for me on the latter days of a sesshin. Could these bands be correlated with breathing?

To test this hypothesis I added a new feature to the Physiology Viewer 2.0 program, making it possible to superimpose the respiration signal on the EEG signal.  Let’s first zoom in on an interval, say from 880 – 1030 seconds. This is the spectrogram we get.

spectrogram for a short interval

We see the bands clearly in the EEG. Now let us add an overlay of the breath signal.

breath overlay on EEG signal

Remember that the respiration and EEG signals have been synchronized. So we see that the growth in high frequency power (left side of yellow bands) begins with the inhalation, continues during the exhalation and then drops off (right side of yellow bands) during the extended exhalation.

Below are spectrograms for a recording done two days earlier (Day 4 of the osesshin).

spectrogram for a segment of zazen

spectrogram with breath overlay

We see the same pattern as before. In fact, when I go back and look at other recordings of the February 2016 sesshin, I see these bands in nearly every case. The bands are most pronounced in the left front sensor, lf, relatively weak in rf and often altogether missing in sensors lb and rb.

Looking back at other recordings from the February 2016 sesshin, I find that MOST of them show the high frequency bands at sensor lf. It is harder to find evidence in the recordings of other subjects, but I did find one case. Subject S12 has 21,000 hours of meditation experience. In this recording he is practicing shikantaza with his eyes closed.

spectrogram for subject S12

Vertical yellow bands indicating changes in intensity of high frequency oscillations (from about 20 – 40 Hz) are evident and correlate with his breathing, becoming less intense during the long exhalations.

Synchronization of EEG and Respiration Signals

Our EEG signal comes from the Muse Headband where each data point is time-stamped using the system clock. Elapsed time is computed by subtracting the initial time when the Record button is pressed. The respiration signal, on the other hand, comes from the Respiration Monitor connected to the LabQuest recorder where timestamps start at zero when the Record button is pressed. Therefore, to ensure that the two signals are in sync, we need to start the recording in each program by pressing their respective Record buttons as close to the same time as possible. An automatic mouse clicker program fit the bill.

Auto Mouse clicker is available from Murgee Softwares. We set up a script that automatically clicked the Record buttons on the two recording programs within a millisecond or so.

Auto Mouse Click from Murgee Softwares



 Leave a Reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>