Science · preliminary

Five people on two coastlines

We sat people in front of the sea, recorded their brain, heart and breathing alongside the sound of the water, and asked whether the body falls into step with it.

This is a pilot. Five subjects, one session each, blocks of a few minutes. It was run to find out whether the recording and the analysis work at all before committing to the full study, and that is the weight the results on this page can carry. Everything below is drawn from the analysis output, and every figure names the file it came from.

Part one

The recording

What was measured, where, and the one thing that has to be true before any of the rest is worth asking.

The field

A microphone, a body, and the sea

Five adults took part in outdoor sessions facing the water, on the Chilean coast and on Mallorca. Each wore 32-channel EEG, a single-lead ECG and a respiration belt, while a binaural microphone beside them recorded the same sea they were sitting in front of. All the streams share a clock.

Each person met the sea three ways, in blocks of three to six minutes: watching it with their hearing blocked, listening to it with their eyes covered, and both at once. Two silent rests bracketed the session, one before and one after, and those are what the nature blocks get compared against.

A rocky cove on the Mallorcan coast, with recording equipment set up facing the water. — Mallorca and the Chilean coast. The sea is not a stimulus played to a participant here. It is the room they are sitting in, recorded at the same time as the body that is in it.

A beach at dusk on the Chilean coast during a recording session. — Mallorca and the Chilean coast. The sea is not a stimulus played to a participant here. It is the room they are sitting in, recorded at the same time as the body that is in it.

The premise

Is there anything in the sound a body could track?

The question it asks

Do the sea and the breath even operate at the same speeds?

Before asking whether the body follows the water, the water has to carry energy at a rate the body could plausibly follow. It does. The slow swell of the audio sits mostly below 0.1 Hz, and breathing sits around 0.25 Hz, inside the band it always occupies. They overlap enough for the question to be worth asking.

Power spectra of the audio swell envelope, respiration and instantaneous heart rate. The swell is concentrated below 0.1 hertz, respiration peaks near 0.25 hertz, and heart rate is broadband and slow. — The one figure here taken from the report as an image rather than redrawn from data, because the underlying spectra were not saved alongside the other results. Shaded bands are the standard error across the fifteen recordings, and the lavender strip marks the usual range of human breathing.

In one sentence
The sea moves slowly enough, and breathing is slow enough, that the two could in principle keep time with each other.

Part two

What came out

Three results the pilot supports, each a comparison rather than an absolute claim, and each shown with all five subjects visible.

Nature against rest

Breathing couples more with the sea than with silence

Pooling the three nature blocks against the two rests gives one comparison per subject. The phase measures, which ask whether two rhythms hold a steady timing relationship, go up during nature. Phase-locking value gives p = 0.027 and the weighted phase-lag index p = 0.0067, both from a model fitted on roughly 440 windowed values.

Those two are the breathing rows below. The same contrast was run against two derivations of the heartbeat as well, and those are shown beside them.

Nature versus rest, on the two phase measures

shown

p < 0.05

run in total

corrected

subjects

Show

Each row compares the three nature blocks against the two silent rests, as a linear mixed-effects model on the windowed values with a per-subject random intercept, over roughly 440 windows. Dots right of the line mean more coupling during nature, filled ones are p < 0.05. The pilot ran twelve such contrasts across four metrics; these are the six using the two phase measures, and none of the p-values are corrected for the twelve. Read the rows for direction and confidence rather than comparing their lengths. Source: nature_vs_rest_stats.csv.

Blind to · multiplicity. Twelve comparisons were run and none of these p-values are corrected for that. At this size the honest reading is a direction worth following, not a result to bank.

Which sense

Listening does it; looking does not

The question it asks

Does it matter which sense the sea arrives through?

Among the three conditions, listening with eyes covered produces the strongest respiratory coupling, around twice what watching or watching-and-listening give. Adding vision does not add to it. It takes it away.

Two measures are shown because they fail differently. The weighted phase-lag index throws away the zero-lag part of the relationship, which is the part a shared artefact would produce, so when it agrees with the plain phase-locking value the result is harder to explain away. Here they agree.

Five subjects, three ways of meeting the sea

watching

0.146

listening

0.309

both

0.142

listening ÷ watching

2.12×

Measured by

Whole-block coupling between breathing and the slow swell of the audio, one line per subject. Both measures ask whether the two rhythms hold a steady timing relationship, and both put listening well above the other two conditions. They are worth seeing separately because wPLI ignores the zero-lag part of the relationship, so agreement between them is harder to explain away. The sample is five people, so the group line is drawn through five points and nothing is hidden behind it. Source: coupling_summary_all.csv.

In one sentence
Listening to the sea couples breathing to it about twice as strongly as looking at it does, and adding sight to sound does not help.

The pilot also ran coherence, cross-correlation and mutual information, which answer differently. The four families of coupling measure, and what each is blind to, are worked through properly in Learn: Attunement.

Timing

And they do it at the same point in the breath

This is the cleanest thing five subjects can show. For each person we take the breathing phase at which their coupling to the swell was strongest, and ask whether those phases agree across people. Under listening they largely do, with four of five subjects internally consistent and the group resultant at R = 0.73, which the report gives as p = 0.060, a trend. Add vision and the agreement collapses to R = 0.22.

Does everyone breathe at the same point in the wave?

group R

0.73

subjects locked

4 of 5

Rayleigh p

0.060

R runs from 0, arrows pointing everywhere, to 1, all pointing the same way. The p is quoted from the report rather than recomputed here.

Condition

Each arrow is one subject: its direction is the breathing phase at which their coupling to the swell was strongest, and its length is how consistently they held it across roughly twenty windows. Coloured arrows are subjects whose own phases cluster significantly; grey ones do not. The black arrow is the group. Under listening the arrows line up and the group resultant is long. Add vision and they scatter, which is the shape you would expect if looking at the sea pulls each person’s timing in a different direction. Source: phase_per_subject.csv.

In one sentence
Something about watching the sea moves each person’s timing in a different direction, while listening to it moves them all to the same place.