There is a breathing rate at which your heart swings hardest — and nobody chose it. It falls out of a five-second lag built into your own circulation. Slow down, breathe with the guide, and watch a live model of your pulse find it.
The loop's resonance is set by one number: how long your blood pressure takes to answer a change in heart rate. In a healthy adult that's about 5 seconds. Drag it and watch the whole curve — and your resonant breathing rate — slide. A shorter delay rings faster.
leading-order estimate 1/(2τ) = 6.0 /min · model peak ≈ 5.4 /min (the lag pulls it just below)
You have probably heard the wellness number: breathe at about six breaths a minute to calm down. It sounds like the kind of thing somebody picked because it's round. It isn't. There is a real, measurable frequency at which the cardiovascular system resonates — rings like a struck glass — and breathing happens to be a clean way to strike it. The instrument above isn't playing you an animation. It is integrating a small model of one feedback loop in your body and showing you, live, where that loop rings.
Your heart rate and your blood pressure are wired into a negative-feedback loop — the baroreflex. Pressure sensors in the walls of your aorta and carotid arteries report blood pressure to the brainstem; when pressure rises, the brainstem slows the heart (mostly through the vagus nerve) to bring it back down; when pressure falls, it speeds the heart up. Negative feedback: every push is met by a pull the other way. It's the same kind of governor that holds a thermostat at temperature.
But the loop is not instant. A change in heart rate takes time to work its way through the volume and elasticity of the vascular tree before it shows up as a change in blood pressure — about five seconds in a healthy adult. That lag is the whole story.
A negative-feedback loop with a built-in delay does not just settle. It rings.
Here is the trick, and you can feel it in the instrument. Suppose your heart rate is gently oscillating up and down. The baroreflex tries to oppose that oscillation — but its correction arrives five seconds late. If your oscillation has a period of about ten seconds, then five seconds is half a cycle. A correction delayed by half a cycle arrives exactly when the thing it's opposing has flipped to the other side — so the "opposing" push now lands in the same direction as the motion. The brake becomes a swing. Negative feedback, plus a half-period delay, equals reinforcement.
That is resonance, and it pins the frequency precisely. The loop rings hardest when the round trip — out and back — takes half a period:
There is the famous number — not chosen, derived, from a single physiological delay. And now slowing your breath to that rate makes sense: you are not "breathing deeply" in some vague way, you are driving the loop at its resonant frequency, the one rate at which a small breath produces the largest swing in heart rate. The swing itself — large, smooth, vagally-driven beat-to-beat variation — is the marker clinicians actually measure.
Because f = 1/(2τ) is the leading-order estimate — it counts only the pure delay. The real loop also has some smoothing (the vascular tree is springy, not a bare pipe), and that smoothing adds a little extra phase lag of its own. The extra lag means the loop reaches its half-cycle condition at a slightly lower frequency. So the model — and the bodies it's modelling — ring a touch below the clean 0.1 Hz, around 5.4 breaths a minute. This is not a fudge to dodge the round number; it's the honest output of the same equation. And it matches the clinic: when researchers measure each person's individual resonance frequency directly, it lands between about 4.5 and 6.5 breaths a minute, clustering near 5.5. "Six a minute" is the memorable round-up of a real, slightly-lower, person-by-person number.
Open the Advanced drawer and drag the delay. A 4-second loop rings near 7 a minute; an 8-second loop near 4. The resonance is not a constant of nature — it's a constant of your plumbing, and it moves exactly as 1/(2τ) predicts.
The page runs this same delay-differential integrator (τ = 5 s, lag T = 0.6 s, loop gain K = 0.9). Offline proof: research/resonance-breathing/verify.mjs — six checks, all pass.
This is a minimal model, not a medical device. Real cardiovascular control is nonlinear and runs three coupled baroreflex loops, not one — Vaschillo distinguished a fast heart-rate loop (≈5 s delay → ≈0.1 Hz) and a slower vascular-tone loop (≈15 s delay → ≈0.03 Hz). This tool models the single heart-rate loop, the simplest system that exhibits the real phenomenon. It reproduces why a resonance exists and where it sits; it does not reproduce your particular physiology, and the y-axis is in arbitrary units, not beats per minute.
What's well-established: the ≈0.1 Hz cardiovascular resonance, that slow breathing near it maximises beat-to-beat heart-rate variability and momentarily raises baroreflex gain, and that individual resonance frequencies fall ≈4.5–6.5/min. What's more preliminary: the longer-term clinical benefits of heart-rate-variability biofeedback — promising across anxiety, blood pressure, and stress in controlled trials, but a younger and still-debated literature. We make the first set of claims here and flag the second as the researchers themselves do.
Use it as a breathing pacer if you like — it is a genuine, honest one. But it is a toy that shows its working, not advice, and not a substitute for anyone with a heart or breathing condition checking with a doctor first.