The science

This page is long on purpose. An app that asks you to take it seriously owes you the evidence, including the evidence that argues against it. The short version comes first. The detail and the references follow, for anyone who wants to check.

The short version.

Slow breathing works. Breathing at about 6 breaths a minute lowers stress, and this is well proven. It is the strongest thing in the app, and it needs no strap, no headphones and no sound.

The pulsing sound gets a response, but that is as far as the proof goes. Your brain does follow a steady pulse. Whether that changes how you feel for long afterwards is not settled. Studies disagree.

Some findings go against the app. Background sound can weaken the pulse. A fixed pulse only suits some people. And the best-proven sound technique of all needs a brain sensor the app does not have. We say so rather than hide it.

The breathing: the strongest ground

Breathing slowly, at about 6 breaths a minute, makes your heartbeat speed up and slow down in a smooth wave with each breath, about one wave every 10 seconds. Doctors call the size of that wave heart-rate variability, and a bigger wave goes with lower stress. This is the best-proven thing the app does, and it works with no strap, no headphones and no sound at all.

HRV biofeedback cuts stress and anxiety, Hedges' g = 0.81. Goessl, Curtiss and Hofmann, 2017, Psychological Medicine 47:2578–2586.

143 studies, typically 10 to 20 minutes per session. Lalanza et al., 2023, Applied Psychophysiology and Biofeedback 48:275–297.

A single 20-minute session raised relaxation; the heart-rate change only approached significance. Toussaint et al., 2021, Evidence-Based Complementary and Alternative Medicine, article 5924040.

The pulsing sound: the brain follows it, but that is where the proof stops

It is well established that when you hear a steady pulse, the hearing part of your brain pulses along with it. A sound that switches on and off (the isochronic and monaural options) does this far more strongly than binaural beats do, which is why the app uses isochronic by default and makes binaural the one you have to choose.

Whether that pulse then changes your state of mind for any length of time is much less certain. The fair summary is "disputed". A 2023 review of 14 binaural studies found 5 in favour, 8 against and 1 mixed. One 2020 study found an 8 Hz pulse had the opposite effect to the one expected.

Ingendoh et al., 2023, PLOS ONE (pone.0286023). 14 studies: 5 support, 8 contradict, 1 mixed.

López-Caballero and Escera, 2017, Frontiers in Human Neuroscience. A well-controlled null result.

Licklider, 1950. Binaural beat salience vanishes above about 1,000 to 1,500 Hz, which is why the app caps the carrier.

Brain speeds: alpha, theta and the rest, in plain words

Your brain gives off tiny electrical ripples all the time. A brain sensor (an EEG: a cap or headband with contacts on the scalp) can count how many ripples a second. Scientists named the speeds with Greek letters. It helps to think of them as gears, fastest first.

Gamma, over 30 a second. Sharp concentration and problem-solving. Psiphase does not use it.

Beta, 13 to 30 a second. Normal awake and busy. You are in beta reading this.

Alpha, 8 to 12 a second. Calm and alert. Eyes closed, relaxed, still aware. This is the settle.

Theta, 4 to 7 a second. Drowsy and drifting. The edge of sleep, daydreams, deep rest. The deep part of a session sits here.

Delta, under 4 a second. Deep sleep. The app goes near it only in the deepest sessions, and always brings you back.

How Psiphase uses them. The pulse you hear is set to these speeds. A typical session starts the pulse at about 10 a second (alpha), eases it down to about 4 (theta) for the deep part, and at the end brings it back up to about 15 (beta) to wake you. Focus sessions go the other way after the settle, to about 16 a second, because focus is more awake, not more asleep. The idea, called entrainment, is that the brain tends to fall into step with a steady pulse. So a pulse that slows gives your brain a slope to walk down, and one that rises brings you back up.

What is known, and what is not. It is well shown that the hearing part of your brain follows the pulse. Whether that moves the rest of your brain into the gear is disputed (see the pulse card above). And the app has no brain sensor, so it can never tell which gear you are in. What it can measure is your breathing, how still you are, and your heartbeat if you wear a strap. So Psiphase never says "you reached theta". The gears are how the sound is designed, not a reading of you.

Three findings that argue against parts of this app

The background sound may work against the pulse. A 2025 study found background noise weakens the brain's response to a pulse. So a heavy background sound is a trade: it covers up your room better, but the pulse gets through less. That is why the app suggests a lighter background when you are somewhere quiet.

A fixed pulse only suits some people. People differ a lot. A pulse that plays the same for everyone will suit some and not others. This is the honest reason for the Adaptive sessions, and the honest limit of everything that runs on the clock.

The best-proven sound technique of all is one Psiphase cannot do. Playing short sounds timed exactly to the brain's own slow waves, during deep sleep, reliably deepens that sleep. The timing is what matters, far more than any steady pulse. Doing it needs a brain sensor (EEG), which the app does not have. That is why the one setting that would depend on it, Brain coherence, is locked in the app rather than faked.

Ngo and Born, 2013; Papalambros et al., 2017, Nature Communications (s41467-017-02170-3).

What Psiphase measured for itself

The sessions were not copied from anywhere and not guessed. We analysed the sound of the full classic Gateway course, all 65 recordings, about 47 hours, to find out what they actually do. Three findings shaped the app:

  • Less talking works better. The voice drops from about 1,503 words in the first recording to 151 in the last, roughly 10 minutes of speech down to under 2. The words are training wheels, taken away as you learn. That is why the voice here says less as you settle.
  • Coming back is deliberate. Nearly every recording ends with the same wake-up sound (a pulse of about 15.7 a second, for anyone checking). So every session here wakes you up properly at the end. In the self-hypnosis sessions that is built in and cannot be switched off.
  • The deep recordings go slower than their labels say. The deepest ones hold a very slow pulse, under 1 a second, for 35 to 40 minutes, and one recording sold as one kind of brain rhythm measured as a slower one. Where the name and the sound disagreed, we trusted the sound.

Ideas we offer to explore: unproven, and labelled that way

Reaching outward, distant perception, a younger time or another life. These are offered as things to imagine and explore, clearly labelled, and kept apart from anything measured. Nothing in the app contacts anything or scores whether a guess was right. If that ever changes, the app will show you how often pure luck would get the same result, right next to your score.

What this app cannot tell you

It has no brain sensor, so it cannot tell what state your mind is in. A heart strap measures your heart, and your heart is not your mind. In a session that runs on the clock, the time it takes to reach each stage is just the clock, not your progress. And one session, measured on yourself, knowing what you hoped would happen, does not prove anything. It is a reading. The app shows you readings and calls them readings.

Safety, factually

Light-triggered epilepsy is set off by flashing light, not by sound. The pulses here are sound only, and nothing on the screen flashes. The usual cautions still hold: not while driving or using machinery, and the guided journeys ask a few safety questions first. More on the safety page.