How do binaural beats work?

A binaural beat is the auditory system doing arithmetic. Two tones enter the ears, travel up the cochlear nerve, meet in the brainstem, get compared with sub-millisecond precision, and emerge as a slow rhythmic signal that climbs to the cortex. That is the entire mechanism in one breath — the rest of this page expands each step.

The full route, in anatomical order, looks like this: the outer ear funnels sound into the middle ear, where the tympanic membrane and the three ossicles convert pressure into mechanical motion; the cochlea converts mechanical motion into ion currents through hair cells; cranial nerve VIII carries those spike trains to the cochlear nucleus of the medulla; the trapezoid body crosses many of those fibres to the superior olivary complex (SOC) on the opposite side; the SOC compares the two ears with microsecond-level precision; the inferior colliculus of the midbrain integrates the output; the medial geniculate body of the thalamus relays it; and the primary auditory cortex of the temporal lobe receives it. The binaural beat is born inside this chain, somewhere between the superior olive and the cortex.

What makes binaural beats unusual is that they are a percept created by the comparator stage. Normal sounds are detected at the cochlea — the ear hears them and the brain interprets. A binaural beat is not present in either ear individually. It only exists once two streams are compared. That is why a microphone cannot record a binaural beat. The "beat" lives between the ears, not in the air.

If you want to test the mechanism while you read, the free web generator at brainwavegenenator.com/generator.html will let you select any carrier pair in real time. Put on stereo headphones, set 200 Hz left and 210 Hz right, and the rhythm you start hearing did not exist until your brain manufactured it.

Begin with the simplest possible case. A pair of sine waves: 200.00 Hz delivered exclusively to the left ear, 210.00 Hz delivered exclusively to the right. Two pure tones, never mixed in the air, kept on separate channels by stereo headphones. The hardware is doing nothing more than playing two notes — almost the same note — into two separate transducers.

Inside each outer ear, the pinna shapes incoming sound and funnels it down the external auditory canal toward the tympanic membrane. The eardrum vibrates at the input frequency. Three ossicles — malleus, incus, stapes — act as a mechanical impedance-matching lever, transferring vibration from air into the fluid-filled cochlea via the oval window. From the perspective of binaural beat physics, this stage is faithful: the middle ear preserves the carrier frequency without significant distortion across the relevant range.

Inside the cochlea, the basilar membrane has tonotopic organisation. Different positions along its length respond preferentially to different frequencies — high frequencies at the base near the oval window, low frequencies at the apex. A 200 Hz tone displaces a region near the apex; a 210 Hz tone displaces a slightly different region only a fraction of a millimetre away. Inner hair cells sitting on the membrane at those positions convert the mechanical motion into receptor potentials, releasing glutamate onto fibres of cranial nerve VIII.

The crucial property that makes binaural beats possible is phase locking. For carriers below roughly 1 kHz, the auditory nerve fibres don't just signal "a tone is present" — they fire spikes that are time-locked to specific phases of the input waveform. A 200 Hz tone produces spike trains where individual action potentials cluster on the rising phase of each acoustic cycle. The temporal pattern of the sound is preserved in the temporal pattern of the spike train. This is why the 100–500 Hz range is the workhorse band for binaural beat carriers: it is the band where neural phase locking is sharpest. Above 1.5 kHz, phase locking degrades; by 4–5 kHz it is essentially gone, replaced by rate coding only.

So at this stage you have two phase-locked spike trains travelling up the auditory nerve from each ear. They have not met. Each ear's signal is still completely independent, riding the ipsilateral cochlear nerve toward the brainstem.

Cranial nerve VIII terminates in the cochlear nucleus, a structure of the medulla that exists in both halves of the brainstem — one for each ear. Each cochlear nucleus contains several sub-divisions: the anteroventral, posteroventral, and dorsal nuclei. Different cell populations there process the signal in different ways. Bushy cells, in particular, preserve the timing of incoming spikes with extreme fidelity using a specialised synaptic structure called the endbulb of Held — a giant calyx-shaped axon terminal that wraps the postsynaptic cell and produces fast, reliable transmission.

This timing fidelity matters because the next operation is going to compare arrival times across the two ears. Any jitter introduced here would limit the precision of binaural processing. Evolution has invested a lot of biochemistry in making sure jitter is minimal.

From the cochlear nucleus, axons project to the superior olivary complex via the trapezoid body. This is the first crossover point: many of those fibres decussate to the contralateral side of the brainstem. The right ear's signal is now travelling toward the left superior olive; the left ear's signal is travelling toward the right. Some fibres remain ipsilateral, so each superior olive ends up receiving inputs from both ears. That convergence is what enables binaural comparison. Until now, each ear was a private channel. The trapezoid body is where the privacy ends.

The trapezoid body is one of the most heavily myelinated structures in the brainstem because conduction delays must be minimised and matched. The interaural difference the system is about to measure can be as small as ten microseconds — a window narrower than a single action potential. Any uneven delay along the path would corrupt the calculation. The anatomy is built like a precision oscilloscope.

The superior olivary complex (SOC) is where binaural beats are born. It has two major sub-nuclei that do different jobs. The medial superior olive (MSO) processes interaural time differences (ITDs) — the microsecond-scale delays between a sound arriving at one ear and arriving at the other. The lateral superior olive (LSO), with input from the medial nucleus of the trapezoid body, processes interaural level differences (ILDs) — loudness imbalances between the ears. For binaural beats, the MSO is the protagonist.

MSO neurons are coincidence detectors. Each receives an excitatory input from the ipsilateral cochlear nucleus and another excitatory input from the contralateral cochlear nucleus. The cell fires most strongly when both inputs arrive at the same time. Different MSO neurons are tuned to different relative delays because the axon lengths from each ear are different — a mechanism formalised in 1948 by Lloyd Jeffress as a "delay-line" architecture. The classic mammalian physiology was characterised by Yin and Chan (1990), who showed that MSO cells respond as a function of interaural phase difference at the carrier frequency. Joris, Smith, and Yin reviewed this in detail in Physiological Reviews (1998), describing the system as the most temporally precise computation in the mammalian brain.

Now apply this to two slightly mismatched tones. Left ear: 200 Hz. Right ear: 210 Hz. Each ear is delivering a phase-locked spike train. But because the two frequencies are different, the phase relationship between them is continuously drifting. Every second, the 210 Hz signal completes exactly ten more cycles than the 200 Hz one. The coincidence-detector population in the MSO sees its preferred input pattern recur, and disappear, and recur, on a precise cycle: ten times per second. The output firing rate of the MSO population is therefore modulated at exactly 10 Hz — the difference between the two carriers.

Stated as math: if the left carrier is f_L and the right is f_R, the beat frequency f_b = |f_L − f_R|. For 200 and 210 Hz, f_b = 10 Hz. For 100 and 104 Hz, f_b = 4 Hz. For 240 and 246 Hz, f_b = 6 Hz. The carriers can be almost anything within the phase-locking range; what counts is their difference.

This is the crucial mechanistic point. The 10 Hz signal does not exist anywhere in the acoustic world. It is not in the left ear's input. It is not in the right ear's input. A microphone placed at either ear would record only the original 200 or 210 Hz tone. The 10 Hz signal is created by the brain's act of comparing the two. It is, in a literal sense, an emergent property of the comparator circuit.

The output of the SOC ascends via the lateral lemniscus to the inferior colliculus (IC) in the midbrain tectum. The IC is the major auditory integration centre below the cortex; almost every ascending auditory pathway makes an obligatory stop here. It receives binaural information from both superior olives, monaural information from the cochlear nucleus, and descending feedback from the cortex.

Within the IC, the slow modulation generated by the SOC is preserved and amplified. IC neurons can be selective for amplitude-modulated stimuli, with many cells tuned to specific modulation frequencies in the range of a few Hz to about a hundred Hz. The binaural beat is, functionally, an amplitude modulation of the MSO output. So the modulation-tuned neurons of the inferior colliculus are well suited to encode it. This is also the level at which the auditory steady-state response is generated — the rhythmic EEG signature that researchers use to measure binaural beat entrainment from the brainstem.

This is where it matters to be honest about the data. Some of the earliest frequency-following response (FFR) work — notably Smith, Marsh, and Brown in 1975 — measured envelope-following responses to binaural beats using scalp electrodes and concluded that the response originated from subcortical (brainstem and midbrain) generators. Galbraith et al. (2003) refined the picture and showed that the FFR has contributions from multiple levels. The practical takeaway: the beat is detectable in neural activity long before it reaches the auditory cortex. Whatever happens at the cortex is built on a foundation already prepared by the brainstem and midbrain.

From the IC, axons project to the medial geniculate body (MGB) of the thalamus — the auditory thalamus. The MGB is the gateway to cortex. It gates and routes the signal, modulates it according to attentional state, and projects it to the primary auditory cortex on the upper surface of the temporal lobe. The MGB also has reciprocal connections back to the IC, so the system is not strictly bottom-up; cortical and thalamic feedback can influence what gets through.

Now the signal reaches the primary auditory cortex (A1), buried in the transverse temporal gyrus of Heschl. A1 is tonotopically organised — the same logic as the cochlea, but mapped onto cortical sheet — and it responds to the spectral content of incoming sound. The slow modulation generated downstream now arrives as a rhythmic input to A1, and some of A1's intrinsic oscillators begin to phase-align with it. That alignment is what we mean by "cortical entrainment."

The frequency-following response (FFR) is the EEG-measurable signature of this alignment. Pratt and colleagues (Clinical Neurophysiology, 2010) recorded cortical FFR to binaural beat stimuli and confirmed that bilateral auditory areas synchronise to the beat frequency. Karino et al. (NeuroImage, 2006) used magnetoencephalography (MEG) to show that listening to a binaural beat produces sustained, rhythmic activity in both left and right auditory cortices, time-locked to the beat. These studies establish, fairly clearly, that cortical neurons do follow the binaural beat — at least in the lower frequency bands.

But the literature is not unanimous, and overstating the cortical effect would be dishonest. Lopez-Caballero and Escera (Frontiers in Neuroscience, 2017) directly compared subcortical FFR to binaural beats and to monaural amplitude-modulated tones. They found that for beat frequencies above about 4 Hz, the brainstem-level FFR to binaural beats was substantially weaker than older studies had suggested, and weaker than the response to acoustically modulated stimuli. The most parsimonious reading: cortical entrainment to binaural beats is real and measurable; brainstem-level FFR exists but is smaller and more variable than early enthusiasm implied; the effect is genuine, but it is not as overwhelming as some marketing copy claims.

The entrainment is also selective. The slow envelope at the beat frequency is what propagates upward; the carrier-frequency oscillations stay in the lower auditory system. The cortex doesn't synchronise at 200 Hz or 210 Hz — those are too fast and too acoustic. It synchronises at 10 Hz, the difference. That is why the carrier choice has so much latitude and the difference choice is everything.

Once the rhythm is established in A1, neighbouring cortical regions can be drawn along. There is suggestive evidence that adjacent areas — parts of parietal cortex and prefrontal areas active during attention — show some modulation at the beat frequency during longer exposures. This is the threshold at which auditory entrainment begins to spill into cognitive state shift. The further you get from primary auditory cortex, the weaker and more variable the effect. The honest limit: the auditory cortex follows the beat reliably; the rest of the brain follows it conditionally.

You can probe this directly. The mobile app's lab tools include an isochronic generator (known to produce a stronger cortical FFR than binaural beats) and a multi-stage sweep that lets you feel your subjective state change as the beat frequency moves through the bands.

The mechanism above is real, but it is also fragile. A lot can go wrong between the headphone and the cortex. If you have tried binaural beats and felt nothing, one of these is usually the culprit:

• Hearing asymmetry. The MSO does coincidence detection on two phase-locked spike trains. If one ear has even mild conductive or sensorineural hearing loss, the two streams are imbalanced — in amplitude, in latency, or in fidelity — and the comparator output degrades. Asymmetric hearing is a quiet but common reason binaural beats feel weak.
• The wrong kind of headphones. Bone-conduction headphones and open-ear designs leak each channel across to the other ear. So do speakers, obviously. Even modest crosstalk weakens the illusion. Standard sealed over-ear or in-ear stereo headphones are the most reliable setup.
• Carrier too high. Phase locking degrades above roughly 1 kHz and is essentially gone by 4 kHz. A binaural beat constructed from a 3 kHz pair of carriers will produce a much weaker percept than the same beat built on 200 Hz carriers, because the neural substrate of the calculation is no longer firing in step with the waveform.
• Volume too loud. Comfort matters. High SPL recruits the acoustic reflex and stress responses; the cortex spends its bandwidth coping with intensity rather than following a gentle 10 Hz rhythm. Moderate, sustained volume — just above background noise — works better than loud.
• Too short a session. First-time listeners sometimes expect an immediate, dramatic effect. Cortical FFR can be present within seconds, but the subjective shifts associated with sustained protocols (anxiety reduction, sleep onset, focus state) typically need 10–30 minutes per session, and weeks of practice before the effect feels reliable.
• Sensory overload during listening. Doing demanding cognitive work — coding, reading complex material, holding a conversation — can suppress the auditory cortex's willingness to follow a slow rhythm. Background listening during light tasks works for some; intense multitasking generally does not.
• Individual variation. Phase-locking precision is not identical across people. Age, history of noise exposure, and individual neural variability all matter. Roughly 10–20% of listeners in research samples report minimal subjective effect even under ideal conditions. Some of that is genuine non-response; some is expectation mismatch.

If you've ruled out the first six and still feel nothing, you may simply be in the smaller responder category. That is allowed. Isochronic tones (a single pulsed tone, not requiring binaural integration) work for some non-responders and are produced by the lab tools in the mobile app.

Most binaural beat recordings use carriers in the 200–240 Hz range. This is not arbitrary, and it is not aesthetic. It reflects three constraints converging on a narrow band.

First, neural phase locking. As discussed above, phase locking in the auditory nerve and brainstem is sharpest below roughly 1 kHz and degrades steeply above 1.5 kHz. The whole binaural-beat mechanism depends on phase-locked spike trains feeding into a coincidence detector. So the carrier must be in the phase-locking range. That sets an upper bound around 1 kHz, with comfort margin below.

Second, comfort. Carriers below about 100 Hz feel muddy and rumbling, especially over modest headphones. They also tend to mask one another more than higher carriers do. Carriers above about 400 Hz start to feel bright, and the upper end of the comfort range — 500 to 1000 Hz — is fatiguing for the long sessions binaural protocols typically require.

Third, the difference frequency (the beat) is essentially independent of the carrier. A 10 Hz beat sounds the same whether you build it from 100 Hz / 110 Hz, 200 Hz / 210 Hz, or 400 Hz / 410 Hz carriers — the beat frequency is the only thing the higher pathway "sees." So you might as well choose the carrier that is most comfortable and best phase-locked. That is 200–240 Hz.

A reasonable default for most listeners: 200 Hz base carrier, with the beat added as the difference frequency. For sleep protocols, drop to a 100–150 Hz carrier for a softer feel. For gamma focus work, a 300–400 Hz carrier is often more pleasant than 200.

The beat frequency — the difference between the carriers — is the parameter that maps onto the classical EEG bands. These bands aren't arbitrary divisions; they correspond to genuine, distinct modes of cortical activity, each associated with characteristic states of consciousness and arousal.

• Delta (0.5–4 Hz) — the slowest band. Dominant during stage-3 NREM sleep and the deepest meditative states. Targeted by sleep and recovery protocols.
• Theta (4–8 Hz) — the boundary between waking and sleeping. Active during REM, deep meditation, and hypnagogic imagery. Targeted for relaxation, memory work, and creative incubation.
• Alpha (8–13 Hz) — the resting rhythm of the awake brain with eyes closed. The most-studied band in EEG and the easiest target for binaural entrainment. Used for calm focus, stress relief, and flow-state entry.
• Beta (13–30 Hz) — the working rhythm of task engagement, vigilance, and externally directed attention. Used for productivity and active concentration.
• Gamma (30–100 Hz) — the fastest physiologically meaningful band. Associated with perceptual binding and insight. Often targeted around 40 Hz, the most-studied gamma frequency, though clinical evidence remains preliminary.

For the band-by-band protocol detail, see the companion page /brainwave-frequencies-explained.html.

Working backward from mechanism: once you know that the beat frequency drives entrainment and the carrier choice just needs to sit in the phase-locking range, choosing a protocol becomes straightforward.

Sleep. Target delta. A 2–3 Hz beat on a 150–200 Hz carrier, 25–45 minutes at low volume, headphones, eyes closed, lights down. Many sleep protocols descend from theta to delta over the first ten minutes to ease the cortex toward sleep onset.

Calm focus. Target alpha. A 10 Hz beat on a 200–220 Hz carrier, 20–40 minutes during light work — reading, low-intensity writing, planning. Alpha is the most reliable band for first-time listeners; the calm-but-alert shift is easy to notice.

Primed cognition. Target gamma. A 40 Hz beat on a 250–400 Hz carrier. Treat as experimental: the cortical FFR is real, but the cognitive-performance literature is preliminary and individual response varies. Short sessions (15–20 minutes) before a focused task work better than long background exposure.

Meditation depth. Target theta. A 5–7 Hz beat on a 150–200 Hz carrier, run during a practice that is already familiar to you. Theta beats deepen practice rather than replacing it; expecting the audio to do the meditation for you is the wrong mental model.

For the step-by-step session structure — environment, headphones, what to expect minute by minute — see /what-is-binaural-beats.html, which walks through the first 30-minute protocol in detail.

To build these protocols yourself, the web generator gives four free presets and a live real-time engine where you set any carrier and any beat frequency by hand. The mobile app extends that with 40 expert-built presets, 31 ambient layers, and lab tools including a multi-stage sweep that automates transitions (theta-to-delta for sleep onset, alpha-to-beta for focus ramp-up).

• /what-is-binaural-beats.html — the foundational definition and a 30-minute first-session protocol.
• /binaural-beats-science.html — research-heavy overview with peer-reviewed citations across 120+ studies.
• /brainwave-frequencies-explained.html — band-by-band breakdown of delta, theta, alpha, beta, gamma applications.
• /binaural-beats-research-hub.html — curated index of the published literature.
• /frequencies/alpha-waves.html — deep dive into alpha-band entrainment, the most reliable band for first-time listeners.