Of every rhythm the human cortex produces, one has captured more neuroscience attention this decade than the rest combined. Forty hertz — a single oscillation, forty times a second — sits at the bottom edge of the gamma band, and at the centre of a research story that runs from Tibetan monks in 2004 to MIT mouse models of Alzheimer's in 2016 to ongoing Phase 2 human trials in 2025. This page documents what the literature actually shows — and what it does not.
Cortical activity organises itself into frequency bands, and the fastest physiologically meaningful range — 30 to roughly 100 Hz — is what neuroscientists call gamma. It is not a single rhythm. It is a regime: fast, broadband, locally generated, and tightly coupled to the spike timing of inhibitory interneurons.
Gamma is the brain's solution to a hard problem. When you see a red ball, the colour is processed in one cortical area and the shape in another. The percept "red ball" requires those two streams to be tagged as belonging to the same object. The leading hypothesis — articulated by Wolf Singer and others in the 1990s — is that gamma-band synchrony is how the cortex binds features into objects. The same rhythm, in the same window, says: these signals are about the same thing.
Gamma is not a frequency. It is a way the cortex tells itself what belongs together.
One technical distinction matters here. Evoked gamma is the brief, time-locked burst that follows a sensory stimulus — visible within 50–150 ms of a flash or a tone. Induced gamma is the sustained, higher-frequency activity that emerges during cognition and is not phase-locked to any external event. The MIT entrainment work targets evoked gamma; classical meditation and memory studies usually concern induced gamma. They are related but not identical.
This is also why gamma is the most debated band in EEG. It is small in amplitude, easy to contaminate with muscle artefact, and harder to record cleanly than alpha or beta. Some of the noise in the gamma literature is, literally, noise.
The reason 40 Hz dominates the popular conversation about gamma is a series of papers from Li-Huei Tsai's laboratory at MIT's Picower Institute. They are remarkable, important, and routinely overstated. The honest summary is that the mouse work is real, the human work is ongoing, and the consumer story has run ahead of the data.
The mouse-to-human translation gap is real and important. Mice that "develop Alzheimer's" carry human transgenes that produce pathology faster and more uniformly than the disease in patients; their brains are 1,000× smaller; their immune systems differ. A clean effect in a transgenic mouse is encouraging biology, not a therapy. 40 Hz audio is worth listening to. It is not a treatment for dementia.
Long before the MIT papers, cognitive neuroscience had identified gamma as the rhythm of memory encoding. Recordings from human hippocampus during list-learning consistently show that items that will later be remembered are encoded with greater gamma power than items that will be forgotten. The signal is most reliable in the 35–50 Hz range — right around 40 Hz.
The deeper story involves a relationship called theta-gamma coupling, popularised by Lisman and Jensen (Neuron, 2013). Within each theta cycle (~5–8 Hz), a handful of gamma cycles ride on the slower wave. Each gamma cycle is thought to carry one "item" of working memory. Capacity — the famous seven-plus-or-minus-two — falls out of the geometry: how many gamma cycles fit inside one theta cycle.
If theta is the page, gamma is the line. You write a memory line by line.
Roux and Uhlhaas (Trends Cogn Sci, 2014) extended this to working memory: distinct cognitive contents appear to be represented by gamma cycles nested in different phases of the slower theta or alpha rhythm. This is now the dominant model for how the brain juggles multiple items at once.
None of this proves that listening to a 40 Hz binaural beat will improve your memory. It does mean that 40 Hz is not an arbitrary number. It is the frequency at which a great deal of the brain's information-binding work actually happens.
Each slow theta cycle (cyan) contains roughly six to eight gamma cycles (amber). Each gamma cycle can carry one item of working-memory content. The geometry of nested rhythms is the geometry of cognition.
In 2004, Antoine Lutz and Richard Davidson published the paper that put gamma into wellness culture. Long-term Tibetan Buddhist practitioners — monks with 10,000 to 50,000 hours of practice in non-referential compassion meditation — produced self-generated gamma activity at amplitudes rarely seen in untrained subjects (PNAS, 2004). The signal was strongest over fronto-parietal regions and was sustained for minutes at a time.
It is the most replicated "elite-state" finding in contemplative neuroscience. It is also routinely misread. The monks were not listening to 40 Hz. They were producing it, internally, through decades of trained attention. The relationship between externally driven gamma (what binaural beats and 40 Hz audio offer) and self-generated gamma (what advanced meditators achieve) is genuinely open.
The monks did not need headphones. That is the point — and the problem.
A reasonable working model: entrainment may help a beginner taste a state that experienced practitioners reach unaided, the way training wheels approximate the experience of riding a bike. Whether the borrowed gamma teaches the brain to produce its own is the question the next decade of research will answer.
A fair version of this question is more useful than the comfortable one. The honest answer is: measurably yes — but the effect is weaker in gamma than in lower bands, and it depends on protocol details that most consumer apps ignore.
Two practical implications follow. First, monaural and isochronic 40 Hz produce larger evoked responses than binaural beats at the same frequency — if you care most about cortical entrainment per se, those formats are stronger. Second, even where binaural-driven gamma is small, the experiential and cognitive correlates can be present, suggesting some of the value lies in arousal, attention, and entrainment all working together rather than in any single mechanism.
Forty hertz is the question gamma asks of consciousness.after Singer · Lisman · Tsai / on cortical binding
Gamma is a daytime, eyes-open, pre-task frequency. Treated as such, it tends to feel like a gentle nudge into clarity. Used wrongly — at night, paired with passive consumption — it tends to feel like restlessness.
Listen to 40 Hz binaural — carriers around 200 Hz / 240 Hz — for roughly 20 minutes before a cognitively demanding block. Sustained attention seems to benefit more than brief exposure. Sessions much longer than 45 minutes show diminishing returns.
Read, code, plan, draft. Visual engagement during gamma exposure appears to create a multisensory entrainment effect — the same principle the MIT Martorell paper used. Closed-eye listening is fine but less generative.
40 Hz is energising. Listening to it within two hours of sleep is the wrong tool. Switch to delta (0.5–4 Hz) or low theta (4–6 Hz) for sleep onset; reserve gamma for the hours when the cortex is supposed to be loud.
| Goal | Duration | Time of day | Pair with |
|---|---|---|---|
| Focused work | 20–40 min pre-task |
Morning / early afternoon | Single demanding task; no notifications |
| Memory encoding | 15–25 min during study |
Whenever you're learning | Active recall, flashcards, spaced repetition |
| Open meditation | 20–30 min sustained |
Mid-morning | Open-eye gaze or non-referential awareness |
| Pre-presentation | 10–15 min priming |
Just before | Light cardio · breath work · low caffeine |
| Sleep | — Do not use 40 Hz · switch to delta or low theta — | ||
Audio 40 Hz at normal listening volumes is well-tolerated by healthy adults. The cautions worth taking seriously concern combined audio-visual protocols, specific medical histories, and reasonable expectations about what the research does and does not yet support.
No. The mouse work from Tsai's lab at MIT (Iaccarino 2016, Martorell 2019) is real and important — it shows that 40 Hz light and sound can entrain cortical gamma and shift microglial state in transgenic mouse models of amyloid pathology. Human trials, including Cognito Therapeutics' GammaSense and the Sound-of-Life audio-only trials, produced mixed results in 2024–2025. Some structural and functional endpoints improved; others, including key cognitive composites, did not. As of mid-2026, 40 Hz stimulation remains a research therapy. Anyone with a diagnosis should treat it as adjunctive at most, and only under clinical supervision.
Acute subjective effects — alertness, perceptual sharpness, a sense of being "in the room" — tend to appear within 10–20 minutes for responders. EEG signatures of gamma entrainment can be detected sooner. Cognitive effects on memory and attention, where present, are small and most reliable when 40 Hz is paired with the task you want to improve. Some people respond on the first session; others notice nothing for a week of daily use and then suddenly experience a shift. Individual variation is large.
No. Gamma is the brain's most energising band — listening to 40 Hz in the hours before bed is the opposite of what you want. For sleep, use delta (0.5–4 Hz) or low theta (4–6 Hz). 40 Hz belongs to the morning and to pre-task priming, not to wind-down. If you find 40 Hz "relaxing", you may be experiencing the focus-related state rather than true sedation — still not what the parasympathetic system needs at 10pm.
Beta (13–30 Hz) is the working rhythm of focused, externally directed attention — the rhythm of solving, reading, writing, coding. Gamma (30–100 Hz, anchored around 40 Hz) is faster and is associated with feature binding: the moment disparate cortical regions agree on a single percept, the integration of a stimulus into a unified experience, the click of insight. In practice, beta-band binaural is the workhorse for sustained task focus; 40 Hz gamma is a sharper, more cognitive-integrative state that wears thin after about 45 minutes.
No, but it's the only way to see entrainment directly. Subjective reports correlate only weakly with measured gamma power. Consumer EEG devices (Muse, Emotiv, OpenBCI) can pick up frequency-following responses in the gamma band, though the signal is small and prone to artefact from jaw and eye muscles. For most users, the practical test is simpler: does the listening session reliably precede a clearer block of work, and does that effect survive a few weeks of honest journaling? If yes, it's doing something useful for you. If no, the frequency may not be your responder profile.
Audio 40 Hz at safe listening volume is well-tolerated. The non-trivial caveat is photosensitive epilepsy, which is contraindicated for any flickering-light protocol — including the original MIT light setup. Pregnant users, under-18s, anyone with active migraine or a vestibular disorder, and anyone with a diagnosed cognitive condition should talk to a clinician before a prolonged daily protocol. Stop immediately if you experience headache, nausea, dissociation, or visual disturbance.
Open the free web generator, set the target frequency to 40 Hz, put on headphones, and start a 20-minute morning block. Or install the mobile app for offline 48 kHz synthesis, expert-built gamma presets, and progressive timers tuned for cognitive work.