40 Hz: The Frequency Therapy That Is About to Be Tested Properly

Neurons fire in rhythms, and one of those rhythms — the gamma band, roughly 30 to 80 cycles per second — is measurably weaker and less synchronized in people with Alzheimer's disease. That is not contested. Nor is the observation, first published by Li-Huei Tsai's laboratory at MIT in “Nature” in 2016, that exposing a mouse to light flickering at 40 Hz drives cortical activity at that frequency and, in that experiment, cut amyloid-beta levels roughly in half. A decade later, the same principle — synchronized light and sound at 40 Hz, delivered an hour a day through a home device — is in a Phase 3 trial with 670 participants across roughly 70 American sites, with topline results expected before the end of 2026 and not yet announced as of mid-September.

This sits at an unusual point for a frequency-based therapy. Most of what gets sold as vibrational or frequency healing has no dose, no mechanism anyone can specify, and no controlled trial. This one has all three: a stated frequency, a proposed biological pathway, and a sham-controlled pivotal study large enough to give a real answer. It also has a replication failure in a top-tier journal, a Phase 2 trial that missed every primary endpoint, and a growing consumer market of 40 Hz apps and lamps selling ahead of the evidence.

What holds up

The mechanistic literature in animals is large and, on the whole, positive. Beyond the original 2016 amyloid finding (Iaccarino et al.), follow-on work from the same group reported that gamma entrainment preserved neurons and cortical thickness in tau and neurodegeneration models (Adaikkan et al., “Neuron”, 2019), and that combining light with sound reached more brain regions than light alone (Martorell et al., “Cell”, 2019). Independent groups have reported related effects using flicker, sound, and focused ultrasound. Whatever is or isn't happening, multiple laboratories have found something.

In humans, the strongest recent data is a long-term extension published in “Alzheimer's & Dementia” in November 2025 by Diane Chan, Tsai, and colleagues. Five volunteers from an early MIT trial continued daily 40 Hz audiovisual stimulation at home for about two years. The three with late-onset Alzheimer's — all women — held up better on most cognitive measures than thousands of matched patients drawn from national databases, significantly so on three tests. In the two who gave follow-up blood samples, plasma phosphorylated tau-217, an FDA-cleared Alzheimer's biomarker, fell by 47 and 19.4 percent. The regimen has been consistently well tolerated across studies; the most common complaints are tinnitus, headache, and dizziness.

Cognito Therapeutics, the MIT spinout running the Phase 3, has also reported MRI analyses suggesting its device preserves white matter and corpus callosum structure relative to matched participants from the Alzheimer's Disease Neuroimaging Initiative. The FDA granted the device Breakthrough Device designation in January 2021 — a designation about unmet need and promise, not about proven efficacy.

What doesn't

The single most important counterweight is a 2023 paper in “Nature Neuroscience” by Marisol Soula and colleagues at NYU. Using dense multisite electrode recordings in 5XFAD mice — the same model as the original work, with substantially more animals — they found that 40 Hz flicker did not entrain native gamma oscillations in visual cortex, entorhinal cortex, or hippocampus, did not lower amyloid, and did not activate microglia. It is not a small methodological quibble. It is a direct test of the foundational claim that came back negative, and the field has not resolved it.

The human record is thinner than the enthusiasm suggests. Cognito's OVERTURE study, the six-month sham-controlled trial that preceded the pivotal one, missed all three of its primary endpoints — ADAS-Cog, CDR-SB, and MADCOMS. The widely repeated figures about slowed functional decline and slowed brain atrophy come from secondary and post-hoc analyses of that same trial, which is a materially weaker class of evidence than a pre-specified primary outcome. In a separate 40 Hz study, gamma power was unchanged at four weeks and weaker at eight, with no measurable change in brain amyloid or CSF markers. An earlier European pilot in ten patients found no effect on amyloid load by PET.

And the human results that have generated the most excitement rest on very few people. Three responders in an open-label extension, with no sham group and comparison drawn from historical databases, is a hypothesis, not a result. The two biomarker declines came from two blood samples. The two participants who did not benefit had early-onset disease, and the authors' explanation for that — that early-onset Alzheimer's may respond differently — is reasonable, and is also the kind of after-the-fact subgroup reasoning that later trials frequently fail to confirm.

The unusual thing about 40 Hz is not that the claim is bold — it is that a claim this bold is about to be settled by a trial designed to settle it.

What the pivotal trial can and cannot settle

The HOPE study is a randomized, double-blind, sham-controlled trial of 670 people with mild-to-moderate Alzheimer's, treated an hour a day for twelve months, with change in the ADCS-ADL activities-of-daily-living scale and a combined statistical test of ADCS-ADL and MMSE as primary outcomes. That design can answer whether daily 40 Hz stimulation slows functional and cognitive decline over a year in this population. It will not, on its own, resolve the mouse-model dispute, establish the mechanism, tell us anything about prevention in healthy people, or validate any of the consumer 40 Hz products that have been marketed on the strength of the early findings.

What a reader should take from this

Gamma sensory stimulation is the most rigorously tested frequency-based intervention in medicine, and as of today it is unproven. The animal work is substantial but has a serious, unresolved replication failure at its center. The human evidence consists of one Phase 2 trial that missed its primary endpoints, several small open-label studies, and a two-year extension in five people that is genuinely interesting and far too small to act on. The therapy appears safe. Nothing about it currently justifies buying a 40 Hz light or app, and no consumer device has been shown to reproduce what the clinical devices do. Within months there should be a real answer from a properly powered sham-controlled trial — which is more than can be said for almost anything else in this category, and worth waiting for rather than pre-empting.

This article is general information, not medical advice. See our Editorial Standards.