Light and Sound at 40 Hz: New Review Weighs the Evidence Behind a Bold Alzheimer’s Therapy

A flickering light and a humming tone, pulsing together at 40 times per second, sound more like a meditation gadget than a medical treatment. Yet over the past decade, this simple form of sensory stimulation has become one of the most intriguing noninvasive approaches under investigation for Alzheimer’s disease. The idea rests on a striking biological observation: gamma-band oscillations, the brain’s fastest rhythmic activity spanning roughly 30 to 100 hertz, are impaired in Alzheimer’s disease, and coaxing them back with external stimulation appears in animal studies to reduce amyloid plaques, alter immune signaling, and even promote the clearance of toxic proteins. Now, a systematic review and exploratory meta-analysis published in GeroScience has taken stock of everything the technique, known as gamma entrainment using sensory stimulation, or GENUS, has actually achieved in human beings. The verdict is a careful balance of promise and restraint.

The review, led by Myeonghae Hyun, Yeseung Park, Seo Yoon Park, and senior author Ki Woong Kim of Seoul National University, followed PRISMA 2020 reporting guidelines and searched four major databases, PubMed, Embase, Scopus, and Web of Science, for English-language articles published between January 2016 and December 2025. From an initial haul of 1,310 records, the team winnowed the literature down to just eight studies that met their strict eligibility criteria: human participants aged 60 or older, noninvasive sensory stimulation delivered at approximately 40 hertz through visual, auditory, or combined audiovisual means, and at least one quantitative outcome related to brain structure, function, electrophysiology, or cognition. The included studies ranged from single-session experiments to six-month daily interventions, with sample sizes between 10 and 38 participants, spanning cognitively normal older adults to patients with mild Alzheimer’s dementia. That small and heterogeneous evidence base, the authors emphasize, fundamentally shapes what can and cannot be concluded.

The biological rationale behind GENUS is worth unpacking. Gamma oscillations arise largely from interactions between excitatory pyramidal neurons and fast-spiking parvalbumin-positive interneurons, the inhibitory cells that act as the brain’s metronome. In transgenic mouse models of amyloid pathology, gamma power is reduced in the hippocampus and entorhinal cortex even before plaques become visible, suggesting that gamma dysfunction may be an early pathophysiological event rather than a mere consequence of neurodegeneration. When researchers delivered 40 Hz flicker or sound to these mice, they observed changes in amyloid and tau pathology, glial and immune signaling, synaptic and myelin integrity, and even behavior. One 2024 study reported that multisensory gamma stimulation promotes glymphatic clearance of amyloid, implicating the brain’s waste-clearance system. These preclinical findings provided the impetus for translating the technique to humans, but translation has focused mainly on feasibility, target engagement, and early signals rather than definitive efficacy.

On the structural front, the review found the most tantalizing, and most fragile, evidence. Two randomized pilot trials reported comparable magnetic resonance imaging outcomes. In a three-month trial of daily 40 Hz light and sound stimulation in mild probable Alzheimer’s dementia, Chan and colleagues found less ventricular enlargement and less hippocampal atrophy in the active group than in controls. Agger and colleagues, using invisible spectral flicker in a placebo-controlled double-blinded pilot study, saw a similar directional pattern that only emerged by week 12, with no meaningful separation at week 6. A third study, by Da and colleagues, reported favorable white matter and myelin-related findings after six months of daily stimulation, with the most prominent effects in the entorhinal region, though that study was rated at high overall risk of bias. The review’s exploratory meta-analysis pooled the hippocampal and ventricular data from the two controlled trials using a random-effects model.

The pooled numbers tell a story about statistical fragility as much as about the brain. For ventricular volume, the conventional DerSimonian–Laird analysis yielded a statistically significant effect favoring active stimulation, with a standardized mean difference of −1.330 and a p-value of 0.004, indicating less ventricular enlargement. But because only two studies contributed to the estimate, the authors applied the more conservative Hartung–Knapp–Sidik–Jonkman adjustment with a lower-bound modification, which is recommended when meta-analyses contain very few studies. Under that stricter approach, the confidence interval ballooned from roughly −2.2 to −0.4 to a sprawling −7.2 to 4.6, and the result was no longer statistically significant. The hippocampal estimate, directionally favorable with a pooled g of 0.687, was never significant under either method, and a sensitivity analysis substituting a different follow-up timepoint attenuated both results further. The authors are blunt: these pooled estimates are hypothesis-generating, not evidence of efficacy.

Functional neuroimaging offered a somewhat more coherent, if still preliminary, picture. In the Chan trial, resting-state functional MRI suggested that active stimulation was associated with relative preservation or enhancement of connectivity in networks relevant to Alzheimer’s dementia, most clearly in the medial visual network and posterior cingulate-related connectivity. A separate delayed-start feasibility study in prodromal Alzheimer’s disease, by He and colleagues, found that functional connectivity between the posterior cingulate cortex and precuneus increased after eight weeks of daily flicker, but not after four weeks of no stimulation or only four weeks of flicker. Because that study lacked a sham control, the finding is best read as evidence of network modulation rather than treatment benefit. Fluid biomarkers added another exploratory thread: core Alzheimer’s markers in cerebrospinal fluid, including amyloid-beta 42, total tau, and phosphorylated tau, showed no significant change, but the immune factor profile shifted after eight weeks, including a decrease in TWEAK, hinting at neuroimmune engagement.

The strongest mechanistic support came from electroencephalography. Five studies examined whether the stimulation actually does what it claims, and the answer is generally yes, with important caveats. Entrainment was stronger at lower stimulation frequencies and higher luminance, showing that success depends on careful parameter optimization rather than the mere presence of 40 Hz stimulation. How far the entrained rhythm spreads through the brain appears to depend on each participant’s individual center frequency and on white matter microstructural integrity, suggesting that structural connectivity moderates downstream network engagement. Individual variability emerged as a recurring theme, with some people proving to be strong entrainers and others not, possibly reflecting differences in baseline neural state. The authors note an important transparency caveat here: three of the five EEG studies were authored by members of the review team itself, and excluding them would substantially narrow the parameter-optimization and propagation findings while preserving the core evidence for target engagement.

Cognitive outcomes, the domain patients and families care most about, remain the least convincing. In the Chan trial, the active group improved on a face-name association delayed recall test and showed better circadian rhythm regularity on actigraphy, but global cognitive measures such as the Mini-Mental State Examination and the Montreal Cognitive Assessment did not differ between groups. The Agger trial found directionally lower, but statistically non-significant, scores on an extended cognitive and functional scale at both six and twelve weeks. With tiny samples and divergent measures, the review concludes that cognitive findings from these small exploratory studies do not establish a consistent benefit. The authors also flag methodological vulnerabilities in the underlying trials, including baseline age imbalances of 6.4 and 8.2 years between groups in two studies, a serious concern for outcomes like brain volume that are strongly age-sensitive, and a retrospectively registered review protocol.

Where does this leave the flickering light? The honest answer is that GENUS stands at the classic juncture of translational science: biologically plausible, mechanistically supported in animals, safe and feasible in humans, and demonstrably capable of engaging its neural target, yet unproven as a therapy. The review’s authors frame their synthesis as a rationale for adequately powered, standardized, sham-controlled trials rather than as evidence of therapeutic benefit, and they call for larger studies that harmonize stimulation parameters, outcome measures, and follow-up durations so that results can finally be pooled meaningfully. Until then, the 40 Hz flicker remains one of the most captivating hypotheses in dementia research, a reminder that sometimes the brain’s own rhythms, nudged by nothing more than light and sound, may hold therapeutic potential that rigorous science is only beginning to test.

Subject of Research: Noninvasive 40 Hz sensory gamma entrainment as an investigational approach for cognitive impairment and Alzheimer's disease in older adults

Article Title: Sensory gamma entrainment in older adults with or without cognitive impairment: a systematic review and exploratory meta-analysis

Article References: Hyun, M., Park, Y., Park, S. Y., & Kim, K. W. (2026). Sensory gamma entrainment in older adults with or without cognitive impairment: a systematic review and exploratory meta-analysis. GeroScience. https://doi.org/10.1007/s11357-026-02564-6

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02564-6

Keywords: gamma entrainment, GENUS, sensory stimulation, Alzheimer's disease, cognitive impairment, systematic review, meta-analysis, EEG, hippocampal atrophy, neuroimaging, GeroScience, brain stimulation