Audio Visual Artifacts
Journal 13 Light

Light that moves with sound

The eye and the ear do not vote separately. What the brain does when picture and sound agree, why flicker has hard limits, what the 40 Hz studies show, and why AVA keeps its visuals slow and dark.

In 1976 two psychologists in Surrey filmed a woman saying a syllable, then dubbed a different syllable over the picture. Harry McGurk and John MacDonald recorded her lips forming "ga" and laid the sound "ba" on top. Adults watching the film heard neither. They heard "da", a sound that was not in the room. Close your eyes and the "ba" came back. The paper, titled Hearing lips and seeing voices, ran in Nature, and the effect has carried McGurk's name ever since.

Twenty four years later Ladan Shams and colleagues showed the traffic runs the other way too. They flashed a single disc on a screen and played two short beeps with it. People reliably reported seeing two flashes. One flash, two beeps, two flashes seen. The sound-induced flash illusion, also published in Nature, is a clean demonstration that hearing can rewrite what vision reports, down to counting.

Both findings matter for a room that plays to more than one sense at a time. The brain does not keep the eye and the ear in separate ledgers. It builds one account of what is happening and lets the senses correct each other. When picture and sound agree, the account is stable and the body can settle. When they disagree, some part of you keeps working to reconcile them, whether you notice or not.

The hard limit on flicker

Before any of that, there is a safety line, and it is not a matter of taste. Roughly one person in four thousand has photosensitive epilepsy, and in those people a flashing light in the wrong range can trigger a seizure. The Epilepsy Foundation of America convened a working group in 2005, with the British neurophysiologist Graham Harding among its authors, and published both a review of the evidence and a consensus on what to avoid. The numbers are specific. Flash rates above three per second carry risk. The range around 15 to 25 flashes per second is the most provocative. A flash counts when brightness changes by 20 candelas per square metre or more, when it fills a large part of the visual field, and transitions to or from saturated red are worse than other colors.

Broadcasters adopted the same line. The International Telecommunication Union's recommendation BT.1702, Guidance for the reduction of photosensitive epileptic seizures caused by television, sets limits that television programs are tested against before air. On the web, the Web Content Accessibility Guidelines carry the rule as criterion 2.3.1: nothing on a page may flash more than three times in any one second unless it stays under the size and contrast thresholds.

Those documents exist because a Pokémon episode in 1997 sent several hundred Japanese children to hospital, and because the mechanism is not subtle. Bright, fast, large, red: avoid all four and you are on the safe side of the line. Any room that projects light on people lying down with their eyes open should be able to say how it handles each of them.

The 40 Hz studies, and their limits

There is one line of research where fast flicker is used on purpose, and it deserves an honest account because it is often quoted without one. In 2016 Li-Huei Tsai's laboratory at MIT reported in Nature that mice bred to accumulate the amyloid protein associated with Alzheimer's disease showed less of it in their visual cortex after an hour a day of light flickering at 40 Hz, the frequency of the brain's gamma rhythm. In 2019 the same group reported in Cell that adding a 40 Hz click track to the light extended the effect into the hippocampus and prefrontal cortex and improved the mice's performance on memory tasks. They named the combined light and sound approach GENUS.

Three things need saying next. First, these are mouse studies. Second, the first human data is small. A feasibility trial published in 2022 gave fifteen people with mild probable Alzheimer's a home device delivering 40 Hz light and sound for an hour a day for three months. It was designed to test whether the device was safe and tolerated, and it was, with some encouraging secondary signals. It was not designed to prove benefit, and it does not. Third, in 2023 a separate group published in Nature Neuroscience that 40 Hz light did not entrain the native gamma rhythm in the hippocampus of their mouse models and did not change amyloid there. Two laboratories, two answers. That is what an open question looks like.

So the correct summary of 40 Hz in 2026 is: promising in mice, tested for safety in a handful of humans, contested on mechanism, and delivered in every study under medical supervision with photosensitivity screening, because 40 flashes a second is far above the three per second that the epilepsy guidelines allow for the general public. It is not something a wellness room should improvise.

Bright, fast, large, red. Avoid all four.

Why AVA keeps it slow and dark

We took two conclusions from all of this. The first is that the picture must agree with the sound, because the McGurk and Shams experiments show the brain will force an agreement anyway, and it is better to hand it one than to make it work. On AVA the visuals are generated from the same audio signal that drives the bed. The low band that you feel at 41 Hz under your feet is the same number that swells the image. There is no separate timeline to drift out of step.

The second is that the picture should be quiet. We keep the projected image dim, we keep transitions slow, and we do not use hard cuts or strobes. The image moves the way the low end moves, over seconds rather than frames, and it never approaches the three flashes a second line.

In practice that means four rules for anyone who builds a visual for the room. The image is driven by the slow envelope of the low band, not by every beat, so a kick drum at 120 beats a minute does not become a light pulsing twice a second. Brightness stays low across the whole frame, with no full-screen white. Nothing cuts; every change is a fade measured in seconds. And we avoid saturated red, the color the epilepsy guidance singles out. A visual that fails any of those rules does not play, however good it looks on a monitor.

Rest wants low light for reasons that have nothing to do with epilepsy: a body lying on its back in a dark room with a slow picture above it is a body that has been told nothing is about to happen. That is the message the whole room is built to send.

On AVA, the octagon bed at 29 Nassau Ave plays through 33 transducers in five body zones, from 41 Hz at the feet to 880 Hz at the head, and the visuals projected over the room are rendered live from that same signal, slow and dark by design. If you have photosensitive epilepsy, tell us when you book and we will run your session with the projector off. This is wellness content, not medical advice.

Sources

  1. McGurk H, MacDonald J. Hearing lips and seeing voices. Nature, 1976. pubmed.ncbi.nlm.nih.gov and doi.org
  2. Shams L, Kamitani Y, Shimojo S. What you see is what you hear. Nature, 2000. doi.org
  3. Harding G, Wilkins AJ, Erba G, Barkley GL, Fisher RS. Photic- and pattern-induced seizures: expert consensus of the Epilepsy Foundation of America Working Group. Epilepsia, 2005. pubmed.ncbi.nlm.nih.gov
  4. Fisher RS, Harding G, Erba G, Barkley GL, Wilkins A. Photic- and pattern-induced seizures: a review for the Epilepsy Foundation of America Working Group. Epilepsia, 2005. pubmed.ncbi.nlm.nih.gov
  5. International Telecommunication Union. Recommendation ITU-R BT.1702: Guidance for the reduction of photosensitive epileptic seizures caused by television. itu.int
  6. W3C. Web Content Accessibility Guidelines 2.1, success criterion 2.3.1, Three Flashes or Below Threshold. w3.org
  7. Iaccarino HF, Singer AC, Martorell AJ, et al. Gamma frequency entrainment attenuates amyloid load and modifies microglia. Nature, 2016. pubmed.ncbi.nlm.nih.gov
  8. Martorell AJ, Paulson AL, Suk HJ, et al. Multi-sensory gamma stimulation ameliorates Alzheimer's-associated pathology and improves cognition. Cell, 2019. pubmed.ncbi.nlm.nih.gov
  9. Chan D, Suk HJ, Jackson BL, et al. Gamma frequency sensory stimulation in mild probable Alzheimer's dementia patients: results of feasibility and pilot studies. PLoS One, 2022. pubmed.ncbi.nlm.nih.gov
  10. Soula M, Martín-Ávila A, Zhang Y, et al. Forty-hertz light stimulation does not entrain native gamma oscillations in Alzheimer's disease model mice. Nature Neuroscience, 2023. pubmed.ncbi.nlm.nih.gov
  11. Picower Institute for Learning and Memory, MIT. GENUS: gamma entrainment using sensory stimulation. picower.mit.edu

The 40 Hz studies above were carried out in mice or in small supervised human trials, and one of them found no effect. AVA does not deliver 40 Hz flicker and makes no claim about Alzheimer's disease. The flicker limits cited are published safety guidance, not a guarantee for any individual. This is wellness content, not medical advice.

Book

One room. Eight seats on AVA. Sessions from $65.

AVAUDIO VISUAL ARTIFACTS

AVA, our audio visual artifact, in the listening room at 29 Nassau Ave.

Book a session