Mind & Body · Thursday, 27 August 2026

01 · Briefing · what happened

Shoes that buzz at random make the foot feel more, not less - and the brain does it to itself

Mind & Body 1 min 32 sources

A sense reports nothing below its own cut-off. Add a little randomness and some of the discarded signal gets over the line - which is now being trialled in balance, tinnitus and vision, and sold well ahead of the evidence.

266 mm2

standing sway with the buzz on

down from 438 without it, same test, same people [1]

100-400

millionths of an amp behind the ears

a 2026 review's range, set person by person [3]

trials pooled on vibration and walking

average gain small, and the range for a new trial crosses zero [8]

6 months

how long one balance gain held

in 40 adults averaging 77.7 years, after six weeks of stimulation [5]

At a glance

Switching on a barely-felt buzz under the feet cut standing sway from about 438 to 266 square millimetres in the same test. [1]
The buzz was set at the edge of what each person could feel, measured individually, and ankle muscle activity did not change. The gain was in sensing, not in strength. [1]
The effect has a name: stochastic resonance. Random noise added to a weak signal lifts part of it over the cut-off that would otherwise silence it. [2]
It only works where there is a hard cut-off. In a system without one, added noise just masks the signal. [2]
Too little noise does nothing. Too much drowns the signal. The benefit peaks in the middle, in an upside-down U. [10][12]
The peak sits in a different place for each person, so trials now set the level individually before testing anything. [3][12]
A 2026 review of noise delivered behind the ears settled on 100 to 400 millionths of an amp, in bursts of 10 to 30 seconds. It also says the effect is not universal. Where the inner-ear balance sense was lost on both sides, it helped only those with some function left. [3]
Where it fails, it fails badly. In 14 people with Parkinson's, noise during disturbed walking made body sway worse on the more affected side. [4]
It is not dependable in children either. In eye-movement tasks with children who had varying ADHD symptoms, white noise helped some and did nothing for others. The authors say this challenges the idea that it is reliably useful. [9]
Other senses show it too. Random electrical noise over the visual part of the brain improved reading at the edge of vision, where a steady current did not. Random visual static fed through augmented-reality glasses has been reported to improve sight in age-related macular degeneration. [13][11]
The brain appears to run the trick on itself. After hearing loss it turns up its own internal noise to keep faint sounds detectable, and tinnitus may be the bill for that repair. [6]
The boldest recent claim, that tuned noise could reset immune tolerance, is a hypothesis paper whose authors say direct evidence is lacking. [2]

Forces in play

Evidence for the mechanism High→

the threshold physics is settled and turns up in touch, hearing and vision [1][6][13]

Evidence for the treatments Building↑

the trials run 14 to 40 people, and the largest are still labelled pilots [4][5][30]

Claims running ahead Building↑

a 2025 paper extends noise to immune tolerance and calls itself speculative [2]

Dose has to be personal High→

the best level differs per person, and the wrong level does nothing or harms [3][12][4]

In play Patrick Krauss and Achim Schilling — authors of the noise model of tinnitus, reviewing ten years of it [6] The Erlangen tinnitus group — ran a 72-patient trial fitting each person their own noise level [7] A Spaulding Rehabilitation team at Harvard — proposed extending noise therapy to the immune system, and called it a hypothesis [2] Older adults at risk of falls — the group most of these devices are aimed at [5][17]

How it unfolded

2016 noise is proposed as the reason tinnitus exists, instead of simple misfiring [6]
Ten years on the idea has grown into a wider theory of how hearing copes with loss [6]
2025 a 72-patient trial finds individually fitted low-level noise cuts tinnitus distress [7]
2026 a review sets a standard dose and reporting checklist for noise at the ears [3]
Also 2026 a registered walking trial finds the same idea makes stability worse in Parkinson's [4]

Where this points

The next real test is whether tuned noise beats the sound therapies and insoles already on sale, a comparison the tinnitus review says has never been run. [6]

Also today

9 more stories on this beat.

Long COVID linked to an immune attack on nerves

A Yale and Mount Sinai team tested antibodies from 147 people, reporting in Cell on 28 May 2026. [18] Blood from long COVID patients reacted to human nervous-system tissue more often than blood from healthy people. [18]

Why it matters — It offers a mechanism rather than a symptom list, and in children a one-year study found fatigue in about half of both girls and boys with the condition. [31]
The brain scans, meanwhile, look ordinary

A multi-centre study scanned 140 adults who had lasting neurological symptoms after COVID, at 6 and 12 months. White-matter spots were no more common than in healthy controls, and changes over time were rare. [32]

Why it matters — Real symptoms with a clean scan is the hardest kind of illness to have taken seriously.
The gut-brain line weakens with age in mice

A Stanford and Arc Institute team reported in Nature on 11 March 2026 that old mice did worse on memory tasks. [19] Swapping young mice onto older gut microbes changed how they performed. [19]

Why it matters — Mice are not people, but it names a route from gut to memory that reviews now trace into Alzheimer's, Parkinson's and multiple sclerosis. [27]
Walking slowed Alzheimer's decline, not its plaques

Nearly 300 older adults were followed up to 14 years. Among those starting with high beta-amyloid, low to moderate activity halved the rate of decline, levelling off around 5,000 to 7,500 steps a day. [20]

Why it matters — The protection tracked slower tau build-up, not the amyloid that most current drugs are aimed at. [20]
ME/CFS blood shows an immune system stuck on

Blood from 56 people with ME/CFS and 52 matched controls was tested before and after exercise. The ME/CFS samples carried more inflammatory molecules, signs of oxidative damage and impaired energy production. [21]

Why it matters — It joins chronic pain, where a review argues that worn-out and aged immune cells are the piece nobody has examined. [28]
Insomnia and anxiety tracked fewer immune cells

In young women in Saudi Arabia, insomnia symptoms went with fewer natural killer cells overall, and anxiety symptoms with fewer of the kind that circulate in the blood. [22]

Why it matters — Stress reaches further than mood: another review traces long stress into acne, psoriasis and eczema. [29]
Sleep's steadiness matters, not only its length

A 2026 review argues that day-to-day swings in how long and when people sleep affect heart and metabolic risk in their own right, separately from total hours. [23] A 2025 review sets out how sleep reaches those systems at all: glucose handling, blood pressure, appetite hormones and inflammation. [24]

Why it matters — Nearly all sleep guidance is built around a nightly number, which is blind to the swing.
Buzzing devices are being tried well beyond balance

A pilot in 38 cancer survivors with nerve damage from chemotherapy found low-intensity vibration improved reported sensory symptoms and lower-body function against usual care. A separate open-label pilot ran a 40 Hz buzz with 20 long COVID patients. [14][30]

Why it matters — Both are small and the second had no dummy device, so this is promise rather than proof.
Artificial sweeteners turn out not to be inert

A review finds non-sugar sweeteners shift gut bacteria, cut short-chain fatty acid production and blunt fullness signals, while separating sweet taste from calories in the brain's reward pathways. [26]

Why it matters — The whole premise of a sugar substitute was that the body would not notice it.

02 · Lesson · why it matters

Why a little randomness beats a stronger signal

A sense reports only what clears a fixed line, so the way past it is often not more strength but a little randomness.

How it works

A sense reports only above a fixed cut-off
Everything weaker than the cut-off is discarded, unreported
Add random jitter and some weak moments get pushed over
Enough get through that the pattern becomes readable again
Raise the jitter further and it swamps the pattern instead

The twist

The way past a threshold is not always a stronger signal - sometimes it is a shakier one, because randomness lends the weak moments a lift.

Where you've seen this

Digital sound and pictures

engineers add a faint deliberate hiss so quiet detail survives being rounded to the nearest step

Digital scales and thermometers

a display that only steps in whole units hides slow drift until something jiggles the reading across a step

Pass-or-fail screening tests

the cut-off throws away every near miss, and where the day's measurement lands is partly chance

The catch

The window is narrow and personal - the same amount of noise that helps one system buries the next, and in Parkinson's walking it made things worse.

And the whole of it

Every human sense works this way, and none of them reports where its own cut-off sits. What gets noticed is already the part that cleared an invisible line, helped or hidden by a wobble nobody chose.

03 · Lab · your turn

Over the line

Walk a faint signal past a sense's cut-off by adding noise, and find where more noise starts burying it instead.

04 · Truth · what's really going on

Stripped of the framing

A real, reproducible physics of thresholds is being carried well ahead of its evidence into a market for buzzing insoles and ear devices.

Why it lands — It reverses something everybody already believes, that noise ruins a signal, and a reversal that tidy travels much further than its evidence does.

Claimed

What people said. Not yet a fact.

A Spaulding Rehabilitation team at Harvard Medical School

Tuned noise, microbial or electrical, could restore immune tolerance in conditions such as multiple sclerosis and fibromyalgia. [2]

Their own limitations section calls the work conceptual, hypothesis-driven and speculative, with direct evidence lacking. [2]
The authors of the plantar vibration trial

Eight weeks of white-noise insoles improved dynamic balance control. [1]

That arm of the study ran eight people against eight on silent insoles, which is a signal to test rather than a result to build on. [1]
Krauss and Schilling, reviewing ten years of the tinnitus noise model

Tinnitus is the side effect of the brain usefully raising its own noise after hearing loss. [6]

They also state that the sound therapies built on the model have never been compared head to head with the maskers already sold. [6]
The Erlangen tinnitus group

Individually fitted low-level noise can suppress tinnitus, in some cases completely. [7]

The complete-suppression part rests on what patients reported, and the group given a dummy sound first benefited much less. [7]

Verified

What we could actually stand behind.

The effect needs a system with a hard cut-off; in one without, added noise only masks the signal. [2]

How we checked — Stated the same way in a physiology review, in the vestibular standardisation review, and in the visual stimulation study in the pool. [2][3][13]
The benefit rises and then falls with the noise level, peaking in the middle. [10][12]

How we checked — Two unrelated papers, one on insect signalling and one on human heart-rate responses, both report the upside-down U. [10][12]
The dose has to be set person by person. [3][12]

How we checked — A 2026 review builds individual titration into its recommended protocol, and the heart study found individual variation set the optimum. [3][12]
Noise stimulation made walking stability worse in 14 people with Parkinson's. [4]

How we checked — A registered trial reporting its own negative result: peak body-centre excursion rose significantly on the more affected side. [4]
Vibration's average effect on walking in Parkinson's is small. [8]

How we checked — A meta-analysis of 14 trials; the pooled gain was 0.27 and the range expected for a new trial crosses zero. [8]

Nobody knows

Open questions — ours included.

Whether tuned noise beats the sound therapies and insoles already on sale.

The tinnitus review says that comparison has never been run systematically. [6]
How long any of the benefit lasts.

The longest follow-up in the pool is six months, in a 40-person study its own authors call a pilot. [5]
Who it helps and who it harms.

One review reports it working in healthy and impaired people alike, and failing where the sense is gone entirely; the Parkinson's walking trial found harm. [3][4]
Whether the immune version is real at all.

Nobody has shown the effect in human immune signalling, and the paper proposing it says so itself. [2]

Who gains

Makers of vibrating and textured insoles — The mechanism hands a cheap consumer device a research vocabulary. Insole trials report better standing balance, and a review of below-knee amputees reports vibrating feedback raising walking speed by 41 percent. [16][17][15]
Rehabilitation researchers and their funders — A tuned-noise protocol is far cheaper to trial than a drug, and the 2026 standardisation review exists to speed that pipeline up. [3]
Firms selling into fall prevention — Falls are the leading cause of injury hospital admission in older adults, and that is the market these balance trials point at. [5]
Home rehabilitation programmes — Remote delivery matched in-person delivery in a balance-game trial, which is the finding that lets a programme run without a room full of staff. [25]

Who pays

Older adults afraid of falling — They are the group these devices are aimed at, and the group least placed to judge a 16-person trial. [1][5]
People with Parkinson's — The strongest marketing case and the clearest recorded harm sit in the same population. [4][8]
People with tinnitus — A fitted-noise device costs money and weeks of appointments, and nobody has compared it with the cheaper maskers already available. [6][7]
People with nerve damage from chemotherapy — A 38-person pilot is being read as evidence for a device category, and the pilot's authors call it preliminary. [14]

05 · Hope · carry this

For a century the noise in a measurement was the thing to get rid of. It turned out to be part of how the machinery works, and that discovery cost nothing but a second look at something already filed away.