Subsensory

In 1993, a team at the University of Missouri put electrodes on the tail fans of crayfish and played noise into the water. Not signal — just noise. Random fluctuation. Then they added a weak periodic signal underneath it, one too faint for the crayfish's mechanoreceptors to detect on their own. They found that with the right amount of noise, the signal became detectable. Not despite the noise. Because of it.

The mechanism isn't complicated to describe, even if it's strange to accept. A neuron has a threshold — a level of input it has to reach before it fires. A signal below that threshold does nothing: arrives, fails to cross, leaves no mark. Noise is different from signal. Noise varies randomly, sometimes up, sometimes down. If you add noise to a subthreshold signal, there are moments when the noise bumps the combined input above the threshold and the neuron fires. The timing of those firings carries information about the original signal — not perfectly, but enough. Too much noise and it all blurs out. Too little and you're back where you started. At the right level, you get detection you wouldn't have had otherwise.

This was the first biological demonstration of something called stochastic resonance. The idea had been around in physics since the early 1980s, proposed as a possible explanation for the timing of ice ages. What Douglass and colleagues showed was that a real nervous system, in a real animal, was sensitive enough to exploit it.

What came after is harder to sit with.

If you play white noise through headphones to a person at the right volume — around 70 decibels, roughly the level of a moderately loud conversation — their ability to detect faint touch on their fingertips improves. So does their ability to detect faint visual signals. So does their postural control, which depends on proprioception. The same auditory noise, at the same volume, improves three fundamentally different sensory systems simultaneously. Touch, vision, the body's sense of itself in space — all of them become more sensitive when a specific amount of unrelated noise enters through the ears.

I've been sitting with this. It shouldn't work. The ear is not connected to the fingertip. The auditory nerve is not the same pathway as the somatosensory nerve. These are different channels, different anatomies, different cortical areas. What could auditory noise be doing for touch?

The proposed answer involves brain regions — the superior colliculus, the posterior parietal cortex — that integrate information from multiple senses. These areas are always doing something. Auditory input, even random auditory input, reaches them. It appears to raise a kind of general activation level in circuits that handle cross-modal coordination, and that raised level moves other sensory systems closer to their detection thresholds. Not over the threshold — just closer. So that signals they would have missed, they now catch.

Closer to threshold isn't the same as over it. The noise doesn't create signals that weren't there. It makes the system slightly more available to signals that were always arriving.

There is a shoe insole. This is the part that keeps returning. Researchers built a piezoelectric insole that vibrates at a level below what the wearer can consciously feel — below their own sensory threshold. Elderly participants walked with these insoles and were steadier on their feet. Postural sway decreased by roughly 18 percent. Gait variability dropped. They performed better on balance tests. None of them could feel the vibration. The insole was delivering noise they weren't aware of, to soles of feet they'd lost some sensitivity in, and it was working through a mechanism they had no access to.

Their balance improved, and they couldn't have told you why.

What I keep returning to is what this implies about thresholds. A threshold sounds like a fixed feature of a system — the fence below which nothing passes. But a threshold is a comparison. Input compared to background state. What stochastic resonance reveals is that the background state is adjustable, and adjusting it — even with noise — moves the effective threshold. The fence isn't fixed. It sits on ground that can be shifted.

This makes me wonder about all the signals that are arriving just below threshold, all the time, that never get registered. Not because they're absent. Because the background isn't in the right position. What would it take to move it? And is some of what passes for ordinary perception — the sense that you're detecting what's there — actually a negotiation between signal and background that you have no view of?

I don't know. The crayfish literature doesn't resolve this. Neither does the shoe insole study. But something about the image stays: an elderly person walking steadily on vibrations they cannot feel, their nervous system doing something useful with noise they aren't aware of, the threshold having shifted without their knowledge, the world having become slightly more available to them in ways they can't report.