Indexing the archive…
Your Universe of Digital Possibilities
A signal too weak to cross a detection threshold sits in silence. Add just enough noise and the random kicks lift it over the threshold at exactly the right moments — output spikes cluster at the signal’s peaks, and the SNR of the output is higher than it was without noise. The SNR peaks at an optimal noise level σopt, then falls. Roberto Benzi proposed this in 1981 to explain the 100,000-year ice-age cycle; it was later found in biological neurons, hair cells, and tactile sensors. Evolution apparently found the hump.
A periodic signal of amplitude A sits below detection threshold θ — by itself it never fires the detector. Adding Gaussian white noise σ ξ(t) creates random barrier crossings; at the right σ, crossings cluster at the signal’s peaks and the output becomes correlated with the input. Biological sensory neurons, inner-ear hair cells, and crayfish mechanoreceptors all appear to operate near this optimal noise level.
A subthreshold signal — too weak to cross a detection threshold on its own — can become detectable by adding the right amount of noise. The output signal-to-noise ratio rises with noise from zero, peaks at an optimal amplitude, then falls again. Noise is not just a nuisance; here it is the carrier.
The rate at which thermal fluctuations knock a particle over a potential barrier ΔU, with D = σ²/2 the noise diffusion. In the threshold model of stochastic resonance, this sets how often the noisy signal crosses the threshold — and therefore the output firing rate that tracks the input signal.
Stochastic resonance sits at the rack’s deepest intersection: between The Code (INST·20, where noise is the enemy of channel capacity) and The Walk (INST·19, where random steps are the mechanism). Here noise is neither enemy nor mechanism — it is the carrier. The threshold model is the simplest possible detector, and it is exactly this model — not a more elaborate one — that produces the hump. Inner-ear hair cells, crayfish mechanoreceptors, and muscle spindles all show SR; the noise floor of the sensory system is set near σopt. The most surprising implication: if you remove all noise from a biological sensor, you make it worse. The universe charges you for silence as surely as it charges you for knowledge (see The Demon, INST·39).