Add more noise to noise and what do you get? Less noise.
Bad reception on your TV screen? Garbled voices on your mobile phone? One way to fix it could be to add more interference. Adding more noise to noise can actually quieten things down, new research suggests 1.
"For a long time," say Jose Vilar of Princeton University in New Jersey and J. M. Rubí of the University of Barcelona, Spain, "noise was considered to be only a source of disorder, a nuisance to be avoided." Not any longer.
Vilar and Rubi have developed a mathematical model, based on how nerve impulses pass between cells, which suggests that adding more random variation to an incoming signal can even out the emergent signal.
In the past decade, physicists and engineers have been forced to rethink how 'noise' -- random fluctuations, like the ripples on the sea's otherwise flat surface -- affects a system.
Take, for example, the phenomenon 'stochastic resonance', in which the transmission of a signal, such as an information-laden radio broadcast, is enhanced by the addition of some static. You add some noise to an input signal, and the signal-to-noise ratio of the output improves.
Stochastic resonance has now been identified in a wide variety of systems, from electronic signal processing to the neural circuits of aquatic animals. Our own sensory systems might even use stochastic resonance to capitalize on the naturally 'noisy' environment in which we live.
The effect now identified by Vilar and Rubí sounds deceptively like stochastic resonance; but it is different. They are not concerned with the transmission of signals at all, but simply with how much a system fluctuates at random. If you shake a system that is already fluctuating, you'd expect it to fluctuate more. But under some conditions, say the researchers, it will fluctuate less.
Vilar and Rubí's model describes the fluctuations that occur in the electrical current in an array of ion channels: these are pore-like proteins that allow electrically charged ions to pass through nerve cell membranes. The model predicts a lower output noise (fluctuation in ion-channel current) if the 'input' voltage that controls the opening of the ion channels is more variable.
"I've never seen such a result before," says Alexander Neiman of the Center for Neurodynamics at the University of Missouri, St Louis. But he adds that "it seems to be very general", rather than being a peculiarity of this particular model.
It is hard at this point to give a general explanation of exactly how the strange 'suppression of noise by noise' works, Neiman confesses. "It is basically an interplay of nonlinearities," he suggests. A nonlinear effect does not simply increase in proportion to its cause, but may instead have unpredictable consequences. In this case, a scrambled signal going in might not be simply 'shaking' a system more but might allow it to access completely new kinds of behaviour.
Vilar,J. M. G. & Rubí, J.M. Noise suppression by noise. Physical Review Letters 86, 950 - 953 2001.
About this article
Cite this article
Ball, P. Come on feel the noise. Nature (2001). https://doi.org/10.1038/news010215-12