Stochastic Resonance Can Help Improve Signal Detection

Physicists are discovering evidence that adding random noise in the right way to certain electronic devices and biological systems can counter-intuitively increase the detectability of signals and the transmission efficiency of information. Because of its generic nature, this phenomenon, known as "stochastic resonance," holds universal applications extending from classical and quantum physics to biomedicine.

The term "stochastic resonance" was first coined in 1981 by physicists to describe the annoying hiss in modern communication devices that can hinder the detection of weak signals. However, under certain conditions, it has been shown to actually aid detection. This effect is rooted in three basic ingredients: a source of background noise, a generally weak coherent input, and a characteristic sensory barrier or threshold that the system typically has to overcome in order to perform its usual task.

As reported in a Monday Division of Biological Physics session of the March Meeting, scientists have begun to question whether stochastic resonance manifests itself on a quantum scale as well. Peter Hnggi of the University of Augsburg in Germany has theorized that because quantum noise persists even at absolute zero temperature, the transport of quantum information should be aided by quantum fluctuations as well. In fact, the phenomenon of quantum tunneling — in which a particle "tunnels" through a barrier without going over it — provides the nonlinear system with an additional channel to overcome a threshold. For strongly damped systems, these quantum corrections can enhance the classical SR effect up to two orders of magnitude.

In addition, a series of novel nonlinear quantum stochastic resonance phenomena occur in the deep quantum cold regime, where tunneling in the presence of dissipation and periodic time-dependent perturbations mutually influence each other. Of particular interest to scientists is the effect of driving-induced quantum coherences in damped quantum systems, as well as the phenomenon of a characteristic suppression of nonlinear higher harmonic responses. The latter can be used to "clean" the quantum output distortion that is caused in nonlinearly processed information.

According to Hnggi, these novel effects have now been confirmed experimentally using a superconducting quantum-interference device (SQUID), which constitutes a macroscopic quantum system where quantum tunneling transitions begin to modify and blur the classical stochastic resonance response with decreasing temperature. "The combined application of both time-dependent perturbations and quantum dissipation yields novel possibilities to influence quantum processes," he said. These include strong laser light induced manipulation of electron transfer processes, which can be used to control reactant-product yields in chemical reactions, or to regulate tunneling of atoms or molecules that are deposited on surfaces.

Stochastic resonance has also been demonstrated in complex systems of biological transducers and neural signal pathways. For example, it has been experimentally observed to improve broadband encoding in the cricket cercal system (see related story, page 3). However, the principles of biological amplications are far from understood. "It is only clear that biological amplifiers are unique in their ability to detect small signals in a noisy environment," said Igor Vodyanov (National Institute of Health). However, the possibility that the phenomenon could occur at the sub-cellular level remains open, and Vodyanov reported on the recent observation of noise-enhanced electrical signal transfer in a simple system of voltage- dependent ion channels formed by the peptide alamethicin in a lipid bylayer.

In most experiments with stochastic resonance, the weak and noisy signals were first recorded with electronic instruments, and the resulting data were analyzed by computer. Italian physicist Enrico Simonotto of the University of Genoa began to question whether the human brain was designed well enough to extract small amounts of information from quite noisy stimuli. To find out, he has been conducting experiments with human subjects by presenting them with subthreshold images with added noise.

According to Simonotto, for a fixed, subthreshold image, the addition of fast noise dramatically enhances the quality of the image as reported by the subjects, until an optimal level is reached. Thereafter, the addition of more noise only degrades the image quality. "The experiment indicates that dynamical noise _ noise that is constantly fluctuating with time, as all noise in the real world does _ is more effective than static noise, and the faster it fluctuates, the better," he said. "This may prove useful in enhancing the accuracy and speed of visual perception of information in rapidly changing situations, such as those encountered by the pilots of high-performance aircraft."