Fishing for function in noise.

I n everyday life, noise is usually viewed as being detrimental to detecting signals and transmitting information. We tune out static noise from our radios, pay extra money for cellular phones with clear reception, and move away from a loud pneumatic drill when trying to have a conversation. Engineers spend considerable time and effort designing noise filters, which are common in many consumer goods and scientific instruments. As one of my colleagues remarked, " I have filters in every piece of equipment in my laboratory, including the coffee pot ". But on page 291 of this issue, Russell, Wilkens and Moss 1 challenge the idea that noise is always harmful. Instead they show that, in some cases, noise can be beneficial and that it may improve the functional behaviour of an animal. This work is based on the phenomenon of stochastic resonance, originally proposed by physicists in the early 1980s in the context of global climate modelling 2,3. Stochastic resonance is an effect by which the ability of certain nonlinear systems to detect and transmit weak signals can be enhanced by the presence of a certain level of noise. This effect was limited to physical systems until 1991, when a report in Physical Review Letters 4 described the discovery of stochastic resonance in sensory neurons. John Maddox, then editor of Nature, featured this study in a News and Views article 5 entitled " Towards the brain-computer's code? ". Maddox's article brought stochastic resonance to the attention of the broader scientific community, highlighting the possible benefits of noise in nonlinear systems — in particular, in biological systems. Since then, stochastic-resonance-type effects have been demonstrated in a range of such systems. These include crayfish mechanoreceptors 6 (which detect mechanical signals such as water movement), the cricket cercal sensory system 7 (which detects air disturbances) and human muscle spindles 8 (which detect muscle stretch). This work 4 laid the foundation for the development of noise-based sensory prosthetics 9 and mechanical ventilators 10 , both of which could have considerable clinical implications. Despite such advances, the influence of stochastic resonance has been limited in the biological community. This is partly due to scepticism associated with this counter-intuitive phenomenon, and concerns about its unfulfilled promise. Some have speculated that neurophysiological sensory systems may have evolved to take advantage of sto-chastic resonance as a means for enhancing performance and functionality 6. This is an attractive hypothesis, …