Sharpening the Senses With Neural 'Noise'

Science, 19 September 1997, Volume 277, Number 5333

James Glanz

The science of nonlinear dynamics would seem to be far from the practical concerns of Casey Kerrigan, a physical rehabilitation specialist at Harvard Medical School in Boston. Kerrigan helps stroke patients, along with diabetics and elderly people with "peripheral neuropathy"--a deadening of sensation in the extremities--cope with their condition and relearn simple tasks.

But that effort has led her straight into a collaboration for exploring the nonlinear effect called stochastic resonance (SR). This counterintuitive effect relies on "noise"--any random, or stochastic, background fluctuation--to make a system sensitive to an otherwise undetectable signal. In the last year, a series of experiments has shown that sensory stimulation consisting of mechanical or electrical noise can sharpen everything from the sense of touch to proprioception--the ability to perceive where a limb is in space. Not only do these results offer insights into the normal workings of the nervous system, but they also open new strategies for rehabilitating patients like Kerrigan's, she says. "The sensory loop is so essential [in rehab]," says Kerrigan. "They use the feeling to relearn a motor task. This could really help."

The theory of SR, developed 15 years ago by mathematicians and physicists, describes how an optimum level of noise can boost a signal over a threshold of detection. To see this effect, says Boston University bioengineer James J. Collins, "you need a dynamical system with a threshold; you need a weak signal; and you need noise. And that's it." Consider a coin resting in one of two indentations on the dashboard of a car winding through a mountain road. Forces on the coin might not be enough, by themselves, to push it from one receptacle to another. But if the road is bumpy enough, the lateral component of this "noise" could sometimes allow the regular forces to nudge the coin across. If the road is too rough, though, the coin can move whether the car is turning or not, and the "signal" of the curves gets drowned out.

Similar effects have turned up in physical systems like noisy lasers, superconductors, and electronic circuits. Now add sensory neurons, whose cloud of dendrites "integrate" or add up stimuli until they reach a threshold and the neurons fire. Several years ago, Frank Moss of the University of Missouri, St. Louis, and co-workers monitored how sensory cells on a crayfish's tail fan respond to a weak pressure signal, such as that generated by the approach of a distant predator, in the presence of noise produced by, for example, random water currents. Moss found the classic SR rise and fall as he cranked up the noise, suggesting that the turbulent surroundings of these animals might sharpen their perceptions. More recent experiments have revealed the same sensitivity hump in cricket neurons responding to wind currents and in slices of rat brain subjected to electric fields.

Now Collins and co-workers have demonstrated SR in humans. Subjects rest a finger on a computer-controlled indentation that pulses up and down. Without noise, its action is imperceptible. Then the experimenters add either mechanical or electrical noise to the indentation, and the subjects are asked when they can feel the regular movements. The percentage of correct responses rises, then falls, as the noise gets stronger.

A separate set of experiments, led by Paul Cordo of the Robert S. Dow Neurological Sciences Institute in Portland, Oregon, shows that mechanically jiggling the muscle with a sort of vibrator can enhance a normal subject's ability to sense whether his or her wrist has been flexed by a small amount--the essence of proprioception. "It just makes your jaw drop," says Cordo of the dramatic influence of the tiny, 10-micrometer jiggle.

These experiments have taken "two big leaps," says Jacob Levin of the Massachusetts Institute of Technology, who did the cricket experiments: "One, this work was done in humans; two, it went all the way to the level of perception." Now, researchers are taking it another step: to the level of the neurons themselves. Separate sets of recent experiments led by Cordo and Faye Chiou Tan of the Baylor College of Medicine in Houston suggest that the nervous system may pump up its own sensitivity in this way.

Both groups found that as a muscle is exercised, its sensory neurons can become more sensitive, although they think different mechanisms are at work. Chiou Tan's group detected increasing electrical noise levels in the exercising muscle; the Cordo group traced a humped curve of sensitivity that may have resulted from noise generated in the brain itself during precision movements. Together, the findings suggest that the brain and its peripheral neurons are generating noise and making use of SR. "Everyone is very intrigued and perplexed at this point," says Chiou Tan.

Rehabilitation specialists are already hoping to exploit these effects by developing gloves and socks outfitted with perhaps thousands of piezoelectric noise inducers. If they worked, such devices could help patients maintain an accurate posture and keep them aware of their numb limbs, reducing injuries and infections. Kerrigan, Collins, and Harvard's Lewis Lipsitz plan to take a first step toward testing this promise in a couple of months, when they will repeat Collins's earlier SR experiments in patients. Says Chiou Tan, who also does rehabilitation: "[SR] doesn't have clinical applications yet, but it's getting close."


Volume 277, Number 5333 Issue of 19 September 1997, p 1759
©1998 by The American Association for the Advancement of Science.

Related Item

Mastering the Nonlinear Brain
Science 19 September 1997; 277 (5333):1758 (in Research News)
J. Glanz