Science: On a wing and an echo

For centuries they have been associated with the darker side of human imagination, from Count Dracula's alter ego to an essential ingredient in witches' diabolical potions. For scientists, however, bats represent one of nature's most astonishing creations, flying mammals whose remarkable sense of hearing enables them not only to navigate freely through dense woodland in pitch blackness, but to locate and intercept tiny airborne insects with a speed and accuracy that make modern laser-guided missiles seem primitive.

This pinpoint navigation is achieved by echolocation - similar to a highly sophisticated radar. And while many questions are still to be answered about the phenomenon, advances in information technology are giving researchers an unprecedented opportunity to gain fresh insights into this mysterious, ultrasonic world.

One of the country's experts on bat echolocation is Dr Dean Waters of the University of Leeds. "Echolocation works like a sonar," he says. "The bat uses its larynx to generate an ultrasonic sound wave, which it emits through either its mouth or its nose. The sound wave travels through the air, bounces off an object, returns to the bat's ear and is processed by its brain."

Typically, bat calls are at a frequency of between 20 and 100 kilohertz, compared with 1kHz or so for the human voice. "The bat's hearing mechanisms are similar to ours, except that it hears frequencies that are more than 10 times higher, and it processes the echo information 10 times faster than we could," says Dr Waters.

The time it takes for the signal to return enables the bat to calculate the distance of the object, while the "shape" of the reflected sound wave provides information about the form of the object. For a typical bat, one echolocation call may last a thousandth of a second and be repeated 10 times each second.

Dr Waters is particularly interested in bats that emit "frequency modulated", or FM, calls. Some bats emit calls of constant frequency, whereas in FM calls there are a range of frequencies within the same call. The first portion of the call is usually at a high frequency - such as 100kHz - and goes down to a lower frequency, maybe 20 or 40kHz.

Dr Waters wants to understand precisely how FM echolocators use the information that returns from a single call. "For many years the prevailing school of thought has been that a bat uses its ears just as we use our eyes. By processing the echoes it builds up a kind of acoustic map of its environment, just as we get a visual map of our surroundings.

"I don't necessarily think this is correct. Such computations would be complicated and time-consuming, and in a lot of cases would not be necessary."

Dr Waters suspects that the different frequencies contained in a single FM echolocation call provide the bat with different types of information: the lower-frequency portion of the call tells the bat about the shape of the object, while the higher frequency gives much more precise information about its position.

When foraging, a bat may be echolocating at distances of tens of metres. Higher-frequency sound waves become degraded at these distances, while lower frequencies tend to remain intact. "It could be that the bat identifies potential prey with the lower-frequency part of the call. It then flies towards it, at which point the higher-frequency portion of the echo becomes important in precisely locating the prey. I think this would be a more efficient strategy, and it also fits in better with the way the echo is degraded by the environment."

To test this hypothesis, Dr Waters is creating a unique, elaborate experiment using some very sophisticated electronics, along with computer technology that enable a bat's returning echolocation calls to be manipulated by researchers.

The set-up consists of a Y-shaped wooden platform. A pipistrelle bat (donated by a local bat sanctuary that rescues orphaned and injured animals) is placed at the base of the Y, with a loudspeaker at the end of each prong of the fork.

As the bat echolocates, the outgoing signal is captured by a microphone and returned through one of the speakers more quickly than the natural echo. In this way the bat thinks that one of the speakers is closer than the other.

Dr Waters is training the bat to move towards the "phantom" speaker by rewarding it with morsels of food. Once the bat is trained to do this, the experiment proper can begin. "We can then manipulate the returning signal electronically," says Dr Waters. "We can effectively snip it up into different bits.

"If my hypothesis is correct, when I filter out the higher-frequency sounds and return only the lower frequencies, the bat's ability to discriminate range will decrease. It won't perceive the speaker as being closer. On the other hand, if I filter out the lower frequencies its range-finding ability should not be affected by it."

The success of the experiment hinges on the bat's ability to be trained. Fingers are crossed.

Meanwhile, 200 miles away, researchers in Dr Gareth Jones's laboratory at the University of Bristol are developing new ways of capturing and analysing echolocation calls. The aim is to use the information to improve our understanding of bat ecology. The team is building a neural network - a computer program that can be "trained" to recognise patterns - to try to speed up the process of identifying the "acoustic fingerprints" of the various species' echolocation calls.

"We want to develop a system that can automatically record the call, digitise it, and compare it with a library of calls that have been fed into the computer," says Dr Jones.

The ultimate goal is to be able to do automatic acoustic surveying in the field as a non-invasive way of monitoring bats' feeding areas.

"All British bats are protected, but many have undergone population declines and habitat loss," says Dr Jones. "While the bats and their roosting sites have generally good legal protection, safeguarding their feeding areas has had a lower priority. If we want to conserve bats effectively, it's important that we protect the feeding areas also. Clearly, because bats are nocturnal, they can be difficult to survey. We're hoping that our system will be able to give us useful information about which species feed in which locations."

Not only is modern technology helping us find out more about bat echolocation, it may also be possible to integrate artificial echolocation into sophisticated computerised devices to create new machines. Dr Jones's team at Bristol has created a prototype robotic device that can distinguish simple shapes by echolocation. "Essentially, we generate a bat call digitally on the computer, turn it into an analogue signal and send it out through a pair of speakers. The sound wave bounces off an object and returns. The echo is re-digitised and fed into a neural network." The computer's neural network was successfully "trained" to distinguish between different-shaped objects in various orientations. Such a system could, for example, be used on conveyor belts to sort objects of different shapes.

In Leeds, Dr Waters has just been awarded a grant to try to develop a virtual reality system in which the user becomes immersed in an acoustic world rather than a visual one. This could bring the concept of virtual reality within the reach of people with impaired vision. "Because there is little fundamental difference between ourselves and bats in the way we process acoustic information, if we slowed everything down by a factor of 10 we could potentially get as much information about our surroundings as bats can," he explains.

The idea is for the user to wear a helmet and headphones, but also to be equipped with an "acoustic torch". The torch would send out a virtual sonar pulse, which would reflect off objects in the surrounding environment and return via the headphones. As with bats' echolocation, the returning pulse would convey information about the distance of an object as well as its form.

"Virtual reality is being seen as an important tool for education and training, as well as entertainment," says Dr Waters. "But because it is designed for people with normal vision, those with visual disabilities are effectively excluded. Hopefully, this project will go some way to redressing things."