Light sensors housed in the mechanical body feed the brain sensory information. The brain tissue processes this information to generate command signals which tell the robot's motors which way to turn in response to its environment.

Steve Grand, a expert in artificial life with Cyberlife Research in Somerset, describes the work as "laudably perverse" and likely to bring the world of cyborgs one step closer.

The robot possesses only a few neurons borrowed from the sea lamprey Petromyzon marinus, a primitive eel-like vertebrate. Yet it still displays apparently "complex" behaviours in response to simple light stimuli.

Ferdinando Mussa-Ivaldi of Northwestern University in Chicago and his colleagues at the University of Illinois also in Chicago and the University of Genoa, Italy, describe it as an "artificial animal".

To create the hybrid, the team extracted a lamprey's brainstem and part of its spinal cord under total anaesthetic, and maintained it in an oxygenated and refrigerated salt solution.

The researchers then located a group of a few very large nerve cells known as M�ller cells. These cells, which are easy to access and have been extensively studied, are responsible for integrating command and sensory signals directed to the motor nerves, helping the lamprey orient itelf.

Mussa-Ivaldi and his colleagues then attached electrodes to stimulate the M�ller cells with the sorts of frequencies they would normally receive. Other electrodes monitored the activity at the axons, the output part of the neurons.

The robot itself is a commercially available module called a Khepera and couldn't look less like a lamprey. With two wheels and a body made up of a couple of circular circuit boards it looks more like a wired-up Oreo biscuit than a cyborg. The researchers didn't mount the brain tissue on the robot, but connected it by wire.

When the robot was presented with a number of light stimuli, its lamprey brain responded with a variety of behaviours, such as following the light, avoiding the light and moving in a circle.

The research was originally intended to explore how brain cells adapt to changing stimuli. But Mussa-Ivaldi hopes that learning how neurons can communicate with artificial machines will have other benefits. "We will be able to build better prosthetic limbs and devices for disabled people," he says.

Team member Vittorio Sanguineti of the University of Genoa says the work can also reveal the principles of how the brain learns and how memory works. The work will be presented at Artificial Life 7 in Portland, Oregon, in August.

Kevin Warwick, a cyberneticist at Reading University, believes that it may even one day be possible to have your brain transferred to a robot when your body dies. It would be extremely difficult, "but mapping the entire brain to a robot can't be ruled out", he says. More realistic, he says, is connecting electronic devices such as mobile phones directly into our brains.

This article appeared in the June 10 issue of New Scientist New Scientist. Copyright 2000 - All rights reserved. The material on this page is provided by New Scientist and may not be published, broadcast, rewritten or redistributed without written authorization from New Scientist.