Published online 24 June 1999 | Nature | doi:10.1038/news990624-5

News

The power of thought

Implausible as it may sound, six lab rats in the United States can move a robot arm with the power of thought alone. This finding, which could lead to better, brain-operated prosthetic devices for paralysed patients, is reported by John Chapin of the Hahnemann School of Medicine, Philadelphia, Pennsylvania and colleagues, in the July issue of Nature Neuroscience.

First the rats were trained in the classical way to press a spring-loaded lever (which moves the robot arm in an arc) using their paw, to get a drop of water. This allowed the researchers to ascertain which groups of brain cells were involved in the task. Having thus tracked down their target cell-groups (to parts of the brain called the motor cortex and the thalamus) the researchers implanted arrays of electrodes into these groups to study the roles of individual neurons within them. With these electrodes in place, the researchers had a detailed window on the neuronal activity that gives rise to the bending, pushing and stretching movements that constitute a lever press.

Through a statistical analysis, over many hundreds of trials, of the firing patterns that accompany such an apparently simple movement, the researchers located the neurons responsible for every stage of the action: preparation, flexing the forelimb, extending it and so on. Finally, the team harnessed these neurons for their own ends by wiring them up so that their firing directly controlled the robotic arm without the aid of the lever.

With this new set-up, the animals very quickly learned that there was no need to physically press the lever in order to get water. Within only a few tries they reconfigured their brain activity so that it alone moved the water-bearing mechanical arm.

Although this is not the first time that brain activity has been used to drive a machine, it is the first time that signals so ‘high up’ the motor pathways have been used. Previous devices have used far cruder signals - recorded from muscles in the stump of an amputated limb, or at the surface of the scalp, for example. By contrast, this individual neuron technology should, in principle, offer far greater speed and precision.

Nonetheless, as Eberhard Fetz of the University of Washington School of Medicine, Seattle, Washington points out, there will be considerable, though “ultimately surmountable”, technical obstacles to overcome before this finding will be able to bring relief to disabled humans. After all, allowing rats to control a one-dimensional hinge with bursts of cell activity is one thing; enabling humans to continuously control a multi-jointed prosthetic limb with three degrees of freedom under variable load conditions is quite another.