125Shut your eyes. Now, touch your nose.Chances are you can do this withouteven thinking about it. For this you canthank your sense of proprioception,which is so much a part of us that most of usare unaware that it exists. This ‘sixth sense’ letsour brain know the relative positions in spaceof different parts of our bodies. Without it, ourbrains are lost.Ian Waterman knows how that loss feels.More than 30 years ago he lost this sensealmost overnight, when a flu-like virus dam-aged the required sensory nerves. His musclesworked perfectly, but he could not controlthem. “I lost ownership of my body,” he says.He could no longer stand, or even sit up byhimself, and doctors said he would never beable to do so again. Waterman’s condition arosefrom a disease called acute sensory neuropathyand is so rare that only a dozen or so similarcases are known to the medical literature.Some neuroscientists are taking a cue fromWaterman’s experiences and starting to inves-tigate whether robotic devices controlled bythought alone could be integrated with an arti-ficial sense of proprioception. If so, they rea-son, these ‘neuroprosthetics’ could be made to work in a much more life-like way. What’smore, they hope to gain a deeper understand-ing of how proprioception works, and howthey might be able to manipulate it.Some months after his virus attacked,Waterman, only 19 years old, was lying in bedapplying all his mental energy to the fight forcontrol of his body. He tensed his stomachmuscles, lifted his head and stared down at thelimbs that seemed no longer to belong to him.He willed himself to sit up.Concentrated effortLater, he realized that it was the visual feed-back that allowed his body to unexpectedlyobey the mental instruction. “But the eupho-ria of the moment made me lose concentrationand I nearly fell out of bed,” he remembers. From then on he learnt to compensate forhis deficit in proprioception with other formsof sensory feedback to help him understandwhere his limbs are, and thus control them. Itrequires constant, intense concentration, butnow, despite his profound impairments, he canmanage fairly normal movements. Most of theinput that he relies on is visual — standing upwith his eyes closed is still nearly impossible —but he can also tune in to the tug of a jacketsleeve to work out the direction his arm ismoving. Or to the cool air on his armpit whenhe raises his arm in a loose shirt. Neuropros-thetic engineers are realizing that many sensory feedback signals could be similarlyharnessed.A neuroprosthetic is more accurately calleda brain–machine interface. Hundreds of elec-trodes, fixed into tiny arrays, are placed in oron the surface of the cortex, the thin, foldedouter surface of the brain that controls com-plex functions including the organization ofmovement. The electrodes record the electri-cal signals from the cortex’s neurons and theseare translated by a computer algorithm andused to drive specific actions — the movementof a cursor on a computer screen, for example,or of an artificial limb.In this issue of Nature, two papers1–3demon-strate dramatic progress in the area. A teamconsisting of John Donoghue’s group, based atBrown University in Rhode Island and Cyber-netics Neurotechnology Systems in Foxbor-ough, Massachusetts, implanted 96 electrodesinto Matt Nagle’s motor cortex, the brainregion that processes information about move-IN SEARCH OF THE SIXTH SENSE Implants in the brain could one day help paralysed people move robotic arms and legs. But first,scientists need to work out how our brains know where our limbs are, says Alison Abbott.NATURE|Vol 442|13 July 2006 NEWS FEATURER. MASSEY/GETTY13.7 News Feat Mind Machine MH 11/7/06 9:45 AM Page 125Nature PublishingGroup ©2006NEWS FEATURE NATURE|Vol 442|13 July 2006126ment. Nagle is a quadriplegic patient and thefirst human volunteer to reach this advancedstage of testing (see picture, above). Hooked upto computers and attended by a team of techni-cians, Nagle could move a cursor to issue dif-ferent instructions — for example, to opene-mails or turn down the television. Krishna Shenoy’s group at Stanford Univer-sity, California, has done similar work in anon-paralysed monkey’s premotor cortex, thearea of brain where the animal’s movement-related ‘intentions’ are generated. Using a newalgorithm, the team’s brain–computer inter-face produced results four times faster andmore accurate than previously seen.Closing the loopThe two papers show how closely neuropros-thetics are approaching medical reality. Butalthough moving a computer cursor bythought alone may be dazzling, scientists havelong-term ambitions to make neuroprosthet-ics reproduce more complex functions. Couldpatients direct a robotic arm to pick up a cof-fee cup, for example? “For this, the devicesneed to deliver feedback to the brain — weneed to close the loop,” says Daofen Chen,director of the neural prosthesis programme atthe US National Institute of Neurological Dis-orders and Stroke in Bethesda, Maryland.The brain’s sensory cortex receives signals —proprioception, touch, pain and so on — fromthe body (see graphic), and in response con-stantly modifies its movement-related com-mands. The current generation of output-onlyneuroprosthetics are open-loop systems —with more limitations than Ian Waterman, whocan at least use visual, temperature and tactilefeedback. “Brain–machine interfaces will haveto become interactive,” says Chen. “But nowthat we would like to exploit it, we realize weknow next to nothing about sensory input.” A handful of researchers is starting to try towork out where and how to stimulate the sen-sory nervous system to reproduce the sorts ofinformation that a limb might send to the sen-sory cortex. It is early days: none of their workis published. And as so little is known aboutthe system, there is no obvious place to start.Theoretically, the ‘where’ could be thenerves running from the limb into the spinalcord, or the spinal cord itself (see graphic). Orit could be higher — in the brain’s thalamus,where incoming sensory signals are integratedand redirected to the appropriate part of thecortex, or the sensory cortex itself.The ‘how’ refers to the design of the electri-cal signals to be fed into the cortex. Thesecould mimic the sensory system’s naturalnerve impulses, based on parameters such asfrequency and amplitude. Or they couldinvolve creating artificial