Man and machine

While most brain machine interfaces are restricted to planned movements to one target at a time, it is possible to simultaneously plan movements to two separate targets, reports a study published this week in Nature Neuroscience. These results suggest that brain machine interfaces, which could allow paralyzed patients to move an external robotic limb, can potentially handle the more complex movements required in the real world.

Brain machine interfaces allow direct communication between the brain and an external machine. However, most interfaces thus far are restricted to planning movements to one target at a time, which is not ideal for medical purposes, such as helping paralyzed patients, as the interface would be able to decode only one movement held in memory and risk loss of stability.

Ziv Williams and colleagues recorded brain activity in monkeys that were trained to remember a sequence of two targets, to which they would later have to move a computer cursor. The authors found that each of the upcoming movements was encoded by different population of cells in a single brain area at the same time. They then programmed a computer to decode these patterns of brain activity, producing a signal that could be used to move a cursor to the sequence of locations that the monkey was remembering. The brain machine interface that Williams and colleagues created could do the brain activity decoding quickly enough to allow the task to be done in real time, moving the cursor at around the same time as the monkeys would have.