A baby aspirin-size brain sensor containing 100 electrodes, each thinner than a human hair, that connects to the surface of the motor cortex (the part of the brain that enables voluntary movement), registers electrical signals from nearby neurons, and transmits them through gold wires to a set of computers, processors and monitors. The goal is for patients with brain stem stroke, ALS, and spinal cord injuries to eventually be able to control prosthetic limbs directly form their brains.An earlier version of the BrainGate system helped a young tetraplegic named Matt Nagle control a mouse cursor and operate a very basic prosthetic hand. A 25-year-old locked-in patient named Erik Ramsey, who is participating in the only other FDA-approved clinical trial of a brain-computer interface. Ever since a car accident nine years ago, the only part of Erik's body that has been under his control has been his eyeballs, and even those he can only move up and down. The hope is that he might someday use his neural implant to control a digital voice:
When Erik thinks about puckering his mouth into an o or stretching his lips into an e, a unique pattern of neurons fires--even though his body doesn't respond. It's like flicking switches that connect to a burned-out bulb. The electrode implant picks up the noisy firing signals of about fifty different neurons, amplifies them, and transmits them across Erik's skull to two small receivers glued to shaved spots on the crown of his head. Those receivers then feed the signal into a computer, which uses a sophisticated algorithm to compare the pattern of neural firings to a library of patterns Kennedy recorded earlier. It takes about fifty milliseconds for the computer to figure out what Erik is trying to say and translate those thoughts into sound.Like the BrainGate sensor, Erik's neural implant was inserted into the motor cortex (in his case, the specific region that controls the mouth, lips, and jaw). But Erik's implant only has a single electrode, whereas the BrainGate has 100, which means it should, theoretically, be able to differentiate signals from a far greater number of neurons.