Currently the field of brain-machine interfaces can be divided into two broad categories. There are devices which send signals to the brain, providing sensory substitution, and there are devices which extract signals from the brain to control external devices. In the latter group, progress has been made in enabling subjects with limited control analogous to native motor function. For instance, in 2003 researchers at Emory and Georgia Tech developed a primitive system which allowed a paralyzed individual to spell words by modulating brain activity. More recently, there have been successful experiments replacing limbs with robotic prosthetics controlled telekinetically (see video).