Pilot study on the use of audio feedback for sEMG-based brain-machine-body interface

We recently demonstrated the use of a novel sEMG based brain-machine-body interface in which subjects learned to continuously navigate a cursor to targets by modulating a surface electromyography signal (sEMG) at a single muscle site. Subjects visually observed the cursor position as it moved on the screen. Two-dimensional control was achieved by simultaneously manipulating the power in two separate frequency bands contained within the one single sEMG measurement. Here, we preliminarily investigate the efficiency of audio feedback on a cursor-to-target task, where subjects cannot observe the cursor position. Two pilot subjects had to infer the cursor position from two-dimensional audio feedback presented in forms of tones through a stereo head set. Testing took place over five separate one-hour long sessions. In the last session, audio feedback was removed half-way through testing. Our preliminary findings show that these two subjects could achieve accuracies of up to 70% despite not seeing the cursor moving on the screen. When the audio feedback was removed, performance accuracies dropped significantly and times-to-target increased.