Human-computer interfaces aim to restore some lost independence for individuals with disabilities by allowing them to act on the environment through alternative means. Here, we report performance by four able-bodied pilot subjects on a human-computer interface, which is based on a single surface electromyography signal recorded from the temporalis muscle on the head. The experimental task was 'Center Out' model, where a cursor was directed from the center of a screen to targets of varying sizes and locations around the center. The cursor rotated by default and subjects contracted the muscle to stop its rotation and move forward. The main performance metric was the Fitts' Law throughput (bits/s). We show that our single-signal device produced throughput scores that are higher than a comparable interface relying on multiple signals.