Three games were chosen for the experiment. The subjects played each with both a mouse and then again with eye tracking software.
Orientation in a FPS was tested using Jake 2, a Java port of Quake 2. Players had to work their way down a hall and through a door, taking down five enemies with a pistol and unlimited ammo. One problem with using the eyes, researchers found, was The average completion time was 40 seconds for eye trackers, but only 30 seconds when a mouse was used. Survey data suggests the subject's overwhelmingly felt the mouse was easier to use and preferred its use in the future.
Targeting moving objects gave subjects using eye tracking software the most trouble. Playing a variant of Missile Command, those using a mouse earned a score 130% better than the "eye trackers." Subjects found the mouse easier to use (92%) but were split on enjoyability and future preferences (58% each).
Communicating with an avatar, surprisingly works better with eye tracking software than a mouse. Playing Neverwinter Nights, subjects opened two chests and walked through the door. Although those using a mouse completed the level faster, those who controlled the character with eye movement enjoyed it more (83%) and two-thirds of the subject prefer eye tracking over mouse usage.
We really enjoy the attempt at innovating control schemes and enhancing emotions. One idea we enjoyed, discussed early in the research, is cascaded (MAGIC) pointing, which is a hybrid of eye movement and mouse usage -- the subject looks at a screen and the mouse follows along, so that you only have to make minute adjustments with the mouse to get the desire spot. We love thinking of its application to the RTS genre.
Read (.pdf file)