1. How Eye Trackers Use Invisible Light to Find Where You Are Looking
Most eye-tracking devices use a type of light called near-infrared, which is invisible to the human eye, along with cameras[10]. A light source placed in front of the screen shines this light into your eyes. The camera records two things: the center of your pupil and a small reflection point on the surface of your cornea, called a glint or Purkinje image[10].
The key is that the camera does not just look at the direction of your black eye. When your eye rotates, the position of the pupil moves, but its relationship to the reflection point changes. By calculating the angle and distance between these two points, the computer figures out which spot on the screen you are looking at[10].
Think of it like shining a light on a marble: the smooth surface reflects a bright dot. The cornea is also smooth and reflects light in a similar way, though real eyes are much more complex than marbles. Near-infrared light is used for two main reasons: first, because it is invisible, it does not distract the user; second, it is less affected by changes in room brightness than normal visible light[10].
2. Why Eye-Tracking Devices Need Calibration Before You Type
Before you can type with your eyes, most devices require a step called calibration. During this process, you look at dots that appear one by one on the screen[10]. The system records where your pupil and reflection point are when you look at known positions on the screen. This data helps link your eye angle to specific screen locations.
Calibration is necessary because everyone’s eyes are shaped differently. Even if two people both "look right," the shift between their pupil and reflection point will be slightly different for each person[10]. However, calibration is not a one-time fix. If the user changes, or if their body moves during use, errors can occur, requiring the process to be repeated[10]. Manufacturers also advise using the same state as when you registered (for example, wearing glasses versus contact lenses) to ensure accuracy[10].
Eye tracking is a tool that must be carefully adjusted for each individual. Once the system knows where you are looking, the next challenge is handling the act of "looking" itself.
3. The Midas Touch Problem: Why Every Glance Could Trigger a Command
If your eyes acted like a mouse cursor, every glance would click a button. This creates a problem known as the "Midas touch," named after the myth of King Midas, whose every touch turned to gold[10]. In 1990, researcher Robert J. K. Jacob described this issue in a paper, noting that people look at things constantly without intending to send commands[10].
A common solution is the "dwell" method. This system only registers a selection if you stare at an object for a set amount of time[10]. Research suggests that when people look steadily at an object, their eyes remain relatively still for about 200 to 600 milliseconds (0.2 to 0.6 seconds)[10]. The system waits longer than a normal glance before reacting.
However, there is a trade-off between speed and accuracy. If the wait time is too long, communication feels slow and unnatural; if it is too short, accidental selections increase[10]. There is no single perfect number for everyone. The hardest part of designing eye-tracking systems is separating "looking" from "choosing."
4. How People with ALS Use Eye Movement to Communicate
This technology is often used by people with Amyotrophic Lateral Sclerosis (ALS). In advanced stages of this disease, when breathing muscles and limbs are paralyzed, eye movement often becomes a way to communicate[10]. Eye movement often remains usable late in the disease.
Not everyone can use their eyes in the same way. As symptoms progress, some people have a limited range of eye movement. To help them, researchers have developed letter boards that move toward the user’s gaze, allowing them to select all letters with less eye movement[10]. These devices are customized so users can operate them within their specific range of vision.
In Japan, there is a system to subsidize assistive devices. Communication aids for people with severe disabilities have been covered by this program since 2006[10]. Initially, few practical eye-tracking products existed, but as more commercial options became available, applications for support increased[10].
5. Typing, Drawing, and Controlling Robots with Eye Tracking
The OriHime eye is a communication tool that lets users select letters from a keyboard using eye tracking or switches, then reads the text aloud[10]. With practice, users can type about 30 to 40 characters per minute. Since one minute has 60 seconds, this works out to roughly 1.5 to 2 seconds per character, including the time spent staring to select and moving eyes to the next letter[10].
Developed through trials with ALS patients, this device is also used to control OriHime, a remote-presence robot[10]. It allows people who cannot move their bodies to look around through the robot’s camera and work from home while using artificial respirators. Note that these examples were reported in 2018, so prices and availability may have changed since then.
Hiroyuki Sakaki, an artist with ALS who cannot hold a pen or speak but can move his eyes, continues to draw using this technology[10]. His work has inspired children at special support schools to start drawing with eye tracking too. This tool does more than convey feelings; it sparks new challenges for others.
6. Test the Dwell Time Trade-Off Between Speed and Errors with a Friend
You can test the trade-off between speed and errors using paper and a timer. Draw four large circles on a sheet labeled "a, i, u, e" (or any letters). Have a friend secretly write down which circle they want to choose on a separate piece of paper. Then, ask them to stare at that circle while you watch their eyes. When the agreed time passes, say, "You chose this one." Finally, check their secret note.
Try two rules: first, "select after staring for 3 seconds," and second, "select after staring for 1 second." While staring, ask your friend to briefly glance at another circle. With the 1-second rule, accidental selections will likely increase. With the 3-second rule, you may get it right, but waiting feels frustrating.
This tug-of-war between speed and accuracy is called a trade-off in research papers[10]. After experiencing it yourself, you can look up the original studies to learn more. Check current guidance from reliable sources.