Gaze detection is a technology that determines exactly where a person is looking, typically by tracking eye movements with a camera and infrared light. It works by shining a beam of near-infrared light toward the eye, measuring how that light reflects off the surface, and using that reflection to calculate the direction and rotation of the eye. The result is a precise, real-time map of visual attention.
How the Hardware Works
At the physical level, a gaze detection system has two core components: an infrared light source and a camera. The infrared light, produced by small LEDs, is directed at the eye. Because infrared is outside the visible spectrum, you don’t notice it. The camera captures how this light bounces off the cornea and pupil, creating a bright reflection point called a “glint.” By measuring the position of the glint relative to the center of the pupil, the system calculates the angle and direction of your gaze with high precision.
Professional-grade systems achieve accuracy within about 0.8 to 1 degree of visual angle in the central field of view. To put that in perspective, one degree of visual angle is roughly the width of your thumb held at arm’s length. That level of precision is enough to pinpoint which word on a screen you’re reading or which element of an image caught your eye. Any eye tracker intended for production use undergoes safety testing by qualified professionals, since prolonged infrared exposure near the eyes requires careful engineering to stay within safe limits.
Two Approaches to Estimating Gaze
The software behind gaze detection generally takes one of two paths. Model-based approaches detect specific landmarks around the eye, like the edges of the iris and the corners of the eyelid, then fit those points into a three-dimensional model of an eyeball. The system essentially builds a virtual replica of your eye and uses geometry to figure out where it’s pointed.
Appearance-based approaches skip the 3D model entirely. Instead, they feed a raw image of the eye and face into a machine learning algorithm that has been trained on thousands of examples of people looking at known locations. The algorithm learns to associate the way eyes look in a flat image with a specific gaze direction. This approach is less physically precise but more flexible, which is why it powers most webcam-based systems that don’t have specialized hardware.
Webcam Systems vs. Lab-Grade Trackers
You don’t necessarily need expensive equipment to track gaze. Recent webcam-based systems have closed much of the gap with dedicated hardware. In controlled comparisons, a webcam-based tracker achieved accuracy of about 1.4 degrees, only about half a degree worse than a research-grade system like the EyeLink 1000. Precision values for the lab system were roughly 50% better, but that margin is smaller than many researchers expected.
Webcam systems do have real limitations. Their sampling rate typically matches standard video frame rates, around 30 frames per second, which is too slow to capture rapid eye movements like microsaccades. They also can’t reliably measure pupil dilation because standard RGB cameras lack the infrared sensors needed for that. And accuracy drops when a person moves their head, since there’s no specialist present to reposition them. That said, webcam trackers actually handled head tilting better than the EyeLink in one comparison, losing only about 2% of data to head roll versus over 12% for the lab system, which was designed for a fixed head position.
What Gaze Data Actually Measures
Raw gaze coordinates become useful through a handful of standard metrics. Time to first fixation measures how many seconds pass before your eyes land on a specific area of interest, revealing how noticeable or easy to find something is. Fixation duration captures how long your eyes rest on that area once they arrive, which reflects cognitive effort: the longer you stare, the harder your brain is working to process what’s there. Fixation count tallies how many times your eyes return to an area, indicating sustained interest. And scanpath patterns trace the overall route your eyes travel across a scene, showing the sequence of attention.
These metrics are often visualized as heatmaps, where areas that attract the most visual attention glow in warm colors and ignored regions stay cool. The combination of timing, duration, and spatial data gives researchers and designers a window into attention that self-reports can’t match, because people are often unaware of where they actually looked.
Applications in Design and Marketing
Gaze detection has become a standard tool in user experience research. Designers use it to test whether people notice a critical button on a website, how quickly they find a navigation menu, or whether an advertisement draws attention away from the content a user came for. Time to first fixation on a call-to-action button, for example, tells a designer whether the layout is guiding attention effectively or forcing users to hunt.
In marketing, gaze data reveals what packaging elements shoppers actually look at on a shelf, which parts of a video ad hold attention, and where readers’ eyes go on a print layout. Because these measurements capture involuntary behavior, they’re considered more reliable than asking people what they noticed after the fact.
Diagnosing Neurodevelopmental Conditions
Gaze patterns are also emerging as diagnostic markers for conditions like autism. Children on the autism spectrum consistently show reduced fixation on socially important stimuli, such as faces, eyes, and gestures. Their total fixation duration on these social cues is shorter, and a metric called Distance-to-Reference shows they tend to fixate farther away from central facial regions compared to neurotypical children. Their eye movements also display more variable spatial exploration and atypical patterns of rapid eye jumps between points of interest.
These differences are measurable and consistent enough across studies that researchers are working toward using gaze detection as an objective screening tool. The appeal is that it requires no verbal response from the child, making it useful for very young children or those who can’t yet communicate reliably through language.
Communication for People With Paralysis
For people who have lost the ability to move and speak, gaze detection can restore communication entirely. Eye tracking communication devices are used in the later stages of conditions like ALS, where a person may have full paralysis and no ability to produce speech. The system translates deliberate eye movements into text or commands: a user looks at letters on a screen to spell words, selects pre-built phrases, or controls a computer interface using only their gaze.
A study of 35 people with late-stage ALS found that these devices produced a significant increase in communicative ability and quality of life compared to both having no device and using simpler tools like letter boards. For someone who is fully conscious but locked inside a body that no longer responds, gaze-based communication can be the only channel to the outside world.
Privacy Risks and Legal Gaps
Gaze data carries a unique privacy concern: it captures behavior that is largely involuntary and unconscious. You can choose not to speak or not to click, but you can’t easily control where your eyes flicker or when your pupils dilate. These signals reveal interest, arousal, confusion, and preference in ways that are nearly impossible to suppress. If a company is collecting this data, you may never know it’s happening.
The legal landscape hasn’t caught up. In the United States, no laws explicitly govern the collection, use, or resale of eye tracking data. State biometric privacy laws like Illinois’ BIPA define protected data narrowly and typically only cover biometric information used for authentication, like unlocking a device with a fingerprint. Gaze data collected to analyze preferences, interests, or emotional responses falls outside that definition. Even the EU’s GDPR, which is broader in scope, has no established interpretation covering the kind of biometric profiling that virtual reality headsets with built-in eye trackers can perform.
This gap creates real risks. Because gaze data reveals unconscious reactions, it could be used for targeted manipulation, nudging purchasing decisions or shaping content based on what your eyes reveal you’re drawn to, without your awareness or consent. Legal scholars have flagged this as a distinct category of harm to personal autonomy, one that existing privacy frameworks were not designed to address.