eyetracking

Eye tracking measures eye position and movements. Modern systems use video cameras and infrared illumination to locate where a person is looking, relative to a display or environment. The resulting data are used to infer visual attention, gaze dynamics, and cognitive processing.

Common technologies include video-based, remote or head-mounted trackers that estimate gaze from the pupil center and corneal reflections. In some research, scleral search coils provide very precise measurements but are invasive. Systems vary in accuracy, sampling rate, and robustness to head motion, and they require calibration to map data to gaze coordinates.

Output metrics include fixations (periods of relative gaze stability), saccades (rapid gaze shifts), gaze position, dwell time, scanpaths, and pupil dilation. Researchers often generate heat maps and gaze plots to summarize gaze distribution. Data may be synchronized with stimuli, responses, or other physiological signals.

Common experimental paradigms include free-viewing tasks, reading studies, and scene perception experiments, as well as gaze-contingent paradigms in which the display changes in response to gaze. Eye tracking is also used in usability testing, marketing research, and human–computer interaction to study attention and interaction patterns.

Applications span psychology and neuroscience, market research, usability and product design, driving and aviation safety, and clinical assessment. Eye tracking can enable hands-free interaction, such as eye-typing or gaze-based selection, and can be combined with neurophysiological measures.

Key data-quality considerations include accuracy and precision, typical sampling rates from tens to thousands of hertz, and the effects of blinks, eyelids, glasses reflections, and head movement. Calibration drift and occlusion can degrade data. Data processing often involves event detection algorithms to label fixations and saccades.

Ethical considerations include informed consent and privacy, since gaze data can reveal attentional states and interests. Data handling should minimize re-identification risks and comply with applicable regulations.