What is eye-tracking?

Beata Lewandowska

6 min read

Eye-tracking is a state of the art process which tracks users' eye movements. In other words, it's a process of measuring eyes position over time, so measuring where someone was looking (where was his/her gaze point) and for how long. A device or software that makes those measurements is called an eye-tracker.

Eye-tracking types

There are several techniques used in eye-tracking studies. Nowadays, the most popular and widely used are video-based solutions, where a webcam records eye movement, but it's always good to know the alternative methods.

Eye-attached eye-trackers

Special contact lenses with a mirror or magnetic field sensor. When the eye moves, the lens moves with it, which causes signal changes, i.e., changes in the intensity of the magnetic field. By measuring changes in this signal, we can obtain data about eye movement.

Khaldi, A., Daniel, E., Massin, L. et al. A laser emitting contact lens for eye tracking. Sci Rep 10, 14804 (2020). https://doi.org/10.1038/s41598-020-71233-1

Pros:

• extreme sensitivity and accuracy,
  
• measurements can be done in vertical, horizontal and torsion directions,
• the best choice for studying the dynamics and underlying physiology of eye movement.

Cons:

• lenses can slip during the eye movements,
  
• wearing lenses isn’t comfortable for everyone,
• a test must be done in a laboratory (in person),
• analyzing data requires experience,
• costs of such studies are high.

Electric potential measurement

This methodology is based on two pairs of electrodes placed on the skin around the eyes. Two dipoles, negative at the retina and positive at the cornea, generate an electric potential. During the eye movement from the center towards the periphery position, the retina approaches one electrode and the cornea approaches the opposing one. This causes changes in the potential fields that can be translated to eye movements.

https://www.semanticscholar.org/paper/A-Survey-on-Methods-and-Models-of-Eye-Tracking%2C-and-Gaur-Jariwala/a7646f0d2d4aa845004fe866216dcacd54655920/figure/0

Pros:

• allows detecting eye movement in the darkness, different light conditions, or even while the eyes are closed (so it can be used in sleep research),
  
• blink detection is possible,
• even miniature saccades can be detected,
• requires low computational power,
• suitable MatLab libraries are available and ready to use.

Cons:

• because of the potential drifts and variable relations between the signal amplitudes and the saccade sizes, it's hard to measure slow movements and to detect gaze direction,
  
• it may be uncomfortable to wear the equipment,
• the test has to be done in a laboratory,
• analyzing data obtained requires experience.

Optical tracking

Light (usually infrared) falls on the eye's surface, where it is reflected. Then an optical sensor (i.e. a camera) detects the reflected light and analyze it to extract eye rotation from changes in reflections. These solutions usually use the corneal reflection and the center of the pupil to create a vector that indicates eye movements over time.

http://www.eyeresponse.com/

By Z22. Derivative work of the above file created by Björn Markmann. - File:Auge iris braun brown eye human menschlich Bjoern markmann.JPG and Own work, CC BY 3.0, https://commons.wikimedia.org/w/index.php?curid=38803307

Pros:

• non-invasive,
  
• comfortable for testers (nothing to wear on eyes),
• high accuracy can be achieved,
• easy to use.

Cons:

• optical eye-trackers are expensive (even a few thousand dollars),
  
• subjects have to be tested in a laboratory,
• some experience in analyzing data is required.

The optical eye-trackers can be divided into:

• screen-based eye-trackers (a respondent sits in front of a monitor with an attached or built-in tracking device),
  
• eye-tracking glasses (wearable, tracker mounted onto frames, so it allows to move freely, but they can shift or fell off).

Tobii Pro 3 Glassess

Online webcam eye-tracking (a subcategory of optical eye-tracking) Instead of using infrared light, a commonly used webcam analyzes the participant's eyes image to detect the position of eyes and pupils (it requires calibration at the beginning where the system learns how the participant's eyes look like while watching certain parts of the screen).

Pros:

• costs - the most affordable solution of all,
  
• research can be conducted 100% remotely and fast (no need to invite participants and everyone can complete the test at the same time),
• can be connected with external online tools,
• doesn’t require much experience in analyzing data (anyone can understand it).

Cons:

• less accurate than other solutions (i.e., RealEye accuracy is ~113 px - https://www.realeye.io/whitepaper/),
  
• eye movement can be tracked only on the screen (a screen-based interface),

What is eye-tracking used for?

Eye-tracking makes it possible to track people’s subconscious behavior. It’s commonly used in cognitive psychology, neuroscience (including neuromarketing studies for shelf testing and package design), or graphic design. It allows to track where people look, for how long, how fast they notice things, and in what order.

General eye-tracking metrics

• Gaze point - a raw sample (a point of where someone was looking) of eye-tracking data. Each device has a sampling rate expressed in Hz (the higher, the better), i.e., a 60 Hz sampling rate means that each second 60 samples are collected.
  
• Fixation - a cluster of gaze points that are very close in time and space. Fixation lasts for long enough for a person to focus and process what he/she sees.
  
• Saccade - the movement of eyes between fixations (without processing what is seen).
  
• Heatmap - graphical representation of collected data (gaze points or fixations) density. Values are represented as colors - red color (hot) indicates the most visual attention.
  
• Recording - a video that shows the participant's gaze point on the stimuli over time (how the tester’s eyes were moving over the item).
  
• AOI - Area Of Interest is an analytical tool that allows calculating quantitative eye movement measures. It contains metrics like:
  
• Time To First Fixation/Gaze - how fast participants look at this area for the first time,
• number of fixations/gazes,
• time spent - how much time participants spent by looking at this area,
• ratio - how many of all the viewers looked at this area at least once.