What Eye-Tracking Can and Can’t Tell You About Attention

Eye-tracking solutions are delivered today in several forms. Dedicated hardware-software systems are used in labs to measure eye movements with great precision. Eye-tracking glasses use wireless connectivity to enable mobile eye-tracking in shopping and other “free range” environments. And webcam-based eye-tracking allows thousands of participants to be tested inexpensively, in the comfort of their homes, anywhere in the world, with almost immediate results.

Most dedicated eye-tracking systems use infrared (IR) light to find and trace eye movements and other important measures such as pupil size and distance for the stimulus source. This is true for both in-lab systems and mobile eyewear.

Webcam-based systems are different in that they can only access the visible light captured by the webcam. This creates a number of technical challenges, not the least of which is finding the participant’s eyes in the visual field of the camera and then calculating where those eyes are looking. The result is a trade-off of accuracy for convenience and cost.

Webcam-based eye-tracking is still less precise than dedicated IR systems, both temporally and spatially, but new developments in gaze recording technology like geometric analysis of corneal reflections are shrinking the performance gap between IR-based and IR-free eye-tracking. (Wedel, 2018).

What eye-tracking measures

Eye-tracking benefits from a rich array of well-established metrics, often built directly into software and automatically calculated in near real-time. These metrics allow researchers to develop a very detailed picture of consumer attention, as well as clues about other cognitive reactions, both conscious and unconscious, to marketing, product, and brand stimuli. (For an overview, see Genco et al., 2013).

Eye movements and fixations

Where and when our eyes move from one area of a visual field to another is the core response measured by eye-tracking. Key variables are fixations, saccades, and gaze paths.

Fixations are periods when the eyes are relatively stationary because they’re taking in information. Fixations range in duration from 60 milliseconds when reading to several hundred milliseconds when examining photographs or images. Some of the most commonly used fixation-related metrics are number of fixations, number of fixations on each Area of Interest (AOI), total number of fixations, average fixation duration, total fixation duration, time to first fixation, and number of repeat fixations.

Saccades are rapid eye movements that occur between fixations. They represent periods of visual search during which specific information acquisition is not taking place. Direction and distance of saccades are indications of shifts in understanding, attention, or goals. Regressive saccades (moving back to a previously viewed area) often represent confusion or a lack of understanding. Long jumps indicate that attention has been drawn from a distance. Sudden shifts in saccade direction may represent a change in the viewer’s goals or an indication that the stimulus isn’t aligning with the viewer’s expectations.

Gaze paths represent the summation of fixations and saccades over time. Straight and rapid gaze paths indicate efficient and targeted visual navigation of the stimulus, while longer, more circuitous routes often signify confusion or a lack of direction in the viewing task.

Eye blinks

Blinking is captured by eye-tracking and can be a good indicator of attention, because we blink less when paying attention and more when attention wanes. Two metrics, blink rate (the average number of blinks per minute) and blink duration (how long the blink takes), can therefore be useful indicators of attention allocation over a period of time.

Blink synchronization (the extent to which people blink at the same moment) across an audience of viewers can be used as a moment-to-moment measure of engagement with a video program or narrative. Because each viewer’s brain is independently deciding when to “take a break” with a blink, synchronization of blinking becomes an indicator of when the video is effectively commanding the audience’s attention, and whether it’s effectively communicating its intended narrative transitions (Nakano et al., 2009).

Pupil dilation

Pupillometry is the measurement of pupil size and changes in pupil size. Most eye-tracking systems measure pupil size as a byproduct of monitoring eye movement. Unlike goal-directed visual search, pupil dilation is not under voluntary control. Our pupils dilate in response to attention and interest, emotional arousal, and cognitive load, so pupil diameter changes can be a useful measure of selected cognitive and emotional reactions to stimuli in real time.

Pupil dilation occurs in the presence of both positive and negative emotionally-arousing stimuli, so it is not a good indicator of emotional direction (valence). Also, because pupil dilation is highly responsive to changes in external brightness, it is an unreliable measure for materials that change in brightness over time, like videos. It is most useful for studying static images, like print ads, whose brightness remains constant throughout the viewing experience.

Why eye-tracking is popular

Eye-tracking provides researchers with many benefits, often without requiring deep expertise to implement. Among its advantages are:

• It is excellent for measuring visual attention, cognitive “load,” goal pursuit, and implicit preferences.
• It uses well-established and well-understood metrics, often calculated automatically.
• It does not require placing of sensors.
• It can be implemented with mobile eye-trackers, enabling mobile testing.
• It can be synchronized with other methods, like facial coding, EEG, and fMRI.
• As a webcam-based method, it is inexpensive, scalable, and delivers results in short timeframes.