What is eye tracking?

You want to know whether people find the important information on a package, website or sign. Eye tracking makes their viewing behaviour visible. Its value lies in connecting the recording to your question: what must people find, what task are they performing and which decision should the research support?

Eye tracking measures the direction and movement of the eyes over time, often to estimate where someone is looking. The reporting guideline by Dunn et al. (2024) addresses how to document such recordings.

Use the results, for example, to investigate whether information is looked at and the viewing sequence someone follows. Plan additional questions or tasks when your intended conclusion concerns understanding or preference.

Eye tracking is a family of methods with a long research history. An influential classic is Eye Movements and Vision by Alfred L. Yarbus (1967). It illustrates how instructions for viewing an image relate to viewing patterns.

This makes the task an important part of your brief. Asking someone to find delivery costs creates a different research situation from letting them freely view a product page. Write out both instructions before deciding which fits your question. Yarbus is a historical reference here, without implying that he invented eye tracking.

Many video-based systems use cameras, infrared light, the pupil and a reflection on the cornea to estimate gaze direction. Technical approaches vary; Hansen and Ji (2010) describe them in their review.

The output consists of successive measurements with timestamps. Analysis connects these to the research question. For your brief, the central issue is which screen, object or part of the environment the recorded gaze must be assigned to reliably.

This article uses the following practical definitions:

• Fixation: a period in which gaze is relatively stable within the chosen reference frame. The eye is not completely motionless.
• Saccade: a rapid eye movement that shifts gaze to another position.
• Area of interest (AOI): a defined analysis region, such as a price, product or sign.
• Scanpath: the sequence of gaze positions or selected eye-movement events over time.

The descriptions of fixations and saccades draw on the classic work of Yarbus. AOI and scanpath are operational definitions here. Ask the research plan to specify how software identifies events and defines regions, especially during scrolling or when objects move.

Calibration establishes the relationship between the eye signal and known gaze positions. Research by Nyström et al. (2013) shows that factors including calibration method and eye characteristics relate to data quality.

Then ask for validation: a separate check of recorded gaze against known targets. Request reporting of missing data and any recalibration too. These topics appear in the reporting guideline by Dunn et al..

Make this specific to your materials. Must the analysis distinguish two neighbouring buttons, or does it concern large shop zones? Ask what quality that requires and which parts of the recording were usable. A stated sampling frequency alone does not answer those questions.

The following table is a practical planning aid. Define the calculation beforehand; terminology and definitions can vary between analyses.

Choose a primary measure that fits the decision. If several measures are proposed, ask what each contributes. Keep missing recordings separate from valid recordings in which the area was not fixated.

A heatmap uses colour to display a chosen measure spatially, such as accumulated fixation duration. Start with the legend: what is included, over which period and for which participants? This is an operational definition of the visualisation for this article.

For comparisons, request the same scale and calculation. Then examine participant-level values and the relevant sequence. Without further explanation, a red area does not answer whether the design works better. Use the image as an entry point into the analysis and ask for conclusions supported by the predefined measure.

Choose the setup around the action you want to investigate. This table supports an initial brief; it is not a ranking of devices.

The first two rows describe recording setups. EEG is an additional measurement method that can be combined with eye tracking depending on the study design.

Visual attention can shift without a corresponding eye movement. This distinction appears in the classic research of Posner, Snyder and Davidson (1980).

Make your intended conclusion explicit. To assess whether someone understands an instruction, add a comprehension task. For preference, include a choice or rating. To investigate where someone gets stuck, combine the viewing pattern with task performance and a focused follow-up conversation.

Treat prolonged looking as a pattern whose explanation needs testing. Set out which additional outcome would fit interest, comparison or difficulty with the information. This makes the interpretation open to scrutiny.

Some eye trackers also record pupil size. Mathôt (2018) describes its relationship with light, viewing distance, arousal and mental effort, among other factors. A larger pupil is therefore not an independent preference score.

If pupil measurements form part of the proposal, ask which explanation will be tested and how conditions will be made comparable. The report should clearly separate conclusions based on gaze position from those based on pupil data.

Dimigen et al. (2011) recorded eye movements and EEG simultaneously during reading. Their methodological discussion covers synchronisation and interference from eye movements, among other issues.

First define what the combination should add to your study. Which viewing events should be linked to which EEG analysis? What comparison supports the intended conclusion? Ask for that explanation before adding another measurement method. Read EEG explained for background on that recording.

Fictional research example, without measurement results. An online retailer is considering where delivery information should appear on a product page. The decision concerns which of two layouts better supports finding that information.

A possible task is: “Find out when this product could arrive.” Mark the relevant information area in both designs beforehand. Choose time to first fixation as one gaze measure and also ask the participant to state the delivery time. Define how to handle participants without a fixation and unusable recordings.

Request both the gaze measure and the task answer in the report. If they do not provide a clear picture, explain which design decision remains unsupported. This example illustrates study planning; it is not an existing Neurofactor case.

1. The decision: which design or information do you want to improve?
2. The situation: who is looking, with which task and at what moment?
3. The comparison: which alternatives will you compare, and how will participants or presentation order be allocated?
4. The analysis: which gaze measure, regions and additional task answer your question?
5. The deliverable: request measurement quality, findings, uncertainty and recommendations within the scope of the situation studied.

Start with the information or action you want to investigate. Choose a suitable recording approach, define the analysis and test the interpretation with additional data. This gives you a basis for judging whether an eye-tracking report supports your decision.