Understanding Pro Lab's eye tracking metrics

Tobii Pro Lab Screen based projects

Learn what each Pro Lab eye tracking metric measures

In Pro Lab, the term "metric" is used to define the different measures that can be calculate from the recording data. These measures can be exported in different table/file formats that can either be used to get an overview of the data and extract summary statistics, or to organize the data for further processing in statistical software platforms such as R or SPSS.

For best practice, and unless your study is of an explorative nature, measures should be defined during the planning and design phase of the study. This will help you to evaluate and choose the right tools in the software, and consequently be able to estimate how much time and resources you will need to invest in the analysis phase. For example, if a specific measure is not part of the metrics output of your analysis software, you will most likely still be able to calculate it by exporting the underlying recording data, and process it with the help of external software (e.g. Excel, or Matlab). This will however require more time and knowledge of scripting.

Pro Lab calculates metrics using the following recording data:

  • eye movement states and events (e.g. fixation and saccade metrics)
  • participant actions (e.g. keypress, mouse clicks)
  • stimuli (media) and snapshot events (e.g. media start and stop times)
  • co-registered signal states and events (e.g. galvanic skin response metrics, sync events)

This data can be further agregated and sorted accross independent variables by selecting and exporting:

  • participant variables and participant groups
  • AOI tags and groups

Interval and event metrics

Select these types of metrics to calculate measures related to the occurrence of certain actions, non-eye movement related behaviors or stimuli states that are recorded during the test.

Events are markers that identify when instant relevant behaviors occur during the recording. Events can be generated automatically by Pro Lab (RecordingStart/Stop, ImageStimulusStart/Stop, sync events), or entered by the researcher/analyst during or after the recording (Custom events). Events have an associated timestamp, the exact time the event marker was applied. They can be counted but they do not have any duration per se since they simply mark a meaningful point in time. When two events are associated with each other as start and stop marker s, they create an interval.

Available measures for events:

  • Number: number of custom events that occur during an interval of time.
  • Time to first event: Elapsed time between an interval start event until the first. time a custom event.

Intervals are sections of the recording, defined by a start and end event, that identify important occurrences that span over a period of time. The concept of an interval is an integral part of the Times of Interest (TOI) concept. When you create a custom TOI to capture the sections of the data that are of interest for your analysis, you determine the start and end of that TOI by selecting start and end events, that have been logged during, or after the recording is made. Every time these events occur, a start and stop event, they will generate an interval that is then aggregated under a single TOI. Just like events, intervals can be generated automatically by the system or manually by the researcher/analyst by creating custom TOIs. Intervals can be counted (for example use the Count measure for the start event to determine how many times it occurred) and their duration can be measured.

Available measures for intervals: 

  • Duration: Elapsed time between the interval start and end events.
  • Start: the time when an interval start event occurs.

Fixation and saccade metrics

Select these metrics to obtain measures based on eye movements, more specifically, slow movements, fixations, fast movements or saccades. These measures are based on the pre-processed data generated by the Pro Lab Gaze filter’s different functions.
Note: if you set your Gaze filter to raw data, every valid eye tracking sample is treated as a fixation.

Fixations can be defined as the periods of time where the eyes are relatively still, holding the central foveal vision in place so that the visual system can take in detailed information about what is being looked at. In Tobii Pro Lab, fixation is a sequence of raw gaze points, where the estimated velocity is below the velocity threshold set in the I-VT gaze filter.

Since fixations are made up of multiple gaze points and they have a start and end point, each with a timestamp, we can measure at which point they occur in time and their duration. The fixation’s location on the screen/stimulus is described by a single set of spatial coordinates.

Available measures and important concepts:

  • Number: number of fixations that occur during an interval of time, and in a target Area of Interest (AOI).
  • Duration: the elapsed time between the first gaze point and the last gaze point in the sequence of gaze points that makes up the fixation.
  • Time to first fixation: the elapsed time between an interval start event until the first fixation occurs in that interval and in the target AOI.
  • First fixation: The first fixation in the target AOI.
  • Whole fixation: is when all the gaze points that constitute the fixation, are located within the TOI. i.e. the fixation starts and ends within the TOI interval.
  • Partial fixation: is when part of the gaze points that constitute the fixation, are located outside the TOI. i.e. either the fixation starts or ends outside of the TOI interval.

Note: When calculating fixation metrics, Pro Lab does not distinguish between whole and partial fixations.


Saccades are the type of eye movement used to move the fovea rapidly from one point to another. This movement starts with an initial fast acceleration until the eye reaches a peak velocity. After reaching the peak, it starts to decelerate until the eye reaches the target location. In Tobii Pro Lab, a saccade is a sequence of raw gaze points, where the associated velocity lies above the velocity threshold set in the I-VT gaze filter.

Like fixations, saccades are made up of multiple gaze points and they have a start and end point each with a timestamp, so we can measure at which point they occurred in time and their duration. The location of a saccade is described by two sets of spatial coordinates, the start location, that corresponds to the location of the centroid of the fixation that precedes the saccade, and an end location that corresponds to the centroid of the fixation that follows the saccade. In between those two locations, it is also possible to check the velocity associated with each gaze point and determine at which point the velocity reached the highest value.

Available measures and important concepts:

  • Number: the number of saccades that occur during an interval of time, and in a target Area of Interest (AOI).
  • Peak velocity: the highest velocity measured during the saccade.
  • Amplitude: Distance between the centroid of the fixation that precedes the saccade to the centroid of the fixation that follows the saccade.
  • Direction: Absolute angle between a straight line from saccade start to end and a straight line from saccade start along the horizontal axis of the screen (or active display area). These measures use the unit circle as the coordinate system, where 0 degrees is located to the right and the angle grows anti-clockwise.
  • Latency: Latency is defined as the duration in milliseconds of the period between the onset of the TOI interval and the first saccade gaze point.
  • First saccade: The first saccade in the target AOI.
  • Entry saccade: The saccade that precedes the first fixation on the target AOI.
  • Exit saccade: The saccade that follows the last fixation on the target AOI.
  • Whole saccade: is when the saccade starts and ends within the TOI intervals.
  • Partial saccade: is when the saccade, starts or ends outside of the TOI intervals.

Note: Pro Lab will only use whole saccades when calculating saccade metrics, such as the number of saccades in an interval and AOI.

Visits and glance metrics

Select these metrics when you wish to calculate measures associated to general looking behavior and attention to AOIs . Both these metrics measure the general gaze activity associated to an AOI.

Visits are defined as the portion of gaze data (including missing data) between the start of the first fixation on the AOI until the end of the last fixation on the AOI, before an exit saccade.

Available measures and important concepts:

  • Number: number of visits that occur during an interval of time, and specific to a target Area of Interest (AOI).
  • Duration: the elapsed time between the start of the first fixation on the AOI until the end of the last fixation on the AOI.


Glances (ISO 15007-1) are defined as the portion of gaze data (including missing data) between the end of the last fixation before entering the AOI (includes the entry saccade), until the end of the last fixation on the AOI, before the exit saccade.

Available measures and important concepts:

  • Number: number of Glances that occur during an interval of time, and specific to a target Area of Interest (AOI).
  • Duration: the elapsed time between the end of the last fixation before entering the AOI (includes the entry saccade), until the end of the last fixation on the AOI, before the exit saccade.

Other important concepts

AOI dependent vs. AOI independent metrics

Depending on your study design you might want to calculate metrics that are associated to specific areas of the stimuli, or associated to the stimuli as a whole within a specific period of time. For example, you may want to measure how much time a participant spends looking at advertisement elements, or the eyes of a person talking. The second type of metrics are tightly related to the concept of time of interest interval, for example the average fixation duration during the interval, regardless of what the user specifically looked at.

AOI dependent metrics can says something about the object in the area of interest, while the AOI independent metrics helps you to put this number in context of how the participant behaves in general and when seeing everything else that is visible during this stimulus presentation.

If you select an AOI-based metric, you will get one metric computed per AOI. If it is a saccade metric, then the AOI will determine if a saccade is an “entry saccade” (moving into the AOI) or an “exit saccade” (moving out of the AOI). However, if you choose a metric that is AOI-independent, then you will get a metric that is computed for the entire interval. For example, an average fixation duration that is AOI-independent will give you the average fixation duration of all fixations occurring in that TOI interval, regardless of what they looked at.

Events, intervals and saccade metrics are AOI independent metrics, while AOI visits, glances, AOI fixations and AOI saccades metrics are, as the prefix indicates, AOI dependent metrics. 

Partial and whole fixations and saccades

In this section you will learn how Pro Lab deals with fixations or saccades that start or end outside an AOI or a TOI interval.

Fixation metric calculations in Pro Lab are very inclusive, in that they include all fixations in a Time of Interest (and Area of Interest). All fixations means both whole fixations (they start and end within the TOI), but also partial fixations (they start before the TOI interval starts, so we don’t know for how long that fixation has been going on). On many types of research, this is perfectly fine. If the study participant is looking in one direction and an advertisement suddenly appears there, the participant will see the ad even though they don’t move their eyes and trigger a new fixation.

For saccade metrics, partial events provide little value. If you want to calculate the peak velocity of a saccade, and that velocity peak happens outside of the TOI but the rest of the saccade inside it, it is not obvious that the peak velocity should be included? Same problem for saccade amplitude as well. Usually, experiments are also set up in way that exclude these type of problems, for example by using a fixation cross to constrain what position the gaze at the onset of the stimulus. Also, if you are an eye movement researcher, you are interested in the full and representative event, and the saccade had a certain amplitude (e.g. 20 degrees) even if it was cut by the TOI start so it is only 8 degrees now.

Related, if there is any uncertainty with the start and end of a saccade event, for example the fixation/saccade filter producing some unknown event (fulfilling the criteria of neither eye movement type), then we cannot rule out the possibility of the samples constituting this unknown event are not really a part of the actual saccade. In this case, with a non-fixation event preceding and succeeding the saccade, the saccade is rejected from analysis.

For these reasons, saccades follow strict criteria for when they should be included in the calculations:

  • A saccade must be preceded by a fixation event.
  • A saccade must be succeeded by a fixation event.
  • A saccade must by wholly contained in the time of interest.

Furthermore, some research is interested in both the saccades and the fixations that occur in the same time of interest. For example, some research suggests there are phases of visual exploration, with an early orienting phase with high-amplitude saccades and short-duration fixations, followed by an inspection phase with low-amplitude saccades and long-duration fixations. To explore this in a meaningful way, we want fixations that are similarly strictly defined as the saccades are. Therefore, we also allow the export of “whole fixations”, which are defined as:

  • A whole fixation must be preceded by a saccade.
  • A whole fixation must be succeeded by a saccade.
  • A whole fixation must be wholly contained in the time of interest.

Finally, for fixations that related to particular AOIs, there is an additional criterion that:

  • A whole fixation must be wholly contained in the area of interest.

Related Articles