Purpose of this experiment is to record the participant’s eye-movements while he/she is reading a short text. For each trial a text is presented on the screen. Participant’s task is to read the text and optionally respond to a true/false statement about the text.

This example is a slightly modified version of the reading experiment template provided on the Zep wiki.

  1. Download experiment package: reading_demo.zip

  2. Install the experiment package as described in Installing a Zep experiment script

  3. Run the experiment as described in Running a Zep experiment

On successful completion of the experiment session the experiment results database will contain the results for the session as defined by the experiment. Since this is an eye-tracking experiment it should also have produced an eye-tracking data file (or multiple, depending on the script) in a subdirectory of the experiment folder.

In addition to eye-tracking data (eye samples and gaze analysis events) various other data records can be present in an eye-tracking data file. For instance, the start and end of blocks, trials and stimuli, or other experiment defined messages.

The experiment can also output the bounding-box of each word in the presented sentences. This AOI data can then be used in the post-hoc processing stage where fixations are to be assigned to the stimulus words.

For details about the experiment script please refer to the documentation on the Zep wiki.

Screencast of the experiment

This screencast shows running the experiment from ZepMan. The stimuli are just examples; they might contain translation errors. Video not yet narrated.

This experiment produces trial results in the Experiment Results Database. We can use ZepMan to extract these results from the database: reading_demo-test-1.csv
Since we ran just one session the database will only contain data for that single session. Normally you would extract the database contents after having tested all participants taking part in the study.

In addition to the regular trial results an eye-tracking data file is created: 0001_00_01.bed[1]
Converting this file to CSV format yields: 0001_00_01.csv

The following command was used to perform the conversion:
eyevec-convert 0001_00_01.bed -right=no -left=no -px -r

Post-processing the eye-tracking data

Since the words in the presented texts are relatively small the estimated gaze won’t always point to the word the participant is looking at. Deviations will mostly be visible in vertical direction. They can occur for single fixations, but more often, if the deviations are due to head movement then the fixations in a local temporal interval will have roughly the same vertical error or show an upward or downward slope.

Manual and/or automatic post-processing is generally necessary to check and correct the fixation-to-word assigments. There are several algorithms and tools available for this purpose. Here we will be using use the Fix8 tool (Al Madi et al., 2025). In addition to correcting fixations this tool can also perform a few analyses on the data (a.o. AOI first-fixation duration, AOI gaze duration).

Information about how to convert the eye-tracking data so it can be read by Fix8 will be added here.

Recommended reading

Carr, J.W., Pescuma, V.N., Furlan, M. et al. (2022). Algorithms for the automated correction of vertical drift in eye-tracking data. Behavior Research Methods, 54, 287-310. https://doi.org/10.3758/s13428-021-01554-0

Al Madi, N., Torra, B., Li, Y. et al. (2025). Combining automation and expertise: A semi-automated approach to correcting eye-tracking data in reading tasks. Behavior Research Methods, 57, 72. https://doi.org/10.3758/s13428-025-02597-3

Niehorster, D.C., Nyström, M., Hessels, R.S. et al. (2025). The fundamentals of eye tracking part 4: Tools for conducting an eye tracking study. Behavior Research Methods, 57, 46. https://doi.org/10.3758/s13428-024-02529-7

1. The filename format is dictated by the Zep script. Read as <session>_<list>_<recording>.bed. In general <list> refers to a between-subjects group, and <recording> refers to particular recording made (check the source code).