MultiplEYE WG1: Experiment Implementation

This repository contains the code for the MeRID eye-tracking-while-reading experiment for multiple languages.

Contents

• Installation
• Run the experiment without an eye-tracker
• The result files
• Run the experiment with an eye-tracker

Installation

In order to run the MultiplEYE experiment you will need to complete the following steps

1. Please read the official MultiplEYE Data Collection Guidelines linked on this page: MultiplEYE contribute
2. Download the experiment code from this repository. Either clone via git or download the zip file (see download as zip).
3. Following the MultiplEYE Data Collection Guidelines, prepare the stimulus files which includes the creation of the images
4. Copy the stimulus files to the correct location in the experiment data folder: experiment_implementation/data/[your stimulus folder]
5. Copy the files (3 Excel files) for the participant questionnaire the to the correct location in the experiment data folder. You will have to create a folder for your language and country. The folder should be named as follows: experiment_implementation/data/participant_questionnaire_[LANGUAGE_CODE]_[COUNTRY_CODE]_[LAB_NUMBER]/
6. Optional: if you have created a translation of the interface, you can copy it to this location: experiment_implementation/ui_data/interface_language/[your language]. The name of the file should be experiment_interface_[your language].json.
7. Prepare the environment for the experiment following the guidelines in CONDA_ENVIRONMENT.md
8. Install the necessary packages for your eye-tracker. For EyeLink eye-trackers INSTALL_PYLINK.md. For Tobii eye-trackers, please see develop for Tobii further below.
9. Run the dummy experiment to check if everything is working correctly
10. Run the experiment with an eye-tracker

Download the experiment code as zip file

If you do not have git installed, you can download the code as a zip file from this website. Click on the green button Code and then click on Download ZIP. Unzip the folder in our preferred location.

Note: The repository contains a folder with toy stimuli to test the experiment. Sometimes you cannot unzip these files because the path gets too long. In this case, you can either move the folder to a different location closer to the root or unzip and ignore the files in the toy stimuli folder (i.e. skip the error message that pops up during unzipping). To use the toy stimuli, please contact [email protected] and we can provide them and you can copy them to the correct location.

Run the dummy experiment

The experiment can be run in dummy mode which means that can be run without an actual eye-tracker. If you'd like to run it, make sure you have completed the above steps, and you have your conda environment activated.

Activate the environment:

conda activate merid3.9

Then you can navigate to the root folder of your local clone of the repository (your path should now end with wg1-experiment-implementation. Please find instructions on how to navigate in a terminal in the CONDA_ENVIRONMENT.md in section 4). Run the following command to run the dummy experiment:

python experiment_implementation/start_merid_session.py

In the GUI that will pop up you can tick the box Dummy version to run the experiment in dummy mode.

The result files

The experiment will write log and data files to a newly created results folder for your language and country in the data folder (data/eye_tracking_data...). In there it will create a folder depending on the experiment type. For example, if you run the core session, it will create a folder called core_dataset. Within those folders it will simply create a new folder for each participant. The folder name is the participant ID (three-digit number) and information on the language etc. If you run the script for the core dataset, it will prevent you from running the experiment twice for the same participant. Note that if you run a test session, it will not warn you if you enter the same participant ID more than once. It will just write the files to the same folder.

The naming scheme of the log files in the log_files folder is a follows: [log_file_type]_[session_id]_[participant_id]_[date]_[timestamp].txt.

All logfiles are csv files. Note that the timestamps are relative to the start of the experiment. The experiment starts at timestamp 0. Those log files are mainly used for debugging purposes.

The naming scheme of the edf files storing the eye-tracking data is as follows: [participant_id][language_code][lab_number]s[session_id]. It is different from the naming scheme of the MultiplEYE project, where the country codes are used instead of the session_id because there are only one session per participant in MultiplEYE project, but in MeRID, there are multiple sessions per participant.

Run the experiment with an eye-tracker

In order to run the experiment with an actual eye-tracker you can tick the respective box in the GUI in the lab settings section. The box Dummy version should be un-ticked.

Depending on what eye-tracker you intend to use you need to install the software that comes with the eye-tracker. For EyeLink and tobii you can download the software online for free.

If you use a display PC and an external monitor, please start the experiment on the external monitor and close the display PC. Otherwise the resolution can be weird for some monitors.

Run and develop experiments for EyeLink

The code has been tested with EyeLink eye-trackers and mostly on a Windows presentation PC. However, MacOS should work as well, but it has not been tested as thoroughly. You will need to install pylink a package provided by SR Research if you use EyeLink eye-trackers. Note that using pip to install pylink installs a different package although the names are the same! You have to follow these step-by-step instructions of how to install pylink can be found here: INSTALL_PYLINK.md