Squared tasks of attention control for jsPsych

For questions about this specific implementation, please email Van at vrtliceralde at gmail.

For questions about the task in general, please refer to the original Burgoyne et al. (2023) paper or reach out to Alex Burgoyne directly.

When using this task, please cite Burgoyne et al. (2023) and this version of the task as follows:

Liceralde, V. R. T. & Burgoyne, A. P. (2023). Squared tasks of attention control for jsPsych (Version 1.0.0) [Computer software]. https://doi.org/10.5281/zenodo.8313315

@misc{liceralde23squared,
  author = {Van Rynald T. Liceralde and Alexander P. Burgoyne},
  year = {2023},
  title = {Squared tasks of attention control for {jsPsych}},
  howpublished = {\url{https://doi.org/10.5281/zenodo.8313315}}
}

Description

This is a jsPsych implementation of the Squared tasks of executive attention that were developed by Burgoyne et al. (2023) (Engle Lab Website; OSF).

There are two directories: squared and jspsych. The squared directory is the main directory, and it houses the html executable file to be used to run the Squared tasks (i.e., squared.html). The jspsych directory (v7.3.1) contains all the plugins that are locally referenced in the squared/squared.html file; these plugins are necessary for the program to work. If the user wants the reference to the plugins to be online instead of local, the user can modify the html file to reference the CDN-hosted scripts online (see documentation on jsPsych website). The advantage of referencing the plugins online is that the user will only need to keep the squared directory to run the program successfully; the disadvantage is that the program may not run if the servers are down. To be safer, local reference to the jsPsych plugins is implemented here.

Content

The contents of the jspsych directory can be directly obtained here.

The contents of squared directory are as follows:

File	Description
squared.html	The main executable file. Could be opened using any web browser.
squared.js	The main jsPsych script. Contains the implementation of the tasks, and this file is called in squared.html
squared.css	CSS script containing formatting options to keep interface as similar to the original Burgoyne et al. programs
*.png	Image files for Flanker and Simon tasks that are called in squared.js

How to run

The user has the option to run the program locally or through an online experiment hosting site (e.g., cognition.run, gorilla.sc).

At start-up (after collecting participant ID), the user will be prompted on whether the program is being run online or locally. If local is selected, the participant's data will be saved and downloaded into the local console. If online is selected, the hosting site will keep the data. If the program is run online and local is selected, the program will save the data and downloaded into the remote participant's console (which is not ideal).

If the program is run locally, but online is selected, data will NOT be saved.

You can implement a default of running the task online by commenting out the get_location object in squared.js.

Local

If the user decides to collect data locally (e.g., lab computer; where the data is saved to the computer that runs the task), copying the entire squared_jspsych folder somewhere in the local computer and opening the squared.html file using any web browser should suffice. There is no installation process required other than downloading this folder/repository. Bookmarking the location of the squared.html file on your favorite web browser is recommended for easy access.

Online (cognition.run, gorilla.sc)

If the user decides to collect data using an experiment hosting site, the user will need to:

1. Copy and paste the entire contents of the squared.js file onto the main script window
2. Upload all relevant companion files for the hosting site (e.g., the squared.css, all *.png files, and relevant plugin files from the jspsych directory if needed)
3. Make sure that the hosting site calls on both the main .js script in the main window and the companion files correctly

Gorilla.sc requires further ad hoc modifications to their default settings for the uploaded components to work, and it looks like they're sundowning the "Code Editor" functionality (on which this version is implememented) into a "Legacy" object, so if you're finding difficulties making the necessary modifications for Gorilla, please email Van to send you a Gorilla clone of the tasks.

Randomizing trial order / Keeping trial order constant

Trials are randomized in the Engle Lab program (based on a version downloaded on 2023-02-01), so the default setting here is also to randomize trials across runs. However, the user may be interested in keeping the trial order constant to reduce variability in performance that's not due to individual differences in attention control (e.g., variaiblity from trial order). In this case, the user can uncomment the jsPsych.randomization.setSeed('squaredtasks'); line to set the seed for randomization and keep one random trial order constant across runs of this program.

Output

After doing each task, the program outputs task-specific .csv files (i.e., one each for Stroop, Flanker, and Simon) that contain trial-level information about the participant's responses. If run online, the output will all be combined into one .csv file.

All .csv files contain the following variables:

Variable	Type	Description
rt	numeric	response time for each trial
stimulus	string	html-formatted stimulus that is drawn on participant's screen
response	numeric	indicator for which of the two buttons/choices the participant clicked on; 0: left-hand choice, 1: right-hand choice
trial_type	string	jsPsych plugin shown on the screen
trial_index	numeric	cumulative count of each plugin change shown to the participant
time_elapsed	numeric	system-recorded timestamp at the trial
internal_node_id	string	jsPsych-specific marker of the location of the current trial in the experiment sequence
participant_id	string	unique identifier for participant, collected at start-up
task	string	one of the three squared tasks; stroop, flanker, simon
block_trial_count	numeric	cumulative trial count within the practice and main blocks; refer to this variable for what counts as a "trial" in the task; 0: the task timed out at that trial
practice	numeric	indicator for whether the trial was seen in the practice block; 0: main block, 1: practice block
item	numeric	identifier for the type of trial shown to the participant
stim	string	the string prompt shown to the participant
resp1	string	the string choice presented in the left-hand button
resp2	string	the string choice presented in the right-hand button
correct_response	numeric	indicator of whether the left-hand or right-hand button is the correct-response for the trial; 0: left-hand choice, 1: right-hand choice
condition	numeric	indicator of the type/condition of the trial; 1: FullyCongruent, 2: StimCongruentRespIncongruent, 3: StimIncongruentRespCongruent, 4: FullyIncongruent (see Squared Attention Control Task Trial Types.xls from Burgoyne et al.)
accuracy	numeric	indicator for whether the participant made the correct response for the trial; 0: incorrect, 1: correct
timeout	numeric	indicator for whether the block timed out for that trial, if so, accuracy for that trial is not counted towards final score; 0: block is ongoing, 1: time for block ended during this trial
score_after_trial	numeric	cumulative score of participant. Following Burgoyne et al., participants get a point for each correct response and lose a point for each incorrect response. The participant's final score for the task is the score_after_trial value on the very last row

The following variables are included in the output of the Squared tasks from the Engle Lab website and are thus also included in all the .csv files:

Variable	Type	Description
score_final	numeric	final score of the participant at the end of the main block
meanrt_final	numeric	mean RT of all the trials in the main block
score_x	numeric	score of the participant for condition x calculated as correct $-$ incorrect for trials in condition x, where x is: one of the following: 1: FullyCongruent; 2: StimCongruentRespIncongruent; 3: StimIncongruent; 4: FullyIncongruent
meanrt_x	numeric	mean RT of all trials for condition x (regardless of accuracy)

Task-specific variables

Stroop

Variable	Type	Description
stimcolor	string	the color that the prompt string is shown in; blue, red
resp1color	string	the color that the string in the left-hand button is shown in; blue, red
resp2color	string	the color that the string in the right-hand button is shown in; blue, red

Flanker

Variable	Type	Description
stimsign	string	the symbol/text version of the arrow set image shown as the prompt; e.g., <<<<<
resp1sign	string	the symbol/text version of the arrow set image in the left-hand button; e.g., <<<<<
resp2sign	string	the symbol/text version of the arrow set image in the right-hand button; e.g., <<<<<

Simon

Variable	Type	Description
location	string	the location on the screen where the arrow is shown; left, right

FAQs

1. The task is still accepting responses even if the time has run out. Why is that happening? Is that a problem?

   This is due to the how the endCurrentTimeline() function in jsPsych works: when the current timeline is ended, the program also waits for the current trial to end before moving on to the next timeline. So hypothetically, a new trial can appear at 29.9 sec of the 30-sec practice block (i.e., just as the block is supposed to time out), and the program would wait for a response before moving on the next part of the program, making the practice block technically last for more than 30 sec. Apparently, in jsPsych v7.3.1, that's just how it is (or we haven't gotten a solution).

   To work around this glitch, we had implemented for the timeout variable to be updated from 0 to 1 when the time limit is hit, so that even when a response is technically accepted after the time limit, the associated timeout value for that "last" trial is 1 and these trials can filtered out before any analysis. Moreover, these timeout trials have been marked with a block_trial_count of 0 to further distinguish them from valid trials. Note that the final value of the score_after_trial variable is based on the final valid trial---it is not changed at the timeout trial, so you can take the final value as the participant's actual score for the task. All of the score_x and meanrt_x variables are also only based on the valid trials.

   Long story short, this shouldn't be a problem.

2. The output has blank rows every after response. What are those blank rows for?

   The online implementations of the task record every single event that happens to the screen or internally. Those blank rows represent when feedback is shown to the participant (correct or incorrect) and thus no response is collected, leaving many of the variables blank. Those blank rows can be easily filtered out.