Running the Scream Task + Stop Signal Task Battery

Task Type: Multi-day Last edited time: July 2, 2025 10:23 AM

How to run tasks:

1. Clone kaley’s Conte Tasks Github https://github.com/kaleyjoss/conte-p3-tasks
2. In your terminal, cd to that folder on your local machine
3. Run these lines:

export FLASK_APP=app.py
export FLASK_ENV=development # don't include this if not testing
export FLASK_RUN_PORT=9000 #change this depending on which port you have free
flask run --host=0.0.0.0 # This is if you want it to be accessible to others on your IP address or on whatever server you are using

It should return something like:

 * Serving Flask app 'app.py'
 * Debug mode: off
WARNING: This is a development server. Do not use it in a production deployment. Use a production WSGI server instead.
 * Running on all addresses (0.0.0.0) 
 * Running on http://127.0.0.1:9000 #current machine, not portable
 * Running on http://YOUR_IP_ADDRESS:9000 #your IP address, portable to others

1. Then, for local testing, you can go to a browser and input:

http://127.0.0.1:9000/?PROLIFIC_PID=testuser #change 'testuser' to whatever ID

Timing

Stop signal task:

• Using default settings: ~6mins, ~1min per block
  • average 50s per block my timing, 1min08s maximum (no key presses)
  • 64 trials per block, 4 blocks
• Can adjust to make it 6, 8 blocks → 8min, 10min

Risk Task

• ~12mins (whole task including instructions)
  • ~5min 30s per block, 2 blocks

Links

Github: https://github.com/fredvbrug/STOP-IT/tree/master

Nivturk Documentation: https://nivlab.github.io/nivturk/

Stop signal lit review

Code Changelog (in order from most recent to least recent)

KJ060425

diff nivturk-plugins.js: https://www.diffchecker.com/IWU1CwUh/

• on_success save/redirect functionality
  • fixed by putting copy into stop.html
• risk still works, just uncomment + change main.html

KJ060225

diff stop.html: https://www.diffchecker.com/d8GzPsAn/

diff init.py: https://www.diffchecker.com/8RSYwNcb/

• got save functionality to work effectively

  • added save_data_stopTask to init.py
  • made appendData call save_data_stopTask instead of save_data_php

• now need to get the workerID from prolific to populate

  • took out acqext in timeline for testing (from main.html)
  • added if id == ‘prolific’ logic
  • changed to prolific
  • changed logic of id text parsing so that it wouldn’t overwrite prolific
  • changed initVarJs since that is js 7s

KJ053025

• added a way to link the stimulus images correctly
• found save data php doesn’t work

KJ 052925

• added way to get workerId form other tasks into stop signal task {{workerId}}

KJ 052825

diff experiment.py: https://www.diffchecker.com/Dd10PPd1/

diff main.html https://www.diffchecker.com/ijX51WWR/

• Added risk task to experiment.py @bp.route('/experiments/stop'
  • backend— added serving adaptation
• Added experiment.html from stop task to templates/experiments/stop.html
• added jspsych-6.0.5 to static/lib, changed script src in stop.html to the new path to js and jspych package (replace /js/ with ../../static/lib for packages and ../../static/js/stopfor standalone (not within version folder) js files/plugins)
  • also put configuration .js files in ../../static/js/stop

KJ 052725:

1. Conda environment with 3.9 (3.7 wasn’t available)
2. read all documentation
3. Investigated setup

• investigation into subject id

  WorkerID in NivTurk is set in experiment.py

      ## Record incoming metadata.
      info = dict(
          workerId     = request.args.get('PROLIFIC_PID'),    # Prolific metadata
          assignmentId = request.args.get('SESSION_ID'),      # Prolific metadata
          hitId        = request.args.get('STUDY_ID'),        # Prolific metadata
          subId        = request.args.get('PROLIFIC_PID'),    # This used to be gencode(24), changed so that the data has the subject PID input instead of a random 24dig code
          address      = request.remote_addr,                 # NivTurk metadata
          user_agent   = request.user_agent.string,           # User metadata
          code_success = cfg['PROLIFIC'].get('CODE_SUCCESS', gen_code(8).upper()),
          code_reject  = cfg['PROLIFIC'].get('CODE_REJECT', gen_code(8).upper()),
      )

  Which then defines the path of a log file

   ## Parse log file.
          with open(os.path.join(session['metadata'], info['workerId']), 'r') as f:
              logs = f.read()
  
          ## Extract subject ID.
          info['subId'] = re.search('subId\t(.*)\n', logs).group(1)

  where it can then extract the subId

  • from prolific?

  Stop signal task

  ID in SST is set by experiment.html, and added to the datafile jsPsych.data.addProperties({

    // if enabled below, get participant's id from participant and add it to the datafile.
  
    // the prompt is declared in the configuration/text_variables.js file
  
    var participant_id = {
  
      type: 'survey-text',
  
      questions: [{
  
        prompt: subjID_instructions,
  
        required: true
  
      }, ],
  
      button_label: label_next_button,
      on_finish: function(data) {
  
        var responses = JSON.parse(data.responses);
  
        var code = responses.Q0;
  
        jsPsych.data.addProperties({
  
          participantID: code
  
        });
  
      }
  
    };

  It’s referenced by the app.R script by parsing the datafile

    subj_idx <- data()$participantID

  which assigns to ‘server’ the function function , which calls ‘read.csv’ on input$csvs$datapath

  server <- function(input, output) {
    data<-eventReactive(input$process,({
      do.call("rbind", (lapply(input$csvs$datapath, function(i){
        read.csv(i, header=TRUE, na.strings=c("null"), colClasses=c("correct"="factor"))})))
    }))

  flask redirect—

  • maybe could also sned subject id?
  • or find another flask variable which would allow sending a subject ID

          return render_template('experiments/stop.html',  workerId=session['workerId'], assignmentId=session['assignmentId'], hitId=session['hitId'], code_success=session['code_success'], code_reject=session['code_reject'])

  • the workerId is being sent to the frontend