jsPsychMonkeys

Release monkeys to a jsPsych experiment using the R package {targets}, docker and {RSelenium}.

The goal of this package is to simplify testing a jsPysch experiment, generating synthetic responses. With its sister package jsPsychHelpeR you can process the synthetic responses to help catch issues before any actual humans complete the experiment.

jsPsychMonkeys can run local or online experiments. You can launch as many participants as you want (laws of Physics and time constraints for virtual Monkey's labor still apply), including a parallel horde (your number of CPU cores are the limit). You can also ask the monkeys to take a screenshot of every screen they see, etc.

How to use

First, install the jsPsychMonkeys package:

if (!require('remotes')) utils::install.packages('remotes'); remotes::install_github('gorkang/jsPsychMonkeys')

---

You are ready to release the monkeys!

jsPsychMonkeys::release_the_monkeys(uid = 1:10, 
                                    local_folder_tasks = "~/Downloads/new_protocol/")

---

If you want to create and configure an RStudio project to release Monkeys:

Use the folder parameter to select a specific folder for the RStudio project
Use the uid parameter to set the participants' numeric id's, e.g. uid = 1:10 would launch monkeys 1 to 10 Use the local_folder_tasks parameter to indicate the location of the jsPsychMakeR protocol

jsPsychMonkeys::create_monkeys_project(folder = "~/Downloads/Monkeys/", 
                                       uid = 1:10, 
                                       local_folder_tasks = "~/Downloads/new_protocol/")

---

---

Release the Monkeys!

In the new RStudio project.

If you want a sequential process:

• targets::tar_make()

If you want a parallel horde of monkeys:

• targets::tar_future_make(workers = 2)

You can set as many parallel workers as you want. With targets::tar_make_future(workers = future::availableCores() - 1) you can have as many workers as your computer cores minus 1.

In run.R you can see the typical set of commands needed for a variety of situations.

---

10 Monkeys completing a protocol in parallel. You can use targets::tar_watch(seconds = 10, outdated = FALSE, targets_only = TRUE) to see the live progress:

---

Other options

---

Launch monkeys on a server

You will need a .vault/SERVER_PATH.R file that contains the path where the protocols are located in your server: server_path = "http://URL_OF_YOUR_SERVER/PROTOCOLS_GENERAL_FOLDER/"

With the server_folder_tasks you will set the subfolder where the protocol is located. In the example below the monkeys would go to, http://URL_OF_YOUR_SERVER/PROTOCOLS_GENERAL_FOLDER/1

parameters_monkeys_minimal = list(
  uid = 1:10, # User id's for the participants. 
  server_folder_tasks = "1" # Location of your jsPsych protocol
)

---

Other parameters

There are a few parameters that can be useful:

• uid_URL = TRUE: The uid is passed in the URL (e.g. &uid=1)

• local_folder_tasks = rep("Downloads/tests/test_prototol", 25): Passing a vector of multiple tasks will make the monkeys to complete all of them.

• times_repeat_protocol: How many times a monkey should complete the same protocol (useful for longitudinal protocols or to speed up things)

• time_to_sleep_before_repeating_protocol: How many seconds to wait before reattempting to complete the protocol

• keep_alive = TRUE Keep the docker container alive after completing the tasks

• DEBUG = TRUE Activate DEBUG mode. Lot's of stuff will show up in the console.

• open_VNC = TRUE Activate DEBUG mode and open a VNC container to see the monkey's progress.

• screenshot = TRUE The monkeys will take a picture of all the pages they see. The .png files are stored in outputs/screenshots

• debug_file = TRUE Activate DEBUG mode and store all the console output in the outputs/log

• big_container = TRUE Sets the Shared memory size (/dev/shm) to 2 gigabytes. This is useful to avoid long/complex protocols to crash

• disable_web_security = TRUE If you are running a local protocol that loads external files (e.g. consent form in a html file), you may need this. Only works with chrome.

• console_logs = TRUE Store the browser's console logs. Only works with chrome.

• forced_random_wait = TRUE Will wait a randomly sampled number of seconds on page 4

• forced_seed = 11 Set a random seed so the Monkeys' behavior will be fully reproducible

• forced_refresh = 20 Refresh browser in page 20 (if TRUE is given, it will refresh in a randomly sampled page)