Challenges to conduct the randomized memory trials between the primary task trials.

CSC · August 2020

Hi everyone,

I have a study which is collecting data in the lab. The flow is as the below picture. The primary task requests participants comprehend sentence in each trial. After 6 to 8 trials, participants will verify if they read the sentence in the latest 3 trials. `Memory loop` manages this memory check. We use python code to control the timing and content of memory check.

Because of the global lockdown, we are attempting to move this task to the internet. I found some challenges from the words about Inline Javascript:

The JavaScript workspace is not persistent. That is, if you define a variable in one script, it will not be accessible in another script. Exceptions are the vars object and other variables that are globally available in JavaScript

Our python code used the self-defined variables satisfied our goal. After several times of tests, I realized we cannot migrate python logic to JavaScript. I assume some alternatives would be implemented in this case but I am lack of experience on JavaScript programming. Is there any way we can run the task on line as like in the lab?

Below is the inline python code in our old script:

presetup

### codes from http://stackoverflow.com/questions/27276135/python-random-system-time-seed

import random

from datetime import datetime

random.seed(datetime.now())




### Call empty list for comprehension task

probe_list = []

### Select the index of sentences for comprehension task

a = range(-1,2)*10

random.shuffle(a)

idx_list = [sum(x) for x in zip(range(6,49,4), a)]

### Save the index in the variable

var.set('alt_task',idx_list)

Probe_Collect

### Enqueue the probes into list

probe_list.append(self.get("Probe"))

Probe_import

### Randomly select the probe sentence from list

import random

### Choose from the past 16 trials; otherwise choose from the past 8 trials

#if(self.get('count_Probe_Collect') < 8):

# idx = random.choice(range(len(probe_list)))

#else:

idx = random.choice(range(len(probe_list) - 6, len(probe_list)))




### Export the probe sentence for recognition

rectarget = probe_list[idx]

var.set('rectarget', rectarget)

### Set up correct_response

if(idx > (self.get('count_Probe_Collect') - 3) ):

var.set('correct_response', 'j')

else:

var.set('correct_response', 'f')

Daniel · August 2020

Hi @CSC

It seems quite feasible to translate your code from python to javascript. You don't do anything in Python that you wouldn't be possible in osweb javascript. The only challenging part would be the random functions (shuffle, choice, etc) which javascript doesn't offer by standard, but if you Google them, then you will easily find some ready to use examples.

The JavaScript workspace is not persistent. That is, if you define a variable in one script, it will not be accessible in another script. Exceptions are the vars object and other variables that are globally available in JavaScript

Basically this means that if define a variable in one inline_javascript (e.g. my_var=1 ) then it will not be available in another inline_javascript item. To make them available across several inline JS items, you need to store this variable in the vars object, which is retained between items (e.g. do vars.my_var=1 )

To get you started with JavaScript, you can follow this tutorial: https://osdoc.cogsci.nl/3.3/tutorials/wcst/ and read through: https://osdoc.cogsci.nl/3.3/manual/osweb/javascript/

CSC · August 2020

Thanks @Daniel

I have found the way to manage the javascript code based on the python code. However I am engaged in new issues.

First issue is we recollected the sentences appeared in the past trials in an array, then we pick one as the target in memory_loop . Each sentence is a long string like "There is a book on the table.", and we expected the beginning element of the recollected array is the latest sentence. However, we found inline_javascript accessed the character only. For example, in this loop we picked the first element (say "There is a book on the table") as the target, inline_javascript accessed the first character "T" only.

Second issue is we can not control the display the variable we assigned the value in the inline_javascript. For example, we decided in this memory loop the target sentence is "There is a book on the table." and assign it to the variable rectarget . Although the log showed the appearance of this sentence, the sketchpad never showed the sentence in rectarget .

I tested the feasibility of these methods on the web browser. Is there any wrong in the OpenSesame inline_javascript?

CSC · August 2020

Hi I found the reasons caused the recent problems although I have yet solved them.

According to W3Cschool tutorial ( https://www.w3schools.com/js/js_es5.asp ), in ECMAScript 5.1 str[#] access the character in a string rather than a string in an array. To pass this restriction, `Probe_Collect` used .concat() connect the presented sentences trial by trial. In Probe_import I convert the connected sentences to array in use of .split(). The current problem is the below code can not connect the sentences:

if ( vars.probe_list.length > 0) {
   vars.probe_list = vars.probe_list.concat(vars.probe_list,vars.Probe,"|");
}
else {
   vars.probe_list = vars.probe_list.concat(vars.Probe,"|");
}

Although rectarget is a string but we can not show that on screen.

CSC · August 2020

All right. I have wiped out all the difficulties. Here are the solutions:

*note*: because the original post is too long, I split this post into two parts.

Below clip shows the overall sequence of an experimental trial. I made two major revisions based on two limitations: (1) Conditions to run inline_javascript , "Run if" is available for only one condition. (2) OSWEB ignored loop in a sequence .

Codes in presetup :

// getRndInteger(min, max) returns a random number between min and max (both included):
function getRndInteger(min, max) {
   return Math.floor(Math.random() * (max - min + 1) ) + min;
}

// Call empty list for memory check
vars.probe_list = [];

// Call counters for trial sequence
vars.trial_seq = 0;

// Call memory slot for probe list length
let list_length;

// Call target for memory check
vars.rectarget = [];


// Generate the index of memory check
// save each randomized sequence into each variable
vars.check_0 = getRndInteger(7,9);
vars.check_1 = getRndInteger(10,12);
vars.check_2 = getRndInteger(13,15);
vars.check_3 = getRndInteger(17,19);
vars.check_4 = getRndInteger(21,23);
vars.check_5 = getRndInteger(25,27);
vars.check_6 = getRndInteger(29,31);
vars.check_7 = getRndInteger(33,35);
vars.check_8 = getRndInteger(37,39);
vars.check_9 = getRndInteger(41,43);
vars.check_10 = getRndInteger(45,47);

trial_seq records the sequence number of the current trial. This variable decides in which trial we have to run the secondary task, give a break, and finish the study.

probe_list is to save the presented probes since the first trial. The latest probe will be updated to the head of the list.

list_length counts the length of probe_list at each trial. The maximal length is 6 included the latest presented probe.

rectarget is to save the probe for presentation while the memory check initialize.

check_# refers to the trial number the memory check initialize after the verification response.

CSC · August 2020

Codes in probe_collect

//Concatent the probes into a list
if(vars.probe_list.length == 0){
// Add first probe into the list
vars.probe_list.splice(1,0,vars.Probe);
} else if (vars.probe_list.length > 0 & vars.probe_list.length < 6) {
// Add the last probe into the list till length = 6
vars.probe_list.unshift(vars.Probe);
} else {
// After length = 6, revmove the oldest probe then add the newest probe
vars.probe_list.pop();
vars.probe_list.unshift(vars.Probe);
}
// Store
vars.list_length = vars.probe_list.length;
// Incredimate
vars.trial_seq += 1;

if-else condition control the probes saved in probe_list . This script updates list_length and trial_seq for the next trial.

Codes in Probe_import

// getRndInteger(min, max) returns a random number between min and max (both included):
function getRndInteger(min, max) {
  return Math.floor(Math.random() * (max - min + 1) ) + min;
}

// Locate the probe to be checked
vars.anchor = getRndInteger(0,5);

// Use the first anchor
vars.rectarget = vars.probe_list[vars.anchor];

// Set up correct_response
if(vars.anchor < 3) {
   vars.correct_response = 'j';
}
else{
   vars.correct_response = 'f';
}

anchor saved the position number to be selected as rectarget . That probe saved in probe_list will be assigned to rectarget . anchor also decided the correct response of the current memory check.

There are some other problems I require the assistance. Because they are out of this topic, I'll open the other post to discuss them.

Daniel · August 2020

Hi @CSC,

This looks all good. It is not entirely clear to me if you have solved the problems you were facing or are you still struggling with some aspects (that you are describing in a different post?). Is there anything specifically you still need answers on in this post?

CSC · August 2020

Thanks @Daniel

I have solved all the issues in this post.

Howdy, Stranger!

Categories

Challenges to conduct the randomized memory trials between the primary task trials.

Comments

Howdy, Stranger!

Quick Links

Categories

Challenges to conduct the randomized memory trials between the primary task trials.

Comments