How to ensure different distractor images are shown an equal number of times?

[ellamoeck]

Hi everyone,
I am very new to jspsych and am trying to quickly adapt an existing script for an honours student to collect data online (due to COVID). It's an image-based RSVP program with fillers, distractors, and targets. The fillers, distractors, and targets are randomly sampled from image arrays. This random sampling is great for the fillers and targets (where it doesn't matter if participants see the same image multiple times), but I need to ensure that each distractor image is shown an equal number of times. I think I need to change something from the following code which appears at the end of the script, in which the shuffleArray is what's used for the filler and targets, and the sampleArray for the distractors.
Any help would be greatly appreciated!
Ella

function shuffleArray(myArray) {
    var randNum, tempStore, j;
    for (j = myArray.length; j; j--) {
        randNum = Math.floor(Math.random() * j);
        tempStore = myArray[j - 1];
        myArray[j - 1] = myArray[randNum];
        myArray[randNum] = tempStore;
    }
};

function sampleArray(myArray) {
    return myArray[Math.floor(Math.random() * myArray.length)];
};

[vijaymarupudi]

How are these functions used in the current script?

[ellamoeck]

They are used in the rsvp trial sequence, which I've pasted below. The rewFilenames and neutFilenames are the distractors and StandardFilenames are the fillers. I can send through the full script, if that's helpful?

shuffleArray(standardFilenames);

stimArray = [];
for (ii = 0; ii < itemsPerStream + 2; ii++) {
  // Extra 2 elements in array hold distractor and target stimuli; these will be moved into position later
  stimArray[ii] = standardFilenames[ii];
}

trialDistractType = distractType[trialSelector[trialNumInBlock]];

if (trialDistractType == 1) {
  distractID = sampleArray(rewFilenames);
} else if (trialDistractType == 2) {
  distractID = sampleArray(neutFilenames);
} else if (trialDistractType == 3) {
  distractID = standardFilenames[itemsPerStream];
}

stimArray[itemsPerStream] = distractID;

trialTargetType = Math.floor(Math.random() * 2);
if (trialTargetType == 0) {
  targetID = sampleArray(target0Filenames);
} else {
  targetID = sampleArray(target1Filenames);
}

stimArray[itemsPerStream + 1] = targetID;

return stimArray;

[vijaymarupudi]

Full script will be helpful for context. Could you explain your task in more detail too? I'm having trouble understanding it (I lack the background knowledge!)

[ellamoeck]

For sure. RSVP stands for Rapid Serial Visual Presentation. It's where a stream of images are briefly presented (in this case, it's a stream of 16 images presented for 100ms each). Most of the images in the stream are filler images (pictures of nature/buildings in this experiment), and one of them is the target (a nature/building image rotated 90 degrees to the left or the right). The task is to report which way the target image is rotated. But on 80% of trials, there's also a distractor image which is from one of four categories (though because of what this script was previously used for there's just two categories, "reward" or "neutral", in the attached). Ideally, I would have the four categories of distractors listed in the code, but when I've been trying to do that the code no longer runs. However, by getting each image to show an equal number of times, I thought I could get around the number of categories issue.

Note: There are a few things in the code which I know I don't need (like counterbalancing and some of the instructions), but I kept finding that when I deleted them the code would no longer run. My primary interest is in useable, rather than beautiful, code given the time pressure on getting this study up and running.

Many thanks for your help.

RSVPTask.html.zip

[vijaymarupudi]

So the participant sees a stream of images. When they see a filler, they don't respond. When they see a distractor, they don't respond, but when they see a target, they respond with either left or right, correct?

The actual position of the filler items doesn't not matter, nor do the targets, and both filler items and targets can be repeated. However, you have a specific set of distractors that you'd like to display in a controlled manner from 4 categories. They should not repeat (no distractor after distractor) and should be inserted into the image stream with even numbers of them from 4 categories.

Am I understanding this right?

The sample array function just picks a random image from the distractors list and ignores the category of the distractor. Is that the problem?

[vijaymarupudi]

You'll have to make a 'stack' using the function on the top. You want to provide that function the list of items you want to exclusively add into your stimulus list (i.e. the distractor image array probably). Then everytime you need a unique item from the 'stack', you call it like a function and it will returns the item (and removes it from the stack). You probably want to shuffle the array before providing it to the 'stack'.

Looking at this code, you might want to make two stacks, one for the rewFilenames and one for neutFilenames. And where you're currently calling the stack creator function, you should call the stack.

If you want the 'stack' to be unique experiment wide, make the 'stack' outside of the stimulus function. If you want it unique within a stimulus, them make it inside.

You should call makeDistractorStack with the list of images you want to make a unique stack of, like rewFilenames.

[ellamoeck]

That's fixed the issue! Thank you so much. One more quick question - it worked perfectly when I had each 40 distractor trials (so each distractor image showing once). But I've just increased to 80 trials and on the second round through, the distractor images are no longer showing. Is there a way of looping it so that it will get back and start calling the stack again?

[vijaymarupudi]

Make a stack that's twice as big (so initiate it with an array with 2x the times using Array.concat()). Or make 2 stacks.

[ellamoeck]

Great! Many thanks for your time and patience in providing additional explanations, I really appreciate it.

[vijaymarupudi]

You're welcome!