URL of exp

Description of the problem: I created a pretty large and maybe somewhat heavy experiment involving roughly 700 png files that are displaying according with participants’ responses. Everything seemed to work fine in testing and I launched the study. For some reason, some of my participants contacted me saying that they get a File_Loading_Error while completting the personal info field in the experiment resource download stage - until the message ‘All resource downloaded’ is typically displayed.
As instructed by me, they sent me a screen capture with the error.
Noteworthy, this is a functional experiment. For some participants, there are no problems and I even collected some data.
I suspect that this error may have something to do with the possibly unusually large experiment and their possibly unstable internet connection.
My question is if there is something I can do in order to diminish the number of times that this error is being thrown.
Is this happening to someone else?
I atach an exemplar of the error which was sent from one of my participants. I instructed her to turn of the uTorent downloader - but I have no sign from her in order to know if this worked.

Hi @AndreiCostea, thanks for the error message. I have not seen this one before, and I am not able to recreate the error if I run your task. I have posted your error on GitHub, with a link to the repo, in case @apitiot can shed some light on what is happening.

I’ve also been getting this error, despite my current online experiment being on a much smaller scale. I’ve only been drawing from around x15 .png files at the moment, just in an effort to speed up loading times whilst I’m creating and constantly testing my Javascript.

I think it has something to do with file size / image size. The .pngs being flagged by this error (for me at least) are always the ones that are kinda big. I do plan to standardise the size of all my image stim before actual testing (so making them all 500x500px in Photoshop, rather than uploading the full .png files and letting PsychoPy do the resizing), BUT…once I encounter this error, manually resizing the troublesome .pngs doesn’t fix the issue.

Another possible lead: browser cookies (EDIT: or cache, I don’t know)? That seems odd, I know, but try running your online experiment in an Incognito tab. For me, that consistently avoids the error which shows when I try to load the experiment in a ‘normal’ browser window. I have no idea why this works or what it tells us about what is causing the error - but of course, ideally we don’t want certain Ps being able to access the experiment fine whilst others cannot.

Hello @AndreiCostea and @Jamie,

We have made substantial changes to the server and back-end of Pavlovia over the week-end, which might have positively affected the resource downloading issue you brought up back in May.
Certainly I can run Andrei’s experiment without any problem. Would you mind giving it another go on your end?


1 Like

Hi~I have the same error when try to run my experiment with 18 .jpeg. It stopped at downloading the13th. I’m wondering if there’s a max size of resource that we could apply?

My error message was * unable to download resource: Icon (FILE_LOAD_ERROR)

Hi @apitot ,

I am unable to recreate the error on my machine (and some other I tried while testing) – the task ran perfectly on my machine and the error was initially signaled to me by some participants.

Since your message I contacted and asked some participants who ran into this error to re-run the experiment (10 of them). I already have confirmation from 5 of this 10 people and it looks like the experiment is running smoothly now. I will come back to this thread if errors re-appear.

Great work on your part guys,



Thank you for your kind words. I am very glad to read that all is well!
Best wishes,


I can confirm that I haven’t come across this error again! :slight_smile:

However, I think this issue was mostly due to my browser cache, and from pressing “Run Online” directly from PsychoPy (which runs the experiment in a normal browser window). I use a timestamp on the first page of my exp. to ensure the version I’m testing online is indeed up-to-date with the local version, and this has revealed the online version doesn’t always update properly. Chances are, PsychoPy / Pavlovia is working fine and your exp. is indeed syncing correctly, but your browser cache is the issue.

If anyone is having issues with syncing, and you’re confident your JS coding is correct - try running your experiment in an Incognito window - that’s where I do all my testing now, and it works every time.

  1. Make changes to your PsychoPy exp. locally >
  2. press Sync with web project (but don’t press run online) >
  3. paste your exp. URL in an Incognito tab.

Of course, this is only an issue for us researchers when we’re testing out our changes. Participants will only view the finished version of the experiment - once, presumably - and will not be returning to the same link after changes are made, so browser caches won’t be a problem there.

I am having a similar issue, receiving File_Load_Error on a specific image each time. I have tried changing image size, clearing cache, trying in incognito tab - to the same result each time, on the same image. I wrote about it with a link to my experiment in the following post: Unable to download resources; 502 bad gateway

Any insight is much appreciated…not sure how to go forward from here.

Hello, I am currently having the same problem with my project. I was wondering if you were able to fix it and what you did to fix the issues

Hi @vs265

Sorry for the late reply - busy times :slight_smile:
I encountered this error a while ago so, I am not that sure that I remember the solution correctly. To the best of my recollection, I scaled down all the stimuli - attempting to compress them as much as possible without losing too much quality. A second thing I did is that I split the code component in each trial into two separate components - because my experiments required some 300 lines of JS per trial. After doing both these things, the experiment ran without major incidents aid it used around 560 different images, code and response components.

I do not know if this is helpful, I hope it is though