We have been running an experiment on Pavlovia. We are screensharing with the participant through Microsoft Teams and allowing them to control our mouse to do the task. We are doing that because in part we need to see them do the task, but also because another experiment we are running can only be done on a PC from our computer sharing the mouse, so it’s just easiest to do it this way. We have been doing this on and off for a couple years no problem (and consistently for the last few months with no problems until now). But recently, there has been an issue where some of the drawn images (e.g., a red square) or text will randomly disappear from the screen, the screen will randomly flash black and white, or most recently a frowny face appeared in the corner of the screen. It just randomly happens during the task, and not for all participants.
We did find a discourse about the frowny face, which makes me think it is a memory issue associated with screensharing and a problem with the participants internet connection. When we just test it on our own computer, it’s not an issue (or at least we havent randomly encountered it when we try to see if it bugs). It was originally from an older version of psychopy, so I re-uploaded from the most recent version of psychopy. We have also run it in incogito mode. The issue still happens. There are no images or sounds loaded in, so it shouldnt be computationally intensive to run.
One option would be to run it locally through PsychoPy. The problem is that there is a lot of JS code being implemented, so getting that to work through PsychoPy would be a difficult process (albeit doable). We could also try having the participant run the Pavlovia task from their computer and screenshare with us, that way they arent having to control our mouse (which can be a bit laggy anyways). We have not tried the latter yet, as we are mid-experiment and dont want to change things. But that certainly is better than not getting any data at all (this is pre/posttest assessment after 4 weeks of intensive training…so not having data for the posttest is an absolute killer in terms of resources spent).
I guess I was just wondering if there might be some other solution that we can try that we are not aware of. Again, this has only become an issue in the last 2 weeks.
It’s quick tricky to run PsychoPy while screen sharing. Do you ask them to turn off their video?
I think it’s worth exploring them running it on their computer and screensharing with you rather than having the added overhead of controlling the mouse over the Teams software.
I was also wondering whether you could shift the video call to a phone so the participant is running the experiment on one device and the video call is on another.
So I assume this means you think it is a memory issue, as discussed in the previous discourse? That is, it’s probably not just some random pavlovia bug, despite only just now becoming an issue.
I dont think we have them turn off their video, so we can start with that.
2. If that doesnt work, we will try having them screenshare with us instead.
3. Dont think the phone/video option will work at this point (big change in administration mid-stream). But if there is no resolution, yes, we could just have the complete the experiment on their own computer while we’re on the phone with them.
Will give these options a try and report back if they solve anything. Thanks for the suggestions.
I do think there is a memory issue. If it’s new despite not substantially changing the experiment then I suspect it has been caused by an increased inefficiency of PsychoPy, Pavlovia, Microsoft Team, your operating system or that of the participants. It is difficult to pinpoint exactly which component is to blame (if it is indeed just one). Could Zoom be tried instead of Teams, or another system of remote control?
The first: we screenshared from our computer but had the participant turn off their camera. The white frowny face appeared partway through the task. So it’s not simply the participant’s camera being on that caused the issue. The research assistant restarted the program, had the participant open it and screen share from their own computer, and things worked fine.
The second: Teams was open with another participant on the call, but we werent screen sharing at the time. The background turned white upon startup (it should be black). After putting in the participant ID, the white frowny face appeared. So it’s not simply a Teams screensharing issue, though it could still be in part because Teams is open/running on our local computer. The research assistant restarted the program, ran the program and screenshared from our computer, and everything worked fine.
We moved to having them open the program and control the mouse on their own computer. They screenshare with us so we can see what is going on, but everyone’s cameras are turned off. This seems to have resolved the issue. That is, it seems to primarily have stemmed from originally having them take control of the mouse from our computer (which also introduces a bit of a lag). Still not sure why this issue just started, but at least it’s working properly now.
