URL of experiment: Sign in · GitLab
Description of the problem: For an experiment I am currently piloting on Pavlovia, 2 trials in the experiment appear as having negative reaction times (see attached data file below: cells AF38 and AF79). This only happened for these two trials, out of 700 total trials in the experiment. Out of 4 total pilot experiment being run to date, this issue has occurred in 2/4 of the experiments run.
11_mainONLINE_2020-07-20_13h17.24.867.csv (198.7 KB)
There was a post about a year back of this issue occurring, and per that thread the issue seemed to have been resolved with a release of PsychoPy. My best guess would be that this occurs when participants press a key in the ~16ms between the screen refresh and the presentation of the stimulus? Would this be correct, or any other ideas what may be going on here?
Based on the data file it looks like they are responding just too late for the previous trial and the response is being recorded at the start of the new trial.
Am I right in thinking that you have a 1 second ISI? The negative RTs I saw were just less than -1 second.
You could clear the keyboard buffer just before responses are accepted for the new trial
Yes, you are correct the ISI for this set of trials is 1 second.
So would the solution simply be to add “event.clearEvents()” to a code component in the End Routine tab, in the routine with my trials? Or should I place it in the Begin Routine tab?
Thanks so much for your help!
Following up with this - @wakecarter, is this the solution you recommend? I tried putting event.clearEvents() in both the Begin and End routine code components, but my experiment is still recording negative response times. I’m using PsychoPy 3.1.2. Thank you for your time and consideration!
Please could you show the routine and your component/code that is recording the reaction times?
Of course, thank you so much. Here’s the GitLab URL: Sign in · GitLab. Here’s a CSV file showing the negative RTs (G82 and G87) PARTICIPANT_rfs_thesis_9_11_20_new_2020-09-11_14h46.27.209.csv (11.3 KB). If it’s helpful for you to know, the ITI for these trials was 2s long. Please let me know if there’s any additional information I can provide, and thanks again so much for your input.
In order for others to see the repository, you must change it’s visibility. This can be done from the settings tab when you are browsing the project on gitlab. There will be a “permissions” section which has a “project visibility” option to change the visibility of the project.
Make sure you disable this after wakecarter helps you out.
Thank you so much for the reminder - the project should be public now!
I think that the issue is with your feedback routine.
You seem to be going to great lengths to reset the keyResponse clock (Begin Routine (in Python), Every Frame and End Routine). Why are you doing this? The keyResponse itself seems simple enough, starting at the beginning of a routine and ending after .65s.
I don’t know if it works online, but you could try ticking “discard previous” in keyResponse to delete responses prior to the start of the component, though I’d imagine that this is similar to having event.clearEvents()
Thank you so much for your rapid reply. I really appreciate the time and energy you’re spending towards helping me debug this issue. Checking “discard previous” in keyResponse and adding a new keyboard component during the feedback routine seems to have resolved the error for the time being.