I have designed an experiment in Builder (v2022.1.2). There is a key component defined to calculate RT. The experiment runs perfectly fine when I run the experiment in the current version of Psychopy. The data file is generated in a CSV file, with RT for key response being reported in seconds. However, when I run the same experiment in the latest version of PsychoPy (v 2024.1.1), RT is being reported in values like ‘21.5894215’, ‘21.6388794’ etc. In the current experiment, RT can not be in the range of 21 seconds. So, it can not be a second unit. I am not sure what is the unit of RT being reported in the latest version of PsychoPy. In addition, how can I calculate RT in seconds with already recorded data?
I think that the issue is that the RT is being given relative to the global clock. This should be fixed in 2024.1.2 available early next week.
Not sure if this should be in a new thread.
I have tested the RT calculation on Ubuntu 20.04 and Psychopy 2024.1.2 and have the same problem.
To test I made a two routine keyboard response test.
The first routine is simply a 4 second pause. The second shows “Go” and record the key_resp. The key_resp.rt ~ test routine end (4.3s).
Not sure if this is just on my end.
I can confirm the RT computation not being calculated to the start of the key component in 2024.1.2 (MacOS 13.6.6), I see the same behaviour in 2024.1.2 as in 2024.1.1
It was puzzling me why this was still happening for you both when my tests are running fine but think I’ve tracked it down now! The fix in 24.1.2 fixed it only when using the psychtoolbox backend - in ioHub there was an additional adjustment needed, I didn’t catch it because my tests happened to be using the last backend I tested with (which was psychtoolbox). Will get a .3 release out ASAP with this fully fixed! In the meantime, go to Experiment Settings → Input and set the keyboard backend to “Psychtoolbox” to avoid this bug.
2 Likes
There’s now 2024.1.3 with this hopefully fixed, if you could confirm?
Hello,
I just checked the behaviour with a short simple-response experiment, 10 trials. I found the following response times:
2024.1.2 Input ioHub:
key_resp.rt
2.9790797000168823
5.25073749999865
7.851133000018308
10.571022100019036
13.052549999993062
15.563450300018303
17.82802040001843
20.579928899998777
23.00318299999344
25.139084399997955
2024.1.2 Input Psychtoolbox:
key_resp.rt
0.48188439998193644
0.37089570000534877
0.4493120000115596
0.38295639998978004
0.6155215000035241
0.46478869998827577
0.4702172000252176
0.37770400001318194
0.35765960000571795
0.428477399982512
2024.1.3 Input ioHub:
key_resp.rt
-665.6386526
-665.719861
-665.7245329
-665.6951448
-665.6241216999999
-665.6962546999999
-665.6652124999999
-665.725742
-665.7439535999999
-665.7433785999999
2024.1.3 Input Psychtoolbox:
key_resp.rt
0.4870552999999518
0.3387970999999652
0.31748210000000654
0.3638317999999572
0.3045061999999916
0.28380300000003444
0.25523249999992004
0.3025062000000389
0.3203171000000111
0.3100577999999814
Thus, with input psychtoolbox it seems to be correct for 2024.1.2 and 2024.1.3
With input ioHub it is wrong for both versions. But differently wrong…
This is on windows 10, experiment running on my desktop.
yours - Sven
1 Like
Thank you so much for your response. This was very helpful.
Thank you for sharing the comparison of key_resp.rt in PsychoPy 2024.1.2 and 2024.1.3 versions. It is useful!