Hi
I’m trying to run an experiment that worked perfectly on MPB (2014) Mojave 10.14.5, Matlab 2016b, PTB 3.0.15 on a new machine:
MBP (2020) Catalina 10.15.7 Matlab 2020b, PTB 3.0.17
I can’t seem to pinpoint the failure since there are no errors being returned.
Problem: Response recording fails with fmri buttonbox as input. The responses are recorded correctly if I use the laptop keyboard as the response input.
I looked at the output of PsychHID(‘Devices’) and it’s identical to the old laptop’s output.
I’m using KbCheck to record the responses.
Any ideas of where to look and usual suspects/problems with Catalina OS would be really helpful.
p.s. I know mac is not ideal, this decision is not under my control
The good old days, when Apple still called itself “Apple Computer”, …
So KbCheck does work with the regular keyboard of the MBP laptop? Just not with the fMRI box pretending to be a keyboard? Do KbQueues work / fail the same way? That would rule out the standard bugs we already know and love from Catalina.
Anyway, if you need any further help in diagnosing this issue - without any guarantee of a successful resolution, given the sad state of macOS - please buy priority support, as explained at the top of the forum and on our webpage and provide a suitable authentication code. That would buy you up to one hour of me trying to help you. If you do that, make sure to provide the full detailed output of devs = PsychHID(‘Devices’) to save some time.
That’s true, there could not be a worse and at the same time more expensive choice of hardware and operating system for neuroscience research. But at least your lab won’t be short of money to pay for our support, given somebody consciously chose to buy that machine.
Yes, you are correct, the box pretending to be a keyboard doesn’t get recorded, but the regular MBP keyboard does. KbQueueCheck fails the same way and also works with the regular MBP keyboard.
I’ll ask about purchasing priority support.
I’m also looking at just setting up a VM as a work around, but I’m not sure if that’s a valid solution.
Depends if you care about accurate results or not. Usually a VM is not a solution on any OS, just an extra troublemaker. Certainly nothing we’d support in any way.
-mario
Good-ish news! I’ve been able to replicate the issue on the old MBP running mojave using an actual external keyboard (and not the fmri buttonbox pretending to be a keyboard) as the response input. Not sure where to look now, but at least it makes me think Catalina isn’t the issue…
I do not understand every aspect of this issue, but I also had this problem and our local team was able to fix it. Here is my clumsy attempt to explain.
Using MATLAB on a Mac these days, the inputs from an fMRI button box need to be adjusted. Here’s an article explaining non-auto-releasing (NAR) versus auto-releasing keyboard behavior: MatlabPtbNar < CdiDocs < Foswiki. Our local MRI technicians set up our button box to work in either mode (selected via a switch), and we found that MATLAB on a Mac required the non-standard mode to be able to catch button presses reliably.
Once we changed that, we ran into another problem whereby any trigger sent by the MRI set off an unending chain of constant triggers that drowned out any other inputs. After some tinkering, our MRI technicians discovered that the polarity of the triggers sent from the MRI also needed to be adjusted to work - I do not know what the correct setting was, but if yours has multiple, try each one.
Apologies if this information doesn’t help, but hopefully it provides a few avenues for you to try while troubleshooting.