Description 1:
I am trying to conduct a trial level regression analysis. I have prepared my data and edited my study design in accordance with the tutorial. I precomputed channel measures successfully, but when I go to run my first level analysis there are 10/100 participants who always fail at the design stage. It's always the same 10 participants (T018, T019, T021, T041, T063, T067, T073, T092, T103, T109) but I can't think for the life of me what makes them different that might be causing it. I have performed a regular categorical 2x4 anova on the exact same dataset successfully at the first and second level, with no issues for those particular participants. There must be something going on with the data structure for those 10 participants, but it is only a problem for the response column data. The precise error message is as follows:
Something went bad ... the job has FAILED !
The last error message occured was :
Output argument "X" (and possibly others) not assigned a value in the execution with "limo_design_matrix" function.
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\limo_batch_design_matrix.m at line 350
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\external\psom\psom_run_job.m at line 214
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\external\psom\psom_run_job.m at line 128
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\external\psom\psom_run_script.m at line 437
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\external\psom\psom_run_script.m at line 323
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\external\psom\psom_pipeline_process.m at line 591
File D:\users-nobackup\Matlab Toolbox\eeglab2024.2.1\plugins\LIMO4.0\external\psom\psom_run_pipeline.m at line 417
File C:\Program Files\MATLAB\R2022b\toolbox\matlab\lang\parallel_function.m at line 837
File C:\Program Files\MATLAB\R2022b\toolbox\parallel\remoteParallelFunction.m at line 67
Further debugging involved putting a breakpoint at the offending line and re-running the design job for the offending participants. This yields the following error:
Unrecognized field name "cellArray".
Error in pop_loadset>checkoldformat (line 309)
eegset = EEG.cellArray;
Error in pop_loadset (line 172)
EEG = checkoldformat(TMPVAR.EEG);
Error in limo_batch_design_matrix (line 24)
EEGLIMO=pop_loadset([LIMO.data.data_dir filesep LIMO.data.data]);
Loading the temporary first level .set file into eeglab directly yields the same error.
Steps to Reproduce 1:
Using the first folder I emailed you:
- Load the study file "Trial_Regressor_Analysis_Hands.study" into EEGLab
- Make sure you are using the second Design (design.hands.regressor)
- Click Study > Precompute channel measures
- Tick the ERP box, specify the baseline as [-100 0]. Leave all else as is and run it.
- Click Study > LInear MOdeling of EEG Data > Estimate Model Parameters (channel)
- Check the Split regressions box, keep everything else the same and run it
- See error
Expected Behavior 1:
I expected the first level analysis to run as usual with all participants processing correctly so that I could move on to the second level.
Actual Behavior 1:
Received an error suggesting EEGLab cannot load 10 of my participants for some reason. For whatever reason EEGLab can't read these particular participants, but only when the continuous response variable is part of the design.
Versions 1:
|
|
| OS version |
[Windows 11] |
| Matlab version |
[2022b] |
| EEGLAB version |
[2024.2.1] |
| LIMO version |
[4.0] |
I am kind of at the end of my rope with this error, which brings me to issue #2 (related):
Description 2:
Cyril (@CPernet) was not able to reproduce this error; he found that he could successfully run all participants. I downloaded the exact same versions of Matlab, EEGLab, and LIMO he was using to try to see if this would fix it. When I do this, I cannot even precompute channel measures as I get this popup:
The strange thing is, I looked for this function within the current EEGLab and couldn't find it! So I am really scratching my head and none of my labmates can make heads or tails of this either. If there is any advice or help you have for me I would really appreciate it.
Steps to Reproduce 2:
Using the second folder I emailed you:
- Load the study file "Pain_Trial_Regressor_Hands.study" into EEGLab
- There is only one design on this one, so don't worry about that
- Click Study > Precompute channel measures
- Tick the ERP box, specify the baseline as [-100 0]. Leave all else as is and run it.
- See error popup
Expected Behavior 2:
I expected the precompute channel measures stage to run normally so I could move on to first level.
Actual Behavior 2:
Received the error popup included above, which seems to be referring to an old function no longer in EEGLab?
Versions 2:
|
|
| OS version |
[Windows 11] |
| Matlab version |
[2025b] |
| EEGLAB version |
[develop] |
| LIMO version |
[master] |
Description 1:
I am trying to conduct a trial level regression analysis. I have prepared my data and edited my study design in accordance with the tutorial. I precomputed channel measures successfully, but when I go to run my first level analysis there are 10/100 participants who always fail at the design stage. It's always the same 10 participants (T018, T019, T021, T041, T063, T067, T073, T092, T103, T109) but I can't think for the life of me what makes them different that might be causing it. I have performed a regular categorical 2x4 anova on the exact same dataset successfully at the first and second level, with no issues for those particular participants. There must be something going on with the data structure for those 10 participants, but it is only a problem for the response column data. The precise error message is as follows:
Further debugging involved putting a breakpoint at the offending line and re-running the design job for the offending participants. This yields the following error:
Loading the temporary first level .set file into eeglab directly yields the same error.
Steps to Reproduce 1:
Using the first folder I emailed you:
Expected Behavior 1:
I expected the first level analysis to run as usual with all participants processing correctly so that I could move on to the second level.
Actual Behavior 1:
Received an error suggesting EEGLab cannot load 10 of my participants for some reason. For whatever reason EEGLab can't read these particular participants, but only when the continuous response variable is part of the design.
Versions 1:
I am kind of at the end of my rope with this error, which brings me to issue #2 (related):
Description 2:
Cyril (@CPernet) was not able to reproduce this error; he found that he could successfully run all participants. I downloaded the exact same versions of Matlab, EEGLab, and LIMO he was using to try to see if this would fix it. When I do this, I cannot even precompute channel measures as I get this popup:
The strange thing is, I looked for this function within the current EEGLab and couldn't find it! So I am really scratching my head and none of my labmates can make heads or tails of this either. If there is any advice or help you have for me I would really appreciate it.
Steps to Reproduce 2:
Using the second folder I emailed you:
Expected Behavior 2:
I expected the precompute channel measures stage to run normally so I could move on to first level.
Actual Behavior 2:
Received the error popup included above, which seems to be referring to an old function no longer in EEGLab?
Versions 2: