Single-trial GLM for trace fear conditioning - all parameters zero

[valentina-jelincic]

Dear Bachlab, PsPM users,

I am analyzing data from a trace fear conditioning experiment (very short CS of 100ms, followed by trace interval of 6s, expectancy rating, and the US (omission). The effective CS-US interval is ~10s on most trials. I am trying to extract SCR in the interval between the CS and the expectancy rating (trace interval).

Basic paradigm: differential fear conditioning (two different CS+, two different CS-, electric US)

Trial flow: Fixation cross for 5-7s --> CS for 100ms --> fixation cross for 6s --> expectancy rating until answer --> fixation cross for 1s --> US (omission) --> fixation dot for 2s

Based on the specifics of my paradigm (short CS, long CS-US interval), I chose to model the SCR using GLM. Because I would like to use mixed effects models, I need trialwise estimates. Using this paper by Kuhn, Gerlicher and Lonsdorf as reference, I created a condition file where each CS onset, each expectancy rating onset, and each US onset are included as separate conditions. I ran the batch and exported all parameters.

Now, I am facing two problems:

1. I could not export reconstructed amplitudes, nor use any of the plotting functions under Model Review because of the following error:

Error using [regexpi](matlab:matlab.lang.internal.introspective.errorDocCallback('regexpi')) All cells must be char row vectors.

2. When exporting all parameters (which works), every event (every single CS, rating, and US) has a bf1 and bf2 of zero. I checked what happens when I do combine conditions, and the output there is as expected:

csplus_in recon	csplus_ex recon	csminus_in recon	csminus_ex recon	exp recon	us recon
1.17585313	0.96261459	0.24133225	0.66654362	0.47692827	1.75964341

To be sure, I repeated this with 3 participants, and get the same zero estimates each time, even though the condition-wise GLM works. I am using the same condition file for both GLMs, just slightly tweaked to separate all the onsets in their own cells, so it is not a timing error.

Am I missing something about how the trialwise GLM should be set up? I am using all the default GLM settings. Thank you in advance for any clarity you can provide!

Kind regards,
Valentina

[dominikbach]

Thanks for reaching out.

Which PsPM version are you using?

Could you send us a sample GLM file to check what's happening?

Dominik

[valentina-jelincic]

Hi Dominik, Thank you for the quick reply! I am sorry, I should have specified, I am using PsPM v. 7.0. I am attaching the GLM batch file. I tried to attach one of the condition files, but I am not allowed to send .mat files. Let me know if you want me to share more files, e.g. via FileSender. Valentina

…

________________________________ From: Dominik Bach ***@***.***> Sent: Wednesday, February 11, 2026 14:56 To: bachlab/PsPM ***@***.***> Cc: Valentina Jelinčić ***@***.***>; Author ***@***.***> Subject: Re: [bachlab/PsPM] Single-trial GLM for trace fear conditioning - all parameters zero (Discussion #886) Thanks for reaching out. Which PsPM version are you using? Could you send us a sample GLM file to check what's happening? Dominik — Reply to this email directly, view it on GitHub<#886 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AY4KAPNQA46MN3EPXPDXRS34LMYHPAVCNFSM6AAAAACUXUJH5SVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTKNZWHE4DGMI>. You are receiving this because you authored the thread.Message ID: ***@***.***> [ { ***@***.***": "http://schema.org", ***@***.***": "EmailMessage", "potentialAction": { ***@***.***": "ViewAction", "target": "#886 (comment)", "url": "#886 (comment)", "name": "View Discussion" }, "description": "View this Discussion on GitHub", "publisher": { ***@***.***": "Organization", "name": "GitHub", "url": "https://github.com" } } ] %----------------------------------------------------------------------- % pspm PsPM - Unknown % cfg_basicio BasicIO - Unknown %----------------------------------------------------------------------- matlabbatch{1}.pspm{1}.prep{1}.import.datatype.acq_bioread.datafile = {'E:\FeaRES\Physiology analysis\code_feares\S1_1\scr\pspm_data\019902_FA.mat'}; matlabbatch{1}.pspm{1}.prep{1}.import.datatype.acq_bioread.importtype{1}.scr.chan_nr.chan_nr_spec = 2; matlabbatch{1}.pspm{1}.prep{1}.import.datatype.acq_bioread.importtype{1}.scr.scr_transfer.none = true; matlabbatch{1}.pspm{1}.prep{1}.import.overwrite = 1; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.output.file = 'GLM_trialwise_scr_019902'; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.output.dir = {'E:\FeaRES\Physiology analysis\code_feares\S1_1\scr\pspm_glm_outputs'}; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.output.overwrite = 1; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.chan.chan_def = 0; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.timeunits.seconds = 'seconds'; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.session.datafile(1) = cfg_dep('Import: Output File', substruct('.','val', '{}',{1}, '.','val', '{}',{1}, '.','val', '{}',{1}), substruct('()',{':'})); matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.session.missing.no_epochs = 0; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.session.data_design.condfile = {'E:\FeaRES\Physiology analysis\code_feares\S1_1\scr\pspm_data\019902_FA_GLM_trial_file.mat'}; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.session.nuisancefile = {''}; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.latency.fixed = 'fixed'; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.bf.scrf1 = 1; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.norm = 1; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.filter.edit.lowpass.disable = 0; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.filter.edit.highpass.disable = 0; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.filter.edit.down = 10; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.filter.edit.direction = 'uni'; matlabbatch{2}.pspm{1}.first_level{1}.scr{1}.glm_scr.exclude_missing.excl_no = 'No'; matlabbatch{3}.pspm{1}.first_level{1}.export.datafile(1) = cfg_dep('GLM for SCR: Output File', substruct('.','val', '{}',{2}, '.','val', '{}',{1}, '.','val', '{}',{1}, '.','val', '{}',{1}), substruct('()',{':'})); matlabbatch{3}.pspm{1}.first_level{1}.export.datatype = 'cond'; matlabbatch{3}.pspm{1}.first_level{1}.export.exclude_missing = false; matlabbatch{3}.pspm{1}.first_level{1}.export.target.output.file = 'estimated_SCR_trials_019902'; matlabbatch{3}.pspm{1}.first_level{1}.export.target.output.dir = {'E:\FeaRES\Physiology analysis\code_feares\S1_1\scr\pspm_glm_outputs'}; matlabbatch{3}.pspm{1}.first_level{1}.export.target.output.overwrite = 1; matlabbatch{3}.pspm{1}.first_level{1}.export.delim.tab = '\t';

Hi Valentina,

Could you please send the file directly to me? I’ll take a look at it. Feel free to share them via FileSender or any other file transfer service.

Best,
Bernhard

[valentina-jelincic]

Hi Bernhard,

Thanks, we figured out the issue actually!

The script I had used to split the onsets into separate cells and create names for different Trials had the following line:
names{i} = 'Trial_' + string(i);

However, as noted in the error, all the GLM-related functions require cell values in 'names' to be char arrays, resulting in inaccurate exports. Once I update the line to:
names{i} = convertStringsToChars('Trial_' + string(i));
GLM runs as expected, I can plot the data using Model Review, and the exports make a lot more sense. So, there is no bug or anything, just a class mismatch.

Perhaps it would be worthwhile to emphasize the importance of char arrays in the documentation or tutorials, for those new to coding or coming from R/Python (where string is the default). Or even put a catch in the function to convert any strings to chars. Granted, you would not run into this issue just following the tutorials, because the 'names' variable in the tutorial is created directly as a cell array of chars (with each char array specifying a condition), hence why I had no problems running and exporting the conditionwise GLM. But if you are running a special case GLM like single-trial and need to create variables in a loop, it can happen if you are not very familiar with MATLAB classes. It is probably not an issue that would happen regularly for users though...

Thanks again for the quick reactions, and the work you put in PsPM!

Valentina