I'm very sorry, I said I'd leave you alone but I have one more question.
We record SCR with PSYCHLAB which is the worst program in the world and obviously not supported by anything because all their software is proprietary and there's literally no support for it. I know you can convert PSYCHLAB files to .edf files but my .edf files look weird after I convert them so I'm hesitant to use them. What I HAVE been doing throughout the last couple of years is exporting my SCR data into .txt files where I essentially have 2000 samples before our event markers and then 16000 samples after the shock marker (for the ITIs). When exported into .txt files the output is 3 columns where column 1 is the SCR measurement in microSiemens, column 2 is the shock level when the shock occurs, and column 3 are the event codes indicating the start of the CS+ and CS- trials (and the shock).
I know you can import .txt files with the toolbox. I wouldn't have to change them in order to be able to use the toolbox on them, correct? I would just have to specify the SCR "channel" as 1 and then I can make an onsets file and just work from there? I just want to make sure the multi-column structure of the .txt file won't mess anything up.
Thanks again for your help!
Lauren
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Actually Dominick I might have answered my own question as it seems you can specify columns when using 'text' as file type. I couldn't see any way to delete this post but please disregard. I'm sorry for the superfluous question.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Yes, it is continuous. It seems my adventures are continuing; I just want to clarify something about the non-linear models.
I've imported my SCR data which is a .txt file of samples (we sample at 1000Hz so it is a 900000x1 vector). When I did the GLM just to see if I could do it, I based my timing relative to samples.
Ultimately the non-linear modeling will be better for my purposes so I will stick with the DCM from now on and I'm playing with that at the moment. I set up a single onset file as is stated in the tutorial - with onsets and offsets for each CS trial. I wasn't clear, however, if the onset and offset times in the tutorial relate to seconds or something else. The tutorial input seems to be of sample length (53221x1 vector for a 532.2 duration run) but the 'events' in the 'event_timings_s1.mat' look like seconds except in the info header it says 'markers?' Anyway I set up my own event file with the onset and offset of my CSs based on the samples (e.g. events{1,1}(:,1)=2000, 7500) but that didn't work. So I just divided everything by 1000 to have the event times in 'seconds' and the model now runs. I just want to make sure this is what is needed for event times in a scenario like this (where the data input is in samples) so I'm modeling the right thing.
A final question related to your last sentence above about deconvoling overlapping responses: So in my initial run throughs where I was just testing my input - completely on accident - I had accidently had two trials overlapping (the end of one trial overlapped with the beginning of the next). When I did this I got a warning about overlapping trials; this isn't anything dire, is it? For my test data I'm using right now I will not have overlapping trials but in the future I will. It seems like it's just a warning and will run anyway.
Thank you (again).
Lauren
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
DCM timing is always in seconds. GLM timing can be in seconds, samples, or markers. Regarding your notes on the manual: the units of a channe of type "marker" is given in the field header.sr - if it is 1 then the units are seconds. So the events in the DCM tutorial are read out from a marker channel that is in seconds, and the events are thus in seconds as they should be.
Overlapping trials are not necessarily a problem, but depending on how much overlap and how many responses in the data, the algorithm may not be able to unambiguously estimate the response amplitude. This is why you get a warning. Intuitively, the closer together two external events are, the more difficult it is to assign the ensuing SCR to the correct eliciting event. When event time windows don't overlap in time, the ensuing SCR are likely to be properly assigned.
Hope this helps
Dominik
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I'm setting up a protocol for using this toolbox for our lab. Unfortunately, as indicated above we use PSYCHLAB which is not itself supported by the software. Oddly, converting these files to .edf files (as recommended in another thread) result in .edf files that cannot be read by either EDFbrowser or Polyman so I am hesitant to use them.
So as stated previously, I have outputted a .txt file of a continuous recording with all SCR values in column 1, and 'event markers' in column 3. As the manual states .txt input does not support importing markers. Will be need to import markers at any point for data analysis and if so, is there a way to do this with .txt files? My hack job solution at the current time is just to go into the original file in excel and mark the sample where each trial begins and make an onset file from those and start creating the GLM with those onsets right after importing. Is this an appropriate way to deal with this kind of file structure? I guess I'm really trying to figure out if not being able to import markers will hurt us at any point in the analysis.
Thank you very much for your continued help with my questions.
Lauren
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
the latest version of PsPM now supports importing continuous event data for text files. If you do not have the latest version, you can download it here: https://sourceforge.net/projects/pspm/files/ .
Best
Tobias
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Dominick:
I'm very sorry, I said I'd leave you alone but I have one more question.
We record SCR with PSYCHLAB which is the worst program in the world and obviously not supported by anything because all their software is proprietary and there's literally no support for it. I know you can convert PSYCHLAB files to .edf files but my .edf files look weird after I convert them so I'm hesitant to use them. What I HAVE been doing throughout the last couple of years is exporting my SCR data into .txt files where I essentially have 2000 samples before our event markers and then 16000 samples after the shock marker (for the ITIs). When exported into .txt files the output is 3 columns where column 1 is the SCR measurement in microSiemens, column 2 is the shock level when the shock occurs, and column 3 are the event codes indicating the start of the CS+ and CS- trials (and the shock).
I know you can import .txt files with the toolbox. I wouldn't have to change them in order to be able to use the toolbox on them, correct? I would just have to specify the SCR "channel" as 1 and then I can make an onsets file and just work from there? I just want to make sure the multi-column structure of the .txt file won't mess anything up.
Thanks again for your help!
Lauren
Actually Dominick I might have answered my own question as it seems you can specify columns when using 'text' as file type. I couldn't see any way to delete this post but please disregard. I'm sorry for the superfluous question.
Hi Lauren
from your post I wasn't sure whether you are importing continuous data.
A particular strength of the PsPM approach for SCR is that you can deconvolve overlapping responses - but only if you have continuous data.
Best
Dominik
Hi Dominick:
Yes, it is continuous. It seems my adventures are continuing; I just want to clarify something about the non-linear models.
I've imported my SCR data which is a .txt file of samples (we sample at 1000Hz so it is a 900000x1 vector). When I did the GLM just to see if I could do it, I based my timing relative to samples.
Ultimately the non-linear modeling will be better for my purposes so I will stick with the DCM from now on and I'm playing with that at the moment. I set up a single onset file as is stated in the tutorial - with onsets and offsets for each CS trial. I wasn't clear, however, if the onset and offset times in the tutorial relate to seconds or something else. The tutorial input seems to be of sample length (53221x1 vector for a 532.2 duration run) but the 'events' in the 'event_timings_s1.mat' look like seconds except in the info header it says 'markers?' Anyway I set up my own event file with the onset and offset of my CSs based on the samples (e.g. events{1,1}(:,1)=2000, 7500) but that didn't work. So I just divided everything by 1000 to have the event times in 'seconds' and the model now runs. I just want to make sure this is what is needed for event times in a scenario like this (where the data input is in samples) so I'm modeling the right thing.
A final question related to your last sentence above about deconvoling overlapping responses: So in my initial run throughs where I was just testing my input - completely on accident - I had accidently had two trials overlapping (the end of one trial overlapped with the beginning of the next). When I did this I got a warning about overlapping trials; this isn't anything dire, is it? For my test data I'm using right now I will not have overlapping trials but in the future I will. It seems like it's just a warning and will run anyway.
Thank you (again).
Lauren
Hi Lauren
DCM timing is always in seconds. GLM timing can be in seconds, samples, or markers. Regarding your notes on the manual: the units of a channe of type "marker" is given in the field header.sr - if it is 1 then the units are seconds. So the events in the DCM tutorial are read out from a marker channel that is in seconds, and the events are thus in seconds as they should be.
Overlapping trials are not necessarily a problem, but depending on how much overlap and how many responses in the data, the algorithm may not be able to unambiguously estimate the response amplitude. This is why you get a warning. Intuitively, the closer together two external events are, the more difficult it is to assign the ensuing SCR to the correct eliciting event. When event time windows don't overlap in time, the ensuing SCR are likely to be properly assigned.
Hope this helps
Dominik
Hi Dominik:
I'm setting up a protocol for using this toolbox for our lab. Unfortunately, as indicated above we use PSYCHLAB which is not itself supported by the software. Oddly, converting these files to .edf files (as recommended in another thread) result in .edf files that cannot be read by either EDFbrowser or Polyman so I am hesitant to use them.
So as stated previously, I have outputted a .txt file of a continuous recording with all SCR values in column 1, and 'event markers' in column 3. As the manual states .txt input does not support importing markers. Will be need to import markers at any point for data analysis and if so, is there a way to do this with .txt files? My hack job solution at the current time is just to go into the original file in excel and mark the sample where each trial begins and make an onset file from those and start creating the GLM with those onsets right after importing. Is this an appropriate way to deal with this kind of file structure? I guess I'm really trying to figure out if not being able to import markers will hurt us at any point in the analysis.
Thank you very much for your continued help with my questions.
Lauren
Hi Lauren
the latest version of PsPM now supports importing continuous event data for text files. If you do not have the latest version, you can download it here: https://sourceforge.net/projects/pspm/files/ .
Best
Tobias