I am new to the PsPM software. I would definetly like to use this for my SCR analyses. I have played around with the toolbox a bit with some data from a fear conditioning task. I have a few questions based off of what I am observing at the single subject level.
1) I see that there is a very robust negative deflection for my CS+ condition (see CON+ in the attached figure) and then a later positive deflection. I ran a glm SCRF 1 where I modeled the duration of each stimulus as well (7.5sec instead of 0). What is this strong negative deflection? I expected to see a strong positive response for the conditioned stimulus. Also, if I don't model the 7.5 second event duration (i.e., I keep duration at default of 0), I don't observe that later CON+ positive deflection.
2) Related, would modeling the stimulus events be conceptually similar to using a boxcar modeling function in fMRI?
3) I don't get what the x-axis is modeling. Is that time in seconds? If so, I do not know why it plotted for that long.
4) I do know that I have a lot of baseline time prior to the task events that are very noisey. Would removing pre-task data improve the model?
Basically, I don't get the strong negative deflection for the conditioned stimulus. I would expect the deflection to be opposite. This makes me think that I have done something wrong.
If your CS is as long as 7.5 seconds (is it?) then DCM would be your best bet for analysing the data, as the CS response is unlikely to occur time-locked to the CS (which is what GLM assumes). We use DCM already for 3.5 s CS duration.
Specifying event duration in the GLM specification is inherited from fMRI - but it is not physiologically plausible. While CNS neurons can fire for extended periods of time, sudomotor nerves just don't seem to do that - they fire in sparse bursts of under 1 s duration.
The x-axis in the plot has units of 0.1 s (will be fixed in next version).
I would definitively recommend "trimming" time periods before the first event (e.g. at 5 s before the first event) and after the last one (but keep some 10-20 s to have the tail of the last response).
Hope this helps
Dominik
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Regarding the set-up for DCM, I am a bit confused as to how to set-up the events through the GUI. I am currently entering everything manually. A few questions:
1) To enter the timing files, do I create a new event for each condition like I tried with the GLM (6 events for 6 conditions)?
2) Since this is conditioning task, I should use the 2-column flexible latency approach with event onset in column 1 and event offset in column 2?
3) Do you need an equal number of trials for each condition? I get an error that says 'all vectors must have the same number of rows.' Unfortunately, I have an unbalanced design.
4) But, if I make I try to run it so that each condition has an equal number of events, I get a different error message: "Error in event definition. Either events are outside the file, or trials overlap." I know that there are no duplicate events and all events seem to be well within the length of the data file. Could it be an error in how it is reading the datafile and time? I don't get this message if I enter same files into GLM.
Sorry for all of the questions. I am new to this analysis and toolbox...
Best,
Danny
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
It's difficult to spot the error without seeing the batch script you are using (or the condition file).
However, events in this approach do not correspond to conditions - but to events within one trial (across all conditions). DCM is not explicitly defined about conditions (even though you can specify them, but this info is just stored for later).
I think I get the concept for the events now. I was entering them like the GLM previously and that is why I was getting the error. So basically, for my fear learning study I only have two types of events: 1) CS and 2) US. Using this, the model is running now!
Per the manual, for creating a US event vector, it says that I can enter negative values for the onsets of events that do not have a US. Is that correct?
Thanks,
Danny
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
importantly the model should not be informed about the trial labels - and this includes the presence of a US. The absence/omission of a US should be treated as an event just as well, as it is suggested to elicit SCR.
So you have one matrix for all CS, and one vector for all US/absence of US.
Dominik
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
So instead of having negative values for the onset of the absence of a US, I should include the actual time stamp of when the omission would occur? In other words, if the onset of a US omission is 100 seconds, I would keep it at 100 instead of -100?
I created two different events. The first one was a 2-column matrix with the onsets and offsets for the 44 CS events I have in the task. The second event is a 1-column matrix with onsets for the 44 presence/absence of US events.
Then under the condition section, I entered the trial number that corresponds to my conditions of interest so that I can later run statistics on the parameter estimates.
Thanks again for all the help!
Best,
Danny
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
I am new to the PsPM software. I would definetly like to use this for my SCR analyses. I have played around with the toolbox a bit with some data from a fear conditioning task. I have a few questions based off of what I am observing at the single subject level.
1) I see that there is a very robust negative deflection for my CS+ condition (see CON+ in the attached figure) and then a later positive deflection. I ran a glm SCRF 1 where I modeled the duration of each stimulus as well (7.5sec instead of 0). What is this strong negative deflection? I expected to see a strong positive response for the conditioned stimulus. Also, if I don't model the 7.5 second event duration (i.e., I keep duration at default of 0), I don't observe that later CON+ positive deflection.
2) Related, would modeling the stimulus events be conceptually similar to using a boxcar modeling function in fMRI?
3) I don't get what the x-axis is modeling. Is that time in seconds? If so, I do not know why it plotted for that long.
4) I do know that I have a lot of baseline time prior to the task events that are very noisey. Would removing pre-task data improve the model?
Basically, I don't get the strong negative deflection for the conditioned stimulus. I would expect the deflection to be opposite. This makes me think that I have done something wrong.
Any advice or help would be greatly appreciated!
Best,
Danny
Hi Danny
thanks for your message.
If your CS is as long as 7.5 seconds (is it?) then DCM would be your best bet for analysing the data, as the CS response is unlikely to occur time-locked to the CS (which is what GLM assumes). We use DCM already for 3.5 s CS duration.
Specifying event duration in the GLM specification is inherited from fMRI - but it is not physiologically plausible. While CNS neurons can fire for extended periods of time, sudomotor nerves just don't seem to do that - they fire in sparse bursts of under 1 s duration.
The x-axis in the plot has units of 0.1 s (will be fixed in next version).
I would definitively recommend "trimming" time periods before the first event (e.g. at 5 s before the first event) and after the last one (but keep some 10-20 s to have the tail of the last response).
Hope this helps
Dominik
Hi Dominik,
Thank you for your helpful reply.
Yes, my CS is fixed at 7.5 seconds.
Regarding the set-up for DCM, I am a bit confused as to how to set-up the events through the GUI. I am currently entering everything manually. A few questions:
1) To enter the timing files, do I create a new event for each condition like I tried with the GLM (6 events for 6 conditions)?
2) Since this is conditioning task, I should use the 2-column flexible latency approach with event onset in column 1 and event offset in column 2?
3) Do you need an equal number of trials for each condition? I get an error that says 'all vectors must have the same number of rows.' Unfortunately, I have an unbalanced design.
4) But, if I make I try to run it so that each condition has an equal number of events, I get a different error message: "Error in event definition. Either events are outside the file, or trials overlap." I know that there are no duplicate events and all events seem to be well within the length of the data file. Could it be an error in how it is reading the datafile and time? I don't get this message if I enter same files into GLM.
Sorry for all of the questions. I am new to this analysis and toolbox...
Best,
Danny
It's difficult to spot the error without seeing the batch script you are using (or the condition file).
However, events in this approach do not correspond to conditions - but to events within one trial (across all conditions). DCM is not explicitly defined about conditions (even though you can specify them, but this info is just stored for later).
E.g. event 1: CS (two-column vector), event 2: US/US omission (one-column vector).
Hi Dominik,
I think I get the concept for the events now. I was entering them like the GLM previously and that is why I was getting the error. So basically, for my fear learning study I only have two types of events: 1) CS and 2) US. Using this, the model is running now!
Per the manual, for creating a US event vector, it says that I can enter negative values for the onsets of events that do not have a US. Is that correct?
Thanks,
Danny
Hi Daniel
importantly the model should not be informed about the trial labels - and this includes the presence of a US. The absence/omission of a US should be treated as an event just as well, as it is suggested to elicit SCR.
So you have one matrix for all CS, and one vector for all US/absence of US.
Dominik
So instead of having negative values for the onset of the absence of a US, I should include the actual time stamp of when the omission would occur? In other words, if the onset of a US omission is 100 seconds, I would keep it at 100 instead of -100?
I created two different events. The first one was a 2-column matrix with the onsets and offsets for the 44 CS events I have in the task. The second event is a 1-column matrix with onsets for the 44 presence/absence of US events.
Then under the condition section, I entered the trial number that corresponds to my conditions of interest so that I can later run statistics on the parameter estimates.
Thanks again for all the help!
Best,
Danny
This is correct, Danny, hope it works out! Best, Dominik