Hello Dr. Bach,
I have used pspm to analyse my SCR data with the GLM, and have done my second levels. Now i have some results but also some questions.
Just to be sure, are the parameters shown after the second level, the amplitudes of the SCR?
The results show that only the US really give a signal, but the other CS+ or CS- do not give a big signal. Is there a way to explain this?
Is there a way i can conferm that the participants are conditioned? I want to test if my HC are conditioned on the shock they received at the CS+.
Thanks for your help.
Best regards,
Linde
Last edit: Linde 2015-05-22
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
2nd level analysis shows resulting amplitudes in data units. Note if you z-score the data, these are z-scored units (wrt to the variance of the entire data time series).
There are at least three reasons why CS+>CS- does not give you a big signal: (1) Because people did not learn the association between CS+ and US (this is usually what you want to test), (2) because the CS+ does not elicit a strong CR (depends on your US) and (3) because GLM is not an optimal way of analysing a learning experiment when CS-US SOA is more than 1-2 seconds. In the last case I'd recommend a non-linear model (DCM) as an alternative. Our DCM validation paper (Bach, Daunizeau et al. 2010) demonstrated that DCM is superior to GLM to recover a known CS+/CS- difference when CS+/US SOA was 3.5 seconds.
Hope this helps
Best wishes
Dominik
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello Dr. Bach,
I have used pspm to analyse my SCR data with the GLM, and have done my second levels. Now i have some results but also some questions.
Just to be sure, are the parameters shown after the second level, the amplitudes of the SCR?
The results show that only the US really give a signal, but the other CS+ or CS- do not give a big signal. Is there a way to explain this?
Is there a way i can conferm that the participants are conditioned? I want to test if my HC are conditioned on the shock they received at the CS+.
Thanks for your help.
Best regards,
Linde
Last edit: Linde 2015-05-22
Dear Linde
thanks for your questions.
2nd level analysis shows resulting amplitudes in data units. Note if you z-score the data, these are z-scored units (wrt to the variance of the entire data time series).
There are at least three reasons why CS+>CS- does not give you a big signal: (1) Because people did not learn the association between CS+ and US (this is usually what you want to test), (2) because the CS+ does not elicit a strong CR (depends on your US) and (3) because GLM is not an optimal way of analysing a learning experiment when CS-US SOA is more than 1-2 seconds. In the last case I'd recommend a non-linear model (DCM) as an alternative. Our DCM validation paper (Bach, Daunizeau et al. 2010) demonstrated that DCM is superior to GLM to recover a known CS+/CS- difference when CS+/US SOA was 3.5 seconds.
Hope this helps
Best wishes
Dominik