Dear Vladimir,
I am analyzing some 2*2 designed ERP data. Subjects were presented four kinds of emotional stimuli and they were asked to rate the valence of the stimuli after experiment. Therefore, four contrast images and four rating scores were collected from each subject. I would like to investigate the individual difference on ERP signals associated with behavioral data by using multiple regression in SPM8 as shown below:
images score_a1b1 score_a1b2 score_a2b1 score_a2b2
a1b1_sbj01.img 12 0 0 0
a1b1_sbj02.img 08 0 0 0
a1b1_sbj03.img 06 0 0 0
.... .... .... .... ....
a1b2_sbj01.img 0 05 0 0
a1b2_sbj02.img 0 09 0 0
a1b2_sbj03.img 0 12 0 0
.... .... .... .... ....
a2b1_sbj01.img 0 0 17 0
a2b1_sbj02.img 0 0 05 0
a2b1_sbj03.img 0 0 12 0
.... .... .... .... ....
a2b2_sbj01.img 0 0 0 03
a2b2_sbj02.img 0 0 0 07
a2b2_sbj03.img 0 0 0 13
.... .... .... .... ....
The first column above shows the contrast images (2*2 designed, i.e. a1/a2 * b1/b2) input into the option 'Scans' in the multiple regression design. The other four columns denote the behavioral scores for each kind of stimuli. I am not sure whether it is OK that non-zero values, i.e. scores, and zeros were recruited as covariates for corresponding contrast images and irrelevant images, respectively.
If the multiple regression set above is correct, I would like to find the consistent ERP signals across all four kinds of stimuli, and I would like to know whether a conjunction analysis will contribute to this idea. The conjunction involves the four contrasts from the outcome of the multiple regression.
Is there any way to do the correlation analysis between ERP data and behavioral data? Could ERP data from some ROIs output by Marsbar or some other software?
BTW, the EEG module of SPM8 is really an excellent tool for ERP data analysis.
Bests,
Sun Delin
|