Dear SPMers,
We scanned an event-related fMRI spatial working memory study with two groups of participants. The aim of the study is to analyze differences in working-memory performance & associated neural activity.
The working memory task was a spatial delayed-matching-to-sample task with three levels of difficulty (participants had to remember either 1, 3 or 5 locations).
The task comprised the following 4 conditions:
1. Baseline (fixation cross) = BL
2. WM load 1 (remember 1 location) = WM1
3. WM load 2 (2 locations) = WM2
4. WM load 3 (5 locations) = WM3
Now I’m considering which contrast to compute on the first level for the second level analysis and which first level models to compute.
1. Contrasts for WM1>BL, WM2>BL, WM3>BL
As input for a second level 2x3 ANOVA with the factors group x WM-Load
2. Beta-images for the conditions WM1, WM2, WM3
As input for a second level 2x3 ANOVA with the factors group x WM-Load
3. A parametric-contrast
WM-Load as a parametric repressor in the first level analysis
Should the BL-condition be modeled in this first level analysis?
4. A simple non-parametric contrast WM3>WM1
As input for a between-group t-test on the second-level
Should the conditions WM2 and BL be modeled in this analysis?
Furthermore: would it be reasonable to put all the contrasts in the same first level design matrix?
Thanks & Kind regards,
Ben
|