Dear SPMers, we used a mixed-effects model (within-subject repeated-measures ANOVA) in SPM2 to anlyze fMRI data for 20 healthy subjects. At the second level, we looked for communalities and interactions between two two-level factors (A and B). Significant interactions (-A1 +A2 +B1 -B2) were found in a number of brain areas. We now want to test an additional hypothesis about the regional specificity of the two differences that contributed to the functional interaction. That is, we would like to test wether the second difference (B1-B2) is larger than the first difference (A2-A1) at the peak voxel of some (for instance two or three) significant activation clusters, and vice versa (i.e., A2-A1 larger than B1-B2) at the peak voxel of another significant activation cluster. We would like to carry out these post-hoc comparisons of the two differences outside of SPM, but we are not sure how to extract the individual subject data in which the individual variability of each subject is already accounted for. Thanks for any help and kind regards, Klaus