in our design we have 4 blocks: B1, B2, B3, B4

in each block we have two condition (that repeat in other blocks too): C1 and C2

we want to find locations where C1-C2 is significantly different between B1 and B2.

so I have defined 8 different regressors: B3_C1, B3_C2 ,B4_C1 and B4_C2 are te four that I am not interested in. I have 4 other regressor: 2 for C1 and 2 for C2. in one of C{1,2} I have put 1 and -1 for block type B1 and B2 respectively in valence (3 column format of GLM in fsl full model setup). I call these B1-B2. I have also added B1+B2 for both conditions. I mean I have two regressors for C1 and C2 in B1 and B2 with ones for valence. 

1- what do you think about my strategy?

2- do you have any better way to test this?

sorry if I am not very clear.
WRG,

Ashkan