Hello afni users,

I was running analysis on my data set (125 runs across 8 sessions) in AFNI for the first time using 3ddeconvolve.
My experiment has 3 conditions in blocked design. Baseline: fixation task. A: fixation task with peripheral motion stimulus B:attend to motion. Some percent of the blocks from all three conditions where monkey's fixation has not met the criteria has to be excluded from analysis. These blocks where performance has not met the criteria were considered to be type C blocks.
I ran 3dDeconvovle with conditions A, B, C as my regressors and GLT for A-B. I also included the motion regressors from 3dvolreg.

I am surprised that condition A beta coeff across majority of the voxels are all negative and statistically signfificant with (t <-3 ). even in superior temporal sulcus (STS) for presence of motion. However for condition B, beta coeff looks positive. I computed independentlly a t-statistic for BOLD signal between Baseline line condition and condition A across all voxels. majority of the tvalues are positive and STS voxels show postive t scores.
Can some one see why I might be getting these negative beta values for presence of stimulus for majority of the voxels?

Thanks a lot
Amar