I was trying to look at the activation within an ROI and the activation map does not seem to correspond to the output of 3dmaskave. I used 3dmaskave to calculate the number of active voxels within an ROI:

3dmaskave -mask ROI_allHipp_R+tlrc -drange .01 1000 Patterns_act1+tlrc

and the output was 123 voxels. However, when I visually compared Patterns_act1 to the ROI I could not see where the overlap occurred. The activation map was calculated using:

3dmerge -1thresh 6.21 -1clust 5.2 90 -1dindex 6 -1tindex 8 -prefix Patterns_act1 Patterns.buc+orig
adwarp -apar SPGR+tlrc -dpar Patterns_act1+orig

To view the overlap I used 3dcalc:
3dcalc -a ROI_allHipp_R+tlrc -b Patterns_act1+tlrc -prefix overlap -exp "step(a)+(2*step(b))"

When I viewed the results in AFNI the overlap areas in overlap+tlrc (values = 3) did not correspond with active areas in the activation map (the voxel value in Patterns_act1+tlrc was 0 while the value in overlap+tlrc was 3). The same thing happens if I just apply a step function to the activation map. I am not sure why a voxel is listed as Func=0 in Patterns_act1 but once a step function is applied it goes to one, unless the resampling during the Talairach transformation gave the voxel a value slightly greater than zero (although the afni widget still lists the voxel as 0).