I think it is important to realize that any contrast of 1 -1 and -1 1 produce exactly the same t-statistic, only scaled by -1. SPM results mainly functions to apply a threshold to the t-maps to show what passed the significance criteria you enter.
So, if you load the t-maps as an overlay in some other program, such as MRIcron, the 1 -1 and -1 1 will have identical t-values in every voxel, with the sign reversed. You can therefore never see overlapping activity in these two contrasts.
I don't think you can at all apply the term "deactivation" in either of these contrasts. It is possible both conditions show an task related neural activity. The contrast shows if one conditions shows "larger" activity (as measured by Beta values). Now, things get complicated in that a zero is larger than a negative value. In this case, you can have a "decrease" in neural activity for condition A relative to baseline/rest/etc, and no real change in the condition B, which then shows as "greater activity in A", although perhaps a more accurate explanation is "less deactivation in A and B".
For this reason, I think it is always important to look at the Beta values for each condition for your significant clusters. But then then a negative Beta value must b e interpreted with care, as it is negative with respect to everything else going on in your experiment. So if you have a short ISI, a negative Beta value may not be deactivation at all. I realize that is complicated and not necessarily helpful, but if fMRI was easy it wouldn't be so interesting.
Best of luck,
Colin.