JW-
Well, the command is working, but the overlap of masks (or, non-overlap) is producing all zeros.
Note that there is a difference between isnegative(x) and not(ispositive(x))-- the differentiation is what happens at x==0.
If you want to see where the field of 1s is from the not(ispositive()), you can just make a mask fo that:
3dcalc -a DSET -expr 'not(ispositive(y-5))' -prefix TEST
... and overlay your hippocampus mask; you can play around with where the desired location of the 'split' between post and ant would be.
--pt