There are at least three possible explanations, which are not mutually exclusive:
(a) Motion artifact (as you postulate). However, this should also show up at the top edge of the brain.
(b) Blood inflow artifacts. Huge amounts of blood are being pumped through the Circle of Willis, and this causes a couple of MRI artifacts. Since the inflow rate depends on the phase of the cardiac cycle, and since you are probably not doing cardiac "gating" (sync-ing the MR acquisitions to the hearbeat), these artifacts will appear kind of randomly in the time series.
--(b1) One effect is simply the inflow of "fresh" spins, that have not previously been affect by the RF transmissions.
--(b2) Another effect is that the rapidly moving spins (i.e., water protons in the blood) cause phase changes in the MR signal which are not compensated for in the image reconstruction, which causes a "ghosting" artifact (some image intensity is reconstructed in the wrong place). Ghosting artifact can cause signal fluctuations throughout the slice. Ghosting can be reduced by using a "flow compensated" acquisition method, but this slows down the imaging speed, and so isn't usually done in FMRI.
(c) Respiration causes magnetic field changes (due to movement of the chest in the main magnetic field), which in turn causes MRI signal changes. This is most apparent at the base of the brain (which is closest to the chest).
I'm voting for cardiac effects, without any further information. In any event, the huge amount of effective noise will make it hard to detect any actual FMRI signals in this region. To some extent, filtering out the cardiac and respiratory signals, assuming you have an external measurement of these time series, might help. At present, we don't have a program for doing this filtering; however, there is some effort along this line being made by a student who is visiting here.