I'm not exactly sure, so I skipped it and didn't attempt it. My RPG mobs' faces are on their bodies sorta so uh.. yea.
My best guess is like - say 20 frames -
Frame 0 - normal
Frame 10 heads all the way left
Frame 20 heads all the way right
Test it in-game, see how it works, improve upon concept, eventually (and hopefully), get success. Good luck! (;