The reason it works in HD3D is because it actually sends 2 images at once at a rate of 60hz to be displayed one after the other, which technically is 120hz worth of images. The display then shows each image to your two eyes one after the other, but for normal 120hz, HDMI 1.3 (and 1.4) can't send single images at 120hz. It's known as frame packing, and I have heard of hacks to apply it to 2D, but you obviously have to add a fair bit of latency to make it work.
Anyway, HDMI 2.0 is the first to allow 120hz outside of frame packing (HD3D or hack with latency issues).
So the confusion arises because people see it handles HD3D at 60hz and assumes it handles 120hz without, which it doesn't.