The little troublemaker on the right is Yinlong downsi, an unbelievably primitive marginocephalian from Xinjiang, China. Interestingly, it's name, which means "hidden dragon," is a reference to (what else?) Crouching Tiger, Hidden Dragon, which was filmed partially in Xinjiang. The animal is older even than Psittacosaurus, having been unearthed from unquestionably Late Jurassic strata. The type specimen includes a complete skull and the majority of the skeleton--much of it articulated. What makes Yinlong special (aside from the fact that's it's more primitive than Psittacosaurus) is that is retains many features which may connect it to pachycephalosaurs and, more surprisingly, heterodontosaurs. Yinlong is beyond a doubt a ceratopsian, however, as it has a rostral bone. However, its ornamented squamosal bones where previously thought to be unique to pachycephalosaurs. In addition, Yinlong exhibits enlarged premaxillary teeth and features of the antorbital and temporal skull regions which were thought to be synapamorphies of the Heterodontosauridae.
Yinlong's authors posit a new taxon, the Heterodontosauriformes, to include Heterodontosauridae (as the outgroup) and Marginocephalia, which still includes Pachycephalosauria and Ceratopsia. This is fairly heady stuff, as it would mean that a significant ghost lineage existed prior to the Oxfordian stage of the Late Jurassic, when Yinlong lived. After all, heterodontosaurs proper are known from the Late Triassic/Early Jurassic periods, and Yinlong is an early ceratopsian, meaning that the Marginocephalia must have split prior to the Late Jurassic. Indeed, if the Heterodontosauridae is the outgroup to the Heterodontosauriformes, then we should expect to find early representatives of the Marginocephalia as early as perhaps the Early Jurassic.
That seems pretty cut and dry, really. So Marginocephalia originated farther back than any of us thought? Great. Well, not so great. Just a few weeks ago (you may have read my post on it), Eocursor parvus was discovered and kind of tossed a bone (HA!) into the Heterodontosauriformes scenario. Eocursor is a primitive ornithischian and has the good fortune to be the most complete member of that taxon to have ever been found. More complete than Lesothosaurus by far, in fact. Eocursor seems more derived than the heterodontosaurs, but less so than the Genasauria, the massive ornithischian group that includes thyreophorans, duckbills, and (gulp) the Marginocephalia. It's simply unquestionable that pachycephalosaurs and ceratopsians are among the Genasauria, but Eocursor's authors regale the Heterodontosauridae to the base of the Ornithischia, well below the Genasauria and even the Neornithischia.
So there are still a few options. First, perhaps Eocursor's authors are wrong, and the heterodontosaurs are not more primitive than the Genasaurs. That's certainly a possibility, given the seemingly arbitrary reasons for demoting the Heterodontosauridae (so they've got wide, grasping hands--so what?). Another possibility is that the Heterodontosauriformes is real, but it also primitive, as Eocursor would have it. That would mean that the Marginocephalia went through an incredible amount of convergence with more advanced Genasaurs. It's also possible that Yinlong's authors are wrong, and the Marginocephalia has no connection beyond a little bit of convergence with Heterodontosauridae. This would, again, leave the origins of the Marginocephalia a mystery (other than is forms a sister group to the Ornithopoda).
I suppose the ultimate verdict will depend entirely on where Heterodontosauridae ends up on the Ornithischian family tree. As I understand it, several studies are currently underway as to a further understanding of these strange beasties. Perhaps in a few years, we'll know for sure one way or the other.