The evidence from Mongolia supports the hypothesis that humans process ethnies as natural living kinds (theoretical considerations suggest that they do so at the “species” level). My Torguud subjects have a blood-based model for assigning individuals to ethnies. Beyond this, they consider such assignment to carry implications for ethnic category-based behavior even without any exposure to other members of their ethnic category, and they seem to believe that the ineffable “something” responsible for this is carried somehow “inside.” All of these parallel essentialist thinking in natural living kinds, suggesting that my subjects’ thinking about ethnies is not only primordialist but essentialist and that there is no difference between an ethnic group and a species from the point of view of the schemas that are primed to process them. Processing endogamous norm groups as species, I have argued, was adaptive in the ancestral environment because (1) it allowed us to learn a lot about out-groups in a very inexpensive way, in particular by making inductive inferences about nonobvious properties, and (2) it made possible processes of discrimination that prevented us from incurring the costs of coordination failure. The reason these benefits have been obtained specifically by processing these groups as species results from the fact that ethnies exhibit the most diagnostic features of species: group-based endogamy and descent-based membership. This made it easy for a blind evolutionary process to exapt a preexisting architecture by simply failing to discourage the priming by ethnies of the living-kinds module. This is not, I think, how we think of social categories in general but only how we think of those categories which, as in ethnies, exhibit the strongly diagnostic features of biological species, such as feudal classes and castes.