Over half a century ago, Shepard, Hovland, and Jenkins (1961), described 6 types of category structures, ranging from single feature categories (e.g., all red things), which allow easy generalization to new category members, to arbitrary categories, which must be memorized. I will begin with an overview of their category types as well as older and newer experimental work that documents the ease with which adults and non-human animals are able to learn different types. I will then draw parallels between the Shepard et al. categories and the kinds of generalizations that infants must make to learn a language. Finally, I will attempt to show that infant generalization in several laboratory language learning experiments is consistent with adult and animal learning of the Shepard et al. category structures.