Relational Categories

This chapter is concerned with the acquisition and use of relational categories. By relational category, we mean a category whose membership is determined by a common relational structure rather than by common properties. For instance, for X to be a bridge, X must connect two other entities or points; for X to be a carnivore, X must eat animals. Relational categories contrast with entity categories such as tulip or camel, whose members share many intrinsic properties. Relational categories cohere on the basis of a core relationship fulfilled by all members. This relation may be situation-specific (e.g., passenger or accident) or enduring (e.g., carnivore or ratio). Relational categories abound in ordinary language. Some are restricted in their arguments: For example, carnivores are animals who eat other animals. But for many relational categories, the arguments can range widely: for example, a bridge can connect two concrete locations, or two generations, or two abstract ideas. As with bridge, the instances of a relational category can have few or no intrinsic properties in common with one another. Research on categories has mostly ignored relational categories, focusing instead on entity categories-categories that can be characterized in terms of intrinsic similarity among members, like those shown in Figure 9.1. Further, as Moos and Sloutsky (2004) point out, theories of categorization have often operated under the assumption that all concepts are fundamentally alike. However, as Medin and his colleagues (Medin, Lynch, & Coley, 1997; Medin, Lynch, & Solomon, 2000) have argued, categories are not uniform in character, and the variations support a range of different functions. In this chapter, we contrast relational categories-categories whose members satisfy a specified relational