Categories and Categorization
Surely the most basic cognitive process involved in stereotyping is categorization—or at least that's the common claim (Allport, 1954; Brewer, 1988; Fiske & Neuberg, 1990; Hamilton, 1979; Jones, 1982; Samuels, 1973; Smith, 1990; Tajfel, 1969; Taylor, 1981; Vinacke, 1957). Categorization itself is ubiquitous. Our whole language of nouns (as well as adjectives and adverbs) is built around the fact that we group animals, plants, things, events, and people into categories. Those who argue that people have limited cognitive resources with which to tame a complex environment suggest that categories are helpful in simplifying the world (e.g., Hamilton & Trolier, 1986; Taylor, 1981). To say that an animal is a dog is to encourage its dog qualities, to shellac its Rover individuality—a necessary process in a world with too many unique Rovers. However, others (e.g., Medin, 1988; Oakes & Turner, 1990) have argued that the essential cognitive problems we humans face is not too much but rather too little information, and that categories help us infer information not directly given by our senses. We know that Rover can bite and bark, even if we do not catch her in the act. Both perspectives are helpful. We group things into categories because we expect that the things within a given category will be similar in some ways and different in others from things alien to the category. This gives us predictive control over the environment, a leg up in deciding on appropriate behavior (Anderson, 1991; Ross & Murphy, 1996).
Before stereotyping can take place, an individual must be seen as a member of one or more categories to which stereotypes may apply. In that sense, categorization is a necessary condition for stereotyping to occur. But it may not be a sufficient condition. I can imagine seeing a person as Asian American or as a banker without necessarily deriving any cognitive consequences from that categorization. However, it remains an open question whether categorizations and the accompanying inferences them can be divorced so easily. Why would we develop categories unless we used generalizations about them efficiently and quickly (Schneider, 1992)?
Having said that, I should note that those who operate from a Gibsonian perspective (e.g., Zebrowitz, 1996) do not think that such categorization is always neces