ABSTRACT
Jean Mandler and her colleagues have made important discoveries about infants' categorizing of objects that have the potential for transforming our understanding of conceptual development, word learning, and indeed cognitive development in general. On the basis of their work they have made a set of related theoretical claims:
Conceptual and perceptual categories develop together from the infant period. This view accords with those who distinguish between perceptual and conceptual categories (e.g., Barsalou, 1999; Bornstein, 1984; Nelson, 1985). It contrasts with those theories of conceptual development that assume that categorization on the basis of similarity of perceptual features is primary, and that only later do children learn to categorize on the basis of essential characteristics (Keil, 1987; Smith & Jones, 1993).
The initial conceptual categories are abstract and global. This is perhaps Mandler's most crucial claim, and it is complemented by the third proposal. Both contest accepted formulations based on Rosch's pathbreaking work in the 1970s, which claimed that there is a basic level of categorization of objects that is most “natural” and that is acquired first by children (Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976). This is the level in an object taxonomy where category instances share most distinctive features with each other and are distinct from other categories at that level, thus the features have high cue validity for identifying category instances. In Rosch's view, the critical features of the basic level include perceptual attributes, especially shape, and movement. For example, cars and trucks are each basic level categories within the vehicle taxonomy. Specific kinds—subordinate types—of cars (e.g., Camrys, taxis) share car features but have distinctive features differentiating them from each other. Vehicles share only a few features in common, thus are higher order, not basic level, and the features they share are primarily abstract and functional, for example, “used for transportation.”
Basic level categories are not primary in development. Rosch's work, and that of others, indicated that young children could categorize at the basic level, but not at higher or lower levels. Mandler's counter claim that the basic level is not primary in development is based on findings that infants respond selectively to instances of categories such as animate and inanimate, rather than randomly or to narrower categories, such as dog and cat (McDonough & Mandler, 1998). These findings contrast with habituation studies showing that infants can discriminate perceptually between dogs and cats (discussed in a section to follow). Forming global categories such as animate is abstract, not based on specific perceptual features, a conclusion that goes against theoretical claims that conceptual categories develop from concrete to abstract bases. Mandler is not claiming that children form superordinate categories that include basic level categories; rather basic level categories must be differentiated from global categories.
Conceptual categories are formed in the effort to establish meaning; thus the abstract basis for infant categories involves the role that objects play in events, a relational basis, not a static inherent perceptual basis. This claim is based in part on findings that infants will allow members of one global category (animate) to engage in the same functions in events (such as drinking from a cup) but will not extend this liberty to members of a different (inanimate) global category.
