Do infants reason about the object kind human, as represented by the human body, in a more mature fashion than they do other object kinds? When adults recognize an object as belonging to a certain kind (e.g. dog), this licenses a wide variety of inferences, including kind-specific inferences, like whether the object barks, and more abstract inferences, such as whether the object’s behavior is goal-directed. It also allows adults to track or count individuals in ambiguous contexts (Bonatti et al., 2002; Hall, 1998; Hirsch, 1982; MacNamara, 1986; Spelke, 1990; Xu, 1999, 2007; Xu and Carey, 1996; Xu et al., 1999). For example, if a small, furry, white dog goes out the door and sometime later a small, furry, white cat comes back in, adults know they have encountered two individuals because adults understand that dogs do not become cats. Even within a single point in time, kind designations help adults to individuate objects even when they are partially occluded. A tail and a nose poking out from behind opposite sides of a bush suggest one dog, but a tail poking out from behind each side indicates two dogs. Infants, on the other hand, are considerably less able to use kind information in such situations (Xu, 1999, 2007; Xu and Carey, 1996; Xu et al., 1999).

The current studies examine whether infants reason more maturely when the object in question belongs to the kind human. This possibility was suggested by Bonatti and colleagues (2002) on the basis of infants’ skill with human faces. One suggested explanation for this advantage is the status of humans as the prototypical goal-directed object. The current studies rule out this explanation by showing that (1) when the exemplar object used to represent the kind was a human hand rather than a doll’s face, the advantage disappeared, and (2) this was so even when goal attributions to the hand were deliberately primed. Specifically, this series of studies showed that although 9- to 10-month-old infants recognize and represent the human hand “well enough” to encode its behavior in terms of goals, they do not recognize or represent it “well enough” to track a hand through occlusions on the basis of its appearance alone. We use this dissociation to argue for a lack of integration across different types of representational/inferential systems in infancy, specifically with respect to body representations.