Abstract

representation of the support relation. In contrast, infants who had heard general phrases, novel words, or no words at all during habituation failed to notice the change in relations even for familiar objects; they attended to a change in objects, but not to a change in relations. These studies suggest that language is instrumental in prompting infants to form stable spatial relaiional categories. Consistent with this claim, it appears that in the absence of linguistic guidance, young infants are ready to form a variety ofspatial categories. McDonough, Choi, and Mandler (zoo 3J famili arized nine-to-fourteen-monthold Englishor Korean-learning infants with either pairs of tight containment events or pairs of loose containment events, accompanied only by music. Although the tight-loose distinction is far more central in Korean than in English, both groups of infants were able to extract the category during familiarization Both groups could distinguish the familiar category from the new category when shown novel test pairs [with new objects) consisting of a tight containment event and a loose containment event.3 Hespos and Spelke [zoo4) studied even younger infants and found that five-monthold English-speaking infants can readily form either the English support/containment distinction or the Korean tight fiV loose fit distinction. The infants were habituated either with a single tight containment event or with a loose containment event: for examplg a cylinder entering another container that fit either tightly or loosely. They were then tested with both tight containment and loose containment events (shown sequentially). Infants habituated to tight containment looked longer at the loose containment event and vice versa, indicating that they had abstracted the respective category. More surprisingly, this pattern held up even when infants had to transfer the tighV loose distinction from support to containment. That is, when infants were habituated with either tight or loose support events, and I A11 ages showed a familiarity preference in both ranguages. then shown the tight containtnent and loose containment test events, they looked longer at the novel test event. This suggests that five-month-old infants can form the Korean tight-loose distinction, even when it cuts across the English in/on disttnction. Lining up the developmental studies discussed so fa4 we have a rather perplexing contrast. Five-month-olds showed sensitivity to the tight-loose distinction unmarked in their native language in Hespos and Spelke's [zood study. But in Casasola's [zoo5) study, which also used a habituation paradigm, eighteen-month-olds failed to show sensitivity to the support category which is marked in their language, unless they heard the requisite spatial term. We suggest that this difference may rely on the degree of generalization that the infants needed to make (see also Casasola, zoo8J. In Casasola's study, the habituation events were quite varied and the objects involved were perceptually rich and differed across trials; in the Hespos and Spelke study, the habituation trials utilized highly similar events, both in the motions involved and in the objects (which were varied only slightly). Likewisg the test trials were perceptually quite dissimilar from the habituation trials in Casasola's study (espe' cially in the novel object trials) and perceptually similar to the habituation trials in Hespos and Spelke's study. One might then ask "So which study is right? 'vVhen exactly do infants have the category of support?" We suggest that this is the wrong question. Rather, the better question is "When [and under what learning conditions) can infants form a category of support at a given level of abstraction?" If we consider that performance in these stud" ies derives in part from abstractions formed during the study (rather than solely from preexisting categories), then both kinds of study are informative. We can see the'se studies as spanning a range. At one pole are studies in which the intended relation is perfectly aligned across exemplars with few distracting surface differences (as exemplified in Hespos and Spelke's studies) an ideal situation in which to form a generalization, albeit one that may not apply far beyond the