Pii: S0010-0277(99)00076-1

Perhaps Needham and Baillargeon would be surprised to ®nd that we are in complete agreement on many important points, including: (1) at issue is the nature of infants' representations of events and objects, and how infants parse the world into individuals and kinds; (2) looking time methodologies are a fruitful source of data that bear on these issues; (3) under some circumstances, infants below 12 months of age make use of experiential, physical (including spatiotemporal) and featural information for object individuation; (4) it is of utmost importance to resolve apparent con ̄icts in data and interpretations from different studies; (5) ®nally, our results do seem to con ̄ict with results from experiments by Needham, Wilcox, and Baillargeon, and relevant data from other laboratories are rather mixed (see below). In our experiments (Xu & Carey, 1996; Xu, Carey & Welch, 1999), 10-month-old infants failed to use property/featural information (e.g. the differences in shape, color, or texture) or object kind information (e.g. the difference between a duck and a car) to parse a display into two objects (although under some circumstances, success is obtained at 9 months; see Section 5.1). By 12 months of age, infants succeeded at these tasks. In contrast, Needham and Baillargeon (henceforth N&B) reviewed studies which provide evidence that under some circumstances, infants succeeded in using property/featural information for object individuation as early as 4.5 months. How do we resolve this apparent con ̄ict? In their reply, N&B questioned both our results and our interpretation of the results. The results, however, are highly robust ± the studies of Xu and Carey (1996) had four internal replications of the failure at 10 months, and the shift Cognition 74 (2000) 285±301 COGN I T I O N