This study challenges the use of adult neuropsychological models for explaining developmental disorders of genetic origin. When uneven cognitive profiles are found in childhood or adulthood, it is assumed that such phenotypic outcomes characterize infant starting states, and it has been claimed that modules subserving these abilities start out either intact or impaired. Findings from two experiments with infants with Williams syndrome (a phenotype selected to bolster innate modularity claims) indicate a within-syndrome double dissociation: For numerosity judgments, they do well in infancy but poorly in adulthood, whereas for language, they perform poorly in infancy but well in adulthood. The theoretical and clinical implications of these results could lead to a shift in focus for studies of genetic disorders.