Empiricist and nativist approaches to understanding the origins and development of conceptions of logical necessity are proposed and criticized. A constructivist alternative is proposed, incorporating contemporary ideas regarding the role of structure, mechanisms of development, the metacognitive basis of stages, and the relation of development and learning. Three stages are postulated. The first involves the development of various classification and seriation behaviors. The second (concrete) stage is marked by a reflection on these behaviors, involving the construction of implicit concepts of necessity which, given the truth of certain premises, require certain conclusions. Finally, the third (formal) stage is marked by a further metacognitive reconstruction of logical necessity on a plane still further removed from empirical truth, thereby yielding the concept of inferential validity. Relevant research is briefly reviewed.