The last few decades have seen a rise in nativist accounts of the development of cognition and language. One reason for this is that the field of cognitive development has experienced great gains in the sophistication with which we describe human knowledge, with no corresponding gain in the sophistication of our process accounts. In particular, we argue that our field is saddled with a behaviorist view of similarity. We propose to replace this view with structure-mapping account, in which comparison is seen as a process of structural alignment and mapping. Our main contention in this paper is that the process of comparison constitutes an important force by which similarity-based processes can give rise to rule-governed systems. We will begin by laying out a developmental framework that we call “the career of similarity.” Then we apply this framework first to the general issue of the development of rules in learning and reasoning, and then to language learning, specifically the acquisition of word meaning.