Abstract This paper addresses the issue of how and under what circumstances new information is incorporated into a person's existing world knowledge. We propose that the primary functional characteristic of incorporated information is that it is easy to retrieve and use such information in a context that is different from the one in which it was originally learned. Subjects in our experiments learned a long story that introduced true facts about real but previously unknown terms. We then examined subjects' ability to answer questions about this newly learned information as a function of test context. The effect of test context differed depending on whether subjects were led to believe the material was real or artificial. Two ways of conceptualizing the distinction between compartmentalized and incorporated information are presented, and the pattern of priming in a speeded lexical decision task is used to distinguish between these two classes of model.