Drawing on a review of recent work conducted in the area of pattern generalization (PG), this paper makes a case for a distributed view of PG, which basically situates processing ability in terms of convergences among several different factors that influence PG. Consequently, the distributed nature leads to different types of PG that depend on the nature of a given PG task and a host of cognitive, sociocultural, classroom-related, and unexplored factors. Individual learners draw on a complex net of parallel choices, where every choice depends on the strength of ongoing training and connections among factors, with some factors appearing to be more predictable than others.