Considerable evidence supports the effectiveness of close comparison of examples as a means to promote the induction of schemas that support generalization, especially to novel cases that require far transfer. The ease of comparison would appear to be maximized by presenting the to-be-compared cases in close spatial and temporal proximity. However, find-ings from a number of recent studies have been interpreted as evidence that induction is fostered not by presenting training cases for a single category together (massed practice), but ra-ther by presenting them in an interspersed fashion (spaced practice). We address this apparent paradox in a study in which people are asked to learn the “styles” of furniture de-signs from a small number of examples of different products (e.g., a bed frame) and then classify examples of entirely dif-ferent products (e.g., a chandelier). We contrasted a learning procedure based on comparison of examples presented simul-taneously with procedures involving processing of individual items, either massed or spaced. Study time was minimized, and generalization was maximized, when learning was based on comparison. In a further study we use structural equation modeling to assess the content of the schemas for visual styles that are acquired by comparison processes. We propose that comparison fosters induction, whereas spacing facilitates re-tention and retrieval.