Abstract

mappings (Clement & Gentner, 1991; Gentner, Rattermann, & Forbus, 1993; Holyoak & Koh, 1987). For example, when given analogies to use in solving problems, people are typically fairly selective about choosing analogies that have genuine structural overlap with the target problem (Bassok & Holyoak, 1989; Holyoak & Koh, 1987; Novick & Holyoak, 1991; Novick & Tversky, 1987; Ross, 1987; Ross, Ryan, & Tenpenny, 1989). We have simulated subjects' retrieval patterns with the MAC/FAC simulation (Many Are Called but Few Are Chosen), in which a first stage retrieval process carries out a wide, computationally cheap and structurally insensitive search for candidate retrievals and a later stage (the FAC stage, essentially SME) performs a structural alignment over these candidate analogues (Forbus, Gentner, & Law, 1995). This system does a good job of capturing the phenomena: Retrievals based on surface similarity or on overall similarity are common, and retrievals based on purely relational similarity-the purely analogical retrievals that strike us as clever and creative-also occur, but rarely. Thus we might be tempted to conclude that Kepler differs most radically from others in his fertile access to various prior analogues (Process a). It is certainly plausible that what most distinguishes highly creative thinkers is a high rate of spontaneous analogical retrievals. But another possibility worth considering is that it may be the mapping process (Process b) that most differentiates highly creative individuals. There are two reasons for this speculation. First, the more energetic the mapping process, the more each analogy is likely to reveal its full potential set of inferences. Second, and less obvious, we conjecture that intense mapping may promote fertile access. For if access to prior material depends on a common encoding (Forbus, Gentner, & Law, 1995), then the highlighting, inferencing, and re-representing carried out in the course of pushing analogies may benefit subsequent memory access, as such activities tend to increase the scope of common internal representations (Gentner, Rattermann, Markman, & Kotovsky, 1995 ). This account is consonant with Seifert, Meyer, Davidson, Patalano, & Yaniv's (1995) prepared-mind perspective on creativity, and also with Gruber's (1995) discussion of Poincare. Thus, we suggest that a major reason for Kepler's high rate of ana-