Abstract

When we think of analogies, we often imagine cases where drawing on complex concepts from one domain helps to extend our comprehension of concepts from a different domain. As such, the use of analogies is often associated with creativity and problem-solving. However, there is evidence that the inferences derived through processes of analogical thinking are applicable in other contexts that might not at first appear to involve analogies. For instance, there is evidence that analogical thinking might play an important role in language acquisition (cf. Gentner & Namy, 2006). Furthermore, as discussed below, there is evidence that the processes underlying comparison and the identification of differences are supported by the same framework that is used for drawing analogical inferences. It appears then that the usefulness of analogical processes extends beyond the boundaries of creativity and inferences derived out of complex representations. We suggest that the theories of analogical thinking are useful not only in explaining creativity and problem solving, but also as a tool for understanding the processes underlying comparison, both conceptual and perceptual (Gentner & Markman, 1997). Specifically, we will suggest that one such theory of analogies, be used to describe not only the processes of analogical reasoning but also those involved in comparison. According to Structure-mapping Theory, analogies are understood via process of structural alignment. The alignment of two representations is assumed to proceed via a local-to-global process that begins by placing identical elements (attributes and relations that exist in both representations) into potential correspondences. These correspondences form an initial set of local matches. These local matches are coalesced into structurally consistent connected clusters (called kernels), which are merged together to form one or a few structurally consistent global interpretations. This global alignment facilitates the generation of analogical inferences, but also reveals structural commonalities between the two representations as well as alignable differences. Therefore this process can be used for the identification of specific differences between representations as well as for the generation of Furthermore, the more similar the two representations are, the easier the alignment process becomes, resulting in the faster identification of specific differences. However, in cases where the identification of specific differences is not required, such as when making simple judgments of " same " or " different " , it is possible to shortcut this process by employing a simple heuristic (Markman & Gentner, 2005). Basically, the lower the number of initial local matches, …