Relational similarity as a defining feature of analogies: A questioning from the categorical approach

Abstract

The structure-mapping theory posits that the central component in establishing an analogy is to discover that two situations share a system of similar relations. Two experiments were performed to evaluate the hypothesis of the categorical approach that relational similarity does not constitute a necessary nor a sufficient condition for people to judge that two facts are analogous, and that the determinant factor is whether or not the compared facts belong to a common schema-governed category. In both experiments, 20 psychology students received a base fact and two target facts describing simple events, and they had to evaluate how analogous the base and each of the target facts were. In Experiment 1 both target facts included relations that were not similar to those included in the base fact. However, while in the first target fact the object to which the action described by the relation was applied made the whole fact became an exemplar of the schema-governed category applicable to the base, the object of the second target fact made it became an exemplar of an alternative schema-governed category. Experiment 2 has the same structure than the first one, with the only difference being that base and target relations were similar. In both experiments results showed that the first target fact was considered analogous to the base fact, but not the second one. This evidenced that relational similarity is not necessary (Experiment 1) nor sufficient (Experiment 2) to consider that two facts are analogous, and that, instead, the criterion employed by people to determine that two facts are or are not analogous is category membership. We discussed the implications of these results for the development of theoretical and computational models on analogical thinking.