When alignable differences count against analogical relatedness

LUCÍA MICAELA TAVERNINI; MÁXIMO TRENCH; RICARDO A. MINERVINO

San Diego

Conferencia; International Cognitive Science Conference; 2016

University of California San Diego

The structure-mapping approach to analogical reasoning posits that drawing an analogy consists in discovering that two situations share a system of relations and roles, and that semantic differences between the objects that play the role of arguments of corresponding relations (alignable differences) have little or no effect on judgments of analogical relatedness. According to an alternative account (Minervino, Oberholzer & Trench, 2013) people concentrate on relational matches only when failing to represent the compared situations as exemplars of a schema-governed category. When the compared events are assigned to a shared category of events (e.g., two cases of murder), people?s judgments of analogical relatedness are based on how both situations fare along relevant dimensions of that category (e.g., the cruelty of the murders). Following this perspective, differences between the arguments of two situations could potentially affect the entire representation of the events, thus leading to different judgments of analogical relatedness. In the present study we assessed how the objects that take part in two analogous visual scenes influenced judgments of analogical relatedness under two experimental conditions, one emphasizing the fact that the scenes belonged to a common schema-governed category, and the other emphasizing their shared relations. Participants in the ?similar categories? conditions received a scene in which an agent was exerting an action to an object (e.g., a man giving a pair of socks to a woman). This scene was followed by another scene involving similar actions and characters but a different object, and by the indication that the two scenes were cases of the same category of events (e.g., two ?awards?). For half of the participants the object (e.g., a cup) was chosen such that the new action would be comparable to the base action in terms of a relevant dimension of such category (e.g., the price of the award). For the other participants the object (e.g., a computer) was chosen such that the new situation would be less similar to the base in terms of such dimension. Participants in the ?similar relations? conditions received the same pairs of scenes, but framed under a common label that emphasized their shared relation (e.g., two cases of ?giving?).Results showed that when the compared situations are framed under a common relational category, participants are sensible to alignable differences along the relevant dimensions of that category, thus assigning lower ratings to analogies in which the compared situations show contrasting values on such dimensions. In contrast, this manipulation had no effect when the presentation of the scenes did not emphasize their inclusion into a common schema-governed category, but instead emphasized their shared relational predicate. These results support the adequacy of the category-based approach for analyzing a wealth of realistic situations than have been overlooked by standard approaches.