Evaluation of analogies framed by schema-governed categories: The role of object similarity

MICAELA TAVERNINI; MAXIMO TRENCH; MARIA VALERIA OLGUIN; RICARDO MINERVINO

Paris

Congreso; Analogy 2017, The Fourth Analogy Conference; 2017

University of Chicago & Institut du Cerveau et de la Moelle

Daily-life comparisons can sometimes be done at two different levels: While facts A and B of Figure 1 could be considered analogous to the key fact as cases of the schema-governed category (SGC) awarding (see Minervino, Oberholzer & Trench, 2013), they could alternatively be considered analogous as cases of the lower-level action give. Experiment 1 tested the hypothesis that when two situations are regarded as cases of a SGC, object similarity plays a role in evaluating the quality of analogies. However, we conjectured that the object similarities that count are not of a general type, but similarities in object properties capable of modifying the value that the compared facts displayed on critical dimensions of the framing SGC. Materials were six triplets of images that comprised a base and two target analogs. While one of the targets had general object similarity and distant dimensional value with respect to the base, the other target analog exhibited the opposite pattern. In our example, while the laptop is similar to the mouse as computer devices, they differ in their price, a property that affects the dimension of importance of the awarding. Participants of one group received each triplet framed by a SGC (in the example, awarding), with the task of assessing on a 5-point Likert scale the extent to which they regarded the base analog to be analogous to each of the targets. We expected that under this condition participants would judge Fact B as more analogous to the Key fact than Fact A (a pair of socks is more similar in price to a computer mouse than to a laptop). Participants of a second group received the same scenes but framed in terms of the lower-level action implied (in the example, give). Since evidence shows that general object similarities do not count in analogical relatedness judgments (e.g., Genter & Kurtz, 2006), we predicted that under this condition Facts A and B would be rated as equally analogous to the Key fact, notwithstanding general similarities between the objects of compared situations. A 2x2 ANOVA revealed an interaction between Framing condition and General object similarities: Under the relation framing condition, general object similarities affected analogical evaluation scores, but they had no effect under the SGC framing condition (see Figure 2). As expected, quality evaluations on this group were instead determined by dimensional closeness, which overrode the effect of general object similarities. Although significant, the difference between quality ratings of analogies with close vs. distant dimensional value was not as big as expected. One explanation could be that the materials exhibited an unnatural competition between general similarity and dimensional similarity. To overcome this limitation, Experiment 2 only included the SGC framing condition with target objects that were equally different from the base object in general similarities. Another explanation could be that the task may not have invited a fine-grained evaluation of the comparability of the situations, which we thought that could occur if the comparison was embedded in ?me too? interchanges (Hofstadter & Sander, 2013). Thus, participants in Experiment 2 had to rate the analogability of a pictorial base situation in which, for example, a woman is telling to another one that she won a computer in a contest, and a target situation in which a second woman replied by telling that ?the same thing happened to her? because she won a pair of socks/racing bike in a contest. Compared to Experiment 1, results showed a greater significant difference between quality evaluations of analogies with close and distant dimensional value. Present findings are discussed within the debate about what counts in mapping and evaluation of analogies.