An Investigation of the Inference and Mapping Components of the Componential Theory of Analogical Reasoning.

Sternberg\u27s componential theory of analogical reasoning (1977) provides a series of steps that may be followed in order to solve analogies. According to Sternberg, when given a simple analogy (in the form A:B::C: (D$\sb1$, D$\sb2$, D$\sb3$)) the reasoner: (a) encodes the terms of the analogy, (b) infers the relation between A and B, (c) maps the relation between A and C, (d) applies a relation analogous to the inferred one from C to each answer option, choosing the closer option, and (e) responds. The purpose of this research was to further investigate the inference and mapping processes of the componential theory of analogical reasoning. Subjects were 41 eighth graders of average and above average ability and 41 adults from two undergraduate secondary reading courses. Each subject was required to solve eight target analogies and, in a separate two-part task, to identify the relation(s) within 16 corresponding word pairs. In order to counterbalance the two tasks, subjects were randomly assigned to the analogies first or relations first condition within intact classrooms. In the analogy task half the analogies to be solved were presented in recognition format and half in production format. In the relation task each item represented the word pair corresponding to the inference or mapping relation in the total analogy. Subjects were required to generate a list of possible relations within each word pair. Data were analyzed using a 2 $\times$ 2 $\times$ 3 mixed AVOVA (group $\times$ mode of response $\times$ relations known). A statistically significant main effect for the number of relations known when an analogy was correctly solved was found, p $\u3c$.0001. A statistically significant interaction between group and number of relations known when an analogy was correctly solved was also found, p $\u3c$.0001. There was no statistically significant difference in the probability of accurately solving an analogy whether both the inference and mapping relations or either of them were known. The data obtained here supports the contention that the reasoner is not required to both infer A-B and map A-C before correctly solving an analogy. Further, no single relation is more valuable than another for accurate analogy solution