Cognitive component models for psychometric analogies: Conceptually driven versus interactive process models

Abstract Analogical problem solving, particularly with verbal analogies, has had a major role in shaping theories of intelligence and in providing measurements of it. A variety of cognitive models have been proposed. The models range from simple conceptually driven models, which have had excellent success for modeling laboratory analogies, to more complex interactive models, which are believed necessary to account for performance on psychometric analogies. Three experiments on psychometric analogies examine (1) the explanatory power of the simple conceptually driven models, and (2) the extent and nature of the interactive effects. A modified version of a conceptually driven model, which includes an interactive structural mapping component, was hypothesized to describe the pattern of results obtained from the experiments.