SOME REMAINING PUZZLES CONCERNING ANALOGICAL REASONING AND HUMAN ABILITIES

This thesis investigated three puzzles arising from Sternberg's (1977a, 1977b) componential analysis of analogical reasoning. First, the contribution to latency of each of seven components--encoding, inference, mapping, application, justification, comparison, and response--was assessed. This assessment was an advance over previous studies, because each component difficulty measure was explicitly manipulated by the experimenter, rather than being estimated on the basis of subjective ratings. Also, all seven components were measured, rather than a subset (as was true in some previous research). Second, the predictability of stimulus attributes (i.e., visual features) relevant for analogy solution was varied to assess the effect of attribute predictability on correlations between the latency of analogical reasoning components and tests of psychometrically determined subject abilities. Third, tests of fluid ability and crystallized ability (Horn & Cattell, 1966, 1967; Horn, 1968) were administered to determine subjects' reasoning abilities. Also included were tests of perceptual speed.;Results from 24 subjects who participated in ten hours of individual testing suggested the following: In reference to the first goal, the justification component was found to be by far the most lengthy (1423 msec.). Other components ranged in latency from 664 msec. (encoding) to -142 msec. (comparison, estimated as a negative latency to indicate a "time savings").;The predictability manipulation did not show an effect on the relationship of most components to reasoning ability with one notable exception. Response was found to be strongly related to reasoning in the low predictability condition, but not related to reasoning in the high predictability condition. This was attributed to a metacomponential difference between the two conditions. Also, mapping was found to be a necessary part of the componential model for the low predictability analogies, but not a necessary part for the high predictability analogies. Since the necessity of the mapping component has been a source of some dispute, this specifies a set of circumstances under which mapping may be expected to be present or absent.;Measurement of fluid and crystallized intelligence proved revealing. The feature transformation detection components (inference, mapping, application, justification, and comparison) were found to be strongly related to fluid ability, but unrelated to crystallized ability. In addition, all components were unrelated to perceptual speed ability.