Unities in inductive reasoning.

Two experiments sought to discover sources of communalities in performance on three inductive reasoning tasks: analogies, series completions, and classifications. In Experiment 1, 30 subjects completed an untimed pencil-and-paper test in which they were asked to solve 90 induction items, equally divided among the three kinds of induction items noted above. The subjects' task was to rank-order four response options in terms of their goodness of fit as completions for each particular item. Data sets for the three tasks were highly intercorrelated, suggesting the possibility of a common model of response choice across tasks. Moreover, a single exponential model of response choice provided a good fit to each data set. The single parameter estimate for this model was roughly comparable across tasks. In Experiment 2, 36 subjects completed a timed tachistoscopic test in which they, too, were asked to solve 90 induction items, equally divided among the three kinds of induction items noted above. The subjects' task was to choose the better of two response options as a completion for each particular item. Data sets for the three tasks were again highly intercorrelated, suggesting the possibility of a common model of real-time information processing across tasks. Moreover, a single linear model of response times provided a good fit to each data set. Three of four parameter estimates for this model were roughly comparable across tasks.