| Abstract: | The hypothesis-testing behavior of kindergarten children in discrimination learning was studied in a factorial design with two temporal placements of introtact probes (pretrial and posttrial) and two types of pretraining (unidimensional and multidimensional). Pretrial probes consisting of a request for the child's current hypothesis were administered in a conventional manner at the beginning of each trial with the stimuli in view and prior to the choice response. Posttrial probes were administered also with the stimuli in view, but after the choice response had occurred and feedback information had been provided. Learning-to-learn experience in solving three pretraining problems was given either with unidimensional simultaneous problems or with more complex multidimensional problems like the criterion tasks. The results indicated that posttrial probes produced superior local consistency, win-stay, and lose-shift probabilities, relative to pretrial probes, for both types of pretraining. Posttrial probes, however, facilitated learning and the testing of valid hypotheses only for multidimensional pretraining. These results are consistent with the hypothesis that posttrial probes constrain the child to be locally consistent and therefore improve short-term efficiency in hypothesis testing under both pretraining conditions. However, posttrial probes produce an improvement in long-term efficiency, and therefore in learning, only when other components of a successful strategy are acquired as in multidimensional pretraining. |
