Children's performance on "animal tests" of oddity: implications for cognitive processes required for tests of oddity and delayed nonmatch to sample.

To investigate the ontogenesis of oddity learning, children (16 to 102 months of age) and adults were tested on two versions of the oddity task using non-verbal procedures originally developed for monkeys. On the standard, "one-part" or "simultaneous" oddity task (Experiment 1), young children (16 to 74 months of age) performed more poorly than older children (81-102 months of age) who were as proficient as adults. The delayed mastery of one-part oddity contrasts to mastery, at much younger ages (3 to 4 years of age) of a similar, but two-part task, delayed non-match to sample (DNMS) (Overman, 1990). In Experiment 2, those children from the first experiment who had difficulty in learning the one-part oddity task were tested on a two-part oddity task, and a subset of the subjects was retested on the one-part oddity task, and, finally, given verbal instructions for the one-part oddity task. The two-part oddity task was mastered significantly more rapidly than the previous one-part task; however, children's performance dropped significantly when tested on the one-part oddity task, and finally, children rapidly mastered the one-part oddity task when given verbal instructions. The data suggested that (a) children used different strategies to solve the different versions of the oddity task, (b) the solution for the two-part-task appeared earlier in life than the solution for the one-part task and did not involve the use of the concept of "oddity relations", and (c) in tasks in which stimuli are shown twice, behavior may come under control of the absolute properties of the exemplar stimulus via a simple "win-shift" pattern of behavior. In contrast, in tasks in which all stimuli are presented simultaneously, behavior may be controlled by stimulus relations, the analysis of which has a protracted ontogenetic development.