Effects of Nonexamples on Concept Formation

Concept formation is affected by the examples and nonexamples provided during training, but the degree to which examples and nonexamples should differ is unknown. Two experiments compared concept formation when different kinds of nonexamples were used during training. Both experiments included a within-subject comparison of concept formation following three training conditions: a) nonexamples that were more similar to the examples, b) nonexamples that are less similar to the examples, and c) no nonexamples. Arbitrary concepts were trained in Experiment 1 and concepts akin to those that could be taught in a classroom were trained in Experiment 2. Before and after training, tests with untrained examples and nonexamples measured concept formation. In general, concept formation improved when the concept was taught using nonexamples compared to training with only examples, and nonexamples that were more similar to the examples resulted in the highest levels of concept formation. However, for some participants in both experiments, concept formation was similar across conditions, and concept formation may have been influenced by condition sequence. These findings indicated that the relation between stimuli used to teach concepts affects concept formation. It is recommended that both instructors and researchers consider these relations when selecting stimuli to promote concept formation in instructional and experimental arrangements