Use of multiple exemplars in object concept training: How many are sufficient?

Abstract

In recent years, concern has developed regarding the degree to which training with a limited set of objects will result in generalized responding to novel objects of the same category. The purpose of this study was to expand our understanding of the benefits of training with multiple exemplars. Specifically, training with five as opposed to three good examples was investigated. The difference in the averaged correct response to novel objects between conditions was not significant (p = .08); however the arithmetic difference between the two conditions indicated higher levels of generalization following training with five examples for five of the six subjects. The results are discussed in reference to implications for teaching and for further research regarding the benefit of using multiple exemplars when teaching natural object categories.