Generalized imitation and generalized matching to sample

Almost 40 years ago, Baer and Sherman (1964) pointed out that experimental work in the imitation area often has failed to invoke imitation in its most powerful meaning. In describing the problem, they commented that "the similarity [of the behavior of the model and the behavior of an observer] may lie in the eye of the experimenter rather than in the eye of the observer" (p. 38). Even though they pointed out that imitation itself must involve "generalization" in the sense that it includes novel instances, the title of their article contained the phrase "generalized imitation," which may have contributed to the spread of that term in the behavior-analytic literature. Many behavioral terms are functionally defined and designate a procedure as well as a resulting process or behavioral change. For example, there can be no reinforcement procedure without a corresponding reinforcement process. Similarly, extinction serves both as a name for a procedure and for a resulting behavioral change. However, such dual usage of behavioral terms has not been entirely consistent. Sometimes, the name of a procedure has been kept even when the implied behavioral change failed to occur. For instance, the "identity matching" procedure is often designated as such even in the absence of support for identity as a controlling variable for the subjects' behavior in the task (e.g., Cumming & Berryman, 1965). Similarly, it seems