cover from a food dish. Those budgies that saw a demonstrator use its foot to remove the cover subsequently used their feet to remove the cover; those that observed a demonstrator use its bill to peck or pull the cover off did the same. Students in my laboratory have repeated the Dawson and Foss experiment ( Galef, Manzig, & Field, 1986) and found weaker but similar effects. The Dawson and Foss procedure of requiring imitation of motor patterns, rather than imitation of the location in which an act is to be performed or the stimuli to which behavior is to be addressed, goes a long way toward solving problems of control for other types of social learning. Positive outcomes are, therefore, more clearly indicative of "true imitation," of "learning to do an act from seeing it done," than positive outcomes in more commonly employed procedures. Dawson and Foss's work with budgies seems among the most convincing of the scores of laboratory experiments on learning by imitation. I would encourage the adoption of their paradigm for use with other species and behaviors in future work on the question of the occurrence of true imitation in animals. (See also Denny & Clos, this volume.)
It is somewhat surprising that almost 100 years of study of social learning in animals has failed to produce a clear answer to the question of whether animals can in fact learn "to do an act from seeing it done," whether they can, in Thorndike's sense, truly imitate. Although a few studies of social learning (e.g., John, Chesler, Bartlett, & Victor, 1968; Herbert & Harsh, 1944) seem to provide unequivocal evidence of imitation learning, successful experiments have rarely been independently replicated and the majority of attempted demonstrations of imitation have failed to provide convincing evidence of the phenomenon.
There is still a pressing need for investigations that proceed beyond identification of an effect of social interaction on behavior acquisition to analysis of the conditions under which such social learning occurs. It is clear from information collected both in field and laboratory ( Galef, 1976) that social interaction can play an important role in modifying the behavior of animals, both facilitating the acquisition of useful patterns of behavior and increasing the probability that behaviors already in an individual's repertoire will be performed. Although such observations may in themselves satisfy chose interested in demonstrating functions of social interaction in the production of adaptive behavior, they represent a challenge to students of causation or mechanism. Analysis of the behavioral processes supporting social influences on behavior has not proceeded far beyond the listing of examples undertaken by Thorndike in 1898. Our vocabulary may be richer than Thorndike's but our level of understanding of the behavioral processes involved in social learning remains similar to his.
Study of social learning offers opportunities both for integration of functional and causal analyses of behavior and for synthesis of field and laboratory