An alternative explanation for the early session effect (i.e., Group Same vs. Group Different) is that the demonstrator may have continued to instruct the observer where to peck by orienting discriminatively to the observer's keys on red and green test trials. For example, if a Group Same demonstrator saw the observer's keys illuminated red, it may have raised its head toward the observer's top key with an attempt to peck it, which could have caused the observer to orient to, and spontaneously peck the top key (correct location). If the demonstrators of Group Different behaved similarly on say "red trials," the observers may have oriented to and pecked the bottom key (incorrect location). Further research is needed to determine whether the observers of Groups Same and Different would transfer differentially when instructional control is prevented by using a deferred test procedure.
The mechanisms that can produce imitative behavior in pigeons range from simple elicitation of pecking by a feeding conspecific to copying of a response pattern that is arbitrarily related to an observed consequence. However, it appears that the most complex mechanism of imitative behavior that can be identified in pigeons is stimulus valence transformation. Valence transformation theory holds that the observer's response topography is determined by the learned association involving the manipulandum (predictive stimulus) and the reinforcer, rather than by a perception of the demonstrator's response topography per se. Although some experimenters have confounded response copying effects with instructional effects ( Edwards et al., 1980; Hogan, 1986; Hogan & Fry, 1986; Zentall & Hogan, 1976), valence transformation theory can explain more parsimoniously why the observers pecked the manipulandum since (a) the observer's peck response was not arbitrarily related to the observed reinforcer cues and (b) key pecking by pigeons can be learned rapidly when the "instructional" stimulus is a nonconspecific's action ( Neuringer & Neuringer, 1974; see also Groesbeck & Duerfeldr, 1971).
It is notable that a clear and reliable experimental demonstration of copying an arbitrary response in pigeons remains to be documented. Despite the fact that some experimenters have established conditions in which response copying could have occurred it did not. Epstein ( 1984, Experiment 3), for example, included a control condition in which an observer was permitted to watch a demonstrator that had been conditioned to turn circles, presumably for food reinforcement. The condition was included to test for the presence of spontaneous imitation of the circling response, but none was seen. Instead, the observer pecked the object (a ball) at a relatively high frequency. This pecking was probably induced by the general activity of the demonstrator ( Zajonc, 1965).
I have also casually looked for evidence of response copying and saw none. A pair of pigeons trained to cooperate by simultaneously pecking adjacent key