The mechanisms of imitation

Imitation is a process by which individuals learn to perform a behavior pat tern as a result of observing another animal performing a similar action. D espite a century of research and a great deal of interest, the processes un derlying imitation remain unknown. On the basis of neural network simulatio ns, we put forward a hypothesis for the computational processes that underl ie imitation. This analysis suggests that imitation may occur because the o bservational experience (which includes both the demonstrator's performance of the target behavior and contextual cues) is composed of stimuli that ha ve features in common with cues associated with an individual's past experi ence. This leads us to propose a stimulus generalization hypothesis for the mechanism underlying imitation. We suggest that a stimulus generalization explanation for imitation has been prematurely rejected, because the comple xity and significance of this prior experience have not been fully apprecia ted. The hypothesis may be able to account for many reported cases of imita tion in animals and humans. This analysis allows us to make a number of pre dictions which are accessible to empirical test.