The evolution of imitation: What do the capacities of nonhuman animals tell us about the mechanisms of imitation?
Ludwig Huber
Department of Neurobiology and Cognition Research, University of Vienna, Austria
In this talk I would like to review reports and present new empirical data from studies with non-human animals that have addressed two distinct problems of imitation research. The one favored by cognitive neuroscientists is the ‘correspondence’ problem, asking how is it possible for actions as seen to be matched with actions as imitated? The other, favored by ethologists and comparative psychologists, is the ‘transfer of skill’ problem, asking how is it possible for novel, complex behaviors to be acquired by observation? Despite a plethora of approaches to, and definitions of, imitation, most scholars agree that when an individual replicates an action that it has observed being performed by another individual it requires a matching system that allows conversion of observed actions by others into actions executed by one self.
In order to answer the question on what neuro-cognitive mechanisms would make imitation possible, I critically review the outcomes of so-called "Do-as-I-do" experiments. Because animals were trained to understand and obey a copying command, a great variety of actions could be compared with respect to the matching degree with which they were replicated by the animal observer. This paradigm enabled the investigation of performance differences when using either body-centered or object-centered actions, as well as either familiar or novel actions. As such it is very informative with respect to the underlying imitation mechanism. In addition to the already existing data from studies with great apes, parrots, dolphins and dogs, I will present novel data acquired recently in our lab. In this study we tested a dog with a large number of actions ranging from excessively trained ones to very novel and peculiar ones. Furthermore, I will discuss the methods to evaluate the matching degree or copying fidelity of imitative performances, including Voelkl and Huber's (2007, PLOS One) quantitative/mathematical approach with marmosets and the neuro-cognitive mechanisms that can or cannot explain the high matching degree found in this study.
The second main issue of contemporary imitation research in non-human animals is concerned with the ‘transfer of skill’ problem, asking about the social transmission of behavioral innovations and the cognitive mechanisms that may or may not support cultural learning. The currently dominant view is that observers can only learn novel, complex behaviors if they are able to understand the goals and intentions of the demonstrator and can use context information to select the most efficient means to solve the problem themselves. Recent evidence from chimpanzees and dogs suggest that this form of imitation is not restricted to humans. In our lab, dogs have proved to adjust the extent to which they imitate to the circumstances of the action (Range et al. 2007 Curr. Biol.). This demonstration of intentional control of imitation fits into the recent evidence of intention reading in non-human animals and disproves the claim that it is a unique feature of human imitation.
In conclusion, I shall propose that non-human animals are able to use at least two different forms of imitation. One is an automatic, non-selective and non-intentional copying of movement patterns, which may be based on the same mirroring mechanism as is used in song learning in passerine birds and cetaceans. The other is an intentional, selective process that requires some causal understanding and 'theory of mind' abilities.