Goal-directed and automatic imitation
Humans have an automatic tendency to imitate other people's gestures, postures, accents, and other behaviours. This copying is often described as an unintentional matching of observed movement kinematics to actions that one could carry out oneself, supported, perhaps, by an innate mirroring network. Indeed, it often happens outside of awareness, when attention is directed elsewhere, and under high cognitive load. This conceptualisation contrasts sharply with goal-directed imitation, or "imitation proper", as it is typically studied in children, which describes an a goal-directed, controlled behaviour in which the imitator actively attempts to achieve the same outcome as the model by copying their behaviour.
This project tries to uncover the mechanisms behind automatic and goal-directed imitation, and to test whether both can be described within a common framework. To this end, we are testing both automatic imitation in adults and goal-directed imitation in adults and children. Our recent studies have indeed revealed such common mechanisms, showing that hierarchical action plans that link intended outcomes to required motor programs underlie both.
Studies on goal-directed imitation have shown that watching others to imitate them later involves the online creation of motor plans, in the same way as when planning an action or mentally rehearsing it. As a consequence, the relevant effector systems are available for other processes, so that watching a hand-action impairs usage of own hands and but not their feet, and vice versa when watching a foot action (Bach, Allami, Tucker & Ellis, 2014). Studies in children (with collaborator Marina Wimmer and the Plymouth Babylab) showed that these motor plans are hierarchical, tying these motor actions to the outcomes they produce. When executing the actions later, children would only incorporate these action components into their own behaviour that were, in the models' actions , tied to action success. For example, when watching a model throw a diskus, children would only copy their stepping movement, when stepping lead to better throws of the model.
Our data on automatic imitation suggest that it may be as much under top-down control as imitation proper and governed by similar hierarchical action representations. For example, we have found that it is not fully automatic but requires attention to be directed to the acting body part (Bach, Peatfield & Tipper, 2007). Moreover, it is as outcome driven as imitation proper. Even for actions that are not yet completed, those that are appropriate to achieve their outcome (e.g. a hand open wide enough to grasp a large object) are copied more than less appropriate actions (e.g. a hand that is not open wide enough, Bach, Bayliss & Tipper, 2011). A current study shows that, these outcomes do not even need to be causally related to the action, but can be one of its incidental consequences (e.g. drawing a winning card in a game).
Together, these results suggest that automatic and goal-directed imitation are based on similar hierarchical action representations that link motoric and higher-level goal information. These integrated action representations are created spontaneously when watching others act and, once established, elicit automatic imitation, so that only those with positive outcomes will be imitated, and others are avoided.
Sample experimental paradigm
In this experiment, developed with Marina Wimmer and supported by a British Academy/Leverhulme Trust grant, a child plays a game of discus on the Microsoft XBox, while their body movement is unobtrusively recorded via Microsoft Kinect. We compare the kids' throwing kinematics both before and after watching a model perform the throws. Crucially, the child watches the model performs the actions in two different ways, for example, throwing with or without stepping forward. We vary, across participants, which mode of action produces the better results in the game. MS Kinect allows us to measure whether the child, from watching the model, establishes a causal model of the action, linking the different action aspects to the likelihood of success. If this is the case, they should copy specifically those action aspects that have lead to success for the model.
Wiggett, A.J., Hudson, M., Clifford, A., Tipper, S.P. & Downing, P. (2012). Doing, seeing, or both: Effects of learning condition on subsequent action perception. Social Neuroscience, 7(6), 606-621. PDF
Bach, P., Bayliss, A.P., Tipper, S.P. (2011). The predictive mirror: interactions of mirror and affordance processes during action observation. Psychonomic Bulletin & Review, 18(1), 171-6. PDF
Wiggett, A, Hudson, M, Tipper, S. & Downing, P. (2011). Learning associations between action and perception: Effects of incompatible training on body part and spatial priming. Brain and Cognition. 76, 87-96. PDF
Bach, P., Peatfield, N.A., & Tipper, S.P. (2007). Focusing on body sites: the role of spatial attention in action perception. Experimental Brain Research, 178, 509-517. PDF
Bach, P. & Tipper, S.P. (2007). Implicit action encoding influences personal-trait judgments. Cognition, 102, 151-178. PDF