The ability to predict the actions of other agents is vital for joint action tasks where the actions of two actors must be coordinated in a timely fashion. Recent accounts suggest that action prediction might be implemented by a system that allows actors to generate predictions about perceived actions by emulating those actions using their own motor systems. In particular, it has been suggested that actors are able to use their own inverse models for action to predict what actions another agent might select in order to realise a particular goal, and that actors are able to use their own forward models for action to generate predictions about the unfolding dynamics of perceived actions. The body of this thesis is divided into three sections (Part II–IV) and examines the empirical support for these hypotheses as well as their theoretical implications. In Part II, Chapter 2 confirms that observers are able to generate more accurate predictions when their own movement dynamics more closely match the movement dynamics of the observed action. Further more, the results also suggest that the output of the prediction system is available in a timely fashion suggesting that it can be used for joint action coordination. Chapter 3 examines what factors might assist observers to reconstruct the motor codes responsible for generating perceived actions by specially examining the inìuence of limb and joint in formation on the prediction of perceived actions. The results suggest that limb and joint information allows observers to generate more accurate predictions about observed actions, but that this effect is limited to those observers who have motor experience with the observed actions. Chapter 4 rules out an alternative explanation for the results obtained in Chapter 2, that differences in action prediction can be explained by self-similarity in action production between the action production and subsequent action prediction phases of the experiment. In Part III, Chapter 5 presents a framework for understanding joint action coordination wherein an attempt is made to extend well understood mechanisms of action control into the domain of joint action. This framework might assist with trying to understand the phenomenology of joint action and experiences such as group flow. In Part IV, Chapter 6 examines the role of action prediction mechanisms in a domain other than traditional conceptions of joint action. Specifically, it examines the role of action prediction and action synchronisation in musically evoked emotion. Finally, the implications of this thesis are discussed as they relate to our understanding of the nature of joint action, phenomenology of joint action, functions of the mirror-neuron system, and the role of sensorimotor systems in cognition.
Predicting the actions of other agents