The role of observed errors in performing and learning motor skills

Committing and correcting errors are critical components of motor control and learning. Watching errors can engage error-related processes in the observer, impacting behaviour. Differences between expected and observed actions are thought to elicit a prediction error, updating the observer’s motor plan and causing compensatory biases. These biases are different to better known imitative “contagions”, which can emerge after watching different, yet correct actions. Diverging biases from observation associated with perceptions of errors have not yet been replicated and the necessary conditions are unclear. I investigated how motor behaviour is influenced by correct versus errorful demonstrations and the mechanisms guiding effects. In Chapter 2, different models were compared for observational learning of a golf putting task, isolating watched variability and errors. Although biases were not present, variable demonstrations improved generalizability and when these demonstrations were assumed errorful, self-estimations associated with sensory predictions improved. These data support the hypothesis that observed variability and errors engage different learning processes. In performance contexts in Chapters 3 and 4, perceptions of observed errors were again manipulated. In Chapter 3, experienced golfers showed the expected imitative effects after watching correct demonstrations to a different target but not when these demonstrations were thought to be errors. The absence of biases for errorful demonstrations led to the two aiming studies in Chapter 4. To test what components of the demonstration elicit any biases, perceived correct versus error groups were compared, alongside conditions showing full kinematics or only outcomes. Imitative biases were again shown for the correct demonstrations, independent of seeing the kinematics. However, when these same demonstrations were perceived as errorful, no biases were present. The differing effects of observing correct versus errorful demonstrations indicate that demonstration type matters. However, the absence of compensation across these experiments undermines its reliability and proposed mechanism. Compensatory biases are potentially modified by error surprise and/or individual differences, although more research is required to evaluate moderators. This work adds to theoretical understanding of the processing of another’s error, with implications for practice and performance in social settings.