Action observation and motor imagery in performance of complex movements: evidence from EEG and kinematics analysis.

Motor imagery (MI) and action observation (AO) are considered effective cognitive tools for motor learning, but little work directly compared their cortical activation correlate in relation with subsequent performance. We compared AO and MI in promoting early learning of a complex four-limb, hand-foot coordination task, using electroencephalographic (EEG) and kinematic analysis. Thirty healthy subjects were randomly assigned into three groups to perform a training period in which AO watched a video of the task, MI had to imagine it, and Control (C) was involved in a distracting computation task. Subjects were then asked to actually perform the motor task with kinematic measurement of error time with respect to the correct motor performance. EEG was recorded during baseline, training and task execution, with task-related power (TRPow) calculation for sensorimotor (alpha and beta) rhythms reactive with respect to rest. During training, the AO group had a stronger alpha desynchronization than the MI and C over frontocentral and bilateral parietal areas. However, during task execution, AO group had greater beta synchronization over bilateral parietal regions than MI and C groups. This beta synchrony furthermore demonstrated the strongest association with kinematic errors, which was also significantly lower in AO than in MI. These data suggest that sensorimotor activation elicited by action observation enhanced motor learning according to motor performance, corresponding to a more efficient activation of cortical resources during task execution. Action observation may be more effective than motor imagery in promoting early learning of a new complex coordination task.