Prediction processes in the acquisition of sequence representations

Action-Imagery-Practice describes the repetitive imagination and Action-Observation-Practice the repetitive observation of an action. Both Action-Imagery-Practice and Action-Observation-Practice are assumed to involve similar motor mechanisms as Action-Execution-Practice, resulting in motor learning. To investigate whether these practice styles differ in the acquired representation types, we compared performance of the practice and transfer hand for same, mirrored, and random sequences in pre- and post-tests. All participants practiced a serial reaction time task to auditory stimuli in ten practice sessions. Five separate groups either physically executed the responses (Action-Execution-Practice), imagined the responses (Action-Imagery-Practice), observed keypresses with an animated hand (Action-Observation-Practice), observed animated keys (Observation-Without-Action), or completed a control condition in which they listened to the stimuli (Auditory-Control). Evidence for effector-dependent representations was obtained after Action-Execution-Practice and Action-Imagery-Practice, but not after Action-Observation-Practice and Observation-Without-Action. Although all groups acquired partial sequence knowledge, sequence recognition was more strongly related to kinesthesis than to the tones alone after Action-Execution-Practice and Action-Imagery-Practice. It is concluded that effector-dependent representations can be acquired via Action-Imagery-Practice even though actual feedback is not available. Conceivably, effector-dependent learning might have been provoked by forward models that predict the action consequences in Action-Imagery-Practice, but not in Action-Observation-Practice, where the action consequences were externally presented on screen.