Two issues in action imagery: Inhibition and errors
Abstract: Action imagery can be defined as an internal simulation of actions without actual movements. From a computational viewpoint, action simulation during action imagery might entail inverse and forward models, similar to actual actions. Based on this framework I will outline two issues which I think are promising for further research to enhance the understanding of processes underlying action imagery. First, I will talk about inhibition. If motor commands are generated during action imagery, the question arises how the execution of actions is prevented. Several forms of inhibition (e.g., tonic global inhibition, phasic global inhibition, and effector specific inhibition) seem to contribute to prevent actual actions during action imagery. They follow a different time course and are flexibly applied to prevent actual actions in the most efficient way. Second, I will talk about errors. Actual actions do not always result in the intended consequences, sometimes errors occur. Such discrepancies between the intended and the actual action effects are critical for learning as they provide feedback about the adequacy of the action that was produced. In action imagery, errors can be imagined. However, they occur less often, and only certain types of errors seem to be noticed. It seems that not all aspects of an action are predicted by forward models or that the prediction of action effects is imprecise in action imagery. In conclusion, the conceptualization of action imagery as a simulation of actual actions seems to represent an adequate framework for action imagery. However, in action imagery some crucial differences to actual actions exist and some processes are specific to action imagery.