Traditionally the motor system was thought of as a movement output device that, after brain damage, can impede patients from correctly planning and executing an action. In the last twenty years neuropsychological observations have contributed to develop a new view that attributes higher-level functions to this system. Rapidly, this area of investigation has grown very popular, with motor representations being considered critical for action and object recognition, imagery and even language understanding. At present, we can distinguish between sensorimotor or embodied and disembodied theories. The former predominantly are built on associations of deficits in these abilities while the latter rely on dissociations between deficits. Neuroimaging research has also showed that different action-related functions activate either overlapping brain regions or discrete regions, thus sustaining either association-based or dissociation-based models. Although no convincing theory is to date available to explain both sets of observations, cognitive neuroscientists have begun to explore this intermediate hypothesis space. Here we suggest that concepts like strategy and simulation seem promising in explaining how the cognitive system alternates between action-related functions.