In analyses of the motor system, two hierarchies are often posited : The first – the action hierarchy –is a decomposition of an action into sub-actions and sub-sub-actions. The second – the control hierarchy – is a presumed hierarchy in the neural control processes that are supposed to bring the action about. A general assumption in cognitive (neuro)science is that these two hierarchies match. I will show, however, that two different structuring principles are used to construct the two hierarchies. The action hierarchy is constructed using a part-whole relation between the elements in the hierarchy, while the control hierarchy is structured around causal relations between the elements. I will argue that these two structuring principles are, however, incompatible, and that neither structure can account for both hierarchies. As a consequence, the action hierarchy and the control hierarchy need not, and probably will not, match entirely, suggesting that matching these to hierarchies might not be a viable strategy for empirical research. In addition, an alternative principle to structure a hierarchy can be derived from models in artificial intelligence: higher-level entities are represented more stably than lower-level entities. A hierarchy based on this principle is not subject to the problems that plague the traditional accounts.