Experiments are described in which the spatial relationship between a stimulus and respondent is held constant in terms of visual and body-centred coordinates, while the complexity of the response is manipulated. It is demonstrated that the degree of complexity of an action directed to the same spatial location determines the level of negative priming observed. This result supports the notions that (1) inhibitory selection mechanisms act on action-centred representations, and (2) the level of inhibition is reactive to the relative potency of the evoked action. The results are also discussed in terms of alternative explanations of negative priming. It is concluded that the results are inconsistent with theories that do not involve inhibitory selection mechanisms.