Abstract: The definition of tool proposed by Beck (1980) is still the one referred to in ethology when discussing the question of tool-use in animals, and its pertinence is rarely questioned. However, observations on technical behaviours in animals have multiplied over the last 20 years, and these have profoundly altered our earlier representations. In the present article, we show that Beck's definition is insufficient and that it does not, in fact, work. More generally, we replace a theory of tools with a theory of mediations of actions to account for technical behaviours in animals. We show that a culturally overcharged notion such as that of tool hinders our perception of the diversity and the complexity of tool uses. By speaking of mediations of actions and not of tools, we eliminate the problem of first defining the pertinent object (is it a tool or not?) and are free to concentrate on the means by which the animal externalizes its actions and thus procures greater means of acting on these within a group. In so doing, we prepare the ground for a genuine evolutionary understanding of the dynamics of actions within a given animal population. Whereas, with a few exceptions, ethologists have always separated the question of techniques from that of social behaviour, we emphasize the importance of an ecology of mediations of actions for understanding the structure and dynamics of animal societies, in particular by attempting to rethink such notions as “culture” in the perspective of a general analysis of mediations of actions.