According to Raskin (2000), the way we interact with a product, what we do, and how it responds are what define an interface. This is a good starting definition in one important respect: an interface is not something given or an entirely predefined property, but it is the dynamic interaction that actually takes place when a product meets the users. More precisely, an interface is that interaction that mediates the relation between the user and a tool explaining which approach is necessary to exploit its functions. Hence, an interface can be considered a mediating structure. A useful exemplification of a mediating structure is provided by the so-called stigmergy. Looking at the animal-animal interactions, Raskin (2000) noted that termites were able to put up their collective nest, even if they did not seem to collaborate or communicate with each other. The explanation provided by Grassé (Susi et al., 2001) is that termites do interact with each other, even if their interactions are mediated through the environment. According to the stigmergy theory, each termite acts upon the work environment, changing it in a certain way. The environment physically encodes and stores the change made upon it so that every change becomes a clue that affects a certain reaction from it. Analogously, we might claim that an interface mediates the relation between the user and a tool affording him or her to use it a certain way1. Understanding the kind of mediation involved can be fruitfully investigated from an epistemological point of view. More precisely, we claim that the process of mediating can be understood better when it is considered to be an inferential one.