The question of which kind of entities are really interacting is specifically central and difficult in the study of cooperations between research laboratories and firms, because these cooperations involve various kinds of agents. In France, when there is a contract of cooperation between a research laboratory and a firm, this contract is signed both by the firm and by a big research organization like a university or the CNRS. The contract generally mentions the name of the research laboratory, as it is defined by the university or the CNRS, and also the name of the research leader. For the research side, there are three different levels of action : the liable organization ; the research laboratory as an internal organization inside the university or the CNRS ; and at last a single researcher. It is also known (Amiot, 1996) that there is fourth level, which is the small research team (a dozen of researchers) working with the scientific leader. For the firm, there are also several levels, especially for a group : the group ; a local establishment really cooperating with the laboratory ; a specific team within this establishment ; and at last, the industrial leader of the cooperation. Which is the appropriate scale of analysis for the study of these kind of interactions between heterogeneous agents ? What are the agents really interacting (individuals, small teams, organizations) ? What are the processes by which decision making shifts from a level to another ?