Referring to recent research involving France, Germany and Japan, we will propose some reflections on the "actors of innovation" within firms. This appears to be one of the shortcomings of the expert reports we have just mentioned, which seem to me to adopt the positions of classical economics, where the "actors" are most often absent, to the benefit of structural data, often macro-economic data, which can be modelled or formalized in econometrics. In this case, concrete "actors" are reduced to abstract "agents" inserted into production relationships in terms of investment or resources -even if qualified as “human”; they are thus de-socialized and take on a universal character, just like the economic laws that govern them. In this perspective, it is not surprising that innovation itself has been treated according to the same logic. However, new approaches in the economics of innovation have developed, more open to the dynamic aspects of innovation processes, referring to the forms of collective learning among the different actors, and to the structuring effects of the institutions that contribute to the diversity of "national innovation systems" (Nelson 1993, OECD 1997).