This paper investigates the research and development activity of heterogeneous and rationally bounded firms. The latter conduct this activity through in-house research and by collecting information originating in other firms' spillovers. Thus, research and development activity owes both to independent searching and to interaction diffusing information. We study the conditions under which this idiosyncratic effort yields effects that have either local, system-wise negligible impacts or cumulate to generate significant aggregate ones. In the latter case, global effects feed back upon the incentive to innovate and therefore on the strength of local interaction as well as on autonomous research efforts. It is these dynamic forces that we model. We compare cases in which significant aggregate effects do emerge with cases in which they do not and study their outcome on innovation-directed investment and on long-term growth.