In recent years, the stakeholder approach has been widely applied in the debate on corporate social responsibility (CSR). Although many authors of this approach have reviewed many elements of the model, they have unconditionally accepted several criteria assumed by Freeman (1984) to identify stakeholders. In general, stakeholder authors have assumed that (a) the company establishes dyadic relationships with other agents, and (b) decisions made by a company only have foreseen and direct effects on other agents. These criteria have enabled researchers to understand simple processes. However, they have also prevented researchers from explaining how action comes about, and how responsibility is shared, in many complex processes taking place in contemporary societies. Such complex processes involve many agents, and each decision can generate unexpected effects which accumulate or disseminate. Furthermore, the normative structure governing these processes can affect and/or be affected by the actions of agents. In this study, we propose new criteria to expand the stakeholder model and facilitate the study of CSR in such processes.