My PhD thesis related the growth of a large social system devoted to the treatment of personal problems; i.e. problems successively labelled as madness, mental illness and mental health problems, from the early fifties to the present days. That system involved heterogeneous networks of actors, including scientific experts, established professions, social movements, policy makers, services users and international organisations; instruments and knowledge ranging from psychoanalytical theories to biological knowledge and models of governance. It meant to explain how institutional change happened in that complex social system, by studying past and ongoing reforms, considered as interrelated steps towards complete paradigm shift, including shifting policy means, policy objectives and social organisation. By relying on in-depth analyses of five past reforms, it conceptualised the system as composed of interrelated ecologies, corresponding to different kinds of knowledge of personal problems, whose development was directed by protective and expansion strategies used by two coalitions of actors, holding different kinds of resources to influence the change process. The traditional coalition was embedded in the Belgian institutional system; it referred to medical knowledge of personal problems; and held many institutions delivering residential treatments. By contrast, the reformist coalition was connected to international professional and policy networks stimulating change in OECD-mental health systems; it referred to practical knowledge of social psychiatry and evidence produced by international organisations as the World Health Organisation; and it held non-profit associations delivering community treatments. Cross-regulations exerted through joint-participation of those coalitions in successive reforms caused rapid changes in the system’s structural configuration while hindering change in its social organisation. Thus, we suggested thinking of the issue of change in the system as consisting in setting conditions in which the reformist coalition might extensively use its resources in conducting a new reform starting in 2010. By relying on that assumption, I presented three case studies analysing the devising of that reform at the policy level and its implementing through local networks. Those case studies drew attention to the kinds of knowledge used by key actors and to the way in which they used it in joint-attempts to take the leadership of the reform. Three main findings resulted from those case studies. First, the designing of the policy guide framing the reform consisted in assembling different kinds of knowledge together, including policy learning achieved through past changes in the system and knowledge of alternative care models implemented in OECD-countries, in a way that encouraged sustained enactments of knowledge specific to the reformist coalition, while decreasing the relevance of resources specific to the traditional coalition. Second, local meetings caused by the implementing of the reform enabled adaptations of knowledge embodied by the participants, i.e. multiple actors representing different ecologies and members of one of the two coalitions, to enacted knowledge, i.e. knowledge collectively produced by thinking about concrete means to implement the reform locally. Third, enacted knowledge caused, in turn, adaptations of the care model inscribed in the policy guide to local particularities. Those adaptations did not prevent, however, the global philosophy of the new reform, inspired by social psychiatry, from pervading in local networks, among other by being inscribed in operational documents resulting from meetings. Thus, by following the policy guide through local networks where it was translated into concrete practices, I have been able to indicate new directions in the global process of change in the system, towards a complete paradigm shift, from medical to social psychiatry, and to highlight social and learning processes making it possible.