(How) do flood prone cities build resilience? Towards a learning-sensitive analytical framework.

Urban resilience is almost unanimously identified as an inherently positive guiding principle in the risk reduction policy field. However, limited attention is paid to the learning dimension of resilience-building. To help bridge the gap, this research explores the interplay of learning processes, learning outcomes, and institutional action and investigates how capacities for reflection and collaboration develop in the face of wicked, risk-related problems. The study focuses on post-flood reorganisation processes developed in cities repeatedly affected by more or less severe flood events. It proposes and illustrates an analytical framework to capture dynamics affecting policy-making processes that tackle risk in contexts characterised by high complexity, uncertainty, and political pressure. The framework is tested by looking at reorganisation attempts carried out to face the "eternal flooding" of the Seveso Torrent, which has affected the Northern neighbourhoods of Milan for decades, with over 100 events recorded after 1976. Results from the Milanese case highlight the existence of a learning and policy deadlock, where the impossibility to amend "historical mistakes" in decision-making and patterns of spatial, discursive and governance fragmentation hamper (reflective) action and contribute to policy inertia. This research provides a theoretical background and methodological insights for investigating risk-reduction attempts in their interplay with framing and knowledge- related dynamics and broader relational, discursive, and regulatory factors, thus providing insights into the field of policy analysis.