Template-Type: ReDIF-Paper 1.0 Author-Name: Yeqing Duan Author-Name-First: Yeqing Author-Name-Last: Duan Author-Email: yeqing.duan@svet.lu.se Author-Workplace-Name: Department of Political Science, Lund University Author-Name: Nils Droste Author-Name-First: Nils Author-Name-Last: Droste Author-Email: nd@ifro.ku.dk Author-Workplace-Name: Department of Food and Resource Economics, University of Copenhagen Author-Name: Brian Danley Author-Name-First: Brian Author-Name-Last: Danley Author-Email: brian.danley@geo.uu.se Author-Workplace-Name: Department of Earth Sciences, Natural Resources and Sustainable Development, Uppsala University Title: Modelling land use transition through social learning Abstract: Land use transition toward multifunctional practices is greatly affected by social learning, yet the temporal interaction between learning mechanisms and network structure remains underexplored. This study examines two social learning channels, information exchange and normative pressure, and how network architecture shapes their effects on transition outcomes. We developed SALT (Social learning in Agent-based Land use Transitions), a spatially explicit model that integrates the Consumat framework and reinforcement learning. The model is parameterized using a Swedish forestry context, simulating landowner adaptive decisions under integrated and modular social networks. Results show that the two channels play distinct roles across transition phases. Lack of knowledge limits adoption in early adoption. Individual experience is the main source of knowledge accumulation, and social learning alone cannot close the knowledge gap. As adoption spreads, normative pressure constrains implementation intensity to the prevailing local average, explaining the gap between behavioral and actual landscape changes. Network architecture shapes both channels. Integrated networks widen information exchange and allow alternative-use norms to strengthen over time, while modular networks restrict information circulation and lock in low-implementation local norms. Landscape change organizes along social ties rather than geographic proximity, with architecture determining whether adoption clusters into cohesive blocks or disperses as a diffuse mosaic in the social network. Landowner types contribute differently to behavior change and landscape change across both architectures. These findings suggest that effective transition governance must be tailored to both phase and social context. Early interventions should prioritize technical assistance, while raising the visible norm of implementation intensity matters more as adoption spreads. In modular communities, consolidating norms within communities before extending outreach is more effective than diffuse seeding. Instruments targeting behavior change need to be paired with those that directly support implementation intensity of alternative practice among less conformity-constrained landowners. Length: 40 pages Creation-Date: 2026-04 File-URL: http://okonomi.foi.dk/workingpapers/WPpdf/WP2026/IFRO_WP_2026_01.pdf File-Format: Application/pdf Number: 2026/01 Classification-JEL: C63, D83, Q24, Q57 Keywords: Land use transition; Social learning; Social network structure; Agent-based modelling; Multifunctional landscape Handle: RePEc:foi:wpaper:2026_01