COS 264-5 - Integration of human behavior into dynamic models of environmental systems

Abstract: Natural and human social systems are intimately linked and accounting for these linkages is important to consider implications of environmental policies and alternative natural system management. Most efforts to link human and natural systems have focused on physical processes (e.g., GHG gas emissions, agricultural practices, conservation zones), rather than the human behavioral and social processes that interact with and determine the physical responses. We present a modeling research plan for integrating three modules: biophysical (e.g. physical, ecological), socio-cultural (e.g. human behavior, social dynamics) and socio-metabolic (e.g. natural resources, economics) components. A focus is how the dynamics of individual and group human behavior affect and are affected by natural systems beyond a static scenario approach to allow for dynamic feedback with natural systems. The plan allows incorporation of alternative cognitive, psychological and social theories to account for the human response to the environmental system and the subsequent dynamics of the interacting human/climate/natural resource system. We review divergent approaches to account for social norms, human group formation, and attitudes broadly classified as contagion and cognition, disaggregated spatially to account for regional cultural/political differences. Models associated with this plan have been instantiated for climate change mitigation, infectious disease spread, and are broadly applicable to questions in environmental health and conservation policy. Our review of these efforts summarizes the systems dynamics schemes used to link the three components, focusing on the limited manner in which behavior (risk perception, attitudes, behavioral control, social norms, cultural values) have been considered. Our results provide assessments of qualitative response of these interconnected models accounting for uncertainties in the structural forms of model components in addition to parameterization uncertainties in the assumed functional forms for interactions between state variables. We show how robustness of model structural stability can be feasibly assessed through resampling methods for inclusion/exclusion of model linkages with evaluation criteria of both state space dynamics and state at a terminal-time. We project that, consistent with results from previous model instantiations, regression tree methods allow partitioning of the importance of socio-behavioral components to terminal-state conditions (e.g. mean global temperature at 2100), and clustering methods applied to state-space dynamics provides emergent general classes of trajectories. A take-home message is that feedbacks from inclusion of socio-behavioral components into earth system/environmental models can constrain, rather than exacerbate, model uncertainty for complicated models, providing qualitative model projections representative of alternative futures, helping prioritize social interventions to reduce unintended or harmful human impacts.