Mark G. Orr, Ph.D. - US grants

DESCRIPTION (provided by applicant): This project develops a computational model of individual health behavior. Computational models allow for understanding the explicit psychological mechanisms that underlie behavior. There is a high need for this because, 1) current work in health behavior lacks and understanding of explicit processes, 2) a computational model can mesh with higher and lower level computational models (such as agent-based models, dynamic social network models, and computational neuroscience), 3) aggregate data of health behavior do not allow for making direct inferences about individual level psychological processes. Aims: We propose to build a novel theoretical model of health behavior that combines 1) the psychological, intra-individual processes, 2) the social, inter-individual processes, and 3) the interaction between the two as a complex system. Furthermore, we propose to evaluate this theoretical model against extant epidemiological and psychological data. Key Features of the Proposed Work: (1) Computational Architecture: We will implement the theoretical model as a neural network. This is the first attempt to do so in health behavior with respect to both the individual-level and the social dynamics. This will allow for interfacing with current and future insights into health behavior that come from neurophysiological mechanisms and is highly compatible with other decision-based models such as neuroeconomics and social cognitive neuroscience as well as social network theory and social epidemiological constructs. (2) Study Design: We build a model of the Theory of Reasoned Action (a prominent health behavior change theory) in reference to adolescent sexual initiation behavior. Then, we compare the model behavior to extant empirical work that was collected and designed to test the causal structural model of the Theory of Reasoned Action in reference to adolescent sexual initiation behavior. Finally, we extend the neural network model into a social dynamic process model (with multiple neural networks) that is constrained in its structure by extant social network data. (3) Recommendations for Policy Resistant Behaviors. Individual health behaviors are policy resistant because the underlying mechanisms are difficult to understand without developing explicit computational models. This is compounded when individual level behavior and social dynamics are interdependent. Therefore, policy efforts towards prevention of socially-bound individual-level health behaviors may be under-informed by current methodologies. The proposed work is well suited to elucidating this complex process. PUBLIC HEALTH RELEVANCE: This project proposes a computational model of individual health behavior that will lend insight into the related explicit psychological processes. Furthermore, by embedding this model into a social context, this work provides a computational account of the interdependence of social structure and individual health behavior.