Incentive Design: Learning, Interaction Hierarchies, and Behavioral Traits

Can one induce a desired behavior on another through incentives? This NSF project aims to answer this question through mathematical modeling and analysis. The project is transformative in that it seeks to remove stringent assumptions required for existing incentive design mechanisms that limit its practical use. The questions are motivated via target application contexts. The intellectual merits of the project include a careful combination of many topical areas and wide applicability of the resulting designs in engineering, environmental policy making, and organizational structure, among others. The broader impacts of the project include inter-disciplinary graduate training, research dissemination through publications and seminars, promotion of diversity within the research teams, and engagement with K-12 students. Incentive design is a hierarchical decision-making problem that considers how one rational agent might induce a desired behavior on another through the design of policies. The proposed research will draw on recent advancements from a variety of fields, including control theory, game theory, operations research, machine learning, and behavioral economics to circumvent restrictive assumptions on incentive design. By doing so, the theory will apply much more broadly. Five thrusts of the project will build along the three following themes: (1) Can agents learn each other’s preferences through repeated interactions? (2) How do behavioral traits influence incentive structures? (3) Can incentives be designed to accommodate complex hierarchies of interactions among a population? Besides advancing the theory of incentive design, the project will investigate select applications of the theory in the domain of power systems, specifically in the coordination of distributed energy resources. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.