Modeling normative behavior and enforcement in a multi-agent system based on Dutch Public Transport

Abstract

We live in a normative culture in which people make decisions every day to either follow or break rules and norms. An interesting example of choosing behavior can be found in the public transport system, which as it is a semi-closed system lends itself to being modeled. This human behavior can be modeled to predict and influence the way in which agents and people alike make desirable decisions. This thesis will answer the question ’how can enforcement strategies for public transport be optimized by making use of learning agents’. To answer this question I first had to research different decision-making strategies such as decision theory and reinforcement learning. With this I gained a better understanding of the normative culture of choosing agents, with which this thesis aims to predict norm abiding and norm breaking behavior in a stochastically modeled environment.