Andreas Freivogel

In philosophy, and especially in ethics, reflective equilibrium (RE) is often considered a powerful method for obtaining beliefs that mutually support each other, are justified by evidence, and are backed by good reasons. Beisbart, Betz, and Brun (2021) have introduced a formal model of reflective equilibrium based on the theory of dialectical structures Betz (2013), which they use as a methodological tool to understand the method of reflective equilibrium better. This report is an outcome of the research project 'How far does Reflective Equilibrium Take us? Investigating the Power of a Philosophical Method' and summarizes the findings of assessing the model thoroughly by numerical investigation. We simulate RE processes for a broad spectrum of model parameters and initial conditions and use four different model variants (including the original model). We analyze the dependence of simulation results on different parameters and assess the models' conduciveness towards consistency, and ability to reach global optima and full RE states. The results show that the models' behavior depends crucially on the specifics of the simulation setup. We can, therefore, not draw any general conclusions about the overall performance of the model variants. Rather, the specifics of the context in which an RE model is used must be considered to choose a specific model. Finally, we identify some critical knowledge gaps we cannot close with this report that call for further research into RE modelling.