In Chap. 22, some pragmatic ways to improve the feedback dynamics were described. This chapter will discuss some background. It will turn out that there are many relations to fixed point relaxation techniques, to Markovian processes, to game theory, and to machine learning. For some aspects, it is possible to provide computational evidence about partial aspects. In general, it however turns out that significant parts of ``learning in transportation systems'' is a challenging topic where many open questions remain.