Because emotions make such an important improvement to the system, they are integrated with any new interface between the body and the brain—whether senses or actions. It's also important for different (cognitive) components in the brain to provide sensations, too (for instance, surprise when discovering objects).
Finite state machines are used ubiquitously in game AI, and there are plenty of reasons to use them in nouvelle game AI, too—notably as a phylogenetic (frozen) component to help manage the system. Therefore, finite-state machines can be applied as exercises to many other problems in this book.
Finite-state machines, however, are very static; they suffer from not being able to learn or adapt. Part VII of this book explores reinforcement learning, which can in fact be understood as a learning finite-state machines technique. The weights of the transitions are adjusted according to positive and negative feedback.
The moods created with the hierarchical finite-state machines are used to provide high-level guidance. In humans, emotions are often used to make high-level decisions about what to do, whereas intelligent abilities are used to bring these to life. The moods are put to use in the next part, as guidance for the learning algorithm in our animats.