Over the course of this book, the animats have become relatively intelligent. They are capable of primitive motor control, handling objects, and making some decisions. However, the behaviors are not particularly believable, mainly because the AI focuses on intellectual capabilities.
This part covers emotions, used as a tool to improve the believability of the animats. In the spirit of adding errors to the actions (for instance, noise in the aiming), emotions provide a biologically plausible way to modify the sensors and effectors to affect behaviors. Emotions also provide a high-level bias for decision making.
The emotions improve the simulation of the animat's body, modeling essential human characteristics (such as adrenaline). Not only is this likely to improve the realism of low-level actions, it's also likely to improve higher-level behaviors that arise from the simulation.
As well as affecting existing behaviors, the emotional system provides guidance for intelligent decision-making components. The emotions can be understood as desires and motivations, which are often subjective to each player.
The technology developed in this part also demonstrates systems composed of multiple subarchitectures. In this case, the emotional architecture and the intelligence architecture are mostly independent, with emotions providing influences only during interaction with the body.
Finally, this part covers an AI technique better suited to modeling simple states—and therefore, sequences of actions. Finite-state techniques are often easier to design and provide a greater form of control.
Chapter 36, "Emotive Creatures." This chapter provides an overview of emotions, studying the different approaches in AI and psychological research. We identify the need for emotions in games as a way to interact with human players.
Chapter 37, "Sensations, Emotions, and Feelings." This chapter defines sensations, emotions, and feelings and provides examples of such common in games. This chapter presents an interface used for communicating emotions with the game engine and discusses ways to portray them.
Chapter 38, "Finite-State Machines." This chapter covers finite-state machines from a theoretical and practical perspective. They are a control technique particularly suited to keeping track of states—and therefore sequences.
Chapter 39, "Under the Influence." In this chapter, we use finite-state techniques to model emotions and sensations. This subarchitecture affects the intelligent behaviors by degrading the senses and actions, according to mood.
Chapter 40, "Nondeterministic State Machines." Probabilistic, nondeterministic, and fuzzy models are presented as extensions to finite-state machines in this chapter, each resolving practical issues with the original technique.
Chapter 41, "Hierarchical State Machines." Hierarchical approaches are discussed, and examples of how they are applicable within games are shown—notably how they considerably simplify the design of finite-state machines.
Chapter 42, "An Emotional System." In this chapter, each of the extensions to finite-state techniques is applied to create a hierarchical system that models emotions, feelings, and sensations. The animats also include simple memories and manifestations of their emotions as behaviors.
Chapter 43, "Emergent Complexity." The last chapter of this part discusses the concept of emergence. Different aspects of the behaviors and functionality can be considered emergent. After defining emergence and covering why emergence is desirable, this chapter provides hints to harness the power of emergent phenomena by design.
Emotions rely on other aspects of the AI to provide purposeful behaviors and need the game engine to portray them:
Existing behaviors are required, although they can include various levels of complexity. The architecture providing intelligence is assumed to be self-standing.
Parameters influence the interfaces between the body and the brain, modeling degradation from perfect senses and actions.
Emotions can be portrayed by the game engine using a variety of technologies, not limited to gesture animation, fixed expressions, and text messages.
The following chapter covers the origins of emotions and their different aspects. Understanding provides the foundation for this analysis and specification, which prepares us to develop an artificial system to model emotion.