Summary

The design for this prototype was based on a simple model of sensations and emotions:

• Four primary emotions were chosen to represent the mood.

• The sensations are Boolean triggers that change the mood based on stimuli.

• The emotions are portrayed by degrading of the senses and actions.

Two finite-state components are developed to implement the design:

• A finite-state machine models changes in emotions and keeps track of the mood.

• A finite-state automaton determines the sensations as Boolean conditions from the stimuli.

• Both components communicate by passing messages between each other, to affect the inputs and outputs.

Evaluation of the system reveals the benefits of simple emotions for improving believability, but also technical problems:

• A limited number of emotions are chosen; modeling complex moods becomes increasingly difficult with finite-state machines.

• The system is fully predictable and does not change over time.

• Sensations and emotions are defined as Boolean values.

The next two chapters investigate extensions to finite-state machines that can solve these problems. Fuzzy-state machines can have partial degrees for emotions, probabilistic models make the sensations less predictable, and hierarchical systems simplify the modeling of complex moods.