5.6 Summary

The agent metaphor suggests that programs be built out of components that are designed to be understood in anthropomorphic terms. In order to achieve this, we attempt to endow our agents with purpose, autonomy, and the ability to react to events in their world. This chapter presents a series of agent systems and shows that the metaphor can be used to unite several disparate styles of computation:

• the behavior-module-based, action-selection approach to the generation of creature behavior (Tinbergen, Brooks, Maes);
  
• the dynamics, generative power, and ability to deal with sequential operations found in procedural programming languages (Lisp, Logo);
  
• the ability to integrate declarative and procedural information found in early constraint systems (Sketchpad, ThingLab).
  

DA is the most powerful agent system and integrates these three strands most fully, but it is somewhat hard to understand. The simpler agent systems lack the full power of a programming language, but are simple to understand and lend themselves to the drag-and-drop, mix-and-match construction paradigm of LiveWorld.