Throughout the rest of this book, there are many opportunities to apply neural networks. They can be applied to most pattern-recognition or function-approximation problems. They'll be particularly welcome as an exercise in Part VII, because they prove to be a convenient way to approximate experience learned about the problem.
Part IV builds on the shooting capabilities of the animats. By learning the success rate of different weapons in various situations, it's possible to estimate the best weapon to use in the current scenario. Such weapon selection plays to the strengths of the weapon skills of our animats.
The next part also looks into decision trees. They combine similar capabilities of neural networks with regard to pattern recognition, but also provide the readability and intuitiveness of rule-based systems.