The moving, shooting, and gathering capabilities are exploited using learning algorithms. These benefit from adaptive AI because the most suitable combination of skills can be found automatically—without the designer's direct involvement. When it comes to other behaviors, such as looking around (for instance, focus or scan), learning does not have any advantages. It's much easier to express such capabilities in terms of the moods, without any form of adaptation.
Reinforcement learning would be suitable for the task, but requires setting up a reward signal and setting up the states and actions. The time overheads outweigh the benefits. Not overusing technology in this way is one way to deal with adaptive behaviors, but this topic is discussed extensively in the next chapter.