Observing the World - Behavioral Mathematics for Game AI - page 16

Certainly, there doesn't seem to be a lot we can glean about the behaviors of
owls from observations such as these. Yes, we have compiled a list of possible things
that an owl could do at any one point in time, but it doesn't give us a lot of infor-
mation about when and why .
Often, the secret to the conceptual side of behavioral mathematics is to sit back
and try to see what is really going on. Things are not often as clear-cut as they seem
on the surface. Because of that, there is a layer of abstraction that is less visible but
far more powerful. What sounds cause him to turn his head? Perhaps the owl ruffles
his feathers only when the fan turns on? Is there a pattern to the blinking? What is
he thinking when he moves that foot? Is he even thinking about it at all? Does he say
to himself, “Now I'm going to move my foot�? Probably not. But there is something
going on inside that owl's head on either a conscious or unconscious level, and
the outward signs of that thought process are what makes our owl look and act like
an owl.
What's more important is comparing the actions of one owl to those of an-
other, or comparing the actions of the owl in one situation to his actions in another.
The complexity that differentiates one Poker player from another is not in the
absolute linearity of processing mundane probability statistics, but in the subtle
nuances of human behavior. Many of the things we do that set us apart from one
another are unconscious—or at least born from beliefs that are, themselves, com-
plicated in their origins.
I DENTIFYING F ACTORS
It isn't enough to simply observe actions and reactions. We must identify those
actions for what they are. We must have an explainable correlation between cause
and effect. The difference is similar to the difference between hearing and listening.
We hear things all around us at all times, but often we don't actually listen to what
we hear. When we actually listen to things, we separate out distinguishable sounds
from the aggregate sonic disturbance in the background. The same can be said
about hearing a person speak but not necessarily listening to what he is saying—a
staple of marital discord. (So I've been told. I always listen very attentively to what
my wife says!)
The point is, it is only when we separate different observations from the back-
ground that we can put them into their proper local context. At that point, by iden-
tifying them, we have taken a necessary step toward isolating them for further
analysis. Because there is no shortage of things to observe and identify, the trick is
to determine which of those items are important enough to proceed with.