Game Development Reference
In-Depth Information
6
Rational vs. Irrational
Behavior
One of the problems that von Neumann and others had with their applica-
tion of game theory was the expectation that people behave rationally and,
in doing so, will always attempt to select the best outcome. As we saw from
some of the examples in Chapter 5, this is not always the case. Often, people either
fail to select the best option or even elect
not
to for whatever reason. The result is
the reason for such a difference between normative decision theory and descriptive
decision theory. In that gap lies a whole lot of irrational behavior.
Of course, trying to figure out
what
those behaviors are is a bit of a knotty prob-
lem. Computers are good at figuring out the
rational
answers. Coming up with an
irrational
but reasonable-
looking
answer is another trick entirely. Most of us are
accustomed to the notion that irrationality is something to be avoided or even shunned.
And yet, as we have seen so far—and will continue to explore—irrationality is not
only very real but it is what bestows depth of character on behaviors.
There is a significant problem in trying to work with irrational behavior, how-
ever. While generally there is only one
correct
(i.e., rational) answer to a problem,
the solution set on irrational behavior tends to be a bit wider. That is not to say that
everything that is not the “correct� answer is going to look reasonable. Some things
are
just plain wrong.
If you recall in Chapter 1, I wrote about my beloved pig painting. While my
prone porcine portrayal was less than perfect, it was well within the bounds of
“piggishness.� It didn't have the fifth leg sticking out of the top of its back like a dor-
sal fin. While not perfect, it was reasonably pig-like. There are, in truth, an infinite
number of ways that one could paint a pig, but only a select number of them would
fall within an acceptable range that observers would accept as “looking like a pig.�
Sure, some of them might be categorized as “an odd-looking pig� but would still be
thought of as reasonable enough to not be confused with, say, a horse, an iguana,
or a platypus (although a platypus is confusing enough on its own).
79
