This section provides some justification for using cost functions and
then minimizing their expected value under Formulations
9.3 and 9.4. The resulting framework is
called utility theory, which is usually formulated using rewards
instead of costs. As stated in Section 9.1.1, a cost can
be converted into a reward by multiplying by and then swapping
each maximization with minimization. We will therefore talk about
a reward
with the intuition that a higher reward is better.