How is the “reward matrix” used in Markov Decision Processes?

a) To determine the transition probabilities between states

b) To assign values to each possible state-action pair

c) To calculate the expected return time

d) To create a Markov Chain graph