Probabilities in the action vector

Is it possible to have access to the probabilities of the values of each branch of the action vector? In the case where a chosen action is impossible in the game, I would like to choose the next most probable one.