From the probability tree two fundamental tools arise. The total probability theorem reconstructs the probability of an event by summing the paths that lead to it through a partition of the “causes”. Bayes’ theorem goes the other way: from the observed effect it works back to the cause, inverting a conditional probability. Read geometrically, this inversion is the transformation of a “direct” tree into its “inverted” tree, which contains the same information but answers the opposite question.