Judea Pearl developed graphical methods for probabilistic and causal reasoning. His account of D-separation explains how the orientation of arrows in a directed acyclic graph determines conditional independences implied by its factorization assumptions.[ref]
For a chain , conditioning on blocks the path. For a collider , the path is blocked before conditioning, but conditioning on or an appropriate descendant can activate it. A graphical separation guarantees independence for distributions obeying the graph's assumptions; an unblocked path does not force dependence in every numerical parameterization. These distinctions make the graph a reasoning tool rather than just a picture of observed correlations.
Articles by others on the same topic
Judea Pearl is a prominent computer scientist and philosopher, best known for his work in artificial intelligence, particularly in the fields of probabilistic reasoning and causal inference. Born on September 4, 1936, he is a professor at UCLA and has made significant contributions to the understanding of causal relationships and graphical models.