Discovering causal relationships is a hard task, often hindered by the need
for intervention, and often requiring large amounts of data to resolve
statistical uncertainty. However, humans quickly arrive at useful causal
relationships. One possible reason is that humans use strong prior knowledge;
and rather than encoding hard causal relationships, they encode beliefs over
causal structures, allowing for sound generalization from the observations they
obtain from directly acting in the world.

In this work we propose a Bayesian approach to causal induction which allows
modeling beliefs over multiple causal hypotheses and predicting the behavior of
the world under causal interventions. We then illustrate how this method
extracts causal information from data containing interventions and
observations.