Motivation:
Fundamental natural processes are cyclic (e.g. topsoil maintenance)
Acyclic or unidirectional causal models are unrealistic in terms of describing our interaction with natural systems
How can we represent cyclic causal chains, and predict effects of interventions?