Causation is a concept that plays an essential role in reasoning with
commonsense. It is a property inherent to the dynamics of
changing environments. Therefore, it should follow that causation
should also play an essential role in the theories of action and
change. However, in this article we postulate that causation is a form
of abstraction. As a consequence of this view, one should not
attempt to model causation as a primitive relationship between
entities. Rather, causation should be viewed as an abstract
relationship, which can be derived from observations (as a
correlation) or derived from our knowledge of the structure of the
world.
In this article, we explore the modeling of two types of problems
where causation appears to be necessary. We illustrate the problems with some
simple examples found in the literature. We show that by decreasing
the level of abstraction one can obtain the right results.