Causation vs. constitution: loosening the friction

Theories of mechanistic explanation hold that an explanandum phenomenon (e.g., a car moving) is explained by its constituting entities (e.g., the car engine) and activities (e.g., the engine running). Such mechanistic explanations are often interpreted as 'downward-looking': they explain upper-level phenomena in terms of their lower-level constituents. They differ from causal explanations, which explain effects in terms of their causes, and are thus 'backward-looking'.

Constitution is prima facie different from causation, on the alleged ground that it is a synchronic and mutual dependence relation. Yet, most have proposed to define constitution along the lines of causation and to use causal discovery methods also for constitutional discovery. At the same time, it is a widely held tenet in the modeling literature that causal and constitutional dependence relations must not be mixed in analyzed variable sets. This creates an apparent tension: on the one hand analyzing constitutive relations seems to presuppose means from causal modeling, on the other hand, such means must only be applied to variable sets that are free of constitutive dependencies.