Jan Borner


Causation on Several Levels: Supervenience Relations in Causal Models


Interventionist accounts of causation that are based on causal structural equation models (Pearl (2000), Woodward (2003), Halpern and Pearl (2005)) have come a long way to represent causal relationships in a formally rigorous way. They enabled precise definitions of several causal concepts and made it possible to accurately evaluate causal claims relative to a given causal model. But the accounts have long neglected non-causal relationships of counterfactual dependence, like supervenience or determination. This neglect results in the deficiency that intuitively adequate causal claims on different determination or supervenience levels cannot be evaluated in one and the same causal model. I will call this deficiency the supervenience-related expressivity problem (SEP) of causal models. I argue that SEP is a real problem for causal model based interventionist accounts of causation and not a virtue as arguments of causal exclusion (Kim (2005), Baumgartner (2010)) might suggest. The reason is that causal exclusion arguments do not affect interventionist conceptions of causation. After discussing two previously proposed solutions to SEP that, so I will argue, both have serious weaknesses, I will propose a new solution that employs a framework that was recently introduced by List (2019): systems of levels. I will argue that extending the classical causal model framework by incorporating certain systems of levels, the supervenience-related expressivity problem of causal models can be neatly resolved.

Date / Time / Place

June 23rd / 9:35 / Aula Magna