Beyond Causal Exclusion: New Challenges for Multi-Level Causal Models

The aim of the project is to develop a theory of causation that can adequately describe causal structures that span multiple levels (such as the physical, biological, and psychological levels).

The project is an international collaboration between Prof. Vera Hoffmann-Kolss (University of Bern) and Prof. Thomas Kroedel (University of Hamburg) and will have six members: two PhD students, two postdocs, and the two PIs. The project is funded by the Swiss National Science Foundation (SNSF) and the German Research Foundation (DFG).

 

More about our research:

Factory farming and close human-animal interactions can lead to the emergence of new pathogens. The El Niño event in 1982-1983 was a major contributor to the drought in Ethiopia in the following years. Statements like these suggest that causal relations are often embedded in complex causal structures. If an El Niño event causes a drought, the specific causal chain will be quite long and complicated. Factory farming is only one possible cause of the emergence of new pathogens, and the structure containing it must include various other possible causes and background conditions. Importantly, the causal structures in which these relations are embedded not merely contain a multitude of causal factors, but involve multiple levels of reality, yielding causal interactions at and across the biological, chemical and physical levels and even the social and mental levels.

In recent years, causal models have emerged as the dominant tool for capturing complex causal structures. Initially developed as a formal method for analyzing empirical data, causal modeling techniques soon entered the philosophical discourse and are now regarded as a promising approach to reveal the nature of causation, in particular, the nature of higher-level and multi-level causal relations. While progress has been made in solving the so-called exclusion problem, that is, the issue that causes at different levels seem to compete with one another, applying the causal modeling approach to multi-level structures still poses several challenges: (1) What is the best framework for modeling multi-level causal relations? More specifically, should this framework be probabilistic or deterministic? (2) Multi-level causal models typically contain both causal relations and non-causal, "metaphysical" dependencies. What tools, not yet included in the standard causal modeling framework, are required to cover these additional dependencies? (3) Should the variables we include in a causal model have the same level of granularity, that is, satisfy a proportionality constraint? (4) More generally, how should we choose the variables to be included in a causal model, and does the existence of causal relations depend on this - partly normative - decision?

An integrated approach to multi-level causation that addresses these challenges simultaneously is still lacking. This project aims to fill this gap by bringing together a group of researchers, all experts in one or more of these debates, who will work together to develop a new theory of multi-level causation. Methodologically, we will take a conceptual engineering approach and measure the success of our theory in terms of how well it captures the desiderata of multi-level causal descriptions. This will shed new light not only on the philosophical debate about higher-level and multi-level causation, but also on causal models and our understanding of causation in general.