|
Abstract:
|
Causal inference is an ambitious task. We discuss two difficulties in the context of cognitive neuroscience: the scarcity of interventional data and the challenge of finding the right variables. We outline how invariance across environments may serve as a guiding principle to tackle these problems. A causal model of a target variable generalises across environments or subjects as long as these environments leave the causal mechanisms intact. Consequently, if a candidate model does not generalise, then either it does not consist of the target variable's causes or the variables are ill-suited for a causal description of the problem. We present first ideas how to leverage this principle in the cognitive neurosciences, for example, to obtain a confounding-robust independent component analysis of electroencephalography data that generalises across subjects and facilitates interpretability in a group study.
|
