Causal and anti-causal learning in pattern recognition for neuroimaging
This addresses interpretation challenges in neuroimaging for researchers, but it is incremental as it builds on existing model distinctions.
The paper argues that distinguishing between encoding and decoding models in neuroimaging pattern recognition is insufficient because relevant features have different meanings based on causal or anti-causal relations, concluding that causal inference is essential for interpretation.
Pattern recognition in neuroimaging distinguishes between two types of models: encoding- and decoding models. This distinction is based on the insight that brain state features, that are found to be relevant in an experimental paradigm, carry a different meaning in encoding- than in decoding models. In this paper, we argue that this distinction is not sufficient: Relevant features in encoding- and decoding models carry a different meaning depending on whether they represent causal- or anti-causal relations. We provide a theoretical justification for this argument and conclude that causal inference is essential for interpretation in neuroimaging.