Turtles all the way down? Psychometric approaches to the reduction problem

The question of how different explanatory levels in scientific inquiry are related to each other is known as the reduction problem. This thesis focuses on a specific domain of this question, namely how we should relate brains to (psychological) behaviour. The central position of this thesis is that this question is ultimately a measurement problem. That is, in order to understand the relationship between brains and minds, we need to formulate measurement models that can relate observable variables (e.g. response times, brain activity, brain structure) to the underlying constructs we are interested in (e.g. memory capacity, intelligence or personality differences). Moreover, in the case of relating brains to behaviour, theories from philosophy of mind can be translated into such measurement models, thereby guiding empirical inquiry and simultaneously providing an empirical test of philosophical theories. Further extensions of these ideas focus on the application of representational geometry, whereby the structure of neural and behavioural patterns are used to relate brain and behaviour, and the examination of cases where inferences across explanatory levels goes awry (known as Simpson’s Paradox). Based on empirical applications in several domains it is concluded that supervenience theory, which suggests a fundamentally asymmetrical relationship between brain and mind, is most in line both with theoretical considerations and empirical data.