Welcome to my website! I finished my PhD at the University of Amsterdam at the Department of Psychological Research Methods (PsychoSystems Lab 
Group).
In recent years, the notion has been put forward that mental disorders (such as depression) can be presented as a system of mutually interacting variables. Such a network can be seen as a complex system: a set of variables that influence each other in such a way that critical transitions from one stable state to the other are possible. Complex systems are complex for more than one reason. The system as a whole can quickly grow to be too complex to study in its entirety, and relations between variables can produce critical transitions: a sudden jump from one stable state to another. In the first part of this dissertation, we reduced these complex systems by means of a mean field approximation (MFA), where we approach the complex system as a stochastic cellular automaton (SCA). In both a simulation study and an empirical analysis, we showed that mathemathical approach works well, and that we were able to infer whether patients diagnosed with depression or healthy participants could experience critical transitions.
Relations between variables are often reciprocal, indicating that two variables may influence each other. Establishing a causal relation between two variables is an essential next step; learning the cause of a symptom like concentration problems can help to reduce symptoms like these. However, we cannot unravel all causal relations with only observational data that are often used. By combining observational and experimental data, we may have enough information to properly estimate causal relations. In the second part of this dissertation, we put several algorithms for estimating causal relations to the test in a simulation study, and found that two algorithms – the invariant causal prediction (ICP) and the hidden ICP (HICP) algorithm – in particular hold the best cards for correctly estimating causal relations. In an empirical study, we combined the results of these two algorithms with a literature study in a causal graph approach to create a causal graph of obsessive-compulsive disorder. Although we saw a discrepancy between the algorithms and the literature, we uncovered causal relations that otherwise may have been left unknown, and showed the potential of such an approach for psychopathology.
My research showed that both prediction and explanation are important in psychological research. By putting a system passively or actively under pressure, we may assess where a system can go to, where we may gain a better understanding of the internal workings of a system.
This PhD-project was funded by Yield. My promoters are Lourens Waldorp and Han van der Maas.