Lecture Series (hybrid): Lourens Waldorp

The Centre for Urban Mental Health is organizing a series of hybrid lectures to highlight expertise and current thinking on complexity science and urban mental health.

At the Centre for Urban Mental Health, we aim to unravel new pathways to improve urban mental health that takes into account the complexities and dynamics of mental health problems and mental health disorders in an urban environment.

This is a hybrid lecture format. You are encouraged to attend in-person at our offices in Central Amsterdam (limited tickets avaiable). Otherwise, online participation is always possible.

Schedule:

This is a hybrid Lecture with Lourens Waldorp

Presentation: 30 minutes

Q & A: 15 - 20 minutes

The Q & A session will be based on questions from the audience.

Title:

Causal discovery for psychological networks: combining observational and experimental data

Abstract:

Networks (graphs) in psychology describe the relations between variables (e.g., sleep problems, mood). One motivation for psychological networks is to be able to learn where to intervene. This presupposes that a relation between two variables represents a causal relation, where changing one variable induces a change in another variable. Since psychological networks are often based on purely observational data, discovering causal relations is limited. Borrowing some ideas from biology, we can include both observational and experimental data to improve estimates of causal relations. Here we consider a framework that involves several small interventions (experiments) and no intervention (observations) combined in a single analysis. The method is called perturbation graphs. The induced change in one variable is measured on all other variables in the analysis, thereby assessing possible causal relations. This is repeated for each variable in the analysis. A perturbation graph leads to the correct set of causes (not necessarily direct causes). Subsequent pruning of paths in the graph (called transitive reduction) should reveal direct causes.
We show that perturbation graphs provide a promising new tool for experimental designs in psychology, and combined with invariant causal prediction make it possible to reveal direct causes. As an illustration we apply the perturbation graphs and invariant causal prediction to a data set about attitudes on meat consumption and a time series of a patient diagnosed with major depression disorder.

About:

Lourens Waldorp is Associate professor at the Psychological Methods group of the Department of Psychology at the University of Amsterdam. He studied both psychology and mathematics at the University of Amsterdam, and did his PhD in developmental psychology, also at the University of Amsterdam. His expertise is in probability, statistics, networks and dynamics, with a focus on modelling. He has collaborations with different organisations to model processes (over time) of different kinds of psychopathology. In a recent project with the University of Groningen a causal model was established to more accurately reflect in a network where and how to intervene to improve therapy. In another project with the OLVG hospital, a network model was used to ascertain risk factors of dementia could be interpreted as causal factors.

Host or Publisher The Centre for Urban Mental Health