Although I am interested in many issues in psychology, my work seems to gravitate around a single question: What are psychological attributes? When using words like "depression" or "neuroticism", what do we designate? A traditional answer is: latent variables. Latent variables are structures that are not directly observable, but can be measured indirectly through their manifestations in behavior. Thus, the latent variable of depression is measured through items like "do you feel sad?" and "are you tired?". This idea has been used to formulate statistical models for psychological measurement. Today, the assumption that such a model is true underlies much research in psychology. In early work, I analyzed the assumptions behind this approach. I concluded that, to use measurement models profitably, one needs to interpret the indicator variables (e.g., insomnia, fatigue, concentration loss)as being caused by the psychological attribute (e.g., depression). This requirement is unreasonably strong in many areas of psychology, and this realization drives the second half of my research life (so far). I developed the hypothesis that attributes like depression are not latent variables, but instead are systems of components that reinforce each other (e.g., insomnia -> fatigue -> concentration loss). That is, they form networks . With a team of methodologists, statisticians, and clinical psychologists, I have initiated the Psychosystems Project in which we apply network modeling and complex systems techniques to psychopathology and personality. Most of my research efforts are now dedicated to this project, which has been sponsored by NWO-Vidi and ERC-consolidator grants. Next to psychometrics and network theory, my interests include general methodological issues, philosophy of science, and the history of psychology.