Abstract
Emerging “(computational) systems medicine” challenges neuropsychiatry regarding the development of heuristic computational brain models which help to explore symptoms and syndromes of mental disorders. This methodology of exploratory modelling of mental functions and processes and of their pathology requires a clear and operational definition of the target variable (explanandum). In the case of schizophrenia, a complex and heterogeneous disorder, single psychopathological key symptoms such as working memory deficiency, hallucination or delusion need to be defined first. Thereafter, measures of brain structures can be used in a multilevel view as biological correlates of these symptoms. Then, in order to formally “explain” the symptoms, a qualitative model can be constructed. In another step, numerical values have to be integrated into the model and exploratory computer simulations can be performed. Normal and pathological functioning is to be tested in computer experiments allowing the formulation of new hypotheses and questions for empirical research. However, the crucial challenge is to point out the appropriate degree of complexity (or simplicity) of these models, which is required in order to achieve an epistemic value that might lead to new hypothetical explanatory models and could stimulate new empirical and theoretical research. Some outlines of these methodological issues are discussed here, regarding the fact that measurements alone are not sufficient to build models.
Key words
methodology - qualitative modelling - quantitative modelling - complexity - simulation
