Zentralinstitut für Seelische Gesundheit
J5, 68159 Mannheim
In our department we develop mathematical models of brain function and statistical approaches for estimating these directly from experimental observations. These strongly data-driven models are then used to gain insight into the neuro-dynamical and neuro-computational processes underlying cognitive function, and in particular how these are altered in psychiatric conditions. Although a number of genetic risk, molecular, biochemical, morphological, and physiological factors have been linked to psychiatric disease, we often lack a more causal and mechanistic understanding of how exactly these give rise to behavioral or cognitive symptoms. Yet such an understanding would be highly valuable for devising more effective and individualized treatment options. Mathematical models of neural systems can provide a powerful tool for developing such a mechanistic understanding, and can guide experimental research as well as drug development.
For mathematical model building, we start from electrophysiological, multiple single-unit recordings and behavioral data from psychiatric animal models which we obtain from experimental partners within the Central Institute of Mental Health, and from other international collaborators. We are concerned both with mathematical modeling of psychiatrically relevant neuronal structures per se, as well as with statistical methods for parametrizing these models based on the experimental data, and for judging the physiological validity and predictive power of such models based on strict statistical criteria. In 2014, we primarily worked on the following three methodological developments:
- We have derived a mean-field theory for a system of physiologically valid model neurons that we had developed previously. Mean-field approaches originally come from theoretical physics and enable to analyze key aspects of the neuronal network dynamics in mathematical depth. We have started to use this approach to characterize changes in neuronal dynamics associated with alterations in single cell and synaptic properties experimentally characterized in different genetic animal models.
- We have developed novel statistical approaches for detecting supra-chance spatio-temporal patterns in multiple single-unit recordings from behaving animals. Such statistical methods complement our model-based analyses by unraveling from experimental data important dynamical motifs underlying the computational processes which implement behavior.
- We have developed mathematical models of behavior, based on Bayesian statistical decision and mathematical learning theory, which allow to identify and disentangle cognitive key components underlying (changes in) behavioral performance. Such approaches may also enable to quantitatively characterize common behavioral key ingredients across species, thus facilitate translational research and potentially provide new criteria for psychiatric classification. We have successfully used this approach to characterize the key disturbance responsible for performance deficits in an animal model of depression on a working memory task.
Structure of the Group:
|Group Leader:||Daniel Durstewitz|
|Postdoctoral fellows:||Florian Bähner, Joachim Hass, Georgia Koppe, Claudio S. Quiroga-Lombard, Eleonora Russo, Hazem Toutounji|
|PhD Students:||Carla Filosa, Sadjad Sadeghi, Dominik Schmidt, Grant Sutcliffe|