Institute for Theoretical Neurophysics - Computational Neuroscience

One of the most challenging scientific mysteries of our time is the function of the brain. Which mechanisms in the brain allow us to see, to hear, or to feel? Recent decades of neurobiological, biochemical, biophysical and psychophysical research have provided an increasing flood of information about the neuronal substrate and its abilities. However, simple questions still are far from being resolved. Which spatio-temporal states in the brain are responsible for the perception of an object? How do these states emerge, and on which time scales do they exist? What happens when we recognize an object? How is information learned and stored in the brain? What code does the brain use for processing this information and subsequently generating a meaningful behaviour?

It appears that the neocortex, from an evolutionary perspective thes most recent part of our brains, holds a key to these questions. Our hypothesis is that the information processing relevant for object perception and behavior emerges from the interaction and coordination of precisely organized spatial and temporal patterns of activity and connectivity in the cortex.

Novel experimental techniques like multielectrode recordings and several methods of functional imaging now reveal a range of collective phenomena including a high degree of functional organization of cortex and particular spatio-temporal patterns of neuronal activity which have been demonstrated to reflect Gestalt properties of perceived objects.

In our institute we study biologically realistic models of neural computation. We use methods of nonlinear dynamics and statistical physics to understand the principles leading to the emergence of particular structures in the cortex.

Our approach covers the development and application of methods for the analysis of neurobiological time series using information theoretical methods and hidden Markov models, the simulation of large-scale neuronal networks, the mathematical analysis of collective spatial and dynamical phenomena in these models, and various applications of artificial neural networks to technical problems of system identification and pattern recognition.


Cognium, Room 2470
Hochschulring 18
D-28359 Bremen
Phone: 49 421 218-62000
Fax: 49 421 218-62014

Dr. Udo Ernst
Phone: 49-421 218 62002
Fax: 49-421 218 62014