I am working in the field of computational neuroscience, and my research concerns modelling the processes underlying the neuronal activity in the brain. The main questions of computational neuroscience are how the brain carries out its functions and what kind of biophysical machinery and network architecture enables it to do this. This research involves several levels of description, from studying biophysical processes at single cells and synapses, through neural network modelling up to application of information theory and machine learning to the problem of general principles of brain architecture and functionality.
In my own research I employ all these description levels. My current research projects include the following topics.
1. Modeling short-term synaptic dynamics (BCCN Project A1)
Synaptic transmission is a complex process, in which the responses of postsynaptic neurons to series of presynaptic action potentials can vary systematically on short time scales from milliseconds up to seconds. These systematic changes are referred to as synaptic dynamics of short-term synaptic plasticity. The aim of the project is to provide concise theoretical description of synaptic transmission, which would capture its experimentally observed dynamical properties, such as short-term synaptic depression and facilitation. We formulate a realistic model of short-term synaptic plasticity mainly based on the data obtained at the department (by H.Taschenberger, K.-H. Lin, T. Sakaba, N.Hosoi) from patch-clamp recordings and Ca2+-imaging at the calyx of Held performed in rat brainstem slices. The first goal of the modeling is to find out, what kind of biophysical processes, underlying synaptic transmission, have to be included to explain the input-output relationships of the considered synapses. The second question concerns the consequences of synaptic dynamics for information transmission through single synapses and for the behavior of neural networks. In particular, we apply information theory to study how particular types of synaptic dynamics affect the efficiency of signal transmission at the synapses. On the network level, the question is how the short-term plasticity affects the dynamics and steady states of neuronal circuits in various areas of the brain, such as the auditory pathway, in the visual cortex, cerebellum and hippocampus, and how they might contribute to the functional properties of these brain areas.
2. Encoding of auditory information for human auditory midbrain implants. (BFNT Project 1c).
In the recent decades, the technology in the field of auditory prosthesis has made a considerable progress which enabled, to a certain extent, to restore hearing to patients with severe damages of the primary auditory pathway. Whereas the cochlear implants are established and widely used, patients with damages of e.g. the auditory nerve have to rely on other technologies, such as auditory midbrain implants (AMI) designed for stimulation of the inferior colliculus. In collaboration with the Department of Otolaryngology at the Hannover Medical School (M.Lenarz, H.H. Lim), where the AMIs are currently clinically tested, I am working, together with scientists from Edinburgh and Göttingen (J.M.Herrmann and D.Lyzwa) on the improvement of the algorithm of encoding of the auditory information in the auditory midbrain implants. In particular, we suggest to employ methods of machine learning theory to adapt the stimulation scheme at the inferior colliculus interactively based on perceptual feedback provided by the patients. This would enable to optimize the encoding individually, adapting it to the level of damage and to the particular positioning of the implants in the individual patients.
3. Analysis of voltage-sensitive dye imaging data using methods of machine learning.
Analysis of experimental observations of neuronal activity during processing of sensory information allows understanding the properties of the nervous system and biological constraints under which it operates. In collaboration with the lab of A.Grinvald at the Weizmann Institute of Science, I analyzed the recordings of population activity of visual cortical neurons performed via voltage-sensitive-dye imaging of the primary visual areas of cats and monkeys. In the analysis I used methods of statistical machine learning such as principal component analysis, independent component analysis and self-organized maps to characterize the states of cortical activity and their interaction with sensory inputs and ongoing activity. The latter is of a special interest because it is likely to contain the cognitive component of the brain activity, such as expectation, attention or prior information about the sensory input statistics. My current project, in collaboration with F.Theis (Helmholz Center München), deals with the application of the second-order blind source separation method based on spatial and temporal covariances. This algorithm is expected to improve signal to noise ratio, be useful for artifact cleaning und possibly detection of patterns and features in the neural data.
4. Modeling of population activity in the primary visual cortex.
An important aspect of my research involves neural network modelling. The goal is to build a mathematical framework, which would reproduce experimental data and allow making predictions about behaviour of the modelled system in a natural environment. In particular, the analysis of the imaging data in the primary visual cortex has shown, that activity patterns corresponding to functional states of the network (orientation maps) can emerge in the cortex in absence of sensory stimulation. To model this effect, I build a recurrent network model with synaptic depression, in which the strength of synaptic connections depends on the orientation selectivity of the connected cells and on the preceding activity of the presynaptic cell. The model reproduces steady states and the dynamics of switches between the activity patterns observed in the cat areas 17/18. To study the model properties I use numerical simulations and analytical methods of nonlinear dynamics.