An issue central to the theme of complexity in the dynamics and function of biological networks concerns the inferences that can be made from multiscale measurements ranging from microscopic to macroscopic levels of integration. Our research is based on interdisciplinary approaches ranging from electrophysiology (intracellular sharp and patch recordings, in vivo), network imaging (voltage sensitive dye), psychophysical measurements to functional databasing and phenomenological and computational modeling. The long-term aim is to relate elementary processes of integration (conductance activation) to the emergence of collective « high-order » network properties expressed during low-level (non-attentive) perception.The various research axes are centered on the study of complexity in the dynamics of neocortical networks during sensory processing and percept formation, as well as during functional adaptation and plasticity:
- At the experimental level, reverse engineering techniques will be developed to extrapolate, from the synaptic echoes recorded intracellularly in a single cell, the dynamics of the effective cortical network in which this recorded cell is functionally embedded. Dynamic clamp techniques will be developed with Thierry Bal’s team to connect biological and artificial neurons, in real time in vivo. Multiscale space-frequency-time analyses will be performed on simultaneously recorded microscopic (single cell Vm, synaptic conductances) and macroscopic (VSD imaging, EEG) signals during various levels of anesthesia. A collaboration is under way with Shulz’s team to explore neuroprosthetic applications.
– At the theoretical level, computer-based simulations will be used in two ways : i) neuroinformatics : an integrated database pooling a ten-year period of electrophysiological exploration is developed in collaboration with Andrew Davison’s team, and will serve to generate structural and functional models obtained through international collaborations (Ad Aertsen, Anders Lansner, Wolfgang Maass, Guillaume Masson, all partners in a European integrated project (Facets)); ii) computational neuroscience : large-scale numerical simulations are used to test general algorithms of associative plasticity and predict the dynamic behaviour of constrained recurrent networks, working near the edge of a deterministic chaos.
Our work should open new perspectives on the concept of « ongoing activity » in neural networks, and more specifically « coding efficiency » in sensory neocortex.