Scale analysis can be replicated in the temporal domain, decomposing brain activity in a multitude of dynamic processes with time constants ranging from microseconds to years. It is a current challenge to the field of Systems and Computational Neuroscience to bind in a coherent way these different hierarchies of organization on the basis of experimentally defined descriptors, each of which is endowed with a specific spatio-temporal domain and measurement precision.
An issue central to the theme of complexity in biological systems concerns the inferences that can be made from one level of integration to the next, along top-down (from macroscopic to microscopic) or synthetical bottom-up (from microscopic to macroscopic) axes. The question addressed by the different research themes, at UNIC, and applied in different neural structures and species, is : “To what extent can properties specific to one level of organisation be predicted by those demonstrated at lower levels of organisation?”, or, in other words, is the “whole” the sum of the “parts”? This question applies not only at the structural level but also at the functional level. The existence of distributed local non-linearities in the process of building neural assemblies is a first indication of the complexity of living networks. An additional difficulty in crossing bridges between different organisational levels is the lack of systematic methods for reducing complexity and the absence of models linking different physical observables, such as, for instance, neural activity-based measures made by electrophysiologists with metabolic and haemodynamic measures used for brain imaging.
The research carried out at UNIC is based on interdisciplinary approaches, ranging from electrophysiological techniques (intracellular sharp and patch recordings, dynamic clamp in vivo and in vitro) and functional databases, psychophysical measurements and phenomenological and computational modelling, to network functional imaging (voltage sensitive dye). The various axes of research are all centered on complexity in the dynamics of neocortical networks during sensory processing and plasticity.
A innovative skill developed in the UNIC Department has been the design and use of hybrid neuron networks, interfacing simulated and biological neurons in real-time. This relatively new technology, developed here for use in the mammalian brain, makes use of “dynamic clamp” techniques where the experimenter, while recording in vitro or in vivo from a biological neuron, can implement a virtual synaptic loop in real-time, linking numerically- (or ASIC) simulated neurons and the biological cell. By switching at high frequency (1-10 kHz) between the “record” and “current injection” modes, numerical iteration is used to solve the membrane potential trajectory of the simulated neuronal partner(s), while at the same time recording the impact on the biological cell, over a wide range of parameter sets mimicking excitatory and inhibitory synapses. This technique can be extended to simulation of the influence of on-going contextual network activity (e.g. previously recorded in the intact brain), on the input-output function of identified neurons recorded in vitro, and to the active automatic compensation of the filtering properties of the recording electrode (AEC method), also applicable in vivo.
One application of the study of cortical network dynamics in vitro, in vivo and in computo, is to conceive novel computational architectures inspired by the complexity of living networks, in collaboration with experts in the field of electronic engineering, mathematics and theoretical physics. To this end, UNIC has initiated two European consortia, combining expertise in the field of Biology and Computational Neuroscience with international colleagues possesing complementary expertise in Microelectronic and VLSI circuits, Mathematics, Computer Science and Artificial Vision. This has led to two large projects funded by the European Commission Information Society Technologies Directorate : in the FP5 Future and Emerging Technologies R&D initiative “Life-like perception” (IST-2001-34712 “SenseMaker”; 2002-05) and in FP6 FET-IST initiative “BioI3” (Integrated Project FP6-2004-IST-FETPI-15879 “FACETS”; 2005-09).).