C1 - Neural circuits for flexible gain modulation and inter-areal coordination
Network models [Brozovic et al. 2007] and simultaneous multi-channel recordings [Westendorff et al. 2010] emphasized the role of inter-areal topdown control in context-specific sensorimotor transformations and motor planning. Here we plan to extend BCCN-I approaches by implementing new strategies for the circuit-level modeling of multi-areal interactions. The research strategy of this project is twofold.
On one side, we aim to improve our theoretical understanding of general circuit mechanisms mediating the flexible coordination of the activity of multiple neuronal populations. We will investigate how the interactions between sub-circuits shape the dynamical and functional properties of systems of increasing complexity, from multi-layer cortical columns and hypercolumns [Battaglia and Hansel 2009, Chen et al. 2010] to small sets of interacting brain areas [Battaglia et al. 2009] and whole-brain-scale networks of interlaced simpler structural motifs. Our preliminary results suggest that the effective connectivity between interacting brain areas or local circuits can be dynamically reorganized through the self-organization of coherent neuronal oscillations. Global "effective rewiring" can thus be induced without need for extended changes in actual synaptic connections [Battaglia et al. 2007, Battaglia et al. 2010, Kirst et al. 2010].
On the other side, we intend to analyze experimental data and to develop data-driven models of specific cases of inter-areal interactions. We will focus on a key aspect of inter-areal interactions, dynamic gain modulation, which is common to motor planning [Gail et al. 2009, Gail et al. 2006] and selective visual attention [Treue et al. 1999] (both investigated in BCCN-I projects). A large amount of data has been collected in the labs of A. Gail and S. Treue from simultaneous multi-channel microelectrode recordings from pairs of interacting areas in the primate cortex during cognitive tasks expected to induce strong top-down inter-areal interactions (C3, D2). Simultaneous recordings will be analyzed with statistical tools like Granger Causality or Transfer Entropy in order to detect fast dynamic changes in the inter-areal effective connectivity occurring across different conditions [Witt et al 2010]. The reliability of these methods will be benchmarked through connectivity inference assays based on synthetic neuronal activity time-series [Stetter et al. 2010]. One hypothesis, which will be tested through causal analysis, is that context-dependent motor plans in posterior parietal cortex critically depend on top-down modulation exerted by dorsal premotor cortex [Brozovic et al. 2007].
Available data will contribute to constrain the free parameters of predictive network models. Extensive simulations of circuits involving multiple functional modules with a ring-topology [Battaglia and Hansel 2009] will be used to characterize and predict patterns of neural response modulation in different attentional or behavioral conditions. In silico predictions will thus forerun more challenging future experiments in which the neural circuit dynamics will be experimentally perturbed by in vivo reversible (pharmacological or cryo-loop) manipulations of selected areas.
Is part of Section C