Special GNT Seminars in Theoretical Neuroscience (lnc)

December 5, 2017, 10am
Salle 235 A, 29 rue d'Ulm


10h00 - 10h45: Information encoding in neural networks,Tatjana Tchumatchenko

10h45 - 11h30: Flexible, optimal motor control in a thalamo-cortical circuit model,Guillaume Hennequin

Information encoding in neural networks (Tatjana Tchumatchenko, MPI Brain Research Frankfurt)
To understand the neural code and read out the information neural spikes convey, it is essential to understand how the information is coded and how much of it is available for decoding. To this end, it is indispensable to understand what the basic coding features of spike trains are. Here, we show that temporal pairwise spike correlations fully determine the information conveyed by a single spiking neuron with finite temporal memory and stationary spike statistics. We reveal that interspike interval temporal correlations, which are often neglected, can significantly change the total information. Next, we use this insight to learn whether different signal encoding strategies are comparable in their information coding capacity. Experiments indicate that sensory stimuli can modulate both the mean and the variance of the somatic currents in neurons. Yet, quantifying the efficiency of mean and variance modulations in conveying information at the spiking level remained an open challenge. Here, we calculate the mutual information between signals and spike trains and identify the complete information content and the linearly decodable information fraction for both modulations. We show that the information content about mean modulating signals is generally higher than about variance modulating signals.

Flexible, optimal motor control in a thalamo-cortical circuit model (Guillaume Hennequin, University of Cambridge)
How does the brain control movement? Experiments suggest that the (pre-)motor cortex behaves like an “engine” whose dynamics drive movement, and whose activity must first be initialised into movement-specific states (“the optimal subspace hypothesis”, Shenoy & al, 2013). Both the computational and mechanistic underpinnings of this preparatory process remain poorly understood. Here, we propose a realistic circuit model for movement preparation and execution. We formalise movement preparation as an optimal control problem, under an internal forward model that predicts (future) paterns of muscle activity from momentary cortical preparatory states. We compute the preparatory input to cortex that drives fastest convergence to preparatory states predicted to yield the correct motor outputs. We also request that the control input keeps preparatory activity in an appropriate “output-null” subspace to prevent unwanted, premature motor outputs (Kaufman et al, 2014). Critically, we show that optimal control inputs can be realised via feedback in realistic neural circuit architectures. Specifically, we model cortex as an inhibition-stabilised network, whose dynamics resembles those of monkey M1 during reaching (Hennequin & al. 2014). Optimal movement preparation is accomplished by a thalamo-cortical loop, gated by the basal ganglia. The loop is open by default, closed to drive movement preparation, and eventually re-opened to initiate movement. Importantly, we find that control loops can be flexibly combined to generate movements assembled from a few movement primitives. The model produces naturalistic paterns of preparatory activity, including complex transients early during preparation, and displaying substantial variability in output-null dimensions. Consistent with data, across-trial variability is suppressed during preparation. Moreover, preparation may be as short as 200 ms without loss of motor accuracy, also consistent with recent experimental observations. Our work brings together several threads of experimental research on both cortical and subcortical areas, and offers a new computational, normative perspective on the dynamics of motor circuits.