Our research focuses on fundamental principles of computation in neural circuits. At one level, we study complete neural circuits, seeking to understand how groups of neurons encode algorithms, and compute with them. At a lower level, we build detailed models of neural micro-circuits to study their dynamical repertoire, and seek clues to their computation in health and dysfunction in disease. Tying both together is the need to decode the information contained in experimentally-recorded neural activity.
Our work thus combines three approaches: large-scale dynamic systems modelling, large-scale data analysis, and computational neuroanatomy. Current prominent threads of research include:
(1) Neural ensembles in large-scale neural activity. We recently extended community detection algorithms from network theory to create the state-of-the-art neural activity clustering algorithm, capable of identifying simultaneously the number and composition of neural groups. We are currently applying these algorithms to study
(i) Neural dynamics of the developing striatum [calcium-imaging recordings by Constance Hammond#s lab, INMED, Marseille]
(ii) The organisation of the locomotion central pattern generator network in Aplysia [Voltage-senstive dye recordings by William Frost’s lab at Rosalind Franklin University (Chicago)].
(iii) Population coding in the sensory thalamus and cortex [Multi-electrode recordings by Rasmus Petersen's lab, University of Manchester]
(2) The functions and dysfunctions of the basal ganglia. We have previously developed a unifying theory of their implementation of action-selection algorithms, how action selection is controlled by the neuromodulator dopamine, and how disruption of this action selection underlies symptoms of the disparate basal ganglia-related disorders. Building on this are four collaborative projects:
(i) Paradoxical cognitive enhancement in Huntington’s Disease. With Christian Beste, cognitive neuroscientist at Ruhr-Universitat, Bochum, supplying behavioural and EEG data from control, pre- and manifest HD patients.
(ii) Dopaminergic control of action-outcome learning. With the labs of Peter Redgrave (experimental) and Kevin Gurney (computational) at Sheffield University.
(iii) Probabilistic population coding of action selection. With Mehdi Khamassi at ISIR, Universite Pierre et Marie Curie, Paris.
(3) Computation through coherence. Benchenane et al [Neuron, 2010, 66, 121] showed that periods of high coherence between prefrontal cortex and hippocampus, predictive of correct choices, corresponds to the reorganization of phase relations between neurons across these structures. Together we are constructing models of the linked hippocampal-cortical microcircuits to show how this spike reorganization arises from interactions between the microcircuits, its control by dopamine, and how coherence ultimately enables effective decision-making. With Sid Wiener (experimental), College de France, and Karim Benchenane (experimental), Universite Paris 7.