@ INCM (Marseille) April 11th, 2008
- Laurent Perrinet. From neural activity to behavior: computational neuroscience as a synthetic approach for understanding the neural code., URL . 2008 abstractComputational Neuroscience is a synthetic, inter-disciplinary approach aiming atunderstanding cognition by analyzing the mechanisms underlying neural computations. Wepresent in this seminar our attempt in modeling low-level vision by bridging differentintegration levels, from neural spiking activity to behavior. At the behavioral level, the OcularFollowing Response recorded in the laboratory reveals how the brain may integrate localinformation (moving images on visual receptive fields) to produce a single behavioralresponse (the movement of the eye). Using a probabilistic representation, we provide asimple integrative mechanism that gives the ''ideal'' response to possibly noisy andambiguous information, similarly to a Bayesian approach. This fits well the performancerevealed by behavioral data and may act as a generic cortical ''module''. At the populationlevel, these mechanisms may indeed be implemented for the coding of natural images andwe will show the particular importance of spiking representations and lateral interactions forefficient and rapid responses. In particular, we will present an original unsupervised learningalgorithm that we applied to a model of the primary visual cortex. Finally, at the neuronallevel, I will present work done in the team showing how certain mechanisms at the level ofthe synapse and of the neuron are essential at the population level and how we mayunderstand these mechanisms at the population level. This illustrates the importance ofdynamical processes, distributed activity and recurrent connections to produce a cortical gaincontrol mechanism. As a conclusion, this approach provides useful applications for imageprocessing and possible valorization in future computer architectures. More generally, itproves that the use of a probabilistic representation is a particularly efficient method forbridging biological versus computational neuroscience and illustrates the advantage of suchan interdisciplinary approach..