O.J.-M. D. Coenen, D. Philipona and E. Dognin MODEL OF GRANULAR LAYER ENCODING IN THE CEREBELLUM. Society for Neuroscience Abstracts, 2003

Sony CSL authors: Olivier Coenen, Edouard Dognin, David Philipona

Abstract

Involved in sensorimotor and even cognitive coordination, the cerebellum receives inputs from diverse brain areas. These inputs are recoded by the granule cells, the most numerous cells in the brain, and together with the inhibitory Golgi cells, form the granular layer. A model is presented that suggests that the granular layer performs a recoding of the mossy fibers into a sparse representation that permits efficient noise reduction by the Golgi cells and facilitates learning in the molecular and Purkinje layer of the cerebellum. Plasticity adapts the synaptic weights in the network to align the granule cells responses to statistically sparse directions in the mossy fibers space. Using the sparsity of this encoding, the Golgi cell computes an estimate of the variance of the Gaussian noise component present in the mossy fibers inputs, which is then used through its inhibition for denoising the parallel fiber signals. Hence, although a simple network, we suggest that the granular layer contains different types of signal processing that lead to a robust and efficient representation for coordination. Support Contributed By: EC contract IST-2001-35271 Project Spikeforce (www.spikeforce.org)

BibTeX entry

@INPROCEEDINGS { coenen:03a, AUTHOR="O.J.-M. D. Coenen and D. Philipona and E. Dognin", BOOKTITLE="Society for Neuroscience Abstracts", TITLE="MODEL OF GRANULAR LAYER ENCODING IN THE CEREBELLUM", YEAR="2003", }