Encoding of stimuli in embodied neuronal networks

Abstract

Information coding in the central nervous system is still, under many aspects, a mystery. In this work, we made use of cortical and hippocampal cultures plated on micro-electrode arrays and embedded in a hybrid neuro-robotic platform to investigate the basis of "sensory" coding in neuronal cell assemblies. First, we asked which features of the observed spike trains (e.g. spikes, bursts, doublets) may be used to reconstruct significant portions of the input signal, coded as a low-frequency train of stimulations, through optimal linear reconstruction techniques. We also wondered whether preparations of cortical or hippocampal cells might present different coding representations. Our results pointed out that identifying specific signal structures within the spike train does not improve reconstruction performance. We found, instead, differences tied to the cell type of the preparation, with cortical cultures showing less segregated responses than hippocampal ones. © 2013 Springer-Verlag Berlin Heidelberg.