Spatiotemporal coding in the cortex: Information flow-based learning in spiking neural networks

We introduce a learning paradigm for networks of integrate-and-fire spiking neurons that is based on an information-theoretic criterion. This criterio n can be viewed as a first principle that demonstrates the experimentally o bserved fact that cortical neurons display synchronous firing for some stim uli and not for others. The principle can be regarded as the postulation of a nonparametric reconstruction method as optimization criteria for learnin g the required functional connectivity that justifies and explains synchron ous firing for binding of features as a mechanism for spatiotemporal coding . This can be expressed in an information-theoretic way by maximizing the d iscrimination ability between different sensory inputs in minimal time.