An Information-Theoretic Analysis of Temporal Coding Strategies by Spiking Central Neurons

The brain encodes information in the intervals between the spikes which characterize neural firing events. Therefore it is relevant to study in a timing code how many spikes are necessary for reliably encoding input signals. We analyze the transmission of information, the reliability of signal detection and the coding strategy for the case of central neurons which contrary to peripheral sensory neurons handle input signals assumed to be given by a combination of Poisson spike trains. We consider an integrate-and-fire model of a central neuron which combines diffusion and jump processes. In order to obtain analytical results, we introduce in addition a new Renyi-Information based measure for the discrimination ability of single neurons, which is investigated in the framework of a simple spike response model.