Abstract
This paper examines the synaptic models with interdependence of excitatory presynaptic terminals. In the cases of the presynaptic spike trains such as semi-Markov processes with three types of transition matrices, it analyzes probabilistic and statistic properties of these models. To clarify these properties, the information theory and the serial correlation coefficient are introduced under the consideration that information is carried not simply by the mean frequency but by the detailed sequence of spikes themselves. The average mutual information I (X : Y) is a measure for the description of discrimination characteristics of the presynaptic temporal pattern and the average output information H (Y) is that of the input-out frequency transfer characteristics. In addition, the serial correlation coefficient is a measure for the statistical dependence of interspike intervals.
As a result of this application, the pre-synaptic statistics was found to influence the corresponding post-synaptic discharge. The influence is closely related to the degree of interdependence between the EPSP size and the presynaptic statistics represented by transition matrices of semi-Markov processes.
