Analysis of Fluctuations of Minimal EPSPs in Vitro: Quantal Model

Abstract

Traditional methods of the quantum hypothesis (Chap. 7) were used in previous chapters which assumed the simple binomial model. However, deviations from the binomial predictions were described in some cases and they were explained by a non-uniform release probability for different release sites (Sect. 7.7). In the case of a spatially non-uniform p, the simple binomial model might lead to an erroneous description of the release statistics [85, 175, 635, 806]. The technique of “noise deconvolution” was mentioned in Section 7.3.7 as an alternative to the methods based on the binomial or Poisson models. The deconvolution procedure, although having its own difficulties and limitations [81, 525, 823] (see especially [230] and Sect. 7.3.7), allows to obtain “noise-free” discrete distributions of EPSP amplitudes from experimental noisy histograms without assumptions regarding the distribution of underlying release probabilities and it allows both non-uniform p and different v at different release sites. With application of this technique to the spinal cord synapses (Sect. 7.3.7), the quantal nature of synaptic release was confirmed, but it was found that the amplitudes of spinal EPSPs were not distributed according to the Poisson or (in general) binomial law. A “compound” binomial model was proposed with a uniform v but with a variety of probabilities (p_i) associated with different release sites [823, 1068]. The aim of our last experimental series [1024, 1059, 1060, 1063] was to test the conclusions related to the mechanisms of LTP maintenance, using a variant of the deconvolution technique which does not assume a simple binomial model of transmitter release.