The abstraction we need to use always includes simplifications and omissions, depending on which phenomena we want to study. If we want to describe Earth’s orbit around our Sun, we can consider it point-like at an elementary level and assume pairwise interaction between them. However, for finer details, we need to consider its structure, size, and the disturbing effects of other planets and its Moon. Whether the abstraction of having point-like neurons is valid depends on the targeted phenomenon.
In the initial investigations, the size of the cells was seen to be much smaller than the size of their connections. In addition, the axons were much earlier available for experimental investigations, suggesting that the observed signals originate and terminate in the network nodes. This abstraction might be appropriate (with some limitations, mainly due to the connection speed) until we can develop technical tools to study the internal operation of the network nodes. ”We assume that the dimensions of the cell are small enough so that spatial variations in the membrane potential can be neglected” [24]. Its internal operations and phenomena can only have an artificial timing, its input signals are artificially correlated, and its mystic internal operation produces an action potential as an output signal in a pair-wise interaction.
When starting from ’the so-called point representation of a neuron” [24], admitting that ”such an approximation would be valid, for instance, if we were investigating a small, spherical cell without a significant dendritic tree”, we necessarily conclude that ”individual neurons convert the incoming streams of binary pulses into analog, spatially distributed variables”. This statement attempts to underpin that in the neural networks digital pulses are traveling, which is less then the half of truth. This point of view blocks the interpretation of even the phenomena that are correctly seen and leads to the design of wrong experiments. Among others, it results in the immediate consequence of interpreting neuronal communication as streams of binary pulses, which leads to applying Shannon’s mathematical theory to neural communication, despite Shannon’s sharp opposition [33].