So far, there have been studies that biologically considered the differentiation of neurons, but the mechanism has not been elucidated.The mathematical model of neural networks that has been proposed so far is based on the idea of ​​defining neurons as parts whose functions are known and connecting them by interaction.However, it was not possible to prove the actual differentiation.

　This time, Professor Tsuda and his colleagues devised a mathematical model based on a system consisting of parts whose functions have not been decided, based on the opposite idea to the conventional one.In this model, neurons are self-organized (spontaneous) as a component that "neurons differentiate to transmit input information most efficiently without attenuation", that is, a component that "maximizes the propagation of input information within a network". It is based on the assumption that (order formation).

　When numerical values ​​were input to this model, a model was obtained in which neurons simulated the activity of actual nerve cells.This is called a spiking neuron model, and is a mechanism that artificially reproduces action potentials (spikes) emitted by nerve cells and reproduces them as temporal waveforms.In addition, we were able to observe how the network differentiated into a structure close to the brain.This proved the correctness of the model.

　Based on this result, it is expected that the differentiation of more complicated biological functions will be elucidated, and the utilization of artificial intelligence (AI) will give momentum to the study of the differentiation of neurons that adapt to environmental changes.

Paper information:[Neuroscience Research] A mathematical model for neuronal differentiation in terms of an evolved dynamical system