It is important to investigate the characteristics of nerve nets from the macroscopic viewpoint, in order to elucidate the information processing of the brain composed of an enormous number of neurons. The present paper aims at establishing a mathematical theory of nerve nets mainly from the macroscopic viewpoint.
A method of statistical neurodynamics is proposed in the beginning, ant it is applied to nets of randomly connected neurons. The method is proved to be useful not only for deriving the macrostate equations of large-scale nets and analyzing their behaviors, but also for deriving microscopic features of statetransition. Possible mechanism of association, abstraction and conceptformation is shown, which might take place in large-scale nerve nets.
