After characterizing briefly different theories of memory the author comes to the conclusion that the research into the mechanism of the formation of memory traces (engrams) so far has not brought such data as would confirm significantly enough the Justification of one theory or another or of one hypothesis or another. The process of formation of memory traces Is characterized by the author as a process of the conversion of nervous frequency impulsive (time-arranged, sequential) codes of information into structurally synchronous codes of information. The memory trace is characterized as a recording of this structurally synchronous code enabling the information to be transferred in time and thereby enabling this information to be used again for the direction of neuro-psychic activities and behaviour. Memory is described as being syntagmatically structuralized, hierarchically arranged, relatively closed informatory system of memory traces. By the formation of new memory traces both the universality of this system and the architectonics of its structure are changed, and so are its quantitative as well as qualitative properties. Each new memory trace not only fixes itself, but simultaneously also fits itself into the existing system, whereby this system is being constantly interfered with and re-closed, its structural and functional properties are being changed. Memory enters into a system of those conditions on whose basis the existing programmes of neuro-psychic activities and behaviour are modified and new ones formed (see picture), and this makes it necessary for us to conceive memory not only as a product, but also as an important determining factor of psychic ontogenesis.