Long-term Memory Structure by Hierarchized SOM

The most evolved living organisms are characterized by having a sophisticated central nervous system with memorizing capabilities. In its structure are the neuronal cells. Cognitive memory allows the recording, storage, renewal and reading of information. In this work, memory is characterized by its functional and structural aspects and the main mechanisms for the learning and storage processes are also presented. A hierarchical model based on artificial neural networks is proposed as a conceptual model of information transfer between short and longterm memory. Inspired in recent studies, this model is based on an hierarchized structure, composed by layers of neurons with an appropriated topology and learning mechanism suitable for the retention of information in an organized way. A set of simulations is performed to test and evaluate the performance of this neuronal structure, particularly on how the information is distributed.

I. Introduction

The human brain is the most complex and extraordinary structure ever known. This organ is capable, among many other functionalities, of receiving information from the sensory organs of the peripheral nervous system (hearing, sight, smell, etc.), analysing it, processing it, and generating output stimuli. If necessary, it can store this information for future use [1]. It is imperative for the achievement of cognitive intelligence [2], participating in pattern recognition tasks and facilitating the organization of relevant information, making it available in different ways (speech, writing, movement, among others). All of these processes involve cellular, electrical and chemical neuronal changes in the central nervous system, which are carried out dynamically and at different temporal scales [3][4], When subjected to new excitation signals, these structures may be strengthened, weakened or changed, by creating new synapses [5].