The Significance of the Study of the Brain's Hippocampus for the Progress of Biologically-Inspired Computational Systems

Introduction

The hippocampus is an important area of the brain that contributes significantly to memory and navigation. Alzheimer’s disease and other forms of dementia can affect and even damage the specific area. The hippocampus is also closely related to epilepsy, schizophrenia and post-traumatic stress disorder (The University of Hong Kong, 2017). The complexity of hippocampus can be shown by the fact that hippocampal place neurons do not only represent current location but also, they fire in sequences and these sequences appear to simulate past and future spatial trajectories. The firing sequences match the structure of a complex maze (Jeffery & Casali, 2014). Furthermore, the role of hippocampus in complex brain networks and especially how it integrates with other brain regions is not well understood by scientists (The University of Hong Kong, 2017).

An example of the complexity of the anatomy and of the operations of hippocampus is spatial navigation. Spatial navigation is based on a network of interconnected networks that include the hippocampus, the prefrontal cortex and the basal ganglia. The exploration of the flows of information between these structures will reveal the processes involved to learning complex navigational tasks (Chersi & Burgess, 2015). The hippocampus is long known to be associated with episodic memory (Scoville & Milner, 2000) and the discovery of place cells play a significant role in spatial memory (O’Keefe & Nadel, 1978).

During spatial navigation, cognitive maps are developed. The instantiation of the cognitive maps is done by place, grid, border and head direction cells that are located mainly in the hippocampus. The human hippocampus and the entorhinal cortex support map-like spatial codes while parahippocampal and retrosplenial cortices allow cognitive maps to be related to specific environmental landmarks. The hippocampal and entorhinal spatial codes are used in coordination with the frontal lobe in order to develop navigational routes (Epstein et al., 2017).

The analysis of the spatial navigation capabilities of the hippocampus shows its complexity and this can be the basis for the re-evaluation of current AI systems and the development of new systems based exactly on this complexity. In recent years there has been increment in the complexity and quality of Artificial Intelligence systems. Examples of the evolution of Artificial Intelligence are the Deep Learning Theory, the Deep Reinforcement Learning, the Generative Adversarial Networks, the Lean and Augmented Data Learning, Probabilistic Programming, the Hybrid Learning Models, and the Automated Machine Learning (Rao et al., 2018). Since 2006, deep learning has emerged as a new era of machine learning research ((Bengio, 2009), (Hinton et al., 2006)). Deep Learning is a set of computational methods that allow an algorithm to program itself through learning from a large set of examples that show the desired behaviour (Gulshan et al., 2016). The deep learning techniques impact key areas of machine learning and artificial intelligence.

This paper aims to present how the study of the functions of the brain’s hippocampus can result to the extraction of conclusions that will help in the development of more advanced Artificial Intelligence systems. The paper presents a range of indicative applications from different fields that are based on the characteristics of hippocampus. It then presents arguments on how specific characteristics of the function of the hippocampus can impact the development of biologically-inspired systems. Finally, a range of conclusions are drawn that show how the study of hippocampus can be the basis for the improvement of current biologically-inspired systems.