Coordinated Learning of Entorhinal Grid Cells and Hippocampal Place Cells: Space, Time, Attention and Oscillations

Abstract

The GridPlaceMap neural model explains and simulates how both entorhinal grid cells and hippocampal place cells may develop as spatial categories in a hierarchy of self-organizing maps by detecting, learning and remembering the most frequent and energetic co-occurrences of their inputs. The model simulates challenging behavioural and neurobiological data by embodying several parsimonious neural designs: Similar ring attractor mechanisms process the linear and angular path integration inputs that drive map learning; the same self-organizing map mechanisms can learn grid cell and place cell receptive fields; and the learning of the dorsoventral organization of multiple spatial scale modules through medial entorhinal cortex to hippocampus may use mechanisms homologous to those for temporal learning (“time cells”) through lateral entorhinal cortex to hippocampus (“neural relativity”). Top-down hippocampus-to-entorhinal attentional mechanisms stabilize map learning, simulate how hippocampal inactivation may disrupt grid cells and help to explain data about theta, beta and gamma oscillations.