Building a continuous attractor model of dMEC grid cells using conductance-based neurons

Abstract

The work done in this thesis attempts to model the grid cell network, pivotal in spatial navigation and awareness, using biologically realistic descriptions of neurons. While previous work on this system used rate-based models to simulate the activity of neurons, these models are not able to realistically represent the dynamics of neuronal behaviour. In this thesis, we use conductance-based models of stellate cells and fast-spiking interneurons to construct two types of networks built using a primitive MEC motif. These networks are able to show the formation and movement of an idealized bump of activity along a single dimension. The networks and the translation of the bump are able to show direction- as well as speed-tuning. Theta oscillations are shown to play a role in tunability and control of the system and may play a part in stabilzation of the temporal activity profile. The neuronal networks presented here provide the initial steps in making a full, biologically realistic, model of the grid cell network.