Hippocampal place cells tend to fire when an animal is in a particular location in a given environment, termed a place field. Entorhinal grid cells tend to fire when the animal is on vertices of a regularly spaced triangular grid. While the spatial relationship between the place fields of different place cells is environment specific, the relationship between the firing fields of grid cells in the same subnetwork is the same in every environment. This fact allows universal spatial relationships to be learned once and applied to any arbitrary new environment, exemplifying the benefits of abstraction. This proposal aims to build a model of grid cell development and function based on this framework. Once built, I will explore how such a network could be reused in multiple environments. Finally, I will extend the network to apply to causal learning.