Learning grid cells by predictive coding

Abstract

Grid cells in the medial entorhinal cortex (MEC) of the mammalian brain exhibit a strikingly regular hexagonal firing field over space. These cells are learned after birth and are thought to support spatial navigation but also more abstract computations. Although various computational models, including those based on artificial neural networks, have been proposed to explain the formation of grid cells, the process through which the MEC circuit ${\it learns}$ to develop grid cells remains unclear. In this study, we argue that predictive coding, a biologically plausible plasticity rule known to learn visual representations, can also train neural networks to develop hexagonal grid representations from spatial inputs. We demonstrate that grid cells emerge robustly through predictive coding in both static and dynamic environments, and we develop an understanding of this grid cell learning capability by analytically comparing predictive coding with existing models. Our work therefore offers a novel and biologically plausible perspective on the learning mechanisms underlying grid cells. Moreover, it extends the predictive coding theory to the hippocampal formation, suggesting a unified learning algorithm for diverse cortical representations.

Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.