Event:

Entorhinal grid cells are neurons with remarkable characteristics. Their activity correlates with periodic, hexagonally spaced locations in the animal’s environment. Within the brain, grid cells are co-located with various other types of neurons that also show spatially correlated activity, and their axonal output terminates in regions of the hippocampus that hosts place cells. Together, these properties suggest that grid cells may be part of a specialized neuronal system that facilitates localization and navigation. However, a number of more recent experimental observations show that grid-like activity patterns also occur in correlation with eye movements, sound frequency, or even abstract concepts. This talk presents a computational model of grid cells that entertains the perspective that the observable behavior of grid cells is one instance of a more general information processing scheme that aims at learning a detailed representation of a given input space within a local group of neurons. This alternative perspective might allow one to not only ask new questions about the activity patterns of various entorhinal neurons but may also provide a novel way to think about structural properties of possible inputs to these neurons.