Past events

Grid cells in the rat entorhinal cortex display strikingly regular firing responses to the animal's position in 2-D space, and have been hypothesized to form the neural substrate for dead-reckoning. I will address two theoretical questions that arise from this remarkable experimental discovery: First, how is grid cell dynamics generated in the brain. Second, what information is conveyed in grid cell activity. In discussing the first question, I will focus on continuous-attractor models of grid cell activity, and ask whether such models can generate regular triangular grid responses based on inputs that encode only the rat's velocity and heading direction. In a recent work, we provided a proof of concept that such models can accurately integrate velocity inputs, along trajectories spanning 10-100 meters in length and lasting 1-10 minutes. The range of accurate integration depends on various properties of the continous-attractor network. After presenting these results, I will discuss possible experiments that may differentiate the continuous-attractor model from other proposed models, where activity arises independently in each cell. In the second part of the talk, I will examine the relationship between grid cell firing and rat location, asking what information is present in grid-cell activity about the rat's position. I will argue that, although the periodic response of grid cells may appear wasteful, the grid-cell code is in fact combinatorial in capacity, and allows for unambiguous position representations over ranges vastly larger than the ~0.5-10m periods of individual lattices. Further, the grid cell representation has properties that could facilitate the arithmetic computation involved in position updating during path integration. I will conclude by mentioning some of the implications for downstream readouts, and possible experimental tests.