Grid to Place Cell Connectivity in Eleven Different Rooms
Aggarwal, A.
Show abstract
To understand the grid to place cell connectivity, we took place cell firing data from the Moser lab. The data included single cell recordings from 342 CA3 neurons in 8 animals in 11 different rooms. Of these 342 cells, only 2 fired in all the 11 rooms and over 100 fired in one room. In MATLAB, we created grid cell firing patterns for 4500 grid cells. Connection weights between the place and grid cells were learned using machine learning-gradient descent algorithm. The smaller place fields could be learned from grid cells with single spatial firing frequency. But bigger, multiple and irregular place fields could only be learned from grid cells with multiple spatial firing frequencies. Weights learned were normally distributed with a wider spread and multimodal distribution for rooms with uneven, larger or multiple firing fields. Place cells connected to multi-frequency grid cells are fewer. We conclude that each place cell is connected to single modules of grid cells with similar spatial firing frequency. Our results also show that grid cells resolve the space into spatial distance, orientation, and phase offset. Unique firing patterns of the place cells codify each room with this information.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.