Search Papers | Poster Sessions | All Posters
Poster B99 in Poster Session B - Thursday, August 8, 2024, 1:30 – 3:30 pm, Johnson Ice Rink
A Multi-Region Brain Model to Shed Light on the Role of Hippocampus in Spatially Embedded Decision Tasks
Yi Xie1 (), Jaedong Hwang2, Carlos Brody1,3, David Tank1, Ila Fiete2; 1Princeton Neuroscience Institute, Princeton University, 2Massachusetts Institute of Technology, 3Howard Hughes Medical Institute
We present a multi-region brain model exploring the entorinal-hippocampal region's role in spatially embedded decision tasks. Utilizing an accumulating-tower task, we simulate decision-making processes that reflect the cognitive maps formed within the CA1 region of the hippocampus. Our model integrates a bipartite memory scaffold architecture that incorporates grid and place cells, with a recurrent neural network (RNN) to simulate action selection based on sensory inputs and grid cell representations. We demonstrate that joint encoding of position and evidence information within medial entorhinal cortex (MEC) and CA1 in the model replicates experimental observations of place cell behaviors, and promotes rapid learning. Our findings suggest the grid cells are co-tuned to position and evidence.
Keywords: grid cell place cell evidence accumulation cognitive maps