When and Where: A Model Hippocampal Network Unifies Formation of Time Cells and Place Cells

1. 1The model uses a predictive autoencoder to simulate the hippocampal CA3 region
2. 2The network receives spatial and temporal inputs and generates place and time cells through two dynamical regimes
3. 3The place cells emerge as continuous attractors, while time cells emerge as leaky integrators
4. 4The model provides a unified framework for understanding the neural basis of spatial and temporal cognition

This research benefits neuroscientists and AI engineers working on neural networks and cognitive modeling, as it provides a new understanding of how the brain encodes spatial and temporal information

💡 A single recurrent neural network can generate both time cells and place cells, providing a new understanding of how the brain encodes spatial and temporal information

💡 Unifying time & place cells in a single RNN model! 🤖