Yicong Zheng

The paper shows grid cells decorrelate position representations and act as a pattern separation device, similar to a hash function, enhancing episodic memory.

This study builds a neural network model of the EC-HPC system showing that EC2a supports pattern separation through recurrent inhibition and conjunctive coding, while EC2b/3 supports generalization.

This paper shows that drift in temporal context representations in a hippocampal model enhances recall and explains spacing effects through decontextualization and error-driven learning.

The paper introduces a computational model called Theremin that uses error-driven learning in the hippocampus to improve episodic memory function and learning speed compared to Hebbian learning models.