A coordinated spiking network model of the hippocampus accounts for remapping and inhibitory perturbations.

Authors: Guillermo Martín-Sánchez, William Podlaski, Christian K. Machens

Presentation type: Poster

Abstract

Neurons in the hippocampus show firing tuned to the animal’s location in space. In particular, place cells show monofields in an environment, and grid cells show tuning to a hexagonal lattice of space. We propose a new model that simulates these neuron types and also accounts for remapping, the phenomenon in which place cells’ activity patterns change in response to environment changes. We use a coordinated spiking network that efficiently encodes a continuous signal into spikes and we model remapping as different embeddings of a low-dimensional manifold. Our model, different from other remapping models, also accounts for the recruitment of previously silent neurons to compensate for optogenetically inhibited place cells, and overparticipatory neurons prone to be active in a high number of environments. Overall, our work sheds light on the possible structure of the place code across different environments, draws parallels between the place and grid codes and supports the hypothesis that recruited neurons under inhibition are active place cells in other environments.