Three types of remapping with linear decoders: a population-geometric perspective
Hippocampal remapping, in which place cells form distinct activity maps across different environments, is a robustly-observed phenomenon with many theories and interpretations. Some theories view remapping as the mechanism behind reduced interference between competing spatial memories, whereas others associate it with changes in an underlying latent state representation. However, it remains unclear how these interpretations of remapping relate to one another, and what types of activity changes they are compatible with. To shed some light on these questions, here we propose a neural coding and population geometry perspective to unify and elucidate the mechanisms behind remapping. Assuming that hippocampal population activity can be understood through a linearly-decodable latent space, we show that there are three possible mechanisms to induce activity changes that resemble remapping. Remapping can be due to (i) a true change in the mapping between neural and latent space, (ii) modulation of activity due to non-spatial mixed selectivity of place cells, or (iii) neural variability outside of the latent space that reflects a redundant code. We simulate and visualize examples of these remapping types in a network model, and relate the resultant remapping behavior to various models and experimental findings in the literature. Overall, our work serves as a unifying framework with which to visualize, understand, and compare the wide array of theories and experimental observations about remapping, and may serve as a testbed for understanding neural response variability under various experimental conditions.