Abstract
Hippocampal place cells and entorhinal grid cells are thought to form a representation of space by integrating internal and external sensory cues. Experimental studies show that different subsets of place cells are controlled by vision, self-motion or a combination of both. Moreover, recent studies in environments with a high degree of visual aliasing suggest that a continuous interaction between place cells and grid cells can result in a deformation of hexagonal grids or in a progressive loss of visual cue control. The computational nature of such a bidirectional interaction remains unclear. In this work we present a neural network model of a dynamic loop between place cells and grid cells. The model is tested in two recent experimental paradigms involving double-room environments that provide conflicting evidence about visual cue control over self-motion-based spatial codes. Analysis of the model behavior in the two experiments suggests that the strength of hippocampal-entorhinal dynamical loop is the key parameter governing differential cue control in multi-compartment environments. Construction of spatial representations in visually identical environments requires weak visual cue control, while synaptic plasticity is regulated by the mismatch between visual- and self-motion representations. More gener-ally our results suggest a functional segregation between plastic and dynamic processes in hippocampal processing.