A model for how grid cells contribute to complementary concept representations in the hippocampus

The hippocampus encodes relational memory. The content of hippocampal memories can be spatial or nonspatial (e.g., conceptual knowledge). Moreover, complementary learning systems within the hippocampus encodes distinct levels of representational granularity. For spatial information, these systems encode coarse- and fine-grained details of one’s spatial surrounding. For conceptual information, these systems are hypothesized to encode integrated and differentiated category knowledge. While computational models of this system have elucidated noteworthy mechanisms that the hippocampus may use to achieve these feats, most have considered the hippocampus in isolation. Here, the influence of the entorhinal cortex on complementary learning systems within the hippocampus is modeled as well. The proposed model of the entorhinal-hippocampal circuit offers plausible mechanisms by which it achieves a suite of entorhinal and hippocampal functions that are robustly observed including pattern separation, pattern completion, cognitive maps, and hexagonal firing patterns in entorhinal grid cells. This model offers mechanistic and theoretical insights into the contributions that grid cells make to concept representations and nonspatial cognitive maps in the hippocampus.