A model of joint hippocampal and cortical representation of space and context

Manipulating sensory cues or changing an animal’s task sometimes leads to spatial representation remapping: a phenomenon whereby the preferred location of all place-cells changes. It has been recently discovered (Scheintuch et al., 2020) that a phenomenon akin to remapping occurs even when no manipulation is made to the environment, to the state of the animal or to the behaviour it performs. In agreement with experimental results, we assumed that some non-spatial (contextual) information, which describes the internal state of the animal and occurs in the cortical region of the brain, is responsible for the remapping phenomenon without spatial manipulation. We developed an attractor model, based on a Hopfield network, which allows the coexistence of multiple stable maps for each single environment representing different contexts. Cortical inputs to the hippocampus are then precise enough to initialize the network in the “correct” basin of attraction and, therefore, reproduce an accurate representation of the environment in a certain context.