Nitsch, A. | bioRxiv Science

bioRxiv2024

Integrating episodic and spatial context signals in the hippocampus

Episodic and spatial memory are the two key components of the mnemonic system in humans. Episodic memory enables us to remember events from the past whereas spatial memory enables us to form a map-like representation of the environment. Interestingly, these seemingly different functions rely on the same brain structure: the hippocampus. Yet, how the hippocampus supports both at the same time remains unclear. Here, we tested the hypotheses that the hippocampus supports these two systems either via a common coding mechanism or via a parallel processing mechanism. To this end, we combined functional magnetic resonance imaging (fMRI) with an episodic life-simulation task and a spatial virtual reality task to manipulate episodic and spatial context associations of objects. We then investigated fMRI adaptation effects between these objects as a result of shared contexts. We found that the fMRI signal in the anterior hippocampus scaled with the combined prediction of shared episodic and spatial contexts, in line with the idea of a common coding mechanism. We found no evidence for a parallel processing mechanism, as there were no differences between episodic and spatial effects. The common coding effect for episodic and spatial memory dovetails with the broader notion of domain-general hippocampal cognitive maps.

neuroscience↗

bioRxiv2023

Grid-like entorhinal representation of an abstract value space during prospective decision making

Everyday decisions require us to predict how valuable different choice options will be in the future. Prior studies have identified a cognitive map in the hippocampal-entorhinal system that encodes relationships between states and enables prediction of future states, but does not inherently convey value during prospective decision making. Here, we investigated whether the entorhinal cortex integrates relational information about changing values by representing an abstract value space. To this end, we combined fMRI with a prospective decision making task that required participants to track and predict changing values of two choice options in a sequence. Such a sequence formed a trajectory through an underlying two-dimensional value space. Our results show that participants successfully integrated and extrapolated changes along the two value dimensions. Participants choice behavior was explained by a prospective reinforcement learning model and the degree to which they updated values over time correlated with self-reported navigational abilities and preferences. Crucially, while participants traversed the abstract value space, the entorhinal cortex exhibited a grid-like representation, with the phase of the hexadirectional fMRI signal (i.e., the orientation of the estimated grid) being aligned to the most informative axis through the value space. A network of brain regions, including the ventromedial prefrontal cortex (vmPFC), tracked the prospective value difference between options and the occipital-temporal cortex represented the more valuable option. These findings suggest that the entorhinal grid system might support the prediction of future values by representing a cognitive map, which might be used to generate lower-dimensional signals of the value difference between options and their identities for choices. Thus, these findings provide novel insight for our understanding of cognitive maps as a mechanism to guide prospective decision making in humans.

neuroscience↗