P328 Modelling the impacts of Alzheimer’s Disease and Aging on Self-Location and Spatial Memory
Aleksei Zabolotnii*1, Chrsitian F. Doeller1,2,3, Andrej Bicanski1,3
1Department of Psychology, Max-Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany 2Kavli Institute for Systems Neuroscience, NTNU, Trondheim, Norway 3Wilhelm Wundt Institute for Psychology, Leipzig University, Germany
*Email: zabolotnii@cbs.mpg.de Introduction
Spatial navigation relies on the precise coordination of multiple neural circuits, particularly in the entorhinal cortex (EC) and hippocampus (HPC). Grid cells in the EC play a critical role in path integration, while place cells in the HPC encode specific locations. Dysfunction in these systems is increasingly linked to cognitive decline in aging and Alzheimer’s disease (AD)1. Early AD is characterized by EC dysfunction, including impaired neuronal activity and deficits in spatial navigation, even before neurodegeneration becomes evident2. Similarly, cognitive decline comes with aging and affects navigational computations3. Here we investigate both kinds of deficits in a mechanistic systems-level model of spatial memory.
Methods We extend the BB-model of spatial cognition4 with a biologically plausible variant of continuous attractor network (CAN) model of grid cells5 and investigate the effect of perturbations on grid cells and the wider spatial memory system. Specifically, we investigate the stability against synaptic weight variability, and neuronal loss, the former (to a first approximation) more akin to age-related neural degradation, and the latter mimicking AD-associated neurodegeneration. To quantify the impact of these perturbations, we analyzed the propagation of degraded spatial representations to downstream hippocampal and extra-hippocampal circuits and evaluate changes in the accuracy of self-location decoding from grid cells. Results We demonstrate that our biologically plausible grid cell model can cope with neural loss and changes in synaptic weights, both of which lead to distortions of the activity pattern on the grid cell sheet. Positional decoding degrades gracefully. We also observe the propagation of distorted spatial representations to downstream areas during the imagery-associated mode of the BB-model, as well as deficits in object-location memory. Discussion Our model demonstrates for the first time in a mechanistic model how neural degenerative processes affect spatial accuracy. As damaged EC populations produce distorted activity, it causes imprecise firing of place cells as well as leads to forming distorted memories for locations of novel objects in the environment. Due to changes in the CAN, population activity vectors are unable to provide a correct and unique code for every location in space compared to those in the healthy system, linking our model to the spatial behavior of AD patients and aging adults.
Acknowledgements Aleksei Zabolotnii acknowledges the DoellerLab and the Neural Computation Group. Andrej Bicanski and Christian F. Doeller acknowledge funding from the Max Planck Society References 1. https://doi.org/10.1126/science.aac8128 2. https://doi.org/10.1016/j.cub.2023.09.047 3. https://doi.org/10.1016/j.neuron.2017.06.037 4. https://doi.org/10.7554/eLife.33752 5. https://doi.org/10.1371/journal.pcbi.1000291
