Investigating human grid-cell-like representations and path integration in the context of cognitive aging

Abstract

Grid cells in the entorhinal cortex are a central component of the brain’s spatial navigation circuit. They are largely thought to support the continuous tracking of one’s position in space by integrating self-motion cues, a function called path integration. Analyzing the putative firing of grid cells (i.e., grid-cell-like representations) in human neuroimaging data, however, is a non-trivial endeavor, given the methodological complexity and the absence of standard software tools for this analysis. In Project A of this thesis, we therefore developed the Grid Code Analysis Toolbox (GridCAT), a MATLAB-based open-source software for the automated analysis of human grid-cell-like representations in functional magnetic resonance imaging (fMRI) signals, and made this software openly available to the neuroscience community. In Project B of this thesis, we then applied the GridCAT to fMRI data from young and older adults, in order to investigate age-related changes in grid-cell-like representations. We found that grid-cell-like representations in the entorhinal cortex were compromised in old age, and this effect was mainly driven by a reduced stability of grid orientations over time. Building on this finding, in Project C we then investigated whether compromised grid-cell-like representations might be associated with path integration deficits in old age. Indeed, we found that individual magnitudes of grid-cell-like representations were predictive of age-related deficits in a behavioral path integration task, in which participants had to navigate based on integrating body-based or visual self-motion cues. On the one hand, these findings demonstrate that compromised grid-cell-like representations might be a key mechanism to explain reduced path integration performance in old age. On the other hand, however, it remained to be determined whether and to what extent other sources of error might also contribute to path integration errors in young and older adults. To address this, in Project D we used a computational modeling approach in order to decompose path integration errors into distinct causes that can corrupt path integration computations. We identified internal noise in path integration computations and a biased gain in estimating the speed of self-motion as the main error sources across both young and older adults, with an increase in internal noise accounting for the majority of age-related path integration deficits. Together, the work in this thesis not only advances our understanding of the specific contributors to path integration error, it also helps elucidate the mechanisms that underlie age-related decline in navigational functions.