Open Collections

Abstract

Spatial navigation is increasingly recognized as a sensitive behavioural marker of early Alzheimer’s disease (AD). Yet, many existing paradigms lack ecological validity, examine navigational processes in isolation, and rarely incorporate parametric control over task complexity. These limitations reduce sensitivity to the subtle and heterogeneous behavioural changes expected in healthy aging and preclinical AD. The primary objective of this thesis was to validate a novel immersive virtual reality (VR) paradigm designed to isolate and quantify three navigational processes within a single naturalistic framework: (1) egocentric path integration, (2) cue-based switching between egocentric and allocentric reference frames, and (3) allocentric scene memory. Fifty-five cognitively healthy adults (ages 18–84) completed 30 trials spanning six graded complexity levels. Participants navigated a fog-occluded hallway while tracking their movements and locations of distal landmarks. Seated in a swivel chair, they used physical body rotation for heading changes, preserving vestibular inputs critical for naturalistic navigation. Performance was assessed through pointing tasks under dynamic cue conditions, yielding six error metrics aligned with the targeted processes. Generalized linear mixed-effects models tested effects of age and complexity. Robust age-related increases in error were observed for egocentric path integration and reference frame switching, consistent with selective declines reported in prior literature. In contrast, complexity manipulations did not produce clear effects, likely reflecting confounds with task learning in the fixed-block design. Exploratory covariate analyses further suggested individual differences related to task performance. Together, these findings demonstrate the feasibility and sensitivity of the paradigm in dissociating distinct navigational processes while maintaining both ecological realism and experimental control. At the same time, methodological challenges—including confounding effects of learning, broad complexity definitions, and unbalanced age distributions—limit interpretability and generalization. Refinements such as counterbalanced trial structures, reduced complexity levels, and more precise complexity definitions are recommended to enhance sensitivity and isolate effects of interest. Importantly, the paradigm also generated a rich continuous dataset, including temporal dynamics, head and controller kinematics, and movement trajectories, which remain to be further explored. With design optimizations and validation in at-risk populations, this paradigm holds promise as a scalable, non-invasive tool for detecting subtle navigational deficits in preclinical AD.