Navigating impossible spaces: walking speed as a behavioral index of spatial integration in memory

Human navigation typically occurs in environments that follow stable Euclidean principles, where spatial relations such as distances and angles remain globally consistent. Virtual reality makes it possible to investigate navigation in environments that violate these regularities by creating spaces that are locally navigable but globally inconsistent. Studying behaviour in such environments provides a useful approach for investigating how the brain integrates spatial information and constructs representations of space when global geometric consistency is disrupted. The present study examined locomotor behaviour during navigation in non-Euclidean virtual environments, focusing on walking speed as a behavioural measure of spatial processing. Participants navigated a virtual maze in four experimental conditions that manipulated segment length, turning angles, or both (Control, Length, Angles, Combined). The task included a learning phase, in which participants walked the route, and a route reconstruction phase, in which they reproduced the path from memory. Walking speed was analysed across route segments and experimental blocks using linear mixed-effects models. The study examined whether repeated exposure to the task was associated with changes in walking speed and whether geometric manipulations embedded in the structure of the environment were related to variations in locomotor behaviour across different portions of the route. By analysing walking speed during navigation in non-Euclidean environments, the study aimed to investigate how geometric distortions may influence behavioural responses during route learning and reconstruction, and to contribute to the understanding of how spatial information is integrated by the brain during navigation.