Trails, rails, and over-reliance: How navigation assistance affects route-finding and spatial learning in virtual environments

Many novices struggle with navigation in 3D virtual environments — they frequently get lost and are unable to find objects and locations. In some virtual environments, novices are provided with navigation assistance (e.g., mini-maps, directional markers, or glowing trails) that help them move around in the world. However, it is possible that providing navigation assistance could lead to over-reliance, where the novice’s dependence on the assist means that they never develop a mental model of the environment that would allow them to navigate on their own. To investigate both the benefits and potential risks of navigation assistance in virtual environments, we carried out two online studies in which participants carried out route-finding tasks with different types of navigation assistance. Participants completed training trials, in which they practiced a set of routes with the assist, and transfer trials, in which they had to navigate without the assist. The studies focus on two questions: whether assistance improves performance and user experience when it is present, and whether assistance leads to over-reliance and a drop in performance when the assist is removed. For the first question, both studies found that navigation assistance substantially improved performance and subjective experience while it was present — clearly showing that assistance can improve virtual environments for novices. For the second question, we found mixed evidence regarding the problem of over-reliance: the first study showed no performance differences between the highest and lowest levels of navigation assistance when the assist was turned off; the second study showed that there was a performance reduction when the assist was removed, but that the size of the reduction was much smaller than the improvement provided during training. We found that even when the navigation assist was extreme (e.g., pressing a button to be automatically taken in the correct direction), participants were still able to navigate the trained routes, suggesting that incidental learning does successfully occur in virtual environments. Our studies suggest that designers of virtual environments should strongly consider providing navigation assistance: assists can improve a novice user’s performance and experience by reducing navigation problems, and the risks of over-reliance appear to be small in comparison to the benefits for inexperienced users.