Vestibular damage affects the precision and accuracy of navigation in a virtual visual environment

Abstract

Vestibular information is available to the brain during navigation, as are the other self-generated (idiothetic) and external (allothetic) sensorimotor cues that contribute to central estimates of position and motion. Rodent studies provide strong evidence that vestibular information contributes to navigation but human studies have been less conclusive. Furthermore, sex-based differences have been described in human navigation studies performed with the head stationary, a situation where dynamic vestibular (and other idiothetic) information is absent, but sex differences in the utilization of vestibular information have not been described. Here, we studied men and women with severe bilateral vestibular damage as they navigated through a visually barren virtual reality environment and compared their performance to normal men and women. Two navigation protocols were employed, which either activated dynamic idiothetic cues ('dynamic task', navigate by turning, walking in place) or eliminated them ('static task', navigate with key presses, head stationary). For both protocols, we employed a standard 'triangle completion task' in which subjects moved to two visual targets in series and then were required to return to their perceived starting position without localizing visual information. The angular and linear 'accuracy' (derived from response error) and 'precision' (derived from response variability) were calculated. Comparing performance 'within tasks', navigation on the dynamic paradigm was worse in male vestibular-deficient patients than in normal men but vestibular-deficient and normal women were equivalent; on the static paradigm, vestibular-deficient men (but not women) performed better than normal subjects. Comparing performance 'between tasks', normal men performed better on the dynamic than the static paradigm while vestibular-deficient men and both normal and vestibular-deficient women were equivalent on both tasks. Statistical analysis demonstrated that for the angular precision metric, sex had a significant effect on the interaction between vestibular status and the test paradigm. These results provide evidence that humans use vestibular information when they navigate in a virtual visual environment and that men and women may utilize vestibular (and visual) information differently. On our navigation paradigm, men used vestibular information to improve navigation performance, and in the presence of severe vestibular damage, they utilized visual information more effectively. In contrast, we did not find evidence that women used vestibular information while navigating on our virtual task, nor did we find evidence that they improved their utilization of visual information in the presence of severe vestibular damage.

Keywords: navigation; sex differences; vestibular; virtual reality; vision.

Graphical Abstract

Figure 1

Schematic representation of the VR navigation task. (A) Sample visual scene of a first-person view of the VR environment. (B) Navigation task demonstrating the complete triangle task and the calculation of the angular and linear error. (C) Static navigation task, head and body stationary, navigating with two buttons on the Xbox controller. (D) Dynamic navigation task, schematic showing subject turning in yaw and stepping in place.

Figure 2

Sample trials on the dynamic and static conditions. Data from one BVL and one NC subject; the open circles are the start (home) positions, and the dashed black arrows show the first two visually guided legs of the TCT. The solid lines show the different ‘return-to-home’ trials with the bold X showing the mean end position. x- and y-axes are in units that are equivalent to metres (m), based on the size of the virtual image (Fig. 1A).

Figure 3

Mean ‘angular’ error (inaccuracy) and standard deviation of angular error (imprecision). Box plots for the static and dynamic tasks in NC and BVL men and women. Vertical lines represent the 25th and 75th percentile range; horizontal line bisecting the box represents the median value; the filled dot represents the mean value; and open circles show the value for each individual subject. Statistically significant comparisons (P < 0.05) are represented by filled stars for the Mann–Whitney U-test and a filled diamond for the Wilcoxon signed-rank test. ‘Angular precision’: BVL/male/dynamic versus NC/male/dynamic, Mann–Whitney U-test, U = 71, P = 0.01; BVL/male/static versus NC/male/static, Mann–Whitney U-test, U = 81, P = 0.02; NC/male/dynamic versus NC/male/static, Wilcoxon signed-rank test, V = 40, P = 0.03. Population size: BVL/male = 10; BVL/female = 9; NC/male = 9; NC/female = 10. SD, standard deviation.

Figure 4

Mean ‘linear’ error (inaccuracy) and standard deviation of linear error (imprecision). Box plots for the static and dynamic tasks in NC and BVL men and women, organized like Fig. 3 with the same icons. ‘Linear accuracy’—BVL/male/dynamic versus BVL/male/static, Wilcoxon signed-rank test, V = 53, P = 0.03; ‘linear precision’—BVL/male/static versus NC/male/static, Mann-Whitney U-test, U = 8, P = 0.01. Population size: BVL/male = 10; BVL/female = 9; NC/male = 9; NC/female = 10. SD, standard deviation.

Figure 5

Mean ‘path length’ of final ‘return-to-home’ segment. Box plots for the static and dynamic tasks in NC and BVL men and women, organized like Figs 3 and 4 with the same icons. BVL/male/dynamic versus BVL/male/static, Wilcoxon signed-rank test, V = 55, P = 0.01; BVL/male/dynamic versus NC/male/dynamic, Mann–Whitney U-test, U = 77, P = 0.02. Population size: BVL/male = 10; BVL/female = 9; NC/male = 9; NC/female = 10.