6.2. Materials and Methods 6 101 Figure 6.4: Hallway layouts and mobility trajectories in Experiment 1. The virtual hallway was 44 meters long, and contained 19 obstacle arrangements that were spaced evenly across the length of the hallway, two meters apart. The number of collisions and the trial duration was recorded between the start- and finish line (37 meters distance). Each row displays one of the three unique variations of obstacle placements. The head position in the virtual hallway is plotted of three representative subjects (S10, S11 and S16). In total, nine trials are visualized. The obstacles are displayed as dark rectangles. Collisions with obstacles are indicated with a red circle. meters (i.e., the body dimensions of the simulated participant in the VR environment). In our VR experiment no physical contact was made with walls or obstacles. Despite the vibration feedback from the controllers, positional readjustments were relatively difficult and repeated subsequent collisions occurred frequently. To avoid unnecessary complexity of discriminating subsequent collisions, each collided object was counted only once and collisions with walls were not included in the analysis. In Experiment 2, in the scene recognition task, three subjects did not perform better than chance level across all conditions on average. These subjects were excluded from the analysis. For the visual search task, the gaze directions were analyzed to verify whether subjects found the correct target. Target reports that deviated more than 15 degrees from the correct target direction were considered invalid. Two subjects reported the wrong target object in more than 40% of the cases and were excluded from the analysis. Two different subject did not find a minimum of three targets per study condition and were excluded from the analysis. One target object was not correctly found in a minimum of 40% of the cases and was therefore excluded from the analysis as well. Only search durations of successful searches (targets that were correctly found) were analyzed. We verified that the exclusion of data did not alter the conclusions of the analysis. Statistical analysis For each subject we averaged the endpoints across trials. A Shapiro-Wilk test was used to test if data was normally distributed. To test for significant differences between study conditions, we used a paired samples t-test (for normally distributed endpoints) or a Wilcoxon signed-rank test (for non-normally distributed endpoints). We used a Bonferroni correction to correct for the number of comparisons (three for each endpoint), rendering a required significance level of α=0.0167. To explore potential learning effects over the course of the experiment, we performed a linear regression analysis, also testing for possible interaction effects between the study conditions.
RkJQdWJsaXNoZXIy MTk4NDMw