6 100 6. Gaze-contingent processing improves mobility performance block, we included an additional practice trial with a visual search task in the studio environment. This served two purposes: 1) each time when the simulation condition was switched, the subject had the opportunity to practice and get adjusted to the new simulation condition before starting the scene recognition trial 2) by performing extra practice trials in the studio environment, during the scene recognition session, subjects became increasingly familiarized with the environment before the visual search session. Visual Search Session Roughly the final 30 minutes of the experiment (including breaks) were reserved for the visual search session in the studio environment. By this time, subjects have visited the studio environment at least ten times for a duration of 1,5 minutes per visit. The visual search session consisted of three repetitions of three trials. Each trial lasted two minutes and tested one of the three study conditions. The task was to sequentially search and find as many target objects as possible (either: a pair of shoes, a mug or a stool). At every point in time, there was only one target and each target was only made visible when prompted. Once the participant has found the object, they were requested to indicate its location by looking at it, and pointing at it with the controller while pressing the response button. We used a preset of 14 possible locations where the target could appear and targets appeared at sensible locations To avoid targets from appearing in sight, the new target location was never close to the previous target location. 6.2.7. Study Outcomes Primary study outcomes Several primary outcomes were used to compare the performance across study conditions: For the mobility task in Experiment 1, we analyzed the average trial duration and number of collisions per trial. For the scene recognition task, we measured the classification accuracy and the average trial duration. For the visual search tasks we analyzed the average search duration. Furthermore, in Experiment 2, we recorded a subjective metric after each trial pertaining to the experienced difficulty of the trial. Additional outcomes To evaluate behavioral strategies and differences in visual exploration, we inspected the eye-gaze trajectories in Experiment 2, and calculated the angular velocity of eye and head rotations. Furthermore, we analyzed the questionnaires which were filled out at the end of each block in Experiment 1 and at the end of Experiment 2 to record self-reported observations. 6.2.8. Data Analysis Pre-processing The data obtained in both experiments were analyzed in Python (version 3.11.0). In Experiment 1, five trials were excluded from the analysis because of missing data (a portion of the frames was not saved successfully). The trial duration in the mobility task was calculated as the time between crossing the start and finish lines, which were separated 37 meters. Frames were marked as a collision in case the distance between the horizontally projected head position and the closest obstacle was smaller than 0.225
RkJQdWJsaXNoZXIy MTk4NDMw