Abstract
Abstract: :
Purpose: In recent years eye movement studies in natural environments have become a focus of research. At the same time, virtual reality has become an accepted tool in psychophysical studies. Here we investigate the eye movements of human subjects while walking through our lab and compare them with eye movements of subjects driving throught the "same" hallway in a virtual environment. Methods: 11 subjects took place in the experiment. Their task was to walk along a route of about 50m through our lab. For the experiments in virtual reality we build a model of the floor in a highly detailed manner. In these kind of experiments subject were seated in front of a round, tilted projection screen (horizontal: 163°, vertical: 70°) and could move through the virtual environment with a joystick. In both types of experiments eye movements were recorded (ASL 501 mobil) with an accuracy of about 1°. Instructions about the route were given previously. Subjects performed three trials: either vr–real–real or real–vr–vr. One experiment lasted about one hour, including a calibration procedure of the eye tracker. Results: Subjects in both types of experiments prefer to look on a few objects along the travelled path. Interestingly the number of gazes to the same objects in the real and the virtual environment show a positive correlation. A linear fit has a slope of almost one (f(x) =0.87x – 0.27), the correlation coefficient is 0.85. The majority of gazes are pointed straight ahead in the direction of progression(40 %). In virtual reality there is a stronger tendency to look on the floor (11 % in VR vs. 6 % in real environment). The mean duration of gazes to the same objects in the two types of experiments showed no significant difference (r = 0.85), whereas the total number of gazes in the real environment was much higher as in the virtual environment (1393 in real; 972 in virtual). Conclusions: In virtual reality, subjects look at the same objects along a travelled path as in a similar real environment. In both environments most of the gazes are directed towards the direction of progression, which was also found in other studies. Also the gaze duration in both environments were comparable. The largest difference concerns the number of gazes to the ground floor. This could be due to the different locomotion modalities in the two experiments: In the real hallway we investigate the eye movements of pedestrians, whereas in the virtual environment the movement with the joystick resembles rather driving a car. From driving studies it is known that eye movements are directed in particular towards the street. In summary we can conclude from our results, that virtual reality is a reliable tool for eye movement studies.
Keywords: eye movements • perception • scene perception