关键词: 地标，视觉流，路径整合，虚拟场景，空间认知

Abstract: Path integration refers to the process that people integrate the distance and orientation information depending on their current position and the starting point. Previous studies found that participants could use landmark information to promote the efficiency of path integration during spatial navigation. However, it was inconclusive that whether the optic flow (also used as the texture) could also help to improve the path integration efficiency and whether participants could combine the landmarks and the optic flow to promote the path integration efficiency. In order to explore these issues, the current study used the dynamical 3D-max technology to create the virtual space scenes and explored the difference in the path integration efficiency among 4 kinds of visual scene (pure, texture, landmark and double-clue (texture + landmark)) in the modified homing task. Participants were required to walk from the starting point and reached the end throughout 5 outbound paths. Then they needed to make orientation judgments accurately and get back to the starting point directly. There were 3 kinds of errors when they came back to the starting point (see Figure 2). (1) Angle error, the absolute value of the difference between the correct angle (Ac) and the wrong angle (Ar); (2) Distance error, the absolute value of the difference between the correct distance (Dc) and the wrong distance (Dr); (3) Position error, the distance between the starting point (H) and the homing position (Hr). For the 3 kinds of errors, the smaller the value was, the better the task was accomplished and the higher the path integration efficiency was. In addition, the response time, referred to the time interval between the complication of the walking and making the orientation judgement, was also used to evaluate the path integration efficiency. The shorter the response time, the higher the path integration efficiency. The results showed that the main effect of the visual environment were significant in the response time (F（3, 87）=211.60, p<.001, ηp2=.873), the distance error (F（3, 87）=131.89，p<.001，ηp2=.820) and the position error (F（3, 87）= 10.15, p < .001, ηp2 =.259). The Post Hoc test revealed that the path integration efficiency in the texture condition was significant higher than in the pure clue condition (all p< 0.05), which suggested that the optic flow could improve the path integration efficiency. For the angle error, the main effect of the visual environment was not significant (F（3, 87）=1.52, p=.214). There was no significant difference between the pure condition and the texture condition, suggesting the texture might play little role in the judgement of the angle. Besides, the response time was shorter in the landmark condition than the texture condition (p< 0.05), and the distance error in the landmark condition than in the texture condition (p< 0.05). It reflected an advantage of the landmark in the path integration compared to the texture. But compared to texture condition, the landmark did not show significant advantage for the angle error and the position error, which might attribute to the different landmark types. In addition, the performance in the double clue (texture + landmark) condition was better than the single clue (pure, texture or landmark) condition for the response time and the distance error, suggesting that the participants could combine the two kinds of clues to improve the path integration efficiency. The current study has 2 conclusions: (1) Both the landmark and the texture visual information can promote the path integration, and the landmark had an advantage; (2) The combination of the landmark and the texture can promote the path integration.

Key words: landmark, optic flow, path integration, virtual scene, spatial cognition