The Effects of Training and Visual Detail on Accuracy of Movement Production in Virtual and Real-World Envionments

Abstract

Computer-generated environments have been established as valuable training media for both proximal-field object manipulation and wide-area navigation performance. This study investigates the potential of virtual environments-based computer training for near-field navigation accuracy. Results of this investigation revealed significantly greater performance accuracy in a real-world setting for subjects that were first trained in a virtual environment. In addition, three levels of rendered visual detail were compared for precision of movement within a virtual hallway: (1) untextured polygon surfaces, (2) untextured polygon surfaces with transverse floor gradient lines placed as visual cues, and (3) fully detailed, textured surfaces. Significant effects were found for visual detail in accuracy of movement production. Within the virtual environment, performance error in the undetailed setting was significantly higher as compared to the settings with visual cues and textures. Significant differences in real-world field performance were also found between the groups that were exposed to different levels of visual detail in computer training. Field performance was significantly superior for subjects trained with visual cues as compared to those trained in an untextured virtual setting. Evidence suggests that for movement production accuracy both within a virtual environment and in the real world, fully-detailed textured surfaces are not required in the rendered setting in order to obtain the best performance. Simple surface gradients serve as adequate visual cues for near-field navigation accuracy, while a lack of consistent visual cues can be detrimental to effective movement production.