The effects of sound and priming on user safety decisions in virtual construction simulators

Desktop virtual reality is an economical tool for training large cohorts, such as fulfilling the safety training needs of the construction industry. However, adding more features to a virtual reality simulator would raise its cost of creation. Thus it is important to study what features are worth including in a virtual construction simulator. This thesis investigates whether sound and priming factors are essential to a virtual construction safety training simulator in terms of their value in improving people’s safety decisions in virtual construction simulators and in subsequent situations.There were two stages in this study. For the first stage, a 3D virtual environment (VE) and a corresponding physical environment were created to study the effect of sound on people’s safety behaviours. It was found that the addition of background sound to a visually high fidelity VE enhances people’s sense of presence by creating a high level of normality. Salient sound is helpful in recognizing the related risks. Safe behaviours learnt in a VE can effectively reduce unsafe behaviours in a corresponding physical environment. For the second stage, two 3D VEs were designed to test the priming effects on people’s safety behaviours, including controlling for the background sound. For this thesis priming refers to unconscious stimuli that affect behaviour. It was found that obvious stimuli in a VE with sound help people to recognize more hazards around the stimuli. Priming of safe choices results in people taking fewer risks in the virtual reality simulator, but this does not carry over into subsequent situations. Thus priming factors are not suitable to be added to VR simulators for training purposes. Compared to priming effects, being “injured” in a training simulator with sound results in a greater improvement in the safety behaviour of trainees in subsequent situations.The practical implications of this thesis are that, for safety training purposes, virtual construction simulators involving unsupervised interaction should be designed with sound, without priming factors, and should contain “traps” that the trainees will easily fall into when they make unsafe decisions. Further study can be carried out on how people’s working memory responds to different VEs