Impact assessment of reinforced learning methods on construction workers' fall risk behavior using virtual reality Academic Article uri icon


  • © 2019 Given the nature of construction activities, construction workers usually work in a collaborative way. Thus, interpersonal influences among workers play a crucial role in forming and affecting construction workers' safety behaviors. The social learning literature indicates that interpersonal learning occurs in two opposing ways – positive reinforcement by demonstrating preferred behaviors, and negative reinforcement by demonstrating negative consequences of inappropriate behaviors. Amid theoretical disagreements in the social learning literature, it remains unclear in the construction safety literature how the two reinforced learning methods affect construction workers in safety training. To fill the gap, a human-subject experiment (n = 126)was conducted to investigate people's social learning behaviors in a hazardous construction situation – walking between two high-rise buildings. The experiment utilizes a multi-user Virtual Reality (VR)system with a motion tracking feature. Participants were randomly assigned to one of three groups: control group (no instruction was given), not-falling group (participants observed an avatar demonstrating appropriate walking behaviors), and falling group (participants watched an avatar quickly walking across a plank and falling off). Indicators, including walking time on the plank, walking speed, and gaze movement, were recorded and analyzed to quantify the effects of the two reinforced learning methods. The results indicate that demonstrating information with positive consequences (not-falling group)encourages people to follow the demonstration and maintain normal walking in a hazardous situation. Showing information with negative consequences (falling group)induced participants to walk faster and more irregularly, which further led to more mistakes and unsafe behaviors. This study demonstrates the effectiveness of using VR in safety studies and provides recommendations for better safety training programs.

author list (cited authors)

  • Shi, Y., Du, J., Ahn, C. R., & Ragan, E.

citation count

  • 41

publication date

  • August 2019