Effects of Automation and Transparency on Human Psychophysiological States and Perceived System Performance in Construction Safety Automation: An Electroencephalography Experiment

Abstract

Automation technology has experienced explosive growth in recent decades. However, the construction industry, let alone the field of construction safety, is still among the least digitized globally. Although many types of automation have proven effective in enhancing productivity and accuracy, adoption and acceptance by construction professionals are still limited. Human factors are essential for the success of automation. Therefore, this study investigated the effects of different levels of system automation and transparency, and their interactions, on human states (trust, mental workload, situational awareness) and perceived performance (explainability, satisfaction, usability). An experiment was conducted using a tool to assess scaffolding design safety with varying automation and transparency levels. A between-group design was adopted in which participants were assigned to four groups (a 2×2 matrix of automation and transparency). In a multi-methods approach, human states were measured through questionnaires and electroencephalography, and system performance was measured through a questionnaire. The results indicated that (1) automation level does not have significant impact on human states or perceived system performance, (2) a highly transparent automation system is associated with significantly higher trust and better perceived system performance, and (3) the positive impacts of transparency tend to be more obvious in low-automation systems. Due to perceived complexity, one might think that transparency would be emphasized more in highly automated systems. However, the results of this study shed light on the importance of providing a highly transparent interface for explaining system logic in all automated systems for design safety assessment in construction, particularly those with low automation levels.