Design mechanisms of airport visual guidance systems on passenger wayfinding performance: evidence from causal machine learning and a moderated mediation approach

In airports, effective visual guidance systems are essential for supporting passenger wayfinding under cognitive load and time constraints. However, many existing signage designs lack user-centered optimization, leading to inefficient navigation and decision errors. This study investigates how signage information density and coding format influence passenger cognitive load and decision confidence, with time pressure as a moderator. A two-phase approach was adopted: (1) causal machine learning identified key design factors from real-world airport signage; (2) a laboratory eye-tracking experiment with 60 participants employed a 3 (Information Density: Low/Medium/High) × 2 (Coding Format: Text vs. Text + Graphic) × 2 (Time Pressure: Low/High) mixed design. Visual search tasks measured gaze behavior, accuracy, and subjective ratings. Results show that high information density and text-only formats increased cognitive load and reduced decision confidence, while text + graphic formats improved performance – especially under high time pressure – by lowering visual effort. Mediation analysis confirmed cognitive load as the key mechanism, with time pressure moderating both direct and indirect effects. Findings provide an S – O – R-based cognitive mechanism model and practical guidelines for designing airport signage to enhance wayfinding efficiency in high-density public environments.