How Does Horizontal-Curve Radius Affect Drivers’ Visual and Psychophysiological Load? Evidence from an On-Road Driving Experiment

An on-road experiment with 27 drivers across 50 horizontal curves examined the association between radius and visual psychophysiological workload on a mountainous two-lane highway. Pupil-area change rate, fixation intensity, heart-rate change rate, and heart-rate variability (SDNN) were combined using factor analysis and entropy weighting. Crossed linear mixed-effects models used driver and curve section as random intercepts and radius as a continuous predictor. Smaller radii were generally associated with higher pupil, fixation, and heart-rate responses and lower SDNN. Among linear, quadratic, cubic, and logarithmic forms, the logarithmic comprehensive-load model fit best (AIC = −2511.47; BIC = −2485.43), with a significant radius effect (β = −0.2968, 95% CI: −0.3283 to −0.2653, p < 0.001). Successive-curve inconsistency was not independently significant (p = 0.976). Within 45.83–900.83 m, exploratory workload boundaries were approximately 75, 150, 270, and 500 m. These are preliminary human-factors screening values, not design standards or validated comfort thresholds.