Standard measures of visual acuity do not predict drivers' recognition performance under day or night conditions

This study investigated whether visual acuity or contrast sensitivity, measured under a range of luminance conditions, could predict drivers' recognition performance under real-world day and night road conditions.

Twenty-four participants, comprising three age groups (younger, mean = 21.5 years; middle-aged, mean = 46.6 years; and older, mean = 71.9 years), drove around a 1.8-km closed road circuit under day and nighttime conditions. At night, headlight intensity was varied over 1.5 log-units by ND filters mounted on the headlights. Participants drove around the circuit under five light conditions (daytime and four at night) and were asked to report relevant targets, including road signs, large low-contrast road obstacles, and pedestrians who wore retroreflective markings on either the torso or the limb joints (creating "biological motion"). Real-world recognition performance was measured as percent correct recognition and, in the case of low-contrast road obstacles, avoided. Clinical vision tests included high-contrast visual acuity and Pelli-Robson letter contrast sensitivity measured at four luminance levels.

Real-world recognition performance of all age groups was significantly degraded under low light conditions, and this impairment was greater for the older participants. These changes in drivers' recognition performance were more strongly predicted by contrast sensitivity than visual acuity measured under standard photopic conditions. Interestingly, contrast sensitivity was highly correlated with visual acuity measured under low-luminance conditions. Further analyses showed that recognition performance while driving is better predicted by combinations of two tests: either 1) photopic visual acuity and photopic contrast sensitivity, or 2) photopic and mesopic visual acuity.

These findings confirm that visibility is seriously degraded during night driving and that the problem is greater for older drivers. These changes in real-world recognition performance were better predicted by a standard test of contrast sensitivity than by visual acuity. Still better predictions can be obtained by the use of two vision tests. The implications of these findings for driver licensing standards are discussed.