Driving with Central Visual Field Loss II: How Scotomas above or below the Preferred Retinal Locus (PRL) Affect Hazard Detection in a Driving Simulator

Abstract

We determined whether binocular central scotomas above or below the preferred retinal locus affect detection of hazards (pedestrians) approaching from the side. Seven participants with central field loss (CFL), and seven age-and sex-matched controls with normal vision (NV), each completed two sessions of 5 test drives (each approximately 10 minutes long) in a driving simulator. Participants pressed the horn when detecting pedestrians that appeared at one of four eccentricities (-14°, -4°, left, 4°, or 14°, right, relative to car heading). Pedestrians walked or ran towards the travel lane on a collision course with the participant's vehicle, thus remaining in the same area of the visual field, assuming participant's steady forward gaze down the travel lane. Detection rates were nearly 100% for all participants. CFL participant reaction times were longer (median 2.27s, 95% CI 2.13 to 2.47) than NVs (median 1.17s, 95%CI 1.10 to 2.13; difference p<0.01), and CFL participants would have been unable to stop for 21% of pedestrians, compared with 3% for NV, p<0.001. Although the scotomas were not expected to obscure pedestrian hazards, gaze tracking revealed that scotomas did sometimes interfere with detection; late reactions usually occurred when pedestrians were entirely or partially obscured by the scotoma (time obscured correlated with reaction times, r = 0.57, p<0.001). We previously showed that scotomas lateral to the preferred retinal locus delay reaction times to a greater extent; however, taken together, the results of our studies suggest that any binocular CFL might negatively impact timely hazard detection while driving and should be a consideration when evaluating vision for driving.

Fig 1. Binocular visual field plots (PRL at origin), Reaction Time (RT) boxplots for Central Field Loss (CFL) subjects (24–26 pedestrian appearances at each eccentricity), and individual plots of RT by scotoma occlusion time.

RT boxplot for normally-sighted controls is at bottom left. Box lengths show the 25% to 75% interquartile range (IQR) and whiskers show the maximum extent of cases that are not outliers (values >1.5 times IQR). Detection rates for each eccentricity are below each plot. CFL participant age, binocular contrast sensitivity (CS), binocular visual acuity and driver status are to the left of the plots. CFL5 had a vertical central scotoma and a paracentral lateral scotoma from OS scotoma overlapping OD physiological blind spot.

Fig 2. Central Field Loss (CFL) and Normal Vision (NV) participant mean reaction times by drive type and hazard eccentricity.

CFL participants had longer reaction times than controls at all eccentricities. Error bars represent 95% confidence interval.

Fig 3. Example of a late pedestrian detection event in which the participant’s scotoma (Central Field Loss subject #4) was implicated.

The upper graph depicts the pedestrian’s vertical position and size (black dots), the vertical extent and position of the scotoma at the horizontal position (magenta vertical lines), and the location of the vertical gaze Preferred Retinal Location (PRL) position (blue dots). Lower graph depicts horizontal position and gaze and the scotoma’s horizontal extent. When the pedestrian’s upper and lower limits are both within the vertical scotoma the pedestrian is not visible to the participant. Both the horizontal position of the pedestrian and the horizontal position and extent of the scotoma are relative to the car heading.