Abstract
Aging can cause several issues affecting driving, including visual and attentional loss (e.g., UFOV). However, one aspect of ageing effects upon attention that seems less clear is the use of expectancy, or prior information (e.g., cueing tasks). This study assessed expectancy use in naturalistic driving scenes employing an ideal observer analysis developed for cueing tasks (Eckstein, Drescher, & Shimozaki, 2006). A pilot online study first asked 172 participants where (in 150 driving scenes, greyscale) certain driving-related stimuli (cars, pedestrians, cyclists, signs) would appear. From these responses expectancy ‘heat maps’ were generated by summing and normalizing the Gaussian-blurred responses (σ=1.41°). Then for 75 of the scenes (36.9°x28.1°),driving-related targets were artificially inserted in high- and low-expectancy locations, and 46 older UK drivers (61-85 years) and 49 younger UK drivers (19-29 years) performed a yes/no detection task. Stimuli were presented either in the full scene, or within ellipses removing the scene context (extending 1.41° beyond the vertical and horizontal extents of targets) for either 50ms (young) or 200ms (old). When the younger participants viewed the full scene, they saw a scene without the target (150ms), a blank grey screen (50ms), then the stimulus (50ms), similar to a flicker task (change blindness). For the targets presented without context (ellipses), performance was slightly better in the high-expectancy locations (d’-young-high=1.33, se =0.08; d’-young-low=1.18, se=0.08; d’-old-high=1.35, se=0.12; d’-old-low=1.05, se=0.09). In the full scenes there were also expectancy effects (high-expectancy - low-expectancy hit rates) for both young (0.123, se=0.013) and old (0.075, se=0.018) observers. However, ideal observer analyses found that this expectancy effect could be explained by the d’ difference in the ellipse conditions for the old (but not the young) observers. This suggests that the young (but not the old) observers were using expectancy, and that one potential issue for aging effects in driving is expectancy use.
Meeting abstract presented at VSS 2015