How visual information influences dual-task driving and tracking

Laura Broeker,Markus Raab,Harald Ewolds,Stefan Künzell,Mathias Haeger,Bettina Kretschmann,Otmar Bock

doi:10.1007/s00221-020-05744-8

Abstract

The study examined the impact of visual predictability on dual-task performance in driving and tracking tasks. Participants (N = 27) performed a simulated driving task and a pursuit tracking task. In either task, visual predictability was manipulated by systematically varying the amount of advance visual information: in the driving task, participants drove at night with low beam, at night with high beam, or in daylight; in the tracking task, participants saw a white line that specified the future target trajectory for 200, 400 or 800 ms. Concurrently with driving or tracking, participants performed an auditory task. They had to discriminate between two sounds and press a pedal upon hearing the higher sound. Results show that in general, visual predictability benefited driving and tracking; however, dual-task driving performance was best with highest visual predictability (daylight), dual-task tracking performance was best with medium visual predictability (400 ms). Braking/reaction times were higher in dual tasks compared to single tasks, but were unaffected by visual predictability, showing that its beneficial effects did not transfer to the auditory task. In both tasks, manual accuracy decreased around the moment the foot pressed the pedal, indicating interference between tasks. We, therefore, conclude that despite a general beneficial impact of predictability, the integration of visual information seems to be rather task specific, and that interference between driving and audiomotor tasks, and tracking and audiomotor tasks, seems comparable.

Highlights

The impact of predictability in driving simulationsIt is a matter of common knowledge that car driving requires the handling of multiple tasks at the same time, like lane and distance keeping while watching for other road usersCommunicated by Melvyn A
Post hoc polynomial contrasts show that the relationship between SDLP and predictability was best described by a linear function, t(26) = 3.765, p < 0.001 (quadratic: t(26) = 1.444, p = 0.155)
We did not endorse significant correlations worth reporting between the dependent variables of the two tests (i.e. root mean square error (RMSE) vs. SDLP across the whole route; tracking velocity in four intervals vs. SDLP in four intervals)

Summary

Introduction

The impact of predictability in driving simulationsIt is a matter of common knowledge that car driving requires the handling of multiple tasks at the same time, like lane and distance keeping while watching for other road usersCommunicated by Melvyn A. These costs may differ depending on various external conditions like traffic density or reduced visibility of the road (Mueller and Trick 2012; Trick et al 2010), and external conditions themselves may have a different impact depending on their degree of predictability, e.g., higher density in traffic involving more unpredictable breaks of cars out in front, and darker environments involving less predictability in routing This impact was tested by Lundqvist et al (1997) who compared driving performance for high (good sight; straight roads; preceding car with slightly varying speed) and low predictability (sudden braking of preceding car, sudden appearance of parked car behind a curve; unexpected visual stimuli appearing in field of view) in patients with brain lesions and healthy controls. It has been argued that prediction is an omnipresent principle of human behaviour and that the beneficial effects of predictability in the environment are universal (Blakemore et al 2000; Broeker et al 2017; Northoff 2014), yet evidence for this claim comes from mostly basic tasks

Methods

Results

Conclusion