Using the ISO detection response task to measure the cognitive load of driving four separate vehicles on two distinct highways
DOI: 10.1016/j.trf.2024.02.013
archive: archived pipeline: cataloged verified
Abstract
The ISO Detection Response Task (DRT) is a standard tool for assessing drivers' cognitive load and it has primarily been used to measure the cognitive load of completing non-driving tasks and interacting with vehicle systems. In this study we use the DRT to measure the workload of driving four separate vehicles (a 2019 Tesla Model 3, a 2018 Cadillac CT6, a 2018 Volvo XC90, a 2019 Nissan Rogue) in manual mode and on two distinct roadways (US Interstate Highway 15 and 80) in and around Salt Lake City, UT. Results showed that the unique road characteristics of I-80 resulted in higher levels of cognitive load as demonstrated by the slower DRT response times. Likewise, different levels of workload were found across the four vehicles, with higher workload levels found for one of the four vehicles. This study expands the use of the DRT outside its original area of application, and advances it as a tool to assess the cognitive demand induced by varying road and vehicle characteristics.
Summary
On-road study (Biondi et al., 2024, Transportation Research Part F, Vol 102) using the ISO Detection Response Task (DRT) to measure driver cognitive load across two distinct highways (US Interstate-15 and I-80 in/around Salt Lake City, UT) and four vehicles (2019 Tesla Model 3, 2018 Cadillac CT6, 2018 Volvo XC90, 2019 Nissan Rogue) driven in manual mode. Seventy-one participants (25 female; mean age 40.8) each drove all four vehicles on both highways (~20 min/drive) with order counterbalanced; ~300 vibrotactile DRT stimuli per drive (3-5 s ISI). Each 20-minute drive was split post-hoc into 8 time periods to examine temporal changes in DRT performance. Linear mixed-effects models with highway, vehicle, and time period as fixed factors and participant as random effect. The study extends the DRT beyond its conventional non-driving-task application to assess how road characteristics and vehicle differences modulate driver workload.
Key finding
Significant main effect of highway on DRT RT (chi2(1)=16.15, p<0.001, partial-eta2=0.08): I-80 produced slower RTs (M=498 ms) than I-15 (M=476 ms), with more DRT misses on I-80, consistent with I-80's narrower (2-3 lanes) winding canyon profile vs I-15's flatter 4-5-lane carriageway. RTs increased across the eight time periods (chi2(1)=103.45, p<0.001), and the rate of increase was steeper on I-80 (beta=10.83) than I-15 (beta=7.43). Significant vehicle effect (chi2(3)=71.09, p<0.001, partial-eta2=0.10): the Volvo XC90 elicited 40-50 ms slower DRT RTs (M=521 ms) than the Nissan (477), Tesla (480), and Cadillac (469); a vehicle x highway interaction showed I-80 slowing for the Tesla and Nissan but not the Volvo or Cadillac. No vehicle differences in miss rate. Authors argue interior/UI cannot fully explain the Volvo elevation and suggest vehicle-size-related driver-behavior modulations as a candidate mechanism.
Methodology
On-road experimental study. N=71 (25 F, mean age 40.8, SD 6.11). Within-subjects 2 (highway: I-15, I-80) x 4 (vehicle: Tesla Model 3, Cadillac CT6, Volvo XC90, Nissan Rogue) x 8 (20-min time period) design with order fully counterbalanced. ISO 17488 vibrotactile DRT (Red Scientific device) on left forearm; right-hand microswitch response; ~300 stimuli per 20-min drive at 3-5 s quasi-random ISI; RTs <100 ms or >2500 ms excluded. Vehicles driven in manual mode with all driver-assistance systems disabled. Linear mixed-effects models (lme4 in R 4.1.0) with highway, vehicle, and time period as fixed factors and participant as random effect; likelihood-ratio tests for fixed effects, Tukey post-hoc for pairwise contrasts, partial eta-squared effect sizes.
Sample size: N=71 (25 female, 46 male); mean age 40.8 years (SD 6.11)
Quality score: 5 / 5