This is your brain on autopilot: Neural indices of driver workload and engagement during partial vehicle automation
DOI: 10.1177/00187208211039091
archive: archived pipeline: cataloged verified
Summary
On-road EEG study examining whether SAE Level-2 partial vehicle automation affects neural indices of mental workload (frontal theta) and visual engagement (parietal alpha). 71 participants from two age cohorts (young adults 21-40, n=39; middle-aged 41-64, n=32) drove four production Level-2 vehicles (Cadillac CT6, Nissan Rogue, Tesla Model 3, Volvo XC90) on two Salt Lake City interstates (I-15 high-traffic, I-80 low-traffic) in both manual and partial-automation modes. EEG was recorded with a three-electrode BIOPAC system (Fz, Cz, Pz). A 4-min eyes-closed baseline served as a manipulation check confirming the expected alpha-power increase. Linear mixed-effects models tested level of automation, interstate, age cohort, and vehicle as fixed effects with subject and number of vehicles driven as random effects; Bayes factors quantified evidence for null effects.
Key finding
Engaging Level-2 partial automation produced no detectable change in frontal theta or parietal alpha power relative to manual driving (chi^2(1)=0.20, p=.651 for theta; null Bayes-factor evidence for both bands). Drivers new to the technology remained cognitively and visually engaged with the roadway under partial automation, contrary to under-arousal/disengagement concerns raised by simulator work.
Methodology
On-road experimental study with 2 (Age Cohort) x 2 (Level of Automation: manual vs partial) x 2 (Interstate: I-15 vs I-80) x 4 (Vehicle) factorial design. EEG (BIOPAC, 2 kHz sampling, three midline electrodes Fz/Cz/Pz, mastoid reference, EOG correction via Gratton EMCP) recorded during four 20-min driving conditions after a 4-min eyes-closed baseline. Spectral power averaged in theta (4-8 Hz) at Fz and alpha (8-12 Hz) at Pz via FFT on 1-s Hanning epochs. Linear mixed-effects models in lme4 with likelihood-ratio tests and Bayes factors.
Sample size: N=71 (young adults n=39, mean age 29.07; middle-aged n=32, mean age 52.2); 192 vehicle test sessions
Quality score: 5 / 5