System immersion of a driving simulator affects the oscillatory brain activity

Abstract

Background. Driving simulators allow studying driving behaviour under controlled settings. To reproduce the driver’s behaviour as realistically as possible, we need reliable driving simulators which allow participants to get highly immersed. The property of the technological system delivering the experience is a crucial dimension of immersion. Understanding the effect of different simulator settings on brain activity is vital to ensure that future research in neuroergonomics is reproducible and consistent. This study explores the impact of system immersion on the drivers’ brain activity when operating a conditionally automated vehicle in a driving simulator task.Method. Two groups of participants operated an autonomous vehicle for four sequential 10-minute-long rides while conducting a non-driving secondary task. The high-immersion group (nine participants) operated a fixed-base driving simulator with a 190 ° field of view. The low-immersion group (10 participants) operated the same simulator, but only the front screen was used. Two participants from the high-immersion group were excluded due to simulator sickness. We recorded the brain activity using a 32-channel EEG headset.Results. We used the Mann-Whitney U test to compare the relative mean power over the whole ride in the Theta, Alpha, Beta, low-Beta, and high-Beta bandwidth. We found a significant difference between the groups in the Beta bandwidth (12-30 Hz) at the Cz position; and the high-Beta bandwidth (22-30 Hz) at the Oz, O2, P3, P8, and Cz position. Using mixed ANOVA, we assessed the effect of time (four sequential rides) and group (high- and low-immersion) on relative mean power in the frontal, parietal, occipital, and temporal regions. For the Alpha bandwidth, the two-way mixed ANOVA revealed a significant effect of time in the parietal, temporal, and frontal regions. Neither the group nor the interaction effects were found in other bandwidths. Implementing the Bayesian approach, we found strong evidence against the effect of group on frontal, temporal, and parietal Alpha. Moreover, we found moderate evidence against the effect of time on frontal and parietal Theta, occipital Beta, and occipital high-Beta. We also found moderate or strong evidence against the interaction of time and group on frontal and parietal Theta; occipital, temporal, and parietal Beta; and occipital and parietal high-Beta. Finally, the Bayesian approach suggested equal support for H0 and HA for the effect of group on parietal Beta and parietal high-Beta.Conclusion. Our results suggest that the system immersion of a driving simulator might affect the oscillatory brain activity, especially in the high-beta bandwidth and in the parietal and occipital areas. This finding complies with previous studies on immersion. Moreover, our results indicate that the high-immersion settings could increase the involvement; however, it could also be more stressful. This seems relevant, especially in the light of the increased simulator sickness in the high-immersion group. The Bayesian approach suggests that more data would be needed to draw final conclusions.

Manipulating Drivers' Mental Workload: Neuroergonomic Evaluation of the Speed Regulation N-Back Task Using NASA-TLX and Auditory P3a

“…Higher MW, therefore, leads to decreased P3a amplitude. There is a discussion about whether other factors, such as arousal or top-down attention control, might be a factor influencing the P3a amplitude [1,7]; nevertheless, robust empirical evidence supports the inversed relationship between MW and P3a amplitude due to the competition for resources.…”

Cognitive tasks and combined statistical methods to evaluate, model, and predict mental workload

“…Consequently, it is often measured during the activity and/or after the activity. In recent times, MWL is mostly considered through operational situations such as automotive contexts (Milleville-Pennel and Charron, 2015;Foy and Chapman, 2018;Figalová et al, 2022) or aeronautic contexts like air traffic control (Averty et al, 2004;Mélan and Cascino, 2014;Li et al, 2022). Thus, MWL is usually measured as a consequence of the activity without any a priori certainty of the MWL value in which the person is when performing the task.…”