Informing ethical decisions of autonomous vehicles through video-based choice experiments and brain recordings.

Autonomous vehicles (AVs) are no longer fictional. The success of AVs will depend on how they handle ethical issues in a socially acceptable manner. For example, the decisions that AV should make when there is no way to save everyone (i.e., the trolley problem). A famous cross-cultural Moral Machine experiment evaluated societal expectations about the ethical principles that should guide AV behavior in scenarios based on the trolley problem paradigm and showed that the subjective beliefs of the population play a critical role in the valuation of the ethical aspects of AV behavior. This study aims to expand our knowledge related to the effect of subjective beliefs on the valuation of the ethical aspects of AV behavior. We hypothesize that the brain activity of respondents may provide additional information about the subjective aspects of the valuation process. To test the hypothesis, we carry out video-based discrete choice experiments in which participants choose between two victims/pedestrians of road accidents involving an AV. The experiment is designed such that the AV has no choice but to hit one of two pedestrians approaching from both ends of the road. We vary the socio-demographics, such as the age and gender of the pedestrians, across trials to reveal their impact on the participant’s decisions. We fit a discrete choice model (DCM) to experimental data to uncover the impacts of the considered socio-demographics on choices. Finally, we test if the different neural mechanisms (biomarkers) could explain the valuation of socio-demographics and incorporate them into DCM to better account for the subjective beliefs of the respondents. Our study contributes to multiple fields, including AV research, choice modeling, and psychology. For AV research, we enhance our understanding of societal expectations about ethical aspects of AV behavior. For choice modeling, we advance traditional choice models by including brain signals. For psychology, we reveal mechanisms underlying the perception and valuation of ethical problems.