Dissociating ethical dilemma preferences and actions in autonomous driving by survey and driving experiment

Previous research has revealed the general preferences in ethical dilemmas for autonomous driving. However, people’s legal preference compared to ethical preferences has not been fully understood. In the current study, we aim to reveal the role of legality in ethical preferences for people’s decision and action when facing ethical dilemmas in three steps: data mining, in-person survey and interview, and lab experiment. We first mined the moral machine experiment (MME, Awad et al, 2018) dataset to find the important features and relative importance of legality in people’s decisions facing ethical dilemmas. To find out whether people share the same preference as MME and reveal the reasons of their preferences, we then interviewed 32 participants and found that they valued legal preference as an important factor for ethical preferences in the second step. The question still remains whether people act on their legal or ethical preferences when driving? In the third step, we simulated these ethical dilemmas while driving on autonomous mode in a driving simulator, and examined the bevavioural responses of the participating drivers (n = 62). Interestingly, we found that drivers acted on their ethical preferences (sparing children vs. senior passengers) instead of legal preference (sparing senior following traffic rules vs. children violating traffic rules). We argue that the legal preference requires reasoning and memory of the traffic rules which may take longer time to process, whereas the ethical preference is perceptual, or intuitive, and processed faster for action to take place in emergencies. Our findings shed light on the underlying dual-system processes (rational reasoning and intuition) in ethical dilemmas for autonomous driving and responses in emergency.