Across ordinary traffic simulations, people consistently rated drivers with good character more positively, while reactions to rules and collisions varied by format.
The researchers tested moral judgments using two ordinary traffic scenarios: one involving a mother driving a child to school and another involving a man taking a cat to a veterinary hospital. Each version varied the driver’s character, whether traffic rules were followed and whether the trip ended safely or in a collision.
Professional ethicists and lay participants generally rated virtuous drivers more positively than vicious ones. Expert ethicists also tended to agree in their evaluations, despite favoring different moral theories. Rule-following and outcomes mattered less consistently, and their influence changed between written descriptions and audio-visual simulations.
How people judged the drivers
Across both studies, participants responded strongly to the driver’s character: drivers described as good were judged more positively than those described as bad. Favorable outcomes, such as arriving safely, also generally produced more positive judgments than collisions, but this effect was weaker and depended on presentation format.
The effect of following traffic rules was mixed. It was clear in the study with lay participants, while the expert-ethicist study found different patterns for written and audio-visual scenarios. Audio-visual presentations tended to make character and rule-following more salient, whereas written scenarios placed more emphasis on outcomes. In the expert-ethicist study, participants tended to converge in their judgments even though they endorsed different ethical theories.
Why this matters for vehicle decisions
Autonomous vehicles need to respond to ordinary traffic situations in ways people consider acceptable, not only to dramatic emergencies. The convergence among expert ethicists suggests that some shared judgments might help guide such systems, even when people disagree about the ethical theories behind them.
The findings also suggest that judgments may depend on whether information is encountered in text, video or an immersive simulation. Character-based considerations may therefore need to be considered alongside rules and consequences when researchers study public expectations of autonomous vehicles.
Evidence and open questions
The findings come from two between-subjects quasi-experiments involving professional ethicists in one study and lay participants in the other. Participants judged two low-stakes traffic scenarios presented as written vignettes, computer-generated videos or more immersive audio-visual simulations.
The scenarios involved non-lethal traffic incidents, so the results may not apply to high-stakes situations. The influence of rule-breaking was inconsistent and sometimes reversed between written and audio-visual formats; the researchers say the study could not determine whether this reflected how minor the violations seemed, how salient they were in the text or differences in how participants imagined the scenes. Further research is needed to test these explanations and the findings in more serious situations.
