Abstract
In preparation for unavoidable collisions, autonomous vehicle (AV) manufacturers could program their cars with utilitarian ethical algorithms that maximize the number of lives saved during a crash. However, recent research employing hypothetical AV crash scenarios reveals that people are not willing to purchase a utilitarian AV despite judging them to be morally appropriate (Bonnefon, Shariff, & Rahwan, 2016). This important result, indicating evidence for a social dilemma, has not yet been psychologically explored by behavioral scientists. In order to address the psychological underpinnings of this phenomenon, we developed and tested a novel theoretical proposal – perspective-taking accessibility (PT accessibility). Accordingly, we established that providing participants with access to both situational perspectives (AV buyers can be passengers or pedestrians) in crash scenarios, eliminated the behavioral inconsistency between their utilitarian judgments of moral appropriateness and non-utilitarian purchasing behavior. Moreover, our full PT accessibility induced respondents' utilitarian prosocial judgments and purchasing behavior (Experiments 1a and 1b) and consistent utilitarian preferences across judgment tasks (Experiment 2). Crucially, with full PT accessibility, participants' utilitarian purchasing behavior as well as their willingness to buy and ride utilitarian AVs were informed by their utilitarian moral judgments. Full PT accessibility provides the participants with even odds of being a pedestrian or passenger in crash scenarios, and thus impartiality. It could be argued that full PT accessibility is a new type of ‘veil of ignorance’, which is not based on purposely induced self-interest and uneven risk options (as in Huang, Greene, & Bazerman, 2019), but rather is based on even odds of being a passenger or pedestrian, and therefore with even 50/50 chance to die/live as passenger or pedestrian. Under these circumstances one can measure utilitarian preferences.