A series of six studies published in the Journal of Experimental Psychology reveal that utilitarian agents, who opt to sacrifice a single individual for the greater good, are perceived as less predictable and less moral than deontological agents, who opt for inaction, resulting in 5 people harmed. This finding extends to a non-Western sample of Dani, a traditional indigenous population of Papua.
“I was personally curious about this topic because of how interested humans are in being ‘moral backseat drivers.’ We are intensely concerned with policing the behavior of others. There is a growing perspective in moral psychology originated by Oliver Scott Curry of Oxford, known as ‘morality as cooperation’ that posits that a moral sense evolved in humans in order to facilitate cooperation,” said study author Martin Turpin (@MartinHarry_T), a PhD candidate at the University of Waterloo.
“This moral sense allows us to accomplish far more together than we ever could apart. Naturally, we thought that one factor that is plausibly involved in being seen by others to be a good moral partner would be the extent to which your future behavior – such as, your likelihood of being loyal to or betraying a cooperation partner – could be predicted consistently.”
This research was inspired by “the synthesis of three ideas.” First, cooperation is important for humans to flourish. Second, morality may have evolved to support human cooperation. And third, being able to predict others’ behaviors is necessary to support successful cooperation.
Turpin and colleagues recruited 1988 US residents and 81 traditional people of Papua to partake in six studies. In Study 1, participants read two vignettes featuring a high-conflict moral dilemma involving a trolley, or a bomb, in which a person had the choice of sacrificing one individual to save five others (by actively pushing them on train tracks, or onto a bomb).
Below each vignette was a statement indicating whether this hypothetical individual decided to sacrifice one individual (utilitarian agent) or if they simply did not act (deontological agent). Participants provided ratings of their moral perception of the described agents, as well as their perceived predictability, harm, intentionality, and justifiability of the agents.
Study 2 and 3 largely followed the same procedure, with Study 3 introducing a low-conflict condition. In the low-conflict condition, agents made a choice between pulling a switch (or not) in the trolley dilemma and throwing a bomb outside a restaurant (or not) in the bomb dilemma. Study 4 expanded the list of moral (e.g., goodness, benevolence) and non-moral (e.g., status, wealth) dimensions participants provided ratings for. Study 5 introduced “what would you do?” and “what would others do?” judgements, and a measure of self-reported endorsement of utilitarian principles.
Lastly, Study 6 examined whether the results generalize to a non-Western sample in collaboration with a hypothesis-blind local interpreter from the Dani tribe. Given the Dani’s unfamiliarity with trolleys and railways, they were presented with a modified version of the Trolley dilemma, involving a falling tree.
The researchers found that moral impressions of agents acting in both high and low conflict dilemmas are influenced by perceptions of their predictability. Regardless of the consequences for actions, or violation of proscriptions against killing, participants preferred the more predictable agents.
Importantly, deontological agents were no longer morally preferred when the actions of utilitarian agents were made to appear more predictable. Assessments of predictability were strongly associated with judgments of an agent’s consistency, reliability, intelligibility, and methodicalness.
“One of the reasons people may express skepticism or dislike of people who are willing to make calculated sacrifices for ‘a greater good’ is that such people may subjectively be perceived as unpredictable,” Turpin told PsyPost. “Perhaps a skepticism created by knowing that there is some hypothetical offer which could result in them being willing to defect from a deal or betray you. The more predictable a person seems, the more moral they seem.”
The study author also noted a potential limitation. “We only looked at people making decisions about killing others for a perceived greater good. Since killing people is generally considered to be ‘a bad thing’ we can’t really make any strong claims about people behaving in unpredictably ‘good’ ways.”
“For example, is a friend who gets you a gift spontaneously and unpredictably ‘worse’ than a friend who does such good deeds predictably? Like many things in psychology, there may end up being a stark difference between how predictability as a trait is assessed for ‘bad’ versus ‘good’ actors.”
As for future research, Turpin said, “we still need to get a better sense of what ‘predictability’ as a term really means to people. There are lots of ways you can think about a person being ‘predictable’ – consistent, reliable, easy to understand. In this work we’ve found that predictability seems to be a kind of holistic term capturing lots of these potential concepts, perhaps especially consistency of behavior. But there is still more work to be done in really understanding how ‘predictability’ is interpreted in our moral lives.”
The research, “The search for predictable moral partners: Predictability and moral (character) preferences”, was authored by Martin Harry Turpin, Alexander C. Walker, Jonathan A. Fugelsang, Piotr Sorokowski, Igor Grossmann, and Michał Białek.