People tend to think that others are less at fault when their actions have been influenced by third-party actors, a new Yale study has found.

In a five-part study, researchers presented subjects with different cases in which a person was placed in a situation controlled by a third party who caused them to commit an immoral act. Subjects were asked if their view of the actor’s moral responsibility was impacted by the third party’s intent to have that action happen. On average, participants thought a third party influencing an actor made the actor less at fault. The study is the first to demonstrate the human perception of the conflict between manipulation and free will. The study was published in the journal Cognitive Science on Nov. 9.

“There is this deep causal relationship between the manipulator and the manipulee, so we might not be able to explain exactly what happened in that situation with the distribution of blame,” said Jonathan Phillips GRD ’16, the first author of the paper and a student in philosophy and psychology at Yale. “It’s more of what were the agent’s intentions and how we causally reason about the events that occur.”

Researchers gave participants one of two vignettes, both based on a food shortage in a poor country. In the first version, the government intentionally destroyed farming machinery and burned food stores in order to cause a food shortage. When citizens found out that a neighboring village still had food, they attacked and raided the village, and the government — planning all along to start a war — was pleased. In the second version of the vignette, the government accidentally destroyed farm machinery, unintentionally causing a food shortage. When citizens learned that a neighboring village had food, they raided that village, and the government — having failed in its attempt to create a prosperous economic program — felt bad. After each vignette, participants were asked whether the workers or the government should be blamed for the attack on the village. Participants said that the workers were significantly less at fault in the first scenario.

Phillips added that this study shows that when someone is influenced or manipulated into committing wrongdoing, people often do not see them as fully morally responsible.

According to Fiery Cushman, Harvard psychology professor and director of the school’s Moral Psychology Research Lab, Phillips’ study adds a new layer to a typically two-dimensional situation. Cushman said that since its foundation in the 1930s, moral psychology has only focused on the agent — what the agent did and what the agent meant to accomplish.

“This study shows a third thing is important: Was someone else manipulating [the agent]?” Cushman said.

Even if the agent’s actions and intentions are the same, whether he or she is being manipulated into committing that action changes others’ perceptions of whether or not they are fully at fault.

Cushman described this small nuance of manipulation with an anecdote about a mother and a grandmother. If a mother steals drugs of her own free will, she is morally responsible. But if a mother steals drugs of her own free will at the prompting of the grandmother, the grandmother is now morally responsible. The end result is the same, but the manipulator is seen as the cause of the behavior.

“Our moral intuitions seem designed to direct blame to the person whose behavior we need to change,” Cushman said.

The study has both theoretical and practical implications, Yale cognitive science and philosophy professor Joshua Knobe said. It impacts people’s concept of free will because it showcases people almost as “tools for someone else’s will.” On a more practical level, the study’s findings could be applied to trials concerning entrapment — when police trick a person into committing an illegal act. The hearings for these cases rarely consider intention at all, Phillips said.

This study is one of many on morality and intentions that Phillips hopes to conduct. While this study shows that intentions do influence blame, future research will delve into why this slight difference has such an impact on human perception.

The study was conducted using Amazon’s Mechanical Turk machine, a crowdsourcing tool frequently used in game theory and psychology studies to draw participants from all across the U.S.