In April 2017, I became a defendant—or something close to one.

I was conducting anthropological fieldwork with the Mentawai people of Siberut Island in Indonesia. It was my third week of a long field stay, and I was figuring my living arrangements. My research assistant and I set up a meeting with two brothers, both shamans, to discuss building a small house next to their longhouse. But my plans changed. I decided to live elsewhere, on another clan’s land, and skipped the meeting with the shaman-brothers, thinking it was informal and non-binding. They were friends; I had spent a summer living with them, so I thought they would understand.

The brothers felt differently. A week or so after I missed the meeting, I walked into a discussion among them and their cousins. The older brother was animated, speaking forcefully. At first, I thought he was describing a hunt, but, I wrote in my field notes, “then I realized he was talking about me. Or at least me and [my field assistant].”

He explained how we had inconvenienced them by not coming. Waiting for us, they put off a hunting trip. They had set aside a chicken to sacrifice in a small ceremony but had to cancel that too, eventually killing the chicken for a profane evening meal.

“We demand a tulou payment from you all,” he said [Ku tulougi kai kap.]. Then he asked for 200,000 Indonesian rupiah (15 USD at the time).

I knew about Mentawai justice before, but this was my first intimate encounter with the institution. At its core was tulou, a payment transferred from the offender to the victim, usually paid in local resources like pigs, chickens, durian trees, and sago. Translated by many anthropologists as “fine”, tulou is said to comprise a replacement for damages (silinia) and an additional penalty (tulounia). In the weeks and months following the payment, my intimacy with Mentawai justice grew. Over the course of the year, I conducted 199 interviews, systematically collecting data about 444 cases of wrongdoing, most of which were followed by a tulou payment. Whenever I could, I corroborated cases, either re-interviewing people or asking other parties involved. Knowing my strange fascination with Mentawai justice, people invited me to observe and record negotiations over tulou payments.

In our new study, "Evidence for third-party mediation but not punishment in Mentawai justice", Zach Garfield and I use this data to address a longstanding question in the social sciences: What is the role of third parties in sustaining cooperation, especially in small-scale, decentralized settings? Our findings demonstrate the diversity of ways that societies maintain order.

When I started to study the tulou system, I thought it was enforced by the community. If you wrong someone, you pay them resources. Refuse, I figured, and the community steps in. Even if third parties do not get physically involved, I expected a person’s reputation to plummet following a refusal to pay. 

This impression was inspired by a large literature on social order in small-scale societies. Over the last two decades, researchers have demonstrated how small communities like mining camps and pirate ships create rules and enforce them through coordinated violence. In a similar vein, a huge body of research, spanning theoretical models, experimental games with children, and even studies of East African pastoralists, suggests that humans’ profound levels of cooperation are enabled by third-party punishment. 

Yet across hundreds of transgressions, Zach and I found no evidence of direct third-party punishment. Mentawai victims and aggrieved parties demanded payment. If a transgressor was punished for failing to pay—if someone seized their iron cooking pots, for instance—that punishment was always imposed by victims and never by third parties. 

Getting at indirect sanctions was harder. We could not tell, for instance, whether third parties interacted less with someone after they refused to pay tulou. Still, several lines of evidence suggest that Mentawai justice is regulated less by third parties’ concerns and more by the dyadic relationship between the victim and the offender. Nearly 20% of transgressions were followed by no payment, and payments were less likely when transgressions were among related individuals. When asked why a payment they reported was particularly cheap or expensive, participants frequently pointed to dyadic interests, such as friendship or reciprocity, with 25% mentioning kinship, in particular. With one possible exception, no one cited third parties’ concerns.

Our paper describes two cases in detail. Both were recorded and reviewed with disputants afterwards, and both underscore the importance of dyadic concerns in regulating justice. A man’s great-uncle stole his pig, but the man dropped the case when his uncle threatened to dissolve the relationship. A family whose cow was killed in a deer trap lowered their tulou demands to maintain a good relationship with the offenders. In Mentawai, it seems, penalties function less as communally endorsed norm enforcement and more as costs that offenders and victims negotiate over to reinstate goodwill.

This paper started after I saw Zach present a paper (now published) in the weekly seminar series at the Institute for Advanced Study in Toulouse, where we are both post-docs. Comparing 59 societies, he showed that one of leaders’ primary functions was to resolve conflicts. His cross-cultural work suggested a very different way in which third parties might be involved: mediation. (In my frantic excitement, I sent Zach 7 emails during his hour-long talk on how we might test his cross-cultural findings using the Mentawai data.) Consistent with our impressions, we found that mediators were overwhelmingly third parties: Of 71 cases brought to local mediators and for which we had data, 75% were overseen by individuals who did not share clan membership with either disputant. Strikingly, however, people did not seem to prefer third parties. Rather, they sought out influential shamans and elders, many of whom happened to be third parties. 

It’s easy for many people (ourselves included) to transfer our intuitions about justice in large-scale societies to small-scale settings. We imagine the community as the arbiter of justice, that people must be incensed when wrongdoers go unpunished. But the Mentawai data suggest a different story. Social order can be decentralized, a product less of collective enforcement and more of a network of families bargaining over how to treat each other.