Originally published in Pacific Standard Magazine
By Alan Honick
I’ve been investigating the origins and evolution of the human sense of fairness, along with evolutionary biologist Gordon Orians. We’re trying to gain a better understanding of the role it plays in modern social, economic, and political institutions.
I recently gave a talk on the subject. To begin, I asked the audience members to recollect their first encounter with the concept of fairness. I had formed a fledgling hypothesis, and wanted to put it to the test.
As people raised their hands, I called on them to share their memories. A pattern quickly emerged:
- “I had to take the rap for something my sister actually did!”
- “My parents gave my brother a puppy, and I’m the one who loved dogs. He didn’t even like them.”
- “I came from a family with nine siblings, and we had to fight each other for food.”
- “I was an only child, and I really wanted a brother – all my friends had brothers.”
- “I was foreign, and different, and all the other kids singled me out to pick on.”
Everyone—100 percent of an admittedly tiny sample—remembered their first encounter with fairness in its negative incarnation. My hypothesis is that this pattern holds, if not universally, at least among a very high percentage of individuals.
We apparently begin with fairness as our default position. We’re surprised and affronted when we first face a situation that seems unfair, so that’s what we remember.
It seems we’re predisposed to prefer and expect fairness.
One place where they are looking into this question is the Early Childhood Cognition Lab at the University of Washington. Jessica Sommerville, principal investigator at the lab, and Marco Schmidt, currently at the Max Planck Institute for Evolutionary Anthropology, had collaborated on a research project, Fairness Expectations and Altruistic Sharing in 15-Month-Old Infants, was published in PLOS in 2011.
They performed two experiments with the sample group—one that measured the infants’ sensitivity to fairness, and another their preferences for sharing.
Would any observed inclination toward fairness or altruism be an indicator of an innate influence? At that age, how much of an infant’s behavior is modified by its siblings and parents, and how much emerges from inherited personality traits?
Previous experiments had measured the emergence of sharing behavior in children. In 2008, researchers Ernst Fehr, Helen Bernhard, and Bettina Rockenbach published a paper in Nature called Egalitarianism in Young Children. They created a set of tasks where the subjects—children between the ages of 3 and 8—had to decide to share or not share candy with an imagined partner.
Generally speaking, the researchers found that while most children behave selfishly at age 3, as they progress to age 8, many more “prefer resource allocations that remove advantageous or disadvantageous inequality.”
Only one of their tasks involved altruistic sharing—in other words, if the subjects wanted to make the resource allocation equal, they had to reduce their own supply. In the other tasks, they could make it equal at no cost to themselves. At all ages, the altruistic sharing task elicited the least generous behavior. In all the tasks, sharing behavior rose steadily with age.
The University of Washington researchers thought that, by their design, these sharing experiments could not have detected aspects of fairness that might be evident at an earlier age.
They wanted to explore a quality called third-party fairness. Their paper cites “work suggesting that evaluating interactions between other individuals along the dimension of fairness (third-party fairness) is distinct from being the victim or agent (egocentric fairness) of unfair behavior.”
They wanted to see if they might detect a sense of third-party fairness at 15 months of age. But how, in subjects that can’t yet talk or follow verbal instructions?
Violation of Expectancy
Their first experiment employed a phenomenon known as “violation of expectancy.” It relies on the observation that infants will look longer at something that surprises them than at something they already expected to see.
Violation of expectancy, or VOE, takes advantage of a natural trait—not just in infants, but in adults and animals as well. Imagine that you’re walking along, and you hear an airplane overhead. You look up, and see that the airplane is flying upside down. Would you look longer, than if it was flying normally?
As an experimental technique with infants, VOE has proven to be a robust measure of an infant’s expectations, and how long they look is a reliable measure of VOE. In many different experimental settings, infants look longer when they perceive a mismatch between an outcome and the events that led up to it. Think of a magician pulling a rabbit from a hat; we’re surprised only because he’s shown us the empty hat before he did it.
In their violation of expectancy test, the third-party situation was set up so the infants were passive observers of scenarios that demonstrated fair (equal) versus unfair (unequal) divisions of goods—either crackers or milk. Their reactions were measured in terms of how long they looked at each outcome.
The infant was seated in a parent’s lap, but the parents were instructed to look down and remain silent during the experiment, so they wouldn’t communicate their own reactions to the children. A series of scenarios played out on the video screen as the infants’ reactions were recorded.
Sommerville explained how it worked.
“We showed babies a little story, and in the story they saw someone who had either a bowl of crackers or a pitcher of milk, and there were two recipients sitting at a table. The distributor demonstrated that they had the bowl of crackers or pitcher of milk, and then the two recipients simultaneously asked for some.
“And then at that point…we put up a little black occluding screen, so that babies could see that the person was distributing crackers or pouring the milk, but what they couldn’t see was how much was going where. The idea there was twofold. The first was that we hoped to elicit infants’ expectations about how the story would end, and the second reason…was that we wanted the events to be identical up until the ending. And for the event only to differ in terms of whether it culminated in a fair outcome or an unfair outcome.”
Sometimes they saw two crackers on each plate, and other times divisions of one and three. The glasses either had identical levels of milk, or clearly uneven ones. An unseen observer would record how long they looked at each outcome.
In the clip below, Sommerville shows some of the video they shot of the experiments to Gordon Orians, and explains more about how they worked.
As the video explains, they conducted multiple trials with each subject, including several control conditions, meant to randomize the way the scenarios were presented to the subjects. With 15-month olds, of course, the number of trials is limited.
Nevertheless, a pattern emerged when the looking times were averaged across all the infants, between the normal scenario described above, and the most critical control — presenting the scenario with no recipients at the table. In these trials, the woman who distributed crackers or milk sat at the head of the table alone. Everything else about the scenario was identical, though obviously without the “ask”, since there were no recipients.
With recipients present, the infants looked at the unfair outcome an average of 50 percent longer than the fair outcome. When the situation was depersonalized, and the infants saw only plates or glasses with crackers or milk, they looked at both fair and unfair outcomes for roughly equal amounts of time.
They were interested in trying to understand how a sense of fairness and altruistic behavior might be developmentally related. In the abstract of their paper they write:
“At the proximate level, biologists and social scientists have identified other-regarding preferences—such as fairness based on egalitarian motives, and altruism—as likely candidates for fostering large-scale cooperation. A critical question concerns the ontogenetic origins of these constituents of cooperative behavior, as well as whether they emerge independently or in an interrelated fashion.”
Like the sense of fairness, altruistic behavior is not easily studied in infants. The experiment they devised had the infants sitting in a parent’s lap. They were turned to face a table that had two toys on it — a small doll, and a block. The toys were far enough away from each other on the table that they couldn’t grasp both at once, so they had to choose one. Nearly all of them chose the more interesting one, the doll.
They were turned around away from the table, and both toys were handed to them. Then they were confronted with another experimenter, a young woman unknown to them, who would hold out her hands ask for one of the toys.
In the following video, Sommerville shows Gordon how it works, and relates some of the results.
Their suspicion regarding the developmental relatedness of fairness and altruism was confirmed.
Sommerville explained the results: “In our sample of fifteen month olds, we found that about a third of them shared the toy they had earlier chosen, a third of them shared the unchosen toy…and a third of them didn’t share either toy.
“The most interesting thing was that we found that how infants performed on that test was related to what they did during the earlier looking task. So infants who shared their preferred toy were the ones who most strongly reacted, or showed the greatest attention when that unfair outcome was played.
“So we thought those results were pretty exciting at a couple of different levels, because we had convergent validity, things that you think should go together — fairness and altruism — appear to go together in infants, and also because it gives some empirical teeth to the notion that these two kind of constructs are linked.”
It was interesting—as Gordon observed at the end of the video—that there was such a close correlation in the infants’ performance on the fairness and altruism tests. Roughly one-quarter of the subjects exhibited strong preferences for both altruism and fairness. Moreover, as Sommerville put it, “How quickly they shared their preferred toy is related to their looking time performance.”
In other words, the altruistic sharers were also the most surprised by the unfair distribution of goods. There was nearly a perfect correlation between these two groups.
While that seems to be clear evidence that a sense of fairness and altruism are linked in individual development, it gives only a suggestion that these personality traits may be, to some degree, encoded in our genes. By 15 months environmental influences, such as parenting style and sibling interactions, have undoubtedly begun to shape behavior.
Whatever the roots of causality that underlie these prosocial behaviors, it seems remarkable that a sense of fairness and altruism should be evident so early in life, with such marked variability in its expression. The experiments performed by Ernst Fehr and colleagues demonstrated that prosocial behaviors emerge as children learn to interact within a social milieu. Sommerville’s experiments suggest the possibility that some individuals may develop a strong sense of fairness more naturally.