A couple of years ago a University of Washington researcher who studies how children develop social behaviors like kindness and generosity noticed something odd. The 15-month-old infants in her experiments seemed to be playing favorites among the researchers on her team, being more inclined to share toys or play with some researchers than others.

"It's not like one experimenter was nicer or friendlier to the babies - we control for factors like that," said Jessica Sommerville, a UW associate professor of psychology. She took a closer look at the data and realized that the babies were more likely to help researchers who shared the same ethnicity, a phenomenon known as in-group bias, or favoring people who have the same characteristics as oneself.

"At the time, about half of the research assistants in my lab were Asian-American and the other half were Caucasian, and most of the babies in our experiments are Caucasian," Sommerville said. "We know that by preschool, children show in-group bias concerning race, but results in infants have been mixed."

She and her research team designed a new experiment to test how race and fairness - a social tendency that infants appear to notice - influence babies' selection of a playmate.

The findings, published in the online journal Frontiers in Psychology, show that 15-month-old babies value a person's fairness - whether or not an experimenter equally distributes toys - unless babies see that the experimenter unevenly distributed toys in a way that benefits a person of the same race as the infant.

"It's surprising to see these pro-social traits of valuing fairness so early on, but at the same time, we're also seeing that babies have self-motivated concerns too," Sommerville said.

Forty white 15-month-old babies sat on their parents' laps while watching two white experimenters divide toys between recipients. One experimenter divided the toys equally, and the other experimenter divided the toys unequally, as shown in this video:

Later, when the babies had a chance to choose who to play with, 70 percent of the time infants preferred the experimenter who distributed the toys fairly. This suggests that when individuals are the same race as the infant, babies favor fair over unfair individuals as playmates.

Watch an example of a "choice trial," when a baby chose between two experimenters:

Next, Sommerville and her team asked a more complex question. What would happen when individuals who were of the same race as the infant actually stood to benefit from inequity?

In a second experiment, 80 white 15-month-old infants saw a fair and an unfair experimenter distribute toys to a white and to an Asian recipient. Half the babies saw the unfair experimenter give more to the Asian recipient; and the other half of babies saw the experimenter give more to the white recipient.

When it came time to decide a playmate, infants seemed more tolerant of unfairness when the white recipient benefited from it. They picked the fair experimenter less often when the unfair experimenter gave more toys to the white recipient versus the Asian recipient.

"If all babies care about is fairness, then they would always pick the fair distributor, but we're also seeing that they're interested in consequences for their own group members," Sommerville said.

The findings imply that infants can take into account both race and social history (how a person treats someone else) when deciding which person would make a better playmate.

"Babies are sensitive to how people of the same ethnicity as the infant, versus a different ethnicity, are treated - they weren't just interested in who was being fair or unfair," said Monica Burns, co-author of the study and a former UW psychology undergraduate student. She's now a psychology graduate student at Harvard University.

"It's interesting how infants integrate information before choosing who to interact with, they're not just choosing based on a single dimension," she said.

Sommerville is quick to point out that her findings do not mean that babies are racist. "Racism connotes hostility," she said, "and that's not what we studied."

What the study does show is that babies use basic distinctions, including race, to start to "cleave the world apart by groups of what they are and aren't a part of," Sommerville said.