People often yearn for the innocence of youth, that time of unbiased childlike wonder that came before experience intruded, transforming us into cynical and self-interested adults. But we may be nostalgic for a time that never quite existed, according to a new study that found babies prefer individuals who harm, rather than help, characters who are different from them.

The research -- conducted through clever experiments that drew on rabbit and dog puppets, balls, and graham crackers and green beans -- reveals the possible cognitive roots of the social biases and attitudes underlying violence toward people who are perceived as different.

Led by scientists at Yale University and the University of British Columbia, the researchers posed a complicated social scenario to 9-month-old and 14-month-old babies: If they saw a rabbit puppet who was either similar or different from them in some fundamental way -- in this case, preferring graham crackers or green beans -- would they care how others treated the rabbit?

The researchers already knew two basic things about the choices and preferences of infants. Just like adults, who tend to like people who are similar to them, babies are drawn to others who share their tastes in food and toys. Hollywood movies leverage our impulse to cheer for do-gooder heroes over villains; babies similarly prefer a character that helps someone else climb a mountain rather than pushing them down it, a previous study had shown.

But would babies always, universally, prefer heroes to villains? Or would their preference depend on who was being helped or hindered? The researchers wondered: would they see the enemy of their enemy as a friend?

“I was surprised, and my liberal bleeding heart sunk like a stone, when we found them actually choosing, really robustly, the puppet who punishes” the rabbit puppet that did not share the baby’s preference, said Karen Wynn, a professor of psychology and cognitive science at Yale and senior author of the work, published in the journal Psychological Science.

To do their work, the researchers recruited babies and initially ascertained whether they preferred green beans or graham crackers. Then, the babies watched a rabbit puppet act either similar or different than the baby, saying “Mmm, yum!”or “Ew, yuck!” toward the two foods. Later on, the babies watched another puppet show. In this one, the rabbit puppet played with a ball. At some point, the rabbit dropped the ball, which bounced toward one of two dog puppets -- a helper who retrieved it for the rabbit and a hinderer who snatched it away.

Then, the babies were given a choice: Did they reach for the dog puppet who had helped or harmed? When the rabbit puppet shared their tastes in food, the babies preferred the dog puppet who helped out. When the rabbit puppet was dissimilar, however, they chose the dog puppet who had stolen the ball.

The researchers then decided to see how the babies would choose if they introduced a neutral puppet into the experiment. They found that the older babies, the 14-month-olds, followed the same pattern. They preferred a helpful puppet over a neutral puppet when the rabbit was similar to them, and preferred a neutral puppet over a helpful one when the rabbit was not similar to them. But the pattern of behavior seemed to break down for nine-month-olds, with very few significant differences between how often they chose one puppet over another.

Wynn said that may be because the neutral puppet may be too subtle, and the scenarios may be too much to keep in mind for the younger babies.

What Wynn would like to study next is what further information -- social information, such as watching a parent hug the dissimilar puppet, or other information about shared preferences -- is sufficient to change the babies’ behavior.

“I’m also interested in finding out ways we can talk about similarities and differences between different groups of people, to each other and to our children and babies, that can help us all see the underlying similarities,” Wynn said.