A study conducted five years ago received a lot of attention when it seemed to reveal that babies have an innate moral compass, that people are inherently good and that society is not responsible for the fact that the majority of us do not flagrantly break laws on a daily basis. Now, new research has come out that calls that into question. While it does not go so far as The Onion article that describes children as sociopaths, the study states that the researchers may have read a bit too far into babies' natural preferences.

The original study was conducted in 2007 by researchers at Yale University. Published in the journal Nature, the article was published by Kiley Hamlin and her team, and has since seemingly been reproduced successfully. The study made 6-month-olds and 10-month-old babies watch two interactions. In the first, a climber tried to ascend a hill, only to be either helped or hindered by a second person. The second task asked the infants to pick who they liked better: the helper or the hinderer. The babies overwhelmingly chose the helper.

Not so fast says Damian Scarf and his fellow researchers from New Zealand's University of Otago, in their study published in PLoS One, as the researchers were probably conflating two things. In the original interaction, both the helper and the hinderer bump into the climber, and the researchers say that the babies probably did not like that very much. At the end, in Hamlin's study, the climber jumped for joy at the top of the hill, which Scarf and his team thought that babies liked. According to Scarf, the babies picked the helpers not because they enjoyed helping others and could differentiate between right and wrong, but because after the helper's interaction with the climber, the climber jumped for joy.

So Scarf and his investigative team recreated the study, but with some crucial differences. They performed the climber-helper interaction, but without the jumping at the end. They also performed an additional interaction with the climber and a neutral person. The collision remained in both sets of interaction. Scarf and his team found that babies tended to prefer the neutral person over the helper.

Then Scarf and his colleagues manipulated the task, so that the climber bounced at the end of interactions with the helper and with the hinderer. They postulated that infants would prefer the person associated with the climbers' leap for joy. Indeed, when shown interactions in which the climber bounced after meeting the helper, babies preferred the helper. But babies also preferred the hinderer if the climber bounced after their interaction and babies preferred the helper and the hinderer in about equal numbers if the climber bounced after both interactions.

The study casts doubt on Hamlin's idea that moral compasses are innate, rather than learned.