First, they found that infants as young as 3 months old show racial bias. Now they've found a way to shake up those early preferences -- and perhaps undermine more pernicious racial stereotypes as we grow up.
Research from the University of Delaware had earlier shown that 3-month-olds have a visual preference for faces that match the racial group they see most in their daily lives.
A new study from the same team -- conducted as part of a decade-long international research project and published last month in the journal Developmental Science -- details an experiment that seemed to shift those unconscious biases seen in young children.
Why do babies even have such biases? While infants can distinguish different races, they're less able to distinguish individual faces within a less-familiar racial group. So, for example, a white baby may be able to differentiate among -- and thus prefer -- individual white faces while struggling to differentiate among black or Asian faces.
In other words, these very early leanings seem to be based on visual perception. Actual social biases don't come into play generally until children are around 3 or 4 years old.
The researchers hypothesized, however, that visual biases in infants might be related to the development of social biases in later childhood and adulthood -- and therefore disrupting early visual biases could help dislodge social biases, perhaps for the long term.
"We hit on the idea that if the perceptual and social biases are linked, we might be able to reduce the social bias by perceptual means," Dr. Paul Quinn, professor of psychology and brain science at the University of Delaware and the study's lead author, said in a press release.
To test this hypothesis, researchers in Delaware and China created images of racially ambiguous faces by melding individual photos of African and Asian faces. Some of the faces had pleasant expressions, while others were more severe.
The researchers showed these photos to Chinese children, aged 4 to 6, and asked them to identify whether the faces were Asian or African. The kids tended to rate the happy-looking faces as Asian and the severe-looking faces as African.
Next, the researchers showed the children five different photos of African faces and gave individual names to the faces. This was repeated for 15 to 30 minutes until the children were able to identify the faces by name.
When the kids were shown the photos of racially ambiguous faces again, the bias they had displayed toward their own ethnic group was reduced significantly.
Why? The theory is that the children's perceptual bias for the more familiar-looking faces was undercut when they learned to see the unfamiliar faces as individuals, too.
"It may be that the individuation disrupts the process by which stereotypes are generalized across members of a [racial] category," Quinn told The Huffington Post via email. But he noted that further research is needed to explore how long these effects last and to determine whether the effects are the result of mere exposure.
One question would be whether individualizing people must be done with language -- that is, by naming them. "Would we get it with a unique musical tone for each face?" Quinn said. "Would we get it if we just asked the children to memorize the faces?"
In the real world, the study suggests that personal interactions with people of other races may go a long way in counteracting stereotypes.
"If you want to reduce implicit bias ... provide experience with the other-race group," Quinn said. "In particular, experience that promotes perceiving members of the other-race group as distinct individuals with unique identities."
Other stories in HuffPost's Science of Racism series: