Snap judgments based on appearances may once have been beneficial and evolutionarily adaptive from a health perspective-such as aversions to faces that are unfamiliar to us because they may indicate illness. A genetic disorder, for instance, might come with facial disfigurement. There was once some merit to that concept, but in today’s global community, those snap judgments are at best rarely accurate. Leslie Zebrowitz and Gillian Rhodes have called this the “anomalous face overgeneralization effect.”

Societies have globalized, but instincts to demarcate self and other-and to rush to judgments accordingly-persist. As Obama told The New Yorker last week, “Your job as a citizen and as a decent human being is to constantly affirm and lift up and fight for treating people with kindness and respect and understanding. You should anticipate that at any given moment there’s going to be flare-ups of bigotry that you may have to confront, or may be inside you and you have to vanquish.”

There is, now, a flare up. But the problem can be approached empirically, and it is a solvable one. This month in the journal Nature Human Behavior, Todorov’s group at Princeton has backtracked on some of its prior work about what makes people judge certain faces certain ways, saying that tendencies can quickly be reprogrammed by new exposures.

I’ve covered Todorov’s work on facial bias before (” The Introverted Face “), in which he and fellow psychologists Christopher Olivola and Friederike Funk found that people tend to project characteristics like introversion and extroversion onto people based on features like this:

Todorov first became interested in how these stereotypes came about after studying the judgments people made about elected officials. When people were relatively uninformed, the appearance of a candidate mattered a lot. The image at the top of this piece is a rendering of the face the faces that he found to be least and most attractive on average-and that attractiveness went along with being deemed trustworthy and competent.