Racial bias hard-wired, but can be overcome

By Rong Xiaoqing Source:Global Times Published: 2017/7/27 20:28:39

Illustration: Liu Rui/GT

After Zhang Yingying, a Chinese visiting scholar at the University of Illinois at Urbana Champagne, was reported to have disappeared after she climbed into a stranger's car, an editor of a publication in China asked me to write an article to teach Chinese new arrivals how to tell bad guys from good guys in the US. But I declined. I am reluctant to spread tips that may suggest you can judge people by their appearance or behavior, because it is not only politically incorrect, but also, more often than not, inaccurate.

Zhang's case is the best example. Brendt Christensen, the suspect in her kidnapping, is a 28-year-old white man and a former PhD student and assistant teacher in physics at the university - an image that is unlikely to trigger warning signals in any guidelines for detecting trouble. Before the photo of the suspect was published by the media, and when many people were questioning Zhang's naivety in climbing into the car of a stranger, the image of the stranger in their minds was very likely to be a completely different one.

But charged waters of racial tension aside, it is true that we often have too much bias and too many misjudgments when we consider different racial groups than our own. Many times we cannot even tell one face from another. When the police released Zhang's photo, several people in a nearby town called to report that they'd seen her selling jewelry on the street. That hope unsurprisingly died quickly when police showed Zhang's parents the image of the jewelry seller that they found in a surveillance video.

The jewelry seller's picture was not released. But it is all but certain that she doesn't look like a twin of Zhang in the eyes of most Chinese people. Their confusion between the two Asian girls once again reaffirmed the power of the "cross-race effect."

The phenomenon, which basically can be summarized as being "face blind" to people of other races, has been nagging psychologists and neurologists for a long time. American psychologist Gustave Feingold wrote in his 1914 paper The Influence of Environment on Identification of Persons and Things: "To the uninitiated American, all Asiatics look alike, while to the Asiatics, all White men look alike."

Scientists have conducted numerous experiments since then to tackle the phenomenon. These experiments have proved not only the existence of the "cross-race effect," but also shown that our brain reacts in different ways when we see people of other races and people of our own. These conclusions are already frustrating because they basically confirmed that the root of the "cross-race effect" is genetic and, therefore, out of our own control. What's more disturbing is that the problem is so ingrained that it even finds its way into artificial intelligence.

A 2011 study of facial recognition products conducted by the National Institute of Standards and Technologies (NIST) of the US found that products developed in China, Japan and South Korea have a higher accuracy rate on Asian faces than Caucasian ones, and products developed in France, Germany and the US lead to opposite results. In 2015, a photo app launched by Google backfired when it tagged some photos of black people as "gorilla." 

The good news is that technology has been improving at a fast pace. Google quickly corrected the app. And a 2013 study from NIST found that the accuracy of facial recognition algorithms had improved up to 30 percent since 2010.

Undoubtedly the algorithms didn't correct themselves. It's people who designed them in a better way.

People are able to combat the cross-race effect among themselves, according to some recent psychology experiments. In the results of experiments published in 2000, Kent State University psychologist Daniel Levin asked the participants to detect photos of black faces and white faces among a group of photos and concluded that when we look at people from our race, we tend to pay more attention to their individual features while when we look at people from a different race, we tend to pay more attention to their categorization features, aka race. 

Then, in 2006, Miami University researchers Kurt Hugenberg and Heather Claypool and University of Cincinnati researcher Jennifer Miller had participants conduct a similar face detecting game. But this time, they had two groups of participants. One group of them had the cross-race effect explained and were reminded to pay close attention to the individual features of faces in advance. They found the performance of this group was much better than the other.

So the secret is indeed simple: In order to subdue bias, we need to remind ourselves that we are biased all the time.

The author is a New York-based journalist. rong_xiaoqing@hotmail.com Follow us on Twitter @GTopinion

Posted in: VIEWPOINT

blog comments powered by Disqus