Kang Proposes Behavioral Realist Revision of Affirmative Action

Recent findings in implicit social cognition (ISC) — a science that measures people's subconscious biases — can provide a scientific basis for justifying and revising affirmative action, according to Jerry Kang, a UCLA law professor who explained his "behavioral realist" model of affirmative action at an Oct. 21 talk sponsored by the Center for the Study of Race and Law.

Behavioral realism, Kang explained, takes into account the scientific findings of ISC and uses them to understand how people's subconscious biases affect their behavior in different situations. With this information, lawmakers can craft policies to address current discrimination.

"Officially, we say we're colorblind," Kang said. "What I'm searching for is colorblindness to the infrared frequencies that lurk beneath."

In a paper about behavioral realism and affirmative action that he co-authored with Harvard psychology professor Mahzarin R. Banaji, Kang uses the term "fair measures" rather than "affirmative action" to describe the policies that would use ISC science to counter discrimination. "We don't like the term 'affirmative action' because it has too much baggage," he said.

Traditionally, most justifications for affirmative action have taken one of two approaches: the "backward-looking" approach that focuses on redress for past discrimination or the "forward-looking" approach that focuses on the operational, educational and economic benefits of diversity, Kang said. Fair measures, in contrast, focus on the present.

"It's not about the past, it's not about the future, it's about discrimination going on right now through implicit mechanisms," Kang said.

In one implicit bias test, participants are told to respond to pictures of white faces and positive words and pictures of black faces and negative words. When the rules change — pairing white faces with negative words and black faces with positive words — most participants take a significantly longer time to respond and make more mistakes, Kang said. In another test, subjects play a video game and are told to shoot anyone carrying a weapon. Players typically shoot black characters more quickly than white characters and shoot more unarmed blacks than whites. Similar tests reveal widespread biases against groups such as women, homosexuals, Latinos, and the elderly, Kang said.

Such experiments demonstrate that even people who pride themselves on their colorblindness often hold subconscious biases. "Your explicit self-commitments do not accurately predict your implicit biases," he said. In many cases, tests show that women even exhibit implicit bias against women and black people against blacks.

Another reason Kang cited for the need for fair measures was the widespread phenomenon of "stereotype threat," which occurs when consciousness of stereotypes causes certain groups to overperform or underperform on tests designed to measure merit or intelligence. "When they're told that it's an IQ test, the difference between white and black scores is substantial," Kang said, but when the students are told the test is just a laboratory questionnaire, the difference between white and black students' scores is statistically insignificant.

In another experiment, Kang said, Asian-American women got higher scores on a math test when they were primed, through a series of questions, to think of themselves as Asian, but got lower scores when they were primed to think of themselves as women.

In place of the traditional argument that affirmative action provides positive "role models" to aspiring women and minorities, Kang proposed that prominent women and minorities serve a broader role. According to Kang, those figures of authority become "de-biasing agents" who dispel stereotypes and help rid the entire community of its biases.

Scientific tests demonstrate the importance and effectiveness of de-biasing agents, Kang said. For example, people's implicit bias against African-Americans diminishes if they are shown pictures of culturally powerful and positive black figures such as Martin Luther King Jr.

Because behavioral realists rely on science to measure bias, they also avoid the problem of vagueness when asked how long practices such as affirmative action should continue, Kang said.

"Fair measures should end when implicit bias goes to zero," he said.