In recent court cases involving affirmative action for university admissions, the obvious question became 'when should it ever end?' and how is that not discrimination? Supporters of race-based admissions argued that ending discrimination would mean favoritism.

Favoritism is something less understood than as a form of discrimination, the oft-repeated belief is that discrimination is a hostile act, but a new paper in American Psychologist argues it is even worse than believed. It's a review of other psychology papers, which are overrun with stereotype threats and Implicit Association tests, so the results are not a surprise.

 The authors reviewed surveys (naturally) and articles that had experiments on discrimination from the last five decades. They say they were surprised - meaning we are to believe the work was prospective rather than retrospective, despite the fact that they went back to the 1960s - to find that the discrimination observed in those studies occurred much more often as helping rather than harming someone.

What may be relevant is that the papers they analyzed defined discrimination as based on negative attitudes and hostility, only rarely treating favoritism as a component of discrimination. 

"We can produce discrimination without having any intent to discriminate or any dislike for those who end up being disadvantaged by our behavior," said University of Washington psychologist Tony Greenwald, who co-authored the review with Thomas Pettigrew of the University of California, Santa Cruz.

They myth of discrimination is that it is always something like a black person using racial slurs about white people, or a homophobe yelling slurs at a gay couple. But it is sometimes favoritism instead. Take a manager conducting reviews of two employees, they posit.  Both fall between two performance categories. The manager gives a higher category to the employee whose child is friends with the manager's child, leading to a promotion and salary raise, while the other employee receives a smaller raise and no promotion.

Was the manager consciously discriminating against the second employee? Or did she simply give a boost to someone to whom she had an "ingroup" connection?

"Your 'ingroup' involves people that you feel comfortable with, people you identify with," Greenwald explained. "We usually think first of demographic characteristics like age, race, sex, religion and ethnicity as establishing an ingroup, but there are also ingroups based on occupation, neighborhood and schools attended, among other things. Outgroups are those with whom you don't identify."

Obviously this simplistic scenario falls apart immediately. What if one of the candidates was a pretty girl and the manager is single? Then she might get preference over the family friend. Is that discrimination? If it is, the definition of discrimination is choosing between any people for any reason - which makes a definition sociologically valueless.

Greenwald and Pettigrew contend that doing favors for those like you, rather than inflicting harm on those unlike you, causes the majority of discrimination in the U.S.

"This is not to say that prejudice and hostility are not related to outgroup discrimination," Pettigrew said. "But they are not as central to most discrimination as ingroup favoritism."

Yet, they believe, the social sciences have emphasized prejudicial hostility as the root of discrimination, which calls into question the overwhelming majority of all social sciences papers on race, and their conclusions, and the later articles citing those.

"We looked at how prejudice has been defined in the history of psychology. It has generally been understood as hostility toward outgroups. That's easy to do, because inter-group conflict is an obvious fact of life," Greenwald said. "There are international conflicts, wars, gang battles, labor-management conflicts. When such conflicts are going on it's natural to think of them as rooted in hostility."

Greenwald hopes researchers will change how they study discrimination, because research results have substantial implications both for how discrimination is identified and how it can be ameliorated in employment, health care, education and daily life.

He said overt acts of discrimination began to decline starting in the 1960s following civil rights laws. But prejudicial attitudes didn't necessarily change. What changed is that people were no longer legally allowed to act on their prejudices by, for example, denying housing to blacks or jobs to women.

The co-authors say that racial ingroup favoritism can be very subtle. For instance, if you work in an office that is mostly white and you're asked to recommend someone for a job opening, you're more likely to recommend someone who is like you and the rest of your ingroup.

This sort of ingroup favoritism happens at all ages and in different situations. Greenwald said it can happen on the playground, where children may exhibit ingroup favoritism based on race, economic class, or the same school or sports team.

"Hostility isn't integral to the definition of discrimination; you can treat people differently without being hostile to anyone," Greenwald said. "But it is societally important to understand how discrimination can occur both without hostility and without any intent to discriminate."