Mahzarin R. Banaji, a Harvard social psychologist, studies how people think, and how they think they relate to one another. She’s an expert in the little secrets we all have: those implicit attitudes — sometimes prejudicial — regarding race, age, gender, and similar territories of otherness.

At Harvard since 2002, Banaji is the Richard Clarke Cabot Professor of Social Ethics in the Department of Psychology, and the Carol K. Pforzheimer Professor at the Radcliffe Institute for Advanced Study, where she is a fellow this year.

A decade ago, Banaji developed a series of tests to measure bias. Since then, courtesy of the Internet, about 6 million people have taken one or more of her exams in Project Implicit.

The site reveals an apparent secret garden of prejudices. You can test yourself on race, religion, gender, skin tone, weight, disability, and sexuality.

In a public lecture at Radcliffe’s 34 Concord Ave. building last week (Oct. 3), Banaji gave an audience of about 80 a snapshot of the psychology of ordinary prejudice — what she calls “mind bugs.”

Banaji’s intellectual interest in prejudice dates to her girlhood in South India. She grew up as a light-skinned member of a religious minority, the Zorasterians. That gave her a double lesson in the power of otherness. But what came as a special shock was how the lightness of her skin made her more marriageable than her darker-skinned cousins.

Years later — by way of a Ph.D. from Ohio State University and a teaching career at Yale — Banaji is doing pathbreaking work at Harvard on implicit and explicit prejudices, including the neural mechanisms of self-other comparisons.

Part of her work as a Radcliffe Fellow continues her research on what areas of the brain light up when a person is shown images of someone who looks different. She’s using functional magnetic resonance imaging (fMRI), a noninvasive medical imaging technique that delivers real-time pictures of living tissues.

Humans have a bias for believing what comes most easily to mind. What kills more people every year, Banaji asked, lightning or tornadoes? Most of us would say tornadoes, though that’s not the right answer. This human proclivity for choosing the unconsciously easy answer — “these buggy ways of making decisions,” she said — can lead to errors “that go massively astray.”

Bias, in part, is an artifact of time millennia ago when we lived “in little bands of people [who] looked exactly the same,” said Banaji. In those days, a strong sense of otherness was probably protective.

Today, we have to move through a series of cultures — religious, racial, sexual, and more — in everyday life. That means our atavistic, unconscious judgments may often be in error. Thinking that way creates illusions. In the face of these, said Banaji — in a reference to pioneer psychologist Roger Shepard — “the intellect is completely powerless.”

Our biases are rarely attempts “to be prejudiced in the old-fashioned sense,” she said. They’re more like unconscious reliance on group prejudices that we started acquiring very early.

Infants a month old can tell the difference between one ethnic group and another, and will associate pleasant music with familiar faces.

By age 6, children are frank and forward about their biases, including an explicit preference for people who seem lucky. But by age 10, a social mask — tact? — begins to cover over these automatic feelings. By adulthood, most of us would swear we are not as biased as tests seem to show we are.

Group prejudices are subtle and powerful, and affect all classes and races. “They seem to matter even if we don’t want [them] to,” said Banaji.

She cited a study of job preferences that offered a range of choices based on differences in salary, city of employment, and the personality and gender of the boss. Among other things, the findings showed that young applicants were willing to give up $3,500 a year in salary to get a male boss.

And race? In a 2005 study, Banaji showed separate cohorts of black and white subjects pictures of frightening faces, black and white. In each racial group, fear of the different racial group persisted longer than fear of the same racial group.

Gender bias over the years is weakening, said Banaji in the Radcliffe lecture. But racial bias “hasn’t changed a whit,” she claimed. “The data come in very clearly.”

Another test, on facial recognition, showed how race played into perceptions of how American a person is. Actor Hugh Grant, who is British, was perceived as more American than newscaster Connie Chung, who grew up in Maryland.

“Human behavior,” said Banaji, “is not entirely rational.” Her research into our implicit biases has made her more forgiving in the face of apparent prejudices. After all, she said, “the brain does not ask our permission to do this.”

How malleable is bias? “We are very adaptable, we do change,” said Banaji. Racial prejudice, for instance, is lessened by close personal contact, including interracial dating.

And not all “self-group” biases are bad, she said, including the favoritism we show a child, a neighbor, a workplace, or a sports team.

But biases that worked millions of years ago may not work today, said Banaji. “That world is not our world.”

JFK and the Cuban missile crisis — a new assessment