BIO
Mahzarin Banaji (born in Nagpur, Maharashtra, but raised in Secunderabad, Telangana, India) earned her PhD in Experimental Social Psychology from the Ohio State University (United States) in 1986. Until 2001 she was a professor at Yale University, where she was appointed to the Reuben Post Halleck Chair in Psychology, and since then has been the Richard Clarke Cabot Professor of Social Ethics in the Department of Psychology at Harvard University. She was the first holder of the Carol K. Pforzheimer Chair at the Radcliffe Institute for Advanced Study at Harvard, and from 2011 to 2014 occupied the George A. and Helen Dunham Cowan Chair in Social and Human Dynamics at the Santa Fe Institute, an association that continues to this day. She is a Fellow of the American Academy of Arts and Sciences, the American Academy of Political and Social Science, and the Association for Psychological Science, of which she is a past president. With Anthony Greenwald, she published the book Blindspot: Hidden Biases of Good People (2013).
CONTRIBUTION
The unconscious processes that shape attitudes: implicit bias
“Professors Anthony Greenwald and Mahzarin Banaji developed the implicit association test, which enables reliable measurement of implicit bias and its effects on decision-making”. The “implicit bias” the committee refers to was first put forward by Greenwald and Banaji in their 1995 paper “Implicit Social Cognition: Attitudes, Self-esteem and Stereotypes,” published in the journal Psychological Review. In it, the two researchers described what was known about implicit attitudes and stereotypes, while acknowledging that there was as yet no means of measuring them. “We ended that article with a sentence saying that it would really be nice to have a measure that could assess individual differences in implicit attitudes and stereotypes,” Greenwald recalls today.
Armed with this motivation and the work of several decades, the Emeritus Professor of Psychology at the University of Washington came up with a test that measured reaction times in classifying prompts, a method both easy to use and readily obtainable. He called it the implicit association test (IAT) and convinced Banaji and one of her postdoc students Brian Nosek to apply the method in further research.
“We gave people a chance to experience it and they were very surprised by the results,” he relates. “In the first test, we applied what we called the race attitude IAT, which measures associations of black race and white race with pleasant and unpleasant valence or categories. Taking it myself, I discovered I had a stronger association of black with unpleasant than with pleasant and the reverse for white. And that implicit attitude was one that I didn’t at all want to have and in fact didn’t know I had.”
The IAT allows to measure and better understand attitudes that are hard to measure via self-diagnostic techniques, either because the subjects themselves are not aware of their attitudes or because some prejudices, like racism or sexism, are socially frowned on. “We know that these biases kick in at a very young age, from about two years old. And they are also much more widespread in the population than the explicit biases that people admit to in self-report measures; saying, for instance, that men are not better at science than women.”
This method has served as a starting point for numerous applications in clinical psychology, education, marketing and diversity management, and has been used for data collection in over 2,000 papers. Greenwald himself is currently applying the science in a legal context. “After I retired from teaching, I started on a second career in the law courts, helping people who are suing on the basis of discrimination to win their cases using the concepts of implicit bias.”
The story of IAT and its uses was the subject of the book Blindspot: Hidden Biases of Good People, which Greenwald co-wrote with Mahzarin Banaji. It was subsequently removed from public libraries in a number of U.S. states after being popularized by Hillary Clinton in her 2016 presidential campaign.
From the amygdala to the algorithm: the ubiquity of implicit biases
Mahzarin Banaji is Richard Clarke Cabot Professor of Social Ethics in the Psychology Department at Harvard University. Her research focuses on the disparities between people’s conscious expressions of their values, attitudes and beliefs and the less conscious representation of their mind’s content. As she puts it, “I would say that all the work that I have done in one way or another is to try to get at that invisible but very much present thumbprint of the culture on our brain.” When she convinced Yale University to put the IAT online (in 1998, when the technology was not that widespread), they got 40,000 responses in one month. This tumultuous reception would revolutionize the study of implicit bias. “Every day, I get an email alert about 15 news items that mention the term implicit bias. I would say there is no place I have not seen it used: from the U.S. military to the Museum of Modern Art by way of a garbage collection company. I think that, without knowing, we’ve tapped into something that is fundamental to human nature.”
Besides obtaining an unprecedented volume of data on the implicit biases that affect all corners of society, she has corroborated these results with neuroimaging techniques, observing that the amygdala – the part of the brain that responds to the new or strange – reacts more strongly to black versus white faces the greater the racial bias revealed by the IAT. And she has also been able to show that such biases may not be innate but are nonetheless acquired at a very young age: “Children of six have the same levels of implicit bias as adults.”
Banaji’s work points up a lack of consistency between the values proclaimed at national level and the actual policies that get implemented, as well as between personal values and individual behavior. This has led her to explore the implications of her work for questions of individual responsibility and social justice in democratic societies.
She recently turned her attention to analyzing the presence of these biases in online texts, with worrying results. Using a database of 840,000 words collected in 2014 and 2017, she found that the most frequent associations for “man” or “male” had to do with war and sports, while the words “woman” and “female” were predominantly associated with abuse and pornography, as well as cooking and motherhood. Motivated by these data, she has now focused on analyzing bias in language-based generative artificial intelligence models such as Chat-GPT.
Professor Banaji is now applying the science of social cognition to improve individual decisions and organizational policies. As part of this effort, she has launched a course at Harvard on Outsmarting Implicit Bias, which puts forward strategies to mitigate the effects of implicit bias on individuals and workplace teams. “Becoming educated about bias is imperative and just plain smart. However, I would say diversity training went wrong in a few different ways. It was often preachy. It was rarely evidence based. But when you come to it through the science, humbly, and without judgment, you stand a chance that people with common sense will understand its value.”
