Speaking of Psychology - Can we unlearn implicit biases? With Mahzarin Banaji, PhD

Episode Date: July 13, 2022

The idea that people have biases that operate below the level of conscious thought is uncomfortable. But decades of research have found that many people who would never consciously agree with prejudiced statements against Black people, LGBTQ people or women can nonetheless harbor implicit biases toward these groups and others. Mahzarin Banaji, PhD, one of the pioneers of implicit bias research, talks about where implicit biases come from, the difference between implicit bias and prejudice, and which biases have lessened – and which have not – in recent years.     Links     Mahzarin Banaji, PhD     Speaking of Psychology Home Page   Learn more about your ad choices. Visit megaphone.fm/adchoices

Over the past several decades, the term implicit bias has moved from the relative obscurity of academia into everyday usage. The concept is also at the center of recent debates over how society should think about and teach about racism, prejudice, and discrimination. The idea that we all have biases that operate below the level of our conscious thought may be uncomfortable. But decades of research have found that many people who would never consciously agree, with prejudice statements about black people, LGBTQ people, or women, for example, can nonetheless harbor implicit biases toward these groups and others.

So what is implicit bias and where does it come from? What's the connection between implicit bias and prejudice? How are they different and how are they connected? How do implicit biases influence our behavior? Is it possible to unlearn or overcome our biases? Does educating people about implicit bias help to reduce it, and as a society, are we shedding any of our biases? Do they remain as strong as ever? Welcome to Speaking of Psychology, the flagship podcast of the American Psychological Association

that examines the connections between psychological science and everyday life. I'm Kim Mills. Our guest today is Dr. Mazarin Banaji. Dr. Benaji is a professor of psychology at Harvard University and a pioneer in researching implicit bias. Together with her colleague Anthony Greenwald, she coined the term and ran the first studies on it in the 1990s. Today, Dr. Banaji continues to study the disparity between what people think they believe and their thoughts and feelings that operate on an unconscious level. She is co-author of the 2013 book, Blind Spot, Hidden Biases of Good People, and has won many awards and honors for her work, including most recently the 2022 Atkinson Prize in Psychological and Cognitive Sciences from the National Academy of Sciences.

She also won APA's 2017 Award for Distinguished Scientific Contributions. Thank you for joining me today, Dr. Benaji. Thank you for having me. In the public discussion over implicit bias, the term often gets conflated with prejudice or racism. Can you start our discussion by explaining what implicit bias is and how it's different from bias and prejudice? Implicit bias is definitely not racism. It gets used that way, but that's incorrect. For much of the 20th century, social psychologists who were interested in the topic of prejudice

and how could they not be, the field almost got its start during the war years, during World War II, when larger numbers of scholars and scientists from Germany were coming to the United States. for some reason on the top of their mind was the topic of prejudice. So they studied it in the way in which they could. They asked people questions about who they loved and hated. They asked people more subtle questions about the kinds of policies they might support to try to get at this concept called prejudice. Prejudice to psychologists is an explicit, conscious, aversion or dislike that one person may have

towards groups of people. If I were to say to you, I do not think that women should be teaching at Harvard. That would be an example of something we would call a conscious prejudice. What we've studied and what the modern, the invention of computers that allowed us to do experiments that would get at other levels of our minds, what that work revealed was that a large number of people in our society have genuinely, honestly, come to a point where they believe that they are not biased. And that, I would argue, is true. This is not something that they're making up. This is not something that they're lying about.

They genuinely believe in the ideals of egalitarianism, equal treatment, fair treatment, and they disavow prejudice. They teach their children how to not be prejudiced and so on. And yet, what our evidence showed is that those very same people, I would say people like your listeners, people like myself, we found that we do have what we call implicit levels of bias. It is not at all the same thing as racism or sexism, but it indeed is evidence of an association in our heads that I would call the roots of prejudice. Where do these implicit biases come from? How do we learn them since we're not really aware of

them. You're right to use the word learn because we're not born knowing that African Americans are bad and white people or European Americans are good. We do not, we're not born into the world believing that men should be going out to work and women should be at home. So learning is what happens. But it would be remiss, I think, not to speak a little bit about what we mean here by learning. What we mean here by learning is something that happens extremely fast, quite early in life and then progresses. And that comes not from just the social environment, but by the structure and the architecture of our minds, the way our brain got built over a period of evolutionary history. So we have a certain kind of a brain as humans. And we tend to focus on the conscious aspects of what

our minds allow us to do. I consciously logged in today, you are consciously speaking to me today about the topic of implicit bias. This is what makes humans so unique and different than any other species. But there are many other parts of our brains where work is happening quite outside of conscious awareness. And that's a very smart way for a system to be. You know, a simple computer, not anything like as complex or as magnificent as our brains is doing a lot of work. that's not visible to us on our monitors. It's churning out data. It's putting things together.

It's updating stuff. It's doing all of that work quite outside of our ability to visualize it. Why? That's what makes that computer smart. It's doing a whole bunch more stuff. And it doesn't need to start everything up every single time. These are routines. It has learned we call them compiled programs that are set to deal with certain things that

happen all the time. We have those two. And some of those compiled programs, that we have and that we carry in our minds have the vestiges of the past. They are left inside of our heads and in the ways in which our brain works that come from a time and a place in our history where we lived with people who were just like us. They were genetically identical to us in our own little tribe of 30 people, 150 people, or whatever.

On the other side of the mountain from which our ancestors lived was another group of people who maybe were different from us because they put a different mark on their face or wore a different kind of clothing, but they were genetically also jigs the same as us. And yet when the two met, especially at some resource like food or water, and when it looked like only one of us could have that resource, we learned that we had to do certain things, not very nice things to other people in order to survive. Some of that has been left in our brains. even though our world has changed dramatically. So I always argue that, that, you know, if we just go back to something that's attributed

to Darwin, that it is not the strongest who will survive, nor the most intelligent, but the ones who are most adaptive to change. If we believe that, then what we're speaking about today is that kind of adaptation, because in the old world, perhaps it was reasonable, rational, to be suspicious of people who looked very different from us. But today, the way our world is set up, it's the exact opposite that is being called for if we want to succeed, if we want to adapt.

And that is, if a business person looks at somebody very different from them, I think it would be worthwhile for that business person to say, how can I outsource to that person and make a lot of money for my business? So you see how our moment is calling for a very different kind of approach and a different kind of behavior

than the one that paid off in the long past. And that is part of the reason why we fail, because not only is the stuff invisible, it also, the natural orientation would be towards similarity or familiarity and so on, which today is not necessarily going to be suited. I have to go find people who know things I simply don't know, which means people from different fields,

people from different cultures. So I have to do something that almost goes against my nature. Let me ask you how you got interested in this topic. I understand that you began your career studying memory. So how did you move from memory to implicit attitudes? And where's the connection? So I'm one of those very fortunate people who came to be in a field at a time when clearly something was happening.

Other people had cleared the brush. And I happened to stand on the shoulders of giants. Yes, I was interested in human memory. I had zero interest in social groups or bias. I wasn't looking for research to do on that topic. But in the process of doing research on memory, and this is the fortunate part, a little revolution was happening in the field of human memory. People who had been studying memory for over 100 years since Ebbinghaus in the late 1800s

were studying memory using one kind of measure. we would ask people, what did you eat for breakfast today? Or what happened to you when you were five years old? Or you just learned a list of words in the lab. Tell me what those words are. However different the content may be, what was similar to all these methods, all these questions, was a single method. And that is that the method required us, the human whose memory was being tested, to go into her mind and pull out some of the method. something consciously, because you'd have to do that to answer those questions. Today, we call that form of memory explicit memory, because it requires an explicit action that requires conscious awareness to go into our minds, to pull out something from the past. What was happening at the time I came to graduate school in the 1980s is a recognition that there might be another form of memory that we had. That simply did not respond to questions like that.

These data came from amnesic patients, these data came even from ordinary people who were not amnesics, showing that there is stuff that gets laid down in our brain through our experiences that we may have no recollection of. And yet, if you use the right test, the right probe, you might find evidence for it. And so one such study that I remember picking not quite at random, but almost, to present in a brown, bag on memory that we used to run. My colleague Rob Crowder, who studied auditory memory, and I at Yale University, where my first job was, we ran a little group called the Memory Lunch Group, where we would talk about studies on memory. And I decided to present a study done by a genius cognitive psychologist by the name of Larry Jacoby and his colleagues. And what Larry had done is shown people

names that he had pulled out of a phone book, say a name like Sebastian Weisdorf. And then people who read those names from the phone book went away, and then they came back a day or two later. And now he gave them a new list of names. That list consisted of the names they had seen before, Sebastian Weistorf, but also new names like Timothy Watson that were not famous names, but from the same phone book. And then he added into that list names of some famous people. Wayne Gretzky was such a name at the time he was known, a famous Canadian hockey player, names of athletes, names of film stars, rock musicians, whatever it was. Now imagine you're the subject in this study.

You came on day one and you saw some names and you went away and now you've come back. And before you is a list of names. These studies were done with paper and pencil. And this is a long list of names. It has some old names, some new names. some new names, all not famous, and then mixed in was some famous names. And the task was very simple. This is what I loved about the experiment. Larry said to the subjects and his, you know, just look at each name and just say yes or no, is this the name of a famous person or not. So if people were to

respond accurately, they should say Wayne Gretzky, yes, famous, but Timothy Watson and Sebastian Weisdorf, not famous. But that didn't happen. What he showed, is that indeed Timothy Watson was identified as not famous correctly. But now think about Sebastian Weisdorf. A name you saw the day before appears again, and it has a quality that he called perceptual fluency. When you read the name Sebastian Weisdorf, it looks familiar to you. But your brain doesn't remember whether Sebastian Weisdorf is now a football player

or a rock star or a name you saw the day before. This becomes confusing, and in that moment, implicit memory kicks in and people wrongly identify Sebastian Weisdorf as famous. He called this result becoming famous overnight by just showing that, you know, when things are not consciously, if you were consciously able to recognize

that Sebastian Weisdorf was a name you saw two days ago, I think you would legitimately reject it as a famous name. But because we've forgotten and especially forgotten time information, place information, where did this happen, when did this happen, then the rest of the stuff starts to look familiar and you use that as the basis of this incorrect judgment. I was, I just thought this was a very cool study and I decided to write to Jacoby and get the materials for the study to replicate them. And this is in 1989. because his study appeared in that basic. And I asked him for the material, as we did in those old days,

we used to actually have little postcards that we had printed up and that we would fill in the name of the article and write the address of the person and put it in U.S. mail. And then you would wait. You would check your physical mailbox every day because you are expecting a big fat Manila envelope. And one arrived with all of the materials that Professor Jacoby generously sent to me. As I looked at the materials, I noticed that all the names, you know, famous and not famous were male men's names.

And as I tell the story, you know, I wrote him another postcard. Like, Dr. Jacoby, you only sent me half your stimuli aware of the other half. I didn't actually do that. I did talk to him about it, though. And he said, but women are not famous. And if we, you know, it just, and you know, when he said that to me, it made complete sense. Sure. Like if you want to do a psychological, you know, study, you would have to, on memory, which is not on gender, the question was, why would you bother to get equal, because now you'd have to get equal numbers of famous males and females? It's just not relevant to the study. So often in research, we take that path of finding the most, the best stimuli. And the best stimuli in this case would have been to keep everything constant to use just men's names. Now, I understood that. But this is what I'd love your listeners to think about. Something made me spend an entire summer

coming up with equally famous, I tried, equally famous men and women's names. I was not successful in coming up with equally famous men's and women's names, but I did that. You know, I went to books and I looked at, you know, nothing on the internet, of course, so it was hard to do, but I would test small groups of people and try to come up with names, you know, like Jane Austin or whatever, so that many people would recognize. And so I just did what I thought was a replication of the Jacobi study, but just with both kinds of names in it, expecting, of course, full well that the same result should happen for male names and female names. Why should that be different? Once you had become familiar with a name like Sally Weisdorf, you should now be more

likely to say Sally Weisdorf was famous, just like you did for Sebastian Weisdor. But in six experiments, what I found. I was surprised. Initially I just thought it was a throwaway result and I would just let it go and say you know the women's names did not become famous overnight the way the men's names did. But as I thought about it I began to quiz my subjects. I used to run my own experiments in those days I ran every single subject and I would say to them at the end of the study. So what do you think you used as the rule to identify fame? Not one of them said the gender of the name. So then I started to say to them, do you think you might have

used the gender of the name in making your decision? And they would look surprised and sometimes even offended. What do you think I am a sexist? Of course, I wasn't using the gender of the name. And yet, that was so clearly in the data. I did a presentation at a conference in which I think I took the bold step of saying, I've quizzed, you know, 400 subjects, a book. about this. Not one of them was aware that he or she was using the gender of the name to decide on the person's likelihood of fame. And yet, in their data, we saw this. If this is true, then harm is happening without the person doing the harm being aware that they're doing it. The person being harmed is not going to be aware that she was harmed. What does this mean

for the big question of how we're going to treat each other? in a democratic society where we believe that fairness and egalitarianism and so on are the virtues we care about. And that's where it started. That's pretty amazing. Well, one of the best known ways today to test implicit bias is through something you were involved in creating the implicit association test. How did that come about? How does that test work? So the credit here goes to my colleague Tony Greenwald, who, again, I think even he didn't come to it with an interest in creating something that became the IAT. He was really dissatisfied with the methods that we were using at the time to get at implicit bias that weren't really that powerful. We would get differences

and you can't do, you know, one fame study after another is the way to study implicit bias. Something more general was needed. We did have a method. in psychology. It's called semantic priming. If I flash a word rapidly on a computer screen and the word is something like soldier, let's say. If the word that follows it is John, you'll be much more quickly able to identify John than if the word is Jane. Because Soldier and John go together in our memory and in our minds and our understanding rather than Soldier and Jane. And likewise, you know, if I said something like Nurse and John, you'd be slower to. respond to John after nurse than you would be to Jane. So that was a method. I was using that method.

It was producing good data, but it was inefficient. You know, you would need to give people 800 trials. Imagine a poor subject sitting at a computer, you know, 800 times having to answer these questions and for a whole hour. And then the effect size was small. It was always there. So for the scientists doing research on understanding the human mind, it was perfectly okay. But you, the subject, if you were taking that test, you wouldn't necessarily have the feeling or the experience that you were showing a bias. And what Tony Greenwald did is tried the slightly different way of trying to get at the strength of association between A and D in our minds. The way the IAT or the implicit association test works is actually quite simple. It makes the assumption

that if two things have co-occurred in our experience over and over again, that we'll put those two things together very quickly. Salt and pepper is just a simple example, okay? Bacon and eggs, if you want to stay with food-related things. But mother and father or father and he, mother and she, these are things that have co-occurred in our experience, and therefore they have a strong connection. Neuroscientists have a little phrase that they use to teach people about learning of that kind,

and their phrase is a very useful one. When they speak about neurons, they say, what fires together, wires together. And that's what we are trying to get at. We know that some things have fired together. Can we build a method that will give us an indication of how strongly they're wired together? And so the IAT works in a very simple way. it requires you, and I would say to your listeners, there is no way in which you will appreciate the IAT by listening to what I say as much as you will by taking a test.

Right, which anyone can do, and we'll include a link in the show notes, yeah. So what happens in an IAT, and let's just take something like gender and work, in our experience, even though women have entered the workforce in very large numbers, in our minds, male and career, to the extent that they're more closely associated, should get picked up on a test like the following. You see names of men and women, a name like Peter may show up, a name like Susan may show up. You're told, use two keys on your keyboard. If you see a male name, press the left key. If you see a female name, press the right key. So easy to do. Now we say, okay, we're going to now give you words that mean home and career words. If you see a work like child or marriage or garden or

kitchen, you will press the right key. If you see a career word, a word like office, manager, job, briefcase, desk, things like that, you will press the left key. And now what you're doing the IAT, and this is the critical moment, you combine them. So you say, all right, you're going to see all four kinds of things, male, female, home, career. For now, if it's a male name like Peter or a office-related word like boardroom or a briefcase, press the left key. So you're putting those two together. If it's a female name like Susan or a word like kitchen from home, you press the right key. It's very easy for us to do that because that set of trials is picking up what we think of

as the established stereotypical association. So if you're taking a test when you have to do that, you're whizzing through it. You're making very few mistakes and your speed to response, which is our critical measure, time to respond, call a response latency in milliseconds. But what happens, unbeknownst to you, you're now told, okay, you did very well. Now let's get you to do something a little different. Now put Peter and Kitchen together and put Susan and boardroom together using these keys on the keyboard. Very simple.

And everybody thinks, oh, I should be able to do that. Nobody thinks that they can't do that. I should be able to imagine Peter in the kitchen, Susan in a boardroom. No problem. Until you start to do it. And so people like me who have always had a career, I was born and raised in South India. My mother, in a society where women didn't work outside the home, did, and was a teacher. In spite of all of these direct personal experiences, I can't.

I can't do this. I can't do it as fast as I could do the previous version. And when, you know, so when after those Jacoby experiments that I had done earlier, you know, we had to come up with a name for this thing. And we were struggling with the name. And because I'd been studying memory and the phrase that they were using was implicit memory, to distinguish it from conscious or explicit memory, we called this implicit. bias because it was now showing an ability to do one thing faster than the other as a measure of what I call the thumbprint of the culture on our brain. That's not always consistent with your view of yourself. And that's the dissociation. So in psychology, the term dissociation is an important one when in this case, it's where your data on one measure, When I explicitly ask you, you'll say one thing.

But on then the implicit measure, something different is showing up. So I show no bias if you ask me a survey question about who belongs more at home and who belongs more in the workplace. I will tell you equally men and women. And then you give me this test and it's not anything close to equal. I'm just unable to do it. And we believe the test picks up the thumbprint of the culture on our brain. So is it possible to learn to overcome these implicit biases? So you train yourself out of responding in that way. And when you take the test, it doesn't show up that way anymore.

You're actually able to relate things that wouldn't normally be related because you've done something to rewire your brain. Is that possible? The best way to answer that question is to say, of course. Of course it can be changed because it was learned. So you could make yourself. unlearn it. But the mistake that people make is to think that they can unlearn it easily. Right? You think to yourself, okay, first, I don't want to believe that I have bias. That's the first sort of defense. So you argue and you argue and you argue until you have been hemmed into a box from which you can't escape that shows you that bias does exist in your head. So then you agree.

And then you say, oh, but I'll just tell myself not to be that way. And that's the stake. So we should be aware. So there are many experiments that have been done in my lab recently, but in many other people's labs, where we try to look at what is it that will change an existing implicit bias. The work of Calvin Lai at Washington University is very important here because he set these experiments up as a competition. You, if you were a scientist interested in this question, you could submit what you think is your best intervention to change data on the IAT. I won't go into the details. It's a long, laborious way to take all these people's ideas and turn it into a proper

experiment where every intervention gets a fair shot at changing people's minds. He did it with race, so we can begin with the race IAT, where 70% of white people and even Asians associate white with good and black with bad. you can begin with something like that. And then he asked people, come up with a certain minutes. I forget how long, but your intervention can't be longer than a few minutes. What do you think would work? And it's a lovely way to do science because you're not prejudging what will work and what will not work.

You get to, if you put the work into building that kind of experiment, he'll load it up on the website and get data on it and tell all of us which ones worked and which ones didn't. and a paper was published with many authors to show which ones worked and which ones didn't. So interventions that he tried came in three different categories. Intervention number one, you tell the subject, try to be egalitarian. Do not be biased. Try to treat these two groups equally. The second is you don't say anything to them, but you just give them new associations. So you pair black with good things, white with not so good things, to see

if that will just shift the IAT bias away from the default in our society and towards whatever you've just taught people. If that works, then that tells us that even implicit biases are open to intervention and to change. So that's number two. I call that sort of the associative measure method or kind of like a brute force method. We're just going to give you new associations to see if they will replace the old ones. And the third intervention is something that I think only humans can do.

You know, we can take any other animal other than us, take non-human animals and give them new associations. And they'll learn those, as we would. But there is something unique to humans, and that's language and the ability to be able to soar in our minds and think entirely new thoughts. And so the third kind of intervention is one where you are asked to imagine, imagine. Imagine a world in which you're suffering in some way, and a person from group A, black, helps you, while a person from group B, white, does not help you.

So you imagine that. Does that matter? And the data show that what works is exactly the opposite order of the way I gave you the three categories. If you use that kind of linguistically based thinking, imagining, somebody helps you that produces the biggest change of the IAT. You actually begin to become more pro-black than you were before, or less empty black than you were before.

The second method, the associations, they work, but not as powerfully. And this was a surprise to us, because I had been thinking that that one would work the best. The IAT's measuring associations were messing directly with the association, so that should produce the biggest effect. It does not. This is why I love science, you know, that every day I can make a hypothesis and what are my students will show me, I'm wrong. And now we know something. We know something new and better.

And the worst method, the one that actually has no effect at all, is to tell people to be good people, to be egalitarian and so on. So let me say that this may make it sound like, oh, so it's easy to change. It is easy in the sense that it will last for a short period of time, but it won't last very long. You step out of that experiment and you enter the real world, and the real world is again giving you data that make the same old, same old association between black and bad, white and good, and so your brain snaps back. Now, when young people encounter this result, when they see that, yes, they were able to make change, but the change doesn't last, they get very sad because they want a better world. And I'm not at all sad about that. I think it's telling us the exact truth about how the human mind. mind works. Can you imagine this? We are giving people a tiny little intervention, and it is producing a shift, but the shift does not last. Isn't that a wonderful thing? Because think about all the

apologize for the word, the crap that we hear all the time, you know, imagine if that were to stick forever, you wouldn't want that kind of a brain. So our brains change, our minds change, associations move around, but they always will gravitate to whatever is your cultural default. And so to bring about actual change, society around us has to change. And then we will move, and then the default will be a new default. So I'm not bothered by this result. I would never predict that the effect should last very long because it would be a dangerous thing if the effects lasted long, because we don't get to choose everything we see in here. So it's a good that we get to snap back to something.

So you've been studying implicit bias for decades now, and you have found some evidence that in recent years, some of our biases have decreased. What is happening? How has that happened? How are we measuring it? And what are the biases that are different than they might have been 10 or 20 years ago? It is among the most exciting work that's been happening in my lab. And the person to credit here is a brilliant postdoc, was a graduate student.

Tessa Charlesworth. As you know, we opened the website where implicit bias tests were made available in 1998. And here's the story about the beginning of it. There was very little on the internet. There was no psychology experiment happening on the internet. The internet was not very well developed. People then, even large corporations would have kind of one picture with their name and some URL and some phone numbers. That was what was on the internet. And, you know, here we are, we say, let's put these tests on the internet after four, four years of testing it internally in the lab. Once we felt we had a good handle on what it was doing and all the variations that we needed to do to establish basic issues, we put these tests on a

website and we thought, and I remember very explicitly saying to my two colleagues, Tony Greenwald, and Brian Nosek, who actually built the website, I said to them, wouldn't it be amazing? like we don't have to bring one subject in at a time. And maybe in a year we'll have 500 people who will have taken this test and we'll have so much data and they won't even be, you know, Yale and University of Washington students. So people will actually have to take these data seriously because they will come from real people, not from college students. I mean, we were so naive because in the very first month after the site opened, we had 45,000 completed tests. And we were not only taken aback, we had to figure out a coping strategy because people were now writing to us. Dozens of messages in a given day.

Your test is stupid. You know, I am not, I do not have a race bias. I am a rap musician. How dare you tell me I'm. So this is what we had to deal with. Anyway, to make a long story short, you know, we think something like maybe we don't count anymore, but there at least 40 million tests have been now taking. and we have an awful lot of data. From about 2007, because we were fiddling with things and we were just getting used to how to build it,

we don't believe that the data we got before 2007 were as stable, as reliable, as good as the data from 2007 onwards. So from there onwards, there are lots of data, and the group has put these data on a public website. So anybody can go there. Anybody can take those data and analyze them. and in fact, many people have. So Tessa Charlesworth decided to look at what has happened over time. These are not your typical psychological analyses, right? We do do some longitudinal work.

Developmental psychologists do study what happens to a child as the child develops and so on. But this data set, I think, is so I generous. There is no such database where we've been. been collecting data continuously 24-7 since 2007 to today. So it's just unusual, you know, a gallop poll, however wonderful it might be, is going to go in and in a single moment get data from 2,500 people or whatever and then conclude that Americans do this or that. We're collecting data every second, every minute, you know, of the day and night because we also opened a bunch of international sites. And so truly data are being collected around.

around the clock. And we have, for many tests, take race, which is a very popular test, we now have data from dozens and dozens of people in every little location, every county of the United States. So what we can say is, okay, we've got that big, big fat pool of data. Can we look to see if anything has changed in 2007 to today? And again, I should tell you, I was hesitant about even taking this on because I was so persuaded that implicit bias cannot change in the long term that I thought, okay, do this as an exercise for your statistics class, but it's not really something that, you know, we'll be able to say anything about. We will be able to say, I think.

Nothing has changed. It's implicit bias in my mind, having looked at the data, having studied it for about 20 years by then, more than that, 30 years maybe. I was quite persuaded that we would see flat lines over time 2007 to when we first did the analysis 2016. No change. And maybe we should write about that. Maybe we should say, look, implicit bias is not changing, et cetera. So Tessa not only did the analysis, she did it not just for one or two groups, but for six of them.

She looked at anti-gay bias. She looked at race bias. She looked at skin tone bias, which is a little different than race. bias, of course, age bias, disability, and body weight bias. So there are six different kinds of categories. And what she discovered that surprised us is that on three of those tests, you do see substantial change. I'll begin with the one that showed such big change that we're still gobsmacked from having seen those data. And that's anti-gay bias. So anti-gay bias has dropped off by 64% between 2007 and 2020. How can this be? How can a bias that was so high in 2007,

it was one of the large biases, gay, bad, straight good. How come that bias has been just withering away at this speed? And remarkably, if you look at any demographic group in the country, you can look at elderly people, young people, you can look at people who live in the coasts in the middle of the country, the north, the south, the rich, the poor, the educated, the less educated. You can look at any of these. And we see change in every single group. So everybody's changing on this. Yes, there are two groups that are changing faster than everybody. Your listeners should try to predict what those groups might be. And as they do that, I'll give the answer. Younger people are changing faster.

and people who are self-professed liberals are also changing faster. In fact, those two groups have pulled away from their original bias in 2007 so fast that they are today at neutrality. Wow. For them, gay and straight are equally good and bad. The rest of us will catch up. So one of the nice things about this analysis, these are called time series analysis. And we use a particular model called Arema for those who are. who might want to go look up what it is.

There are lots of limitations to this method. We have to make very sure that we're not just showing this effect because all of a sudden, in recent years, there are many more young people taking the test than they were before. That would not be a correct way to interpret the data to say there is changed. It's not really changed. It's just that the sample has more young people today and didn't have as many young people. So you have to rule out lots and lots of things.

So if they're interested, they should go look at this paper. that was published in psychological science. And what they will see if you get into the data is that this change, we think, is real. This is a real shift in what people are thinking. And when I tell people that these days, they say, oh, yes, of course. I could have predicted that that would happen.

Well, you know, as we know, hindsight is 2020, and they can. But in retrospect, I too now see why this might have happened. So I would love for your listeners while we talk to be thinking about this. Why did sexuality bias change? Because no other bias has changed as fast. Race bias has changed, but by 25%. Okay. So good news is it has moved. But we can ask, why not at the level of 64%. We know that's possible. So why is race slow? Why is skin tone bias going down? Again, by 25%. Why is it not as fast as sexuality? We know that,

gender bias, home and career bias, has come down by about that much, but why so slow? The association that men are for STEM and science and math and women okay for arts and so on, that bias has, but not as much as the anti-gay bias has. There are three biases that have not shifted much at all, in fact. And that's anti-elderly bias, which continues on and is absolutely stable, a little bit lower, but not anything to write home about. Disability bias, just flat since 2007, no change. And bodyweight bias, a bias against people who are overweight, that remains flat, and that looks like the hardest one in some ways in the data that will change. we think these biases are going to be around.

If they move at this rate, we're not going to see change for 200 years. That's our prediction. So then what is it that special about anti-gay bias? And I'd like to put a few ideas out and then see if your listeners would agree. I think that sexuality bias changed because it became personal in some way. That is to say, parents and grandparents, friends, neighbors, cousins, we all discover. the people who were close to us, said to us, I'm gay. That became possible.

And as they began to come out, we would have confronted a state of dissonance, of cognitive dissonance, this old idea in psychology. I believe being gay is bad. Here's my child saying, she's gay. I have to do something in my head to resolve this conflict. And Leon Fester, the genius in the 1950s who wrote about this theory of cognitive dissidents said, we can't maintain these. We have to change one or the other.

Either you have to cut off relations with your child and say, I disown you. Or you can begin to say, I think being gay is not the horrible thing that the Bible told me or whatever, however you came to those beliefs. And what we have found is that maybe it would have gone the other way. I had this happened two centuries ago, but given the time in which we live and the meaning of what religion is to many of us, where we believe that it might be a guide to certain ways of living a good life, but that we may not want to take the Bible. So literally or whatever. It just so happened that people went with love. The love of the child, the love of your neighbor or your friend trumped this other belief.

And so we've seen change. This does not happen for anti-elderly bias, for race. where we are so segregated. And think about race. You know, the two big domains in which we live and do our stuff are home, the neighborhood, the family, and work. And in both those domains, we're deeply segregated by race. We just don't see people across race lines because it gets intersected with socioeconomic

class and so on. So the chance that you would meet that person and have the same kind of. of experience that, you know, straight parents and straight grandparents had to deal with when the time came, that just is not an opportunity that race offers us. So as a society, we have to ask ourselves, what will we do if we care about race bias? The second thing that happened for sexuality is that Hollywood got involved. Institutions are deeply important. To Hollywood, this particular dimension matters because there is a larger number of gay people in the entertainment industry than many other industries. I think they cared about it. So they did something remarkable. They made movies

and they made TV shows, not just ordinary ones, but ones in which the gay character would be cooler and smarter and nicer than the straight people on the show. And I don't know, people watched those and came to see that this was being normalized. And we're very sensitive to that. What do other people in our culture think? If I hear that, you know, 90% of my neighbors are saving energy by doing certain things, I become very competitive and I want to be better than them and I will do it. Much more so than if you ask me to be a good person and care about the environment, that doesn't do very much.

But hearing that other people do. We're very competitive, yes. They're very competitive, especially socially, of course. So I think that that helped a great deal. And then, of course, the hand of government came in. The Massachusetts Supreme Court first said gay marriages legalized. I remember I had moved to Cambridge only a couple years before this happened. And Boston announced that it would open its courtrooms on whatever it was Monday morning at 9 a.m.

and performed the first gay marriages. And Cambridge, the suburb, not wanting to be second, decided that it would always open its courtrooms at one minute past midnight to be the first city to do it. So having lived in a town that was the first and having stood there watching the first gay, married couples walking out, it was clear that this was a moment, something historic had just happened. So now think about this. How many times in our history has change come about at three days.

different levels, the individual, the organizational or institutional, and at the level of governmental policies and the law within a short period of time. Yes, the civil rights movement in the 60s did something at that third level, but I don't think that we changed much at the institutional or individual levels. So it could not have the same impact as I think this did. So this is a very important lesson for us, that if we want to make change, maybe one of the messages from our researchers that try and do it at all three levels. And for that reason, I think we should be pleased that people in our society today use the word systemic so much more than we have. Systemic does not mean systematic. Systemic means that the system needs to get engaged. And I would like to argue

that the system consists of individual people, the system consists of institutions, the system consists of large governmental organizations that can move certain kinds of levers. I cannot say that one is more important than the other, but I would like to argue that if we go for system change, we will see change a lot faster. Well, Dr. Banaji, this has been amazingly interesting.

I really appreciate you're taking the time and explaining all of this to our listeners. I'm sure that they're going to enjoy hearing all of this and and taking the IAT. We'll get you some more data, I promise. Yes, and I'll get lots of little nasty notes at the end of that. Oh, I hope not. No, actually, I should, I should mention that the site is incredibly successful. I mean, the very moving letters that we have received, even from people who initially did not approve of the result that they got. And that is what gives me such faith in human beings, that the same people who wrote to us to say pretty nasty things will write back

later and say, I just had this experience and I wanted you to know that I understood what your stupid test was trying to tell me. Success. Thanks very much for having. It's a pleasure. You can find previous episodes of Speaking of Psychology on our website at www.combeckycology.org or on Apple Stitcher or wherever you get your podcasts. If you have comments or ideas for future podcasts, you can email us at speaking of psychology at APA.org. Speaking of Psychology is produced by Lee Winerman. Our sound editor is Chris Kondyin.

Thank you for listening. For the American Psychological Association, I'm Kim Mills.