Speaking of Psychology - Bonus Episode: How Virtual Reality Can Help Detect Racial Bias in Police Shootings with John Tawa, PhD

Episode Date: November 4, 2019

There's been a great deal of media attention focused on shootings in which a white police officer fired on a black or non-white suspect. Psychology has for years performed research to determine whether racial hostility plays a role in such shootings. These studies have usually entailed having participants sit in front of a computer screen and respond to images of suspects who pop up holding a gun or a benign object such as a wallet or a can of soda. These experiments are helpful. But is there a better way to study this phenomenon so we can curb these types of shootings? Our guest is John Tawa, PhD, of Mount Holyoke College who has developed a new and perhaps more realistic method for testing these responses. Join us online August 6-8 for APA 2020 Virtual.   Learn more about your ad choices. Visit megaphone.fm/adchoices

Hey everyone, it's Caitlin Luna, host of Speaking of Psychology. This episode was recorded during APA's Technology, Mind, and Society Conference, held in October 2019 in Washington, D.C. I was away on maternity leave during that time, so my colleague Kim Mills was a guest host. We hope you like this episode. Hello, and welcome to Speaking of Psychology, a bi-weekly podcast from the American Psychological Association that helps explain how the science of psychology is connected to everyday life. I'm your host Kim Mills. There's been a great deal of media attention focused on shootings in which a white police officer fired on a black or non-white suspect. Psychology has for years performed research to determine whether racial animus plays a role in such shootings.

These studies have usually entailed having participants sit in front of a computer screen and respond to images of suspects who pop up holding a gun or a benign object such as a wallet or a can of soda. These experiments are helpful, but is there a better way to study this phenomenon so we can curb these types of shootings? I'm here today with Dr. John Tawa of Mount Holyoke College, who has developed a new and perhaps more realistic method for testing these responses. Dr. Tawa is an assistant professor of psychology at Mount Holyoke, where he studies the processes that both hinder and help race relations. He is particularly interested in relations between minority groups, such as between blacks and Asians. Dr. Tawa has found that virtual technology is a particularly powerful medium for assessing intergroup behavior. For example, in one of his studies, Dr. Tawa had participants create avatars that resembled themselves and interacted in the virtual world's second life. We'll talk in a few moments about what happened between these groups.

So, Dr. Tawa, let's start by talking about your research into police shootings. Could you describe the design of your study and how exactly it differs from other experiments that don't use virtual reality? Sure. Yeah, I want to just say one thing that I don't intend it to be better than, like a better method than. I think that the research that you described where people sit in front of a computer has been really impactful. I mean, that's sort of, I kind of want to say that's established that that is the case that people are faster to shoot on black suspects. They have greater difficulty differentiating weapons and benign objects in the hands of black suspects compared to white suspects. And archival data, too. I mean, starting in 2015, the Washington Post started just making freely available online every record of any individual killed by the police. And that data is, the raw data is available online. And a quick analysis will show you that black and African Americans are way overrepresented in that database relative to their population.

They're more likely than white suspects to be running or fleeing the scene when they're shot. In fact, about twice as much, twice as likely, and about twice as likely to be unarmed when they're killed. So my point here is that in some ways, what we're trying to do with VR is not necessarily approach that idea. We're sort of taking that as a basis. There's been quite a bit of research that's established that. But what VR does that's really exciting and what it has the potential to do is really immerse people in scenarios where they may experience things like, fear, physiological arousal. They may, for example, have not know necessarily who the suspect is. The suspect might come from behind them or from the side. So there's a sort of unexpected element

that you can't quite get from sitting in front of a computer. And then what are real interest is then within that context be able to look at what kinds of cognitive processes and what kinds of emotional processes are interacting to create bias. And so what do you find? And what is different about the brains of people when they're in these situations? Well, I do want to say this, that what we have now is some preliminary data from a sample about 45 participants. None are police officers, about half are criminal justice majors.

So it's just a preliminary sample. And I just, you know, I do want to say that. But one of the first things, we actually just completed our first analysis of that data, and one of the first things that I was interested in looking at was about how people's beliefs about the concept of race affect, how they interact with suspects in the VR. This has just been an interest of mind, what we call racial essentialism, the extent to which people think of races as naturally distinct from one another and as biologically or genetically different from one another what we're seeing in other research is that that belief set is really

related to a host of really bad intergroup outcomes people are less comfortable interacting with people across different races they're less likely to make friends across races they have more difficulty recognizing other people's faces I mean just a whole whole whole of bad consequences. So yeah, that's one of the first things that I looked at was how people's beliefs about race impacted their shooting decisions. And here's one of the interesting things about behavioral research is that you have so many ways to operationalize outcomes. So this one in particular is something that's been done in those previous studies, which is just to look at reaction time. How fast are people, when a weapon or a benign object is drawn, what's the amount of

time it takes them to decide to shoot? So what we're finding is that when you look at how people interact with the black suspects, that there is a significant correlation between beliefs about race as distinct and quicker reaction times. So the more people think of other races as distinct, that leads to quicker reaction times, but only towards black suspects. When we do the same thing with white suspects, at the moment, we don't have any correlation. That's kind of amazing. But what you're saying then is that you're really not testing with police officers at this point. It's just an average citizen, in a sense, and finding similar reactions.

So is this something that maybe could help an average citizen understand what's happening in these police shootings? Is it used, is the information useful in that respect? That's right. I mean, it is my ultimate goal is to be able to do this with police officers, and that's something that's in the works now. And yeah, that's the ultimate goal. But is this useful for civilian or non-police officers, I think? Absolutely.

one of the reasons that I've in the past been interested in racial essentialism is because I'm an educator and to me that's something that you can just teach. You can just teach people that this is not what race is. And presumably that should have all types of positive effects. Now in the past I've studied things like interactions with diverse others. This is the first time I've really gone into something as, something, you know, specifically related to police violence. But yeah, absolutely.

So how prevalent is this notion of racial essentialism, the idea that race is something biological or genetic? That's a great question. And I can just tell you what I know from a study that I did using, I'm going to call it a national sample, that might be a little generous. It was a sample I collected on Craigslist, basically, targeting major cities in the U.S. The purpose was to get outside of a college student sample.

And to try and get a racially representative sample with age diversity, education diversity. So I ended up with a sample of close to 600 people and asked them a series of questions about, their beliefs about race and not just the question do you think race is biological or not it's it's sort of a call it like a depth of essentialist model so that the the sort of deepest level of essentialism is what we call speciation or the idea of people thinking that races are actually different species or that they different races have their own atom and the for example, that was one item on this test, or the idea that races represent different departure points on the evolutionary scale. So deep level of essentialism. That item in our quote unquote national

sample, that item was endorsed by about 25% of the population. That's amazing. Where would people get that idea? Don't they understand what species are? I mean, I can answer that myself. I guess we don't teach science very well. Sure. I have a thought about that, but let me just kind of scale it down a little bit because the sort of less depth, how you say it, the sort of less deep beliefs were probably endorsed more. So I want to make sure that's clear. Just basic beliefs about what we call genotypic essentialism, the belief that you could, for example, determine somebody's race just looking at their bone structure or through an autopsy, or just by looking at genes or DNA, that was endorsed by about 50%.

Wow. Yeah. Yeah, I mean, where do these beliefs come from? You know, I mean, this is something that I've thought a little bit about when you take surveys, is it a belief or is it, is it like an empty set and then somebody gives you a statement and you hadn't thought too much through it? but it sounds right. I mean, if that's true, I think that's still impactful that somebody could look at a statement

like different races originated at different parts of the earth. And you hadn't thought it through ever in your life. And then you say, yeah, I think that's true. I mean, that means something too. But that's a little bit different than kind of actively promoting it. So that's one thing that's sort of unclear. But I'll tell you that, you know, I'm 42 years old. And the reason I'm telling you that is,

to say that when I was in eighth grade, I remember my science textbook telling us about the three races. Where did you grow up? In Western Massachusetts. Yeah. And in reading that there were Mongoloids, negroids, and kakazoids, and as somebody with Japanese heritage and white heritage trying to understand where I foot in. So I guess what I'm saying is, I mean, I don't think I'm that old for you, too.

But that's what I was taught explicitly. And yeah, that, that, that, so those beliefs were explicitly taught. And then, and then I think we kind of entered like a colorblind era where we just weren't supposed to talk about race. But, but, but we didn't undo those kind of deeply rooted beliefs that I think people held. So when you look at that, when you looked at that question about racial essentialism, demographically or in terms of race and ethnicity, who tended to believe more in the idea of essentialism? Was that white people believe that more?

People of color believed it more? How to split up? People of color in my samples so far, I've had stronger levels of racial essentialism, but it disappears when you control for immigration status. So what I think that might imply is that growing up in a more homogeneous environment, might actually promote those ideas.

And yeah, that's the most sense that we can make out of that. So can you talk a little bit about the research that you did in second life and how you pitted or had people create avatars and then pit against each other? Right. This is sort of my first, like, attempt to use technology to study, to say try to study behavior as it's happening rather than relying on self-report or you know survey and and so in for that study I had people create an avatar in the program second life which is a free more people are familiar with with the program Sims but it's

similar to that, except that all the users are online. And so you can create an avatar that looks relatively realistic, and then you can interact with one another in these virtual environments. And so one of my, I would say theoretical interest is in terms of how perceived competition for resources creates dissension, particularly between minority groups. It has a little bit of an asymmetric effect

in that it impacts my own. minority groups more so than each minority relationship to the dominant group. And so, um, so when people feel like they're competing for resources, they think of resources as zero sum. So more for me means less for you. More for you means less for me. So what are the, the resources in second life? Well, what we had to do is model it, right? So what I did was I had people create these avatars that look like themselves. They would interact in these groups of about 20 at a time. And then in some conditions we called like a utopian condition. Actually, they weren't competing for any resources.

They were just told that at the end of the study, somebody would be randomly picked for a $300 stipend. On the other hand, sometimes what we did was we incur, we said that, well, your job is to kind of assemble a team of people that you think are the most intelligent and hardest working. you're trying to form a business group to simulate this economy. And in the end, the team, the people that form the strongest groups like that have a better chance of winning a $300 gift card. So now suddenly you're interacting with people a little differently because you're evaluating people for how intelligent you think they are, how hardworking you think they are. And it dramatically changes how people interact with one another.

maybe relying on stereotypes, even in the virtual world. And so what we saw, one of the things that I liked about doing this in a virtual environment is that second life is everything's on a grid, like an X, Y coordinate grid. And so you could look at actually physical distances between people and how those distances changed over time. And what we saw was that when people were competing for resources, it was the black and the Asian participants that gradually over time moved away from each other. But when they weren't competing, actually the black and Asian participants moved closer to one

another. And that wasn't actually true about how they moved in relationship to the white participants. So how did they move in relation to the white participants then? There was, let's see, in the resource competition condition, they, let's see, everybody moved away from each other. It was not linear. Let me just say this. So basically what happened was they, in the resource competition condition, there was a,

there was like a coming together. People approached one another. Were they like forming groups in order to succeed? I think what happened was happening was they were like kind of scrutinizing one another. And then it had like this adverse effect where then they pulled apart. So, you know, it was a U shape. If you look at it over time, they would kind of come close together and then they would

pull apart. And so that U-shaped pattern was true for everybody, but it was just significantly stronger for black and Asian participants in relationship to one another. In the utopian condition, this is what we called it. Actually, there was still a gradual, it was more of a gradual movement away, but actually the black and Asian participants in particular had a gradual movement towards one another. So what does this tell us about?

real life. Yeah, that's a great question. You know, that maybe in those cases, there's actually a sense of affinity between minority groups that, you know, the way we modeled these scenarios, what we always had, we tried to model white numeric dominance by always having about twice as many white participants. You know, so you'd have a group of 20, you'd have 10 white avatars slash participants, five Asian, five black, about. Yeah.

And, you know, when there isn't competition, one possibility is that there's a sense of affinity that you are people of color together, your minority group members together. And so there was sort of a sense of camaraderie. As soon as there's competition, suddenly, you know, especially if you think about the way we worded the instructions, like you're trying to find people who are intelligent and hardworking. That was purposeful. That's sort of tapping into these what we call model minority stereotypes of Asian Americans. So suddenly people are, you know, it's really interesting to look at the chat logs because we've, and we've written about this, there's evidence that people are

kind of in a coded way trying to find out if people are Asian because you don't always know. I mean, it's an avatar, right? So, so, hey, who speaks foreign languages? Oh, yeah, what language do you speak? Really interesting. And then suddenly the ones that speak Chinese or Japanese are people are drawn to them. Right. So now in the resource competition, you have you have two groups who are stereotyped really oppositional where Asian Americans are stereotyped as being intelligent and hardworking and African Americans are stereotyped as being unintelligent and lazy. And so suddenly you have an idea set that is pulling two groups apart. And it's just interesting to see it in the in the behavior.

When you watch one of these social events, they were only 15 minutes, by the way, when you watch one, you know, because we did, we would watch them as the participants were interacting. You really can't see anything. And I remember when I was doing the study thinking, like, maybe kind of a bust.

But they were significant. When you, you know, when you actually look at the, you know, at it, at a really acute behavioral level, there is a gradual increase or decrease that are significantly different than the racialized patterns of movement, I should say. Yeah, yeah. So is there an application for this? Or, you know, how would you translate this for some other researcher to use or take it into everyday life? I mean, I think of a couple of directions that we've gone in terms of trying to make it more applicable beyond this particular study. We recently published in behavior research methods.

It's a secondary data analysis from that very data set. But what we were interested in doing was trying to figure out what kind of metrics of intergroup behavior you could derive from what we call spatio-temporal data. So you have data where people are interacting in a physical space over time. And in this case, I mean, it's like people are interacting within meters of one another over seconds because we're recording data at every second. So if you have that kind of data, what kinds of movement patterns can you determine from them? That's a little bit more nuanced than just say moving away or moving towards somebody. So what we actually did was we created overhead still videos of how the nodes moved in relationship to one another.

And from and did like a qualitative observational analysis. And from that we determined, for example, we had one movement pattern, which we called seeking behavior where people would kind of bounce around between small groups. Then the next step was to operationalize that with a formula. and and then see how that metric predicted things like intergroup anxiety, for example, like on a survey. And so that paper that we just put out has a series of four different types of movement. There was a leadership style of movement, a seeking style of movement, a we call it, like homophily or diversity style movement, you know, in terms of people interacting with people unlike them or like them.

And in a way to operation, like people who kind of avoided social interactions, right? And so anyway, the hope is that that's general enough that if anybody has data that is recording spatial locations between people across time, that these. would be you could use those formulas and measure those four things and see how they you know so so that's one one application the other is that I've been I've been doing this for a while is actually as a teacher I've been using second life in my psych of racism in cross-cultural psychology classes because it's an opportunity to like walk in someone else's shoes right right like so if you're white, you know, use this Asian avatar for two hours in second life and see what it's like,

what experience is like. Yeah. What do the students say? Well, I get both. I get some students that it's a really unique opportunity that's unlike nothing else. If we have time, I can give you a brief example of that. I had a student. He was a white student. He was a white student. He's a French international student. He's white. And he was assigned to use a black avatar. And he was in a, one of the things that they had to do in this assignment was they had to go to a pub. It was called the Blarney Stone pub. Of course. Yeah. And it was purpose. And it had a lot of like Irish decor. It was like, you know, I picked it because it had, it would be like a very kind of obviously white context. And so he was in there as a black male and there's a band playing.

So what that means in second life is there's actually people on a stage that are being run by users somewhere in the world. And I don't know how they're doing this, but they're playing music. And then the other thing is, yeah, yeah, yeah. And the other thing is that you can tip the band, like you can give money to the band. And so my student is writing about this in a reflection paper. And he's saying like, oh, you know, I was like, well, I'm not going to tip them because this is like, it's not even my account. this is a class project you know professor tower gave me this avatar there's no money in it um and then he's like oh well so are people thinking i'm not tipping them because i'm black oh right so then he

kind of just goes with it and he has this whole kind of insight that nothing has to happen you know that that that's part of the minority experience is that things don't have to happen for it to be taking up headspace and and troubling you you know and i really like that um He shared that because that made me think like, okay, that's not something you can get from like showing a movie about like minority struggles or something like this. You know, so I was really, you know, so I had some, okay, so you ask me how to students react. So that's those were kind of the best, you know, from a teacher's perspective. But there, yeah, some students just kind of like, this is not realistic. It's ridiculous and, you know.

Wow. It's not realistic, but right, right. But there are some elements. I mean, yeah, it's interesting. So going back for a moment to the shooting experiments that you did, were your results markedly different from what happens with the traditional test where you sit at a computer screen and suspects pop up? Well, so they're markedly different in the sense that we're approaching analysis

in a completely different way. Again, it's less about is there bias and more about trying to figure out what predicts bias. Right. So instead of just trying to see, is there differences in shooting time, it's like what are the various cognitive and emotional things that predict bias? And as I said, where we're at right now is that we've looked at this belief set around beliefs about. race. We've also looked at one of the things we do is we attach a physiological device to measure stress through heart rate variability. And that's another thing that we've seen that when people are stressed, there's a correlation between stress and quicker reaction times towards black

suspects, but that we don't see that same correlation happening with white suspects. The other thing, so that, you know, what's exciting to me about this is that's really, that's just where we're at. Everything else is kind of in the works and trying to figure out much like we did with the, you know, with the second life stuff. Like now it's like try to figure out how to make more nuanced measures of things. It's, that's kind of, you know, I just, I find it fun, but I think also has a lot of applications. But trying to figure out, for example, that, that, the VR set that I have comes equipped with eye tracking. So being able to look at different attentional patterns inside the VR. And then specifically, you know, we have some hypotheses about different ways that stress and attention patterns interact to actually lead to bias.

One of the things that we've seen, you know, so we have the numbers, but we're also watching the screen recordings of the participants interact in these scenes. And one of the things that we're seeing, which is pretty interesting, is that when people become physiologically stressed out, they have these attention patterns that seem to go in different maladaptive directions, one of which is becoming just very disorganized, right? So you can imagine if you're kind of looking left and right real quick, and somebody approaches you from the right, even if that person's approaching you gradually because you're whipping your head around left to right, that person's actually approaching you really quick. And so we're seeing fatal errors in that kind. But we're also seeing people go the other way when they get really anxious is that they lock in, they fixate in a way that's a little bit. You know, if you just kind of decide early on in the scene that this is the suspect and you fixate, now you're sort of not seeing all the other cues that could lead you to make the right decision. So on the other hand, what we're seeing is a pretty, what's seen, and I just want to be clear, this is observation.

So what seems to be, I want to say, healthy or adaptive attention pattern is, you know, having kind of like this flexible but structured attention pattern. You see people looking at a suspect, you know, then looking at their hip to see if they have a gun, back to suspect, looking left, looking right back to the suspect. So it's it's not frozen, you know, but it's, so that's where, you know, that's what we see. And now it's about trying to, how do you operationalize, fixated, flexible, and disorganized eye attention in a 3D, 360 environment? That's what we're working on right now. Wow. It's pretty, pretty cool stuff that you're doing. It's fun in a way.

I mean, it's, you know, I really, I don't mean to say that to say like, it's, you know, I, I really, I don't mean to say like, it's, just for fun. I really have some confidence that it could be very meaningful. That's obviously why I do it. But there's, you know, part of me, it's excited about it. Yeah, sure, sure. Are you offering the program to police departments at this point? Is that, is that a goal? Total open invitation, but I'm having a hard time. I'm getting people to accept. I understand it. I understand it. I mean, it's, I think, and, you know, we've shot ourselves in the foot a bit as psychologist, you know, in terms of being, what are you doing with our data? You know, I get that. Like, so, you know, I have no intention of this. I've even said here just now that my, my,

my goal is not to sort of determine if, if there's bias. It's trying to determine within a person what predicts bias, right? And so, but that's hard to trust that I could go to a police department and say, especially right now. I mean, it's, you know, that I could go to a police department and say, hey, can I assess your officers on this and then that I wouldn't just come out with some report saying that this is a biased police department. I understand that. But that's what I'm hoping for. Great.

Yeah. Right. Well, that all sounds really interesting. So I appreciate you're coming by today and talking with us. This has been edifying. I'd like to spend a little time in second life and maybe try an avatar of a different race and see how that works out. Okay. Just before, it's a weird place. You'll see things you weren't expecting to see. But like a lot of these things, they can be strange and unfamiliar and uncomfortable.

But yeah, I just am excited about the prospect of them for studying behavior. Great. Well, thanks again. And I just want to say before we sign off here, that we'd like our listeners to talk to us. We're open to your ideas, so send us your comments, maybe topics or guests that you'd like to hear us talk about in the future. You can send these ideas to Speaking of Psychology at APA.org. And while you're out there doing things like that, could you give us a rating in iTunes? That's really helpful. Speaking of Psychology is part of the APA podcast network, which includes other informative podcasts such as APA journals dialogue about new psychological research and progress notes, about the practice of psychology.

You can find all our podcasts on Apple, Stitcher, or wherever you get your podcasts. You can also go to our website, www.spicingsoftology.org, and listen to more episodes. I'm Kim Mills with the American Psychological Association.