SIO286: Studies of Racial Bias in Criminal Justice

Transcript

This transcript was generated automatically and may contain errors — the audio is the record.

0:00:29 Thomas: Hello and welcome to Serious Inquiries Only with Thomas and Lindsey. How are you doing, Lindsey Osterman? Dr. Lindsay Osterman.

0:00:36 Lindsey: I'm not bad. Yeah, get it right.

0:00:39 Thomas: Can I add to your official title co-host of Serious Inquiries Only?

0:00:44 Lindsey: Oh, that would be delightful.

0:00:45 Thomas: That... I am so excited. I can't even hold it in another second. Made it official. We got hit... Oh no, I don't want to give people the wrong impression we may not on this podcast host of me, Lindsey, and I wanted to tell everybody that Lindsay is officially on board and, like, uh, she's hired on to the show capacity because she's so great and I love her so much and the content has been amazing, I'm so excited to have you. So, ah, good times, um, you know, it's been a little little slow this month, so this is the first episode in a while.

Sorry if you've missed us, but we've got a great one for you. We're still coming off just the roller coaster of this week. I mean, obviously, with the Chauvin trial and actually getting something resembling justice, wow, feel in all kinds of ways about that. We'll have, obviously, more legal breakdown on opening arguments about that, but we got the science breakdown over here because Lindsay is... is going to take us through some of the science behind bias in the criminal justice system. I can't wait. It's the scientific deep dive that we've come to know and love from Dr. Lindsey Osterman. So there, why don't you get us started, co-host?

0:01:08 Lindsey: I'm super excited. Yeah. Thanks.

0:01:24 Lindsey: I Would be delighted. Yeah, no, this was an interesting thing to prep for because I know some of this stuff, I cover it in Social Psych fairly regularly, but there was a lot that I didn't know as well about some of these findings. So I hope this will be interesting to folks. So yeah, so we're going to be talking about racial bias, and before we get into some of these studies, I wanted to just give a quick primer on these concepts of, like, intergroup bias and prejudice and stereotyping and discrimination and how we talk about them in Social Psych. So, to start with, bias is a general concept.

If we're biased towards something, it means we're partial to it in some way. So a biased coin, right, lands on heads more than it lands on tails or something. And in social psychology, there's a really rich literature examining intergroup bias, which is bias due to some social category that someone belongs to, like gender or race or sexual orientation or something. So intergroup bias is a really multifaceted construct. It can have affective components and cognitive components and behavioral components. So affective intergroup bias we call prejudice, and that means that we may like or dislike someone more or less on the basis of their group membership. So if I really like gay men as a group, I would have a prejudice

0:02:27 Thomas: And who

0:03:01 Lindsey: In favor of gay men.

0:03:01 Thomas: Doesn't? Yeah.

0:03:02 Lindsey: Exactly, right? They're delightful. So cognitive intergroup bias, that's what we call stereotypes, right. So that means that we may be more likely to hold certain generalized beliefs about members of certain groups. So if I'm a white supremacist, then I probably believe that black people as a group are inherently less intelligent than white people. That's a generalized belief based on this.

0:03:23 Thomas: And you might write books about that called like the bell curve. I'm just throwing out weird examples. I don't know why I would say that. And then you might have the author of that book on to like defend how obviously correct it is on your podcast. That would be another. I'm just coming up with examples. That's what my function is here. It's kind of with wild hypothetical examples, not drawn from real life, just making these up, spouting them off.

0:03:29 Lindsey: Yeah. Mm hmm. Yeah, no. But theoretically, yeah, if it was just the facts, I feel like I would have a moral obligation to give a platform to somebody that showed that. Yeah, for sure. So yeah, so affective cognitive components and integrative bias can also be behavioral, and that's what we would call discrimination. So if I refuse to bake a cake for someone on the basis of, say, the fact that they're gay, then I'd be discriminating on that person on the base of their sexual orientation.

0:03:50 Thomas: Yeah, yeah. Okay.

0:04:08 Thomas: So, just clarifying all the different kind of the taxonomy of how this can go, which it sounds like is something you have to do when you're going to do science around a thing. You need to be precise of like, what are we looking at? Are we looking for, you know, attitudes that aren't really expressed, which sounds like that has a fancy name you said already, or are we looking for like attitudes that actually lead to action or so? Is that, that what we're talking about? Kind of dividing out these, these different components of it.

0:04:16 Lindsey: So, yeah, lots of definitions.

0:04:33 Lindsey: Exactly, right? Yeah, parsing them out. And that's a great segue because in addition to those distinctions that I just made around, you know, feelings and thoughts and behavior, social psychologists also make this distinction between explicit and implicit bias, which is exactly what you just said, essentially, right, it's about, well, I'll tell you what it's about. So the specifics of how people make this distinction between explicit and implicit vary a little bit. But I think that a reasonable definition for our purposes is that explicit bias refers to biases that are essentially like consciously identifiable and deliberate as opposed to implicit biases, which I'll come back to in a minute. But again, if I'm a white supremacist and I openly report that I think poorly of black people, or I dislike them, or I will do them harm given the chance, or maybe all of those things, that I'm communicating explicit racial bias, right? Or a less extreme example would be, like, if because of my strong religious faith,

0:04:47 Thomas: Gotcha.

0:05:32 Lindsey: I say that I dislike the homosexual lifestyle, right? I'm communicating explicit prejudice toward gay people. But implicit biases operate more subtly. And another feature of them is that they tend to operate automatically, that's an important feature, and without intentionality. Right. And sometimes they can operate outside of conscious awareness. I don't particularly like calling implicit bias unconscious bias because I think that it gives people a false impression, impression, so um, like, it could give the impression that, like, even though I feel like I, I love gay men and they're my favorite, they're my absolute favorite in the world, that secretly somewhere back in the back of my head there's like a homunculus that secretly like hates gay men and I'm not aware of that, but it's, you know, it's influencing my, my behavior when I'm not looking or something.

And that's, that's really not the idea behind implicit bias, bias. Anyway, so some of the criticisms of implicit bias research seem to be addressing like some version of that kind of caricature. They're taking issue with the idea that we might have these like secret racist motives that we're unaware of or something. But in actuality, I think the important lessons from the implicit bias literature are about the fact that these biases can operate automatically without our consent or motivation or intention. And again, sometimes, times, their effects happen outside of conscious awareness.

0:05:38 Thomas: Right. Right, okay. I'm, I'm not entirely sure I'm getting the distinction because like I could imagine, you knows, maybe I have a, you now what? I'm tired of using other other groups. We'll just use white podcasters. So, so I can imagine being like, I love white podcaster. Some of Lindsay, some of my best friends are white podcaster. Um, yeah, but, and, and I consciously think that, but so it may be a straw man to be like, ah, but you have a little homunculus in there that secretly hates white. Oh shit. Maybe I do actually. Dang. I was going to use this as an example. I realized that I just disproved your point. Cause that is, I do have that homunculous. It's like, God, another white podcaster. No, but I could see it being like, yes, you like this group or whatever. You're cool with this group.

You have positive feelings toward this group, but then you still have biases about their behavior in a way that you might not notice. So like, you know, you might be like, oh, they're always spouting off on their podcast, whatever, something like that, or pick the equivalent stereotype for another group. So is that still, I guess I'm not seeing what the distinction you're making there with that clarification is. Are you saying that's not what's happening or, hmm?

0:07:13 Lindsey: Hey, me too.

0:08:00 Lindsey: Well, so in the example that you just gave, though, right, you can identify the fact that you have mixed feelings about white podcasters, right? And I think that the way that implicit bias is sometimes represented outside of psychology is that you may have absolutely no awareness of the fact that you actually harbor these negative associations with certain groups, and that doesn't have to be. I just think that there, well, I don't just think, there's evidence that there often is awareness of it, right? So for example, like a lot of people have probably heard of the IAT, the implicit association task. That's a popular measure of implicit bias and it measures like the automatic associations between

0:08:20 Thomas: And Oh, that's interesting, so you think there has to be some level of awareness of it, okay. Yeah.

0:08:42 Lindsey: Certain concepts, right, so probably the,

0:08:44 Thomas: Because I could Honestly, really see, as much, you know, as people who listen to the show and hopefully I've demonstrated, like, I want to kind of live feminist values and and perpetrate that, but I could totally see flunky accidentally messing up one of those, like, "Well, actually, the doctor was the mom, you asshole." You know, like I could see messing one of those up and being like, oh, I had a subconscious bias that, you know, the doctor had to be the guy kind of thing.

0:09:00 Lindsey: Exactly. No, exactly, right. And I mean, I think that, like, when you say subconscious, conscious, that's appropriate, because you probably weren't directing attention toward gender as like a salient factor there, but like in retrospect it's like, oh yeah, it's because there's these cultural associations between masculinity and these professional careers, and femininity and these other things. So it's, they're just, they're not unconscious in the sense that they can never be accessed, or that you would never, yeah.

0:09:15 Thomas: Okay. Gotcha.

0:09:29 Thomas: I see. I see that's what you're saying. Okay, they're not locked away in some inaccessible place. Hypothetically.

0:09:31 Lindsey: Yeah. Like not, not in the Freudian sense of, you know, Right, um, and yeah, so on that, on that note, right, so the IAT, there's actually some evidence that people can predict their IET scores with some Degree of accuracy. And so that indicates, right, that the content of these implicit biases is not completely inaccessible, right? I mean, there's one actually that is for men and women and the strength of the associations between men and men and the concepts of STEM versus humanities disciplines.

0:09:37 Thomas: Okay? Oh, okay. Yeah.

0:10:05 Lindsey: And yeah, and I...

0:10:06 Thomas: Well, but that's also just hard science. I mean, the gender imbalance in those fields is like a real thing, though, right? I mean. So the association. OK, sorry. I don't want to jump on what you're saying.

0:10:13 Lindsey: Oh, it is. Yeah. Yeah. Yeah, no, I mean, that's absolutely right. I mean there's real places that these associations come from, right, even if we don't endorse, like, you know, the specific belief that, like men are better at STEM or should be in STEM or something like that and women should be in the humanities more or something, right? Right. But anyway, like I like to use that example in in my social site class, and like, I took it so that I could share my results with them. And I predicted accurately that, yeah, I have the predictable stereotypic association between men and women in those careers. Right. So it's again, it's not inaccessible.

0:10:23 Thomas: Oh, good. You're not about to tell us you predicted you're a flaming racist. Like you just I feel like you should tell us. I just anyway, remember that announcement I made earlier in the show? That's still tentative. Paperwork hasn't gone through just in case Lindsay's like, yeah, I predicted I'm a massive racist. I was right. Right. See, you can predict. No, I'm just kidding.

0:11:02 Lindsey: Yeah. Interview process still underway. Yeah, so anyway, my only point here is that unconscious doesn't necessarily imply that we're not aware of the biases.

0:11:15 Thomas: OK, I think I've got the distinction now. Thanks.

0:11:17 Lindsey: Yeah. But it is true that these automatic associations can influence our judgments and behaviors toward people in ways that can sometimes be difficult for us to identify in the moment or to correct for. And so that's kind of central to the concept concept of implicit here. And one other thing to underline about implicit bias is that these automatic associations, as we've kind of been talking about here, they're often rooted in things like culturally transmitted stereotypes, which means, again, that like essentially is walking around with implicit biases that often they don't endorse, right? They don't want to have those associations and they're there anyway, right? Not.

0:11:33 Thomas: Yeah. Yeah. Well, and I'm sure I'm not wanting to jump on it. We can pause pin this for later or whatever. But like in the history of doing the show, it is amazing how much I've learned in going from, you know, being raised a Fox News conservative to now, how much I've realized, like how self-perpetuating this is, you know, it's like a little bit of bias and then things like we just mentioned like, well, obviously the patriarchy has made it so that more men were doctor only men were doctors for a while. And then the, you know, a few, and then that becomes like, Well, it, okay. My brain hasn't done something wrong per se by identifying that like, yeah, doctors tend to be men. Cause that's the culture you live in. And that perpetuates itself. And it's like a feedback loop.

0:12:10 Lindsey: Yeah. Right.

0:12:38 Lindsey: That's, yes, that is absolutely a super important point, right. It's like, we, a lot of these mechanisms are essentially like picking up on these associations in our environments that come from somewhere, often from these sort of, you know, systemic inequalities on the basis of things like gender and race. And then, yeah, once that feeds into the implicit biases, and then I think that the implicit implicit biases also feed into how we interact with people.

0:13:01 Lindsey: And that has like, yeah, has harmful consequences.

0:13:00 Thomas: And and then you get the Sam Harris's of the world and other people like this saying, no, I see their their genetic and, you know, the the bad Evo psych people, the evil, the anti Lindsey's comic book villain of Lindsey. They're like the evildoer Evo Psych people who are like, no. There's evolutionary reasons why men are better doctors. But when you like my my feeling has been from learning more about this. Once you peel back as much of the like societal bias as you can, and then you look at the real numbers, which is often hard to do because you can't separate like, you know, how, what is this base rate level of difference if there is any between these two things. But even if you could, it seems to be very minuscule. And then possibly the cultural expectation, the biases, the barriers, the systemic barriers, like that could be most of it. You know, that could Be like a large percentage of the difference. And in a more equitable society, maybe you wouldn't see much difference, if any, in some of these things.

0:13:34 Lindsey: Yeah. Oh, yeah.

0:13:59 Lindsey: No, I think that's absolutely right. Really it's a great point. And I wholeheartedly agree with it. And actually some of the things we're going to talk about I think maybe speak to that a little bit. So maybe we can come back to that.

0:14:07 Thomas: But and cool. We'll keep that in our back pockets.

0:14:10 Lindsey: And so then one final thing to emphasize about implicit biases is that they don't have inevitable negative consequences on our behavior toward others, which is one thing that you're going to see in some of the studies that we discuss. And I think it's important to point this out because I think the idea of implicit biases being automatic and unintentional and that everybody has them can sometimes lead to some fatalistic thinking, especially in the realm of racial bias. If they're always there, then what the hell do we do about them kind of thing. But in actuality, the evidence suggests that these automatic biases can be corrected through controlled processes. So if I can recognize the bias, and if I engage in conscious correction for it, I can successfully prevent it from translating into discriminatory behavior. But you may have noticed that there are a couple of ifs in that statement that are kind of important, right?

0:14:57 Thomas: Yeah. Right.

0:15:00 Lindsey: If we're talking about race as a source of bias, I have to recognize race as the source of a bias in the particular decision that I'm making. And then I have to be both motivated and able to engage in effortful conscious control in order to correct for it when I make my decision. And sometimes people aren't aware of race, for instance, as a sort of bias. You actually, in various places before, I think you've talked about those resume studies where they show that if you change the name on a resume in a way that implies either gender or race it can have an impact on decisions to interview those people.

Right, right. And so, like, in the resume studies, I think that it's unlikely that that effect is happening on the basis of gender or whatever it is, because the person is like consciously racist and looking for, you know, ways to sort of disqualify the black candidate in a deliberative way, right? It's probably more the fact of like these implicit associations and the fact that they aren't inspecting gender or race as a relevant variable and how they're evaluating the applications. And when you direct people to do that, then they're able to sort of correct for the biases that those things introduce into decision making, which is, I think, a hopeful note, right. Right.

0:15:07 Thomas: Mm hmm. Yeah, mm hmm. Um well, There have been some issues with blind resumes, you Now, and then they'll, that's what all the kind of, I don't know what to call them again, but the, I guess the right, maybe like the cultural right IDW types will push back and be like, well, they tried that on this one company in Australia and then they didn't hire any women. And you're like, oh, okay. Well, there could be downsides too, again, because there'll be, you know, marginalized groups won't have the same opportunities for experience and that kind of thing. So anyway, but put a pin in all that and we can carry on.

0:16:26 Lindsey: Exactly. Yeah, right. Yeah, there's... there's lots of influences other than, yeah, than this one that we're talking about that need to be corrected as well. But yeah, yeah, so if you're not aware of it you can't correct for it, and also, you know, sometimes people aren't, in aren't able to engage in correction. So if you diminish people's capacity to engage in effortful processing when they're making a decision, so for instance if you sleep deprive them, or if you have them do some demanding task at the same time that they have to make some decision where implicit bias could influence it, then they're less able to correct for the effects of implicit bias.

0:17:08 Thomas: Like if it's a tense armed standoff or something like that per.

0:17:12 Lindsey: Yeah. I like where you're going with that.

0:17:16 Thomas: Interesting. Okay.

0:17:18 Lindsey: Yeah, right? Yeah. And I'm sure that no police officers ever have been sleep deprived as a Result of, yeah. So anyway, anyway, so, yeah, so we're going to work through a few studies, I think, that illustrate the effects of implicit racial bias that have been documented in the realm of both policing and then in sentencing decisions that are really interesting. And obviously, you know, this topic occurred to us to do apropos of nothing.

0:17:24 Thomas: Yeah, true.

0:17:46 Thomas: Yeah. Literally anytime during, let's see, we could do this anytime and they'll be like, oh, they must've done it because of this shooting. And we'll be actually, we did it because it was a different shooting several months ago. It doesn't matter. It's always, it's evergreen, unfortunately.

0:17:53 Lindsey: That's true. Mm-hmm. It's always shootings.

0:18:01 Thomas: Yeah. It's the most depressing thing. Imaginable, but yeah.

0:18:01 Lindsey: Yep. Yeah. We've talked about police shootings before from more of like a bird's eye perspective. Just, you know, the claim that black Americans are shot disproportionately more often than white Americans, right? So we've talked about some of evidence about that. And today, we're going to look at some of the circumstances under which implicit bias might influence the decisions to shoot someone. So we're sort of zooming into these sort of interactions that people have. And then set of studies that I want to talk about. We've also touched on racial biases and sentencing decisions at a couple of points. And we're going to talk a couple really interesting studies, I think, that illustrate when and how implicit bias might operate there as well. So this should be fun in a manner of speaking. So first, I've got these studies on the influence of race on police officers' decisions to shoot armed versus unarmed targets.

0:18:52 Lindsey: So I found this great meta-analysis from 2015. The authors on this meta-analysis are McCauley and Brezin, they examined the evidence from 42 different studies, including 3,427 participants, on whether a target's race impacts people's decisions to shoot or not shoot in a computerized first-person shooter task. So they called it a first- person shooter game, but I think that that's a little bit generous. We're not talking Call of Duty here.

0:19:22 Lindsey: That's what I'm saying. It's just like a series of images that they're showing, but nonetheless. So I've got that.

0:19:21 Thomas: Yeah,. You got one cop in there that's like, "high score, I shot everybody. Blue lives matter," yeah, like, "we can't use this guy, we can yeah,"

0:19:32 Lindsey: Yeah, exactly. No. No, yeah. I have to exclude that guy. So I read this meta-analysis and I found one representative study from the meta-analysis for us to dig into the methods of of, and the findings in this study illustrate sort of the central effects that were going on overall in the meta-analysis. And then there's a couple of additional findings that I think are interesting, and we'll come back to those at the end. This particular study that was included in the metanalysis, this is by Carell and colleagues. It was published in 2007, and it was titled, "Across the Thin Blue Line, Police Officers and Racial Bias in the Decision to Shoot." And so the goal of this investigation was to complement sociological research on racial disparities and actual police shootings of the sort that we've talked about before by providing controlled experimental evidence about the effect of target race in a computerized first-person shooter task.

0:20:27 Lindsey: And this was not the first study to use this first-person shooter task, but I believe it was the first to collect data from actual police officers in this task. So previous studies had just been on community or college samples, but they actually had police officers. And this is important because it's very possible that the results from people who have never received firearms training before might look very different from people who do, presumably police officers. I think that this was some explicit criticism of the prior research that had been raised by people who were critical of the effects. So in study one, they recruited two different samples of police officers. One was from the Denver Police Department, and another was from officers who were attending a two-day police training seminar, and they came from places all across the country.

0:21:16 Lindsey: So this was a national police sample. And then they also obtained a sample of civilians from the communities that the Denver Police Department officers served, and they recruited these folks from the DMV, which I think is kind of a clever way of trying to get a nice cross-section of your communities, right? Right? I mean, we've all got to go to the DMV.

0:21:34 Thomas: I thought it was just like, "hey, you're stuck in here for eternity, want to take a quick test while you're waiting?" Or yeah,

0:21:41 Lindsey: I kind of, you know, I looked for that detail in the study because I wonder if that's what they did. And it may have been. They may have like set up their computer in there and just been like, hey, you're in here for seven hours anyway.

0:21:46 Thomas: I mean, brand new innovation in, you know, social research. Uh, Lindsey, partner with a DMV and come up with some sort of rewards of like, "we'll cut an hour off your wait time if you take this survey or test," or something. Then it's life-changing. We get so much data all the time from a good cross-section of people. I'm here for the million dollar ideas. Bam. Do it. Okay. Moving on. I think there be some governmental reason why they can't. And they're like, sorry, we can't, but it would be awesome if we could. It would be like 14th Amendment or some crap. I don't Due process. I d' n't know how it's the legal show or anything.

0:22:05 Lindsey: Ooh, that's brilliant. No, that'S a great idea. You think they'd do it? No, it would be cool. Yeah that was a good try yeah um so we're gonna our two police examples and we've got our community sample from the DMV. Importantly, in the police samples and the officer samples, these were primarily patrol officers as opposed to like administrative or investigative officers. So, you know, their sort of performance in this task is especially relevant to the situation under investigation, right? Most of the police officer samples were white. So between 64 and 69 percent in the two samples. Samples. And overwhelmingly, they were men.

So 92% of the Denver sample and 88% of national officer sample were men, civilian sample was a little bit more diverse. So only 31% of of the civilian samples white, and it was a little bit closer in terms of being half and half to gender, so only 57% were men so slightly different characteristics there. So the first person shooter game proceeded like this. So during these study sessions, participants would see one at a time, 100 images of young black and white men, 50 black men and 50 white men. These were actual photographs of 50 different people in various poses. And they were holding either handguns or neutral objects like cell phones and Coke cans. And so the images of these men were embedded into these various background scenes like countryside, a city park, in front of a train station, and so on.

There were like 20 of these different background images. And in between each image that contained an armed or an unarmed target, the participant would see an unpopulated background image. So just a city Park or a train Station or whatever with no person in it.

0:24:08 Lindsey: And so they broke up the pictures with the armed or unarmed targets so that there was a break between these shoot, don't shoot decisions that participants were going to have to make. Does All that make sense? Okay. So when the unpopulated image was replaced with an image containing a target, the participant was supposed to decide as quickly and accurately as possible to shoot or not shoot that target on the basis of whether they were armed. So if they saw, yeah. So they're trying to do several things to sort of simulate the richness of the stimuli that cops are encountering on patrol, right, by using all these different background images and all these different photos of people and the different poses and all this stuff. But, you know, there's a degree of complexity that you're just not going to be able to capture in these tasks.

0:24:17 Thomas: Right. Yeah.

0:24:53 Lindsey: And so yeah, inherently, this is a much much simpler decision than the actual decisions that police officers are making in the field. So You know they talked somewhat in the introduction to this paper about the resemblance between this computer task that they created and features of some of the police training that they go through, and I'm not saying that police are actually trained to like, you know, shoot all armed targets or whatever. I don't think that's true, true. But they do have training that tries to sort of increase the accuracy and speed at which they make certain decisions, right? And so if, as you're saying, right, if some of this is implicit in some of that training, then that could potentially be a problem. I don't know for sure that it is. I'm not just saying. Yeah. Okay. So if participants saw an armed suspect, the correct response was to shoot in this task. And if they were holding an innocuous object, the correct response was don't shoot. And they gamified the responses and the task by assigning differential points and penalties for correct decisions

0:25:01 Thomas: Uh-huh. Oh, now

0:25:46 Lindsey: Versus errors.

0:25:46 Thomas: We're talking.

0:25:47 Lindsey: Yeah. So you earned five points for not shooting an unarmed suspect. You lost 20 points for incorrectly shooting an unarmed suspect. I know.

0:25:59 Thomas: I'm just getting ready to hear the points to human lives calculation that we're about to do.

0:26:04 Lindsey: That occurred to me too. Yeah. Yeah,

0:26:07 Thomas: Darkly funny. Sorry. This is, this would be a perfect, like, yeah, dystopian American American movie. Okay. You get minus one point for shooting an unarmed person plus 4,000 million points for shooting either an armed or unarmed, you know?

0:26:07 Lindsey: Mm-hmm.

0:26:22 Lindsey: No, that's terrible.

0:26:24 Thomas: Not I'm sure they set it up better than that.

0:26:25 Lindsey: Yeah, well, you be the judge. Yeah, so you get, you earn five points for not shooting an armed suspect. You lose 20 points for incorrectly shooting an unarmed suspects. You earn 10 points for shooting an armed suspect, you lose 40 points for not shooting an armed suspect. And so you'll notice there's like an asymmetry there between the costs and benefits, and I think the reasoning there was that like they're supposed to assume that if you don't correctly shoot the armed suspect that you're going to be shot, and so they're simulating that higher penalty with the with the 40 points versus the 20. But anyway, the responses here had to be pretty fast. If they didn't make a decision within within 850 milliseconds, which is less than a second. So yeah, within 815 milliseconds of the target appearing, then the trial automatically proceeded to the next image.

0:27:11 Thomas: Sorry. Who are they worried about shooting them? Wyatt Earp? Like is, I get that you want to make the decision faster, but is the person really going to be, uh, or, or Doc Doc Holliday or somebody, you know, like the other person also would have to take the time to shoot you accurately. So I feel like that's a very tough timeline to try to fit into. But oh, well, all

0:27:34 Lindsey: It's not as bad as it might sound. I mean, I know it sounds pretty quick to say that they have to make decisions in less than 0.85 seconds. But that window was based on prior research. They sort of figure out what what reaction time is. And so this is actually kind of a large window, actually, in these kinds of studies, if you're trying to look at if you'RE trying to get any kind of variability in like the accuracy measure, right? So if they're if they'RE making judgments in like, say, two seconds, everybody is going to be 100% accurate all the time. Because that's, yes.

0:27:49 Thomas: Right, right.

0:28:07 Thomas: Well, So I guess my point there is this, this may be a good window when trying to assess biases and stuff like that, and getting ahead of that. But in the real world, I don't think police officers need to know within point eight five five seconds, whether or not they have to kill somebody. That's all like, you probably have a second to sit there and be like, does this, you know, two seconds. Are they going to, are they going to dock holiday, fast draw, shoot you instantly? I don't, I don'T really think we live in a fake wild West where that happens all the time.

0:28:22 Lindsey: No, absolutely. Yeah. Yeah. And that's right. No, no, absolutely not. And I don't think that this detail was one of the things that they were trying to get realism in. They're trying to correct, yeah, they're trying to create a test to get variability.

0:28:45 Thomas: Yeah. Gotcha. They do want to find those biases. Okay. Gotcha,

0:28:48 Lindsey: Yeah. Yeah, and so importantly, right, so if they didn't make a decision in time, that failure to respond accrued a 10-point penalty. And it wasn't counted counted the same way as like incorrectly saying don't shoot, was so, so each participant did 16 training trials and just to familiarize them with the task, and then they did 100 test trials and those were the ones that were actually analyzed, and the researchers recorded both how quickly participants made each decision and their accuracy in the decisions across the 100 trials, and so then they performed analyses to determine whether the race of the targets influenced the decision speed and the accuracy of the shoot-don't-shoot decisions. And so if racial bias affects the difficulty and accuracy in these shooting tasks, you would predict that because armed Black suspects are more consistent with stereotypes about Black men, then participants would be faster and more accurate in making a shoot decision for Black compared to white armed targets.

0:29:46 Lindsey: And conversely, they predicted that participants wouldbe slower and less accurate in and making don't-shoot decisions for unarmed black versus white targets. So here are their results. Overall, participants only provided incorrect responses on 4.7% of trials, so that's a pretty high degree of accuracy. They also timed out on an additional 4.8% of trial, so across all the samples, you got accuracy of something like 90%, which is pretty high.

0:30:13 Thomas: Huh? Yeah.

0:30:14 Lindsey: Both officers and civilians demonstrated a bias toward a shoot response, but the community sample showed a larger bias toward the shoot response than did officer samples.

0:30:25 Thomas: Interesting.

0:30:25 Lindsey: Yeah. And there was also evidence of racial bias, such that participants had a bias towards a shoot decision for black targets, but not for white targets. But this turned out to be driven by the sample, or by, excuse me, it was driven by the community sample. So the two officer samples did not show significantly more bias toward a shoot decision for black targets than toward white targets. It was just the community example.

0:30:51 Thomas: That's interesting.

0:30:52 Lindsey: Yeah. And the researchers interpreted this as to be evidence for the fact that because of their training, police officers are more accurate in their decisions to shoot or not compared to the community. Sample, right? But that's just the results for accuracy, right.

They also collected that information about speed and whether there were biases in the speed of the decisions that they made, so right, overall participants were able to make shoot decisions more quickly than don't shoot decisions, and additionally, there was racial bias in decision time, so participants, officers, and community members alike were faster to shoot armed black targets than armed white targets, and they were slower to decide not to shoot an unarmed black target compared to a white target, So if you take the speed results together with the accuracy results, this is an interesting pattern, Like the right? Reaction time stuff is indicating that the task is more difficult for both officers and civilians when the targets are stereotype discrepant, right, meaning an unarmed Black man or an armed white man, the accuracy thing, right.

So this is clear clear evidence of implicit racial bias in terms of that difficulty. Race is activating these stereotypes related to danger, criminality, or whatever, and this is affecting the decision difficulty. But this bias is translating into behavior only for untrained civilians. They're the only ones who are mistakenly shooting unarmed black targets more often than white targets, right? It's not translating into behavioral for the officers. They are able to correct for that bias. So there's a hopeful thing here, potentially, right? At least the researchers were interpreting this to mean that their training works. It provides them with greater competency in correcting for bias, essentially. You know there's the butt coming. I mean do you have any thoughts about a butt that might be coming?

0:31:49 Thomas: Uh-huh. Yeah.

0:32:33 Thomas: All Well, I could see a number of things. I mean, for one, you talked earlier about the stressful situation stuff, so I could imagine, or the lack of sleep and or the whatever, you know, I could imagine that coming into play. And also the way we have it set up when we apply this to the real world, and obviously, you, I've done episodes on this, and we could always do more episodes on what changes we actually need in policing, because it's, it's so huge, but there's so little, little tolerance for error in the way the system is set up. You know, like, if we had a more, had a more, like, de-escalating police force, you know, then if we had things that gave people more time to be human, like, yeah, there are unconscious biases and there are these things that you're pointing out, possibly there are a slight more difficulty in identifying an unarmed Black man or an armed white man. And for example, if we had a system that allowed people that room, but we don't.

0:32:41 Lindsey: Exactly.

0:33:32 Thomas: I mean, I just listened to this interview on this, I think it's the Slate podcast, What Next? And it was with this guy, Michael Sierra Arevello, I think he's a professor at University of Texas at Austin. And he talked about, he, he talked about the police culture of, like, all the training that he witnessed is every moment could be your last. And he gave this example to the point where police officers drive without seatbelts because they want to be able to pull out their guns unencumbered.

So they're literally putting themselves at way more risk, because if you look at the numbers, they're so much more likely to die because of not wearing a seatbelt. And they do, by the way, they do die from not wearing seatbelt. But because they're laser focused on, on the idea that some bad guy with a gun is going to get them, they're only focused on that risk. So anyway, all this is to say that our system allows for so little error that even these tiny discrepancies that you're seeing in this experiment could be potentially fatal, but I'm sure you're going to give us what the real butt is here.

0:34:04 Lindsey: Oh, wow. Yes. Well,

0:34:36 Thomas: Give us your butt, Lindsay. Oh, I got to edit that.

0:34:36 Lindsey: No, no. I mean, no, no. I mean I think that you've hit on a couple of really important things here. So, again, the fact that they are able to make these corrections under – this task represents essentially like the best case scenario, Right? This is a distraction-free environment. They've got a relatively simple decision that they have to make in these tasks relative to the more complex kind of situations they're in. You've mentioned these features of kind of the culture and the mindset that's going into it. So, like, I think it's encouraging that at the very least, under the best of circumstances, you can correct for these sort of implicit associations. But as you say, like the real world in which these things are happening is a lot more complicated, and officers may not have the sort of the cognitive resources available in these real moments in order to correct for that bias.

0:34:51 Thomas: Yeah.

0:35:29 Thomas: Right. And if the culture of policing is proudly in the no room for error, you know, And by no room for error, I mean no room for allowing even the possible danger toward any possibility of danger toward police lives. That's going to mean shoot first, shoot first. Reduce the possibility. Reduce the allowance for seeing an unarmed person and just assume more people are armed.

0:35:32 Lindsey: Exactly.

0:35:55 Lindsey: Yeah, exactly. Right. Yeah. So those are all pretty big butts, I would think.

0:35:59 Thomas: We like big butts here on Serious Inquiries Only Zomely. And we can't really lie.

0:36:01 Lindsey: Yeah. We must be honest about that.

0:36:03 Thomas: Yeah.

0:36:04 Lindsey: No.

0:36:04 Thomas: You know, I

0:36:06 Lindsey: Okay. So yeah, a couple of other quick little findings from that study. They also looked at a couple of correlations with speed and accuracy and some demographics. And so they looked at response speed as a function of the characteristics of the places they were from, essentially. And they found that officers who served in cities with larger populations, with higher crime rates, and with larger population of racial minority citizens, they showed greater racial bias in their decision times. Again, that wasn't translating into racial bias and accuracy, but it looked like the decisions were even harder as a function of those kinds of factors, right? And they also found that personal endorsement of racial stereotypes also predicted greater racial bias.

0:36:54 Lindsey: So they had some questions in there asking them about like whether they endorsed these sort of stereotypical statements about Black individuals. So explicitly endorsing those decisions also predicted greater racial bias in decision times. That's interesting stuff. All right. So study two, they essentially, they just replicated the effects of study one, but they shortened the time window. I don't have a lot to say about study two. It was basically a replication effect. Study three was interesting. So they used the same procedure, but they examined the effect of practice on the shooter task performance in a sample of undergraduates, and they did this in order to test somewhat more directly their speculation that officers' lesser racial bias in their decision accuracy was due to training. So this sample did the shoot-no-shoot game four separate times in two different study sessions.

0:37:43 Lindsey: So they came in on day one, they did the game twice, and then two days later they came it again and did the the game two two more times on day two, right? So if training accounts for the fact that the officers showed less racially biased decisions, you know, despite evidence that the racial bias is affecting reaction times, then we should see that after practice with the task, the college student's decision accuracy becomes less raciously biased. And indeed, they found results that were consistent with that.

So in the first study session, first day, that is, they saw improvements from the first round of the game to the second round of it and the second round of a game during the first-round participants performance looked just like the community sample did in study one, so participants were more biased to shoot Black targets and less and bias not to shoot white targets, but in the second round the game, the racial bias for decision accuracy disappeared, despite the fact that the racial bias in reaction time remained, so they were still showing the greater difficulty on the stereotype incongruent trials, but their accuracy improved, improved, essentially, in the second study session, the one that that took place two days later, they essentially found the same pattern as they did on day one.

So during the first game session, their their racial bias and accuracy went essentially back to what it was the first time that they played the game. So just having played the gain before didn't seem to eliminate the accurate, the racial bias in the accuracy, right? But then by round two, they again displayed no significant racial bias. And in fact, they were even less biased than they were in the second game that they played on day one, if

0:39:13 Thomas: So they're just getting better at the game, essentially, but not the second time. OK,

0:39:17 Lindsey: Yeah, it's like that gap, right, is sort of eliminating most of the benefit of practice, right? So if you put two days in between the first training, then they look more like novices when they come in on day two. And then they show show better practice effects during the second round, right? So essentially, right, this is indicating that this minimal amount of training or practice doesn't eliminate.

0:39:46 Lindsey: It's not enough, right? You need more training.

0:39:46 Thomas: Like you'd need to keep re-upping it.

0:39:48 Lindsey: Yeah, Right. And it also indicates that training and practice doesn't illuminate the automatic activation of racial stereotypes, right? It can make you better at correcting for them, but everybody is showing, regardless of training, officer samples, samples, community samples, everybody is showing this racial bias in terms of how difficult the task is, as indicated by the response times. Pretty interesting. Oh, and the researchers also noted in the discussion of this paper that they were in the process of collecting longitudinal data on police officer training by looking at their performance on this first-person shooter task when they enter the police academy and then comparing their performance as they progress through their training. And so they were still in the

0:39:48 Thomas: Yeah. Hmm. That's

0:40:26 Lindsey: Middle of collecting this data when they published this paper, but they did note that some preliminary findings that when these recruits entered the academy, they showed racial bias and shooting accuracy that was identical to the community samples. And they were anticipating that after they went through training, they would show evidence of being able to correct for it, essentially. As I said at the beginning, right, the central findings in this paper were consistent with the overall findings from this 42-study meta-analysis of racial bias in these first-person shooter tasks. So consistent evidence was found for racial bias and reaction times across these 42 studies, but they did not find consistent evidence across the 42 studies for a consistent racial bias indecision accuracy, right? So that was consistent with, you know, the overall effects and the meta-analysis.

0:41:12 Lindsey: Now, some additional findings that they talked about in the meta analysis that I think you might find interesting. The belief in stereotypes about black people positively correlated with task difficulty when the target was black. So that was another finding that I mentioned from that one study that's shown up in other studies. But two, and this is the one that I Think you might Find interesting. Participants from states with more more permissive gun laws had more liberal shooting biases. So they had a pronounced bias toward the decision to shoot, right? So I don't know exactly what to make of that, but I think it's interesting. It's possible that means we ought to get rid of guns. I don't know. What do you think?

0:41:42 Thomas: Weird.

0:41:51 Thomas: That's, I don't know. That's hard to believe. Like, I don't... What would be the idea there? Would it be that like there's more you're quicker to think people have guns or you are more like into shooting guns because you are in a state that likes gun? Like, what is the thinking there?

0:42:09 Lindsey: I'm not sure. I mean, that seems like a reasonable speculation, speculation, um, that was one of the speculations that the authors gave, um, and I mean there's another possibility that it has to do with sort of the general political attitudes in in states that um, have more permissive gun laws, but they actually controlled for that in their analysis and still got this association between the the gun car or gun control permissiveness and, yeah, so I I'm not sure what to make of that, but I do think it's interesting.

0:42:23 Thomas: Right. Huh. Yeah. Because I, yeah, That just seems really weird because, you know, I have all kinds of friends from this very state, where I think you would say California is probably not like the most permissive on guns. I don't know, have a keen sense of that among states, but like you would think liberal California wouldn't be like the super gun toady state, and this really feels like whether or not you are part of that culture or not, maybe that's what it is, like if you're more likely to be part of that because, you I'm from California and I don't like guns and don't have a gun, don't do anything with guns, but I also grew up with people who are all about guns on a zillions of them, and we would both fall under the same, like, California, state level of permission for gun laws in this study. So it's just, yeah, I'm not sure what to make of that. It's kind of weird.

0:42:37 Lindsey: Mm. Hmm.

0:43:23 Lindsey: Yeah, I'm not either. I think I'd have to do some more reading. But yeah, I don't know. I just thought I'd throw that out there. Seems. So all of this seems to validate the presence of implicit bias, right, in terms of the evidence with the reaction times. But it doesn't seem to be showing up in the actual behavior, right? That decision to shoot or not shoot. So now the question is, how do we get from this research to this well-documented effect in field data in terms of officer-involved shootings about this racial bias, right?

0:43:25 Thomas: Okay. Yeah. Yeah.

0:43:51 Lindsey: So we've previewed this a little bit, but, you know, people are fully capable of correcting for implicit bias if they recognize the bias and they're motivated to correct for it and if they have the cognitive resources available to do that correction, right. But if they don't have that motivation, if racism is a salient factor in the decision, if they don't the cognitive resources available, then they're not going to be able to correct for it, next study I want to talk about, they're looking at that cognitive resources piece, right? So if officers don't have the cognitive resources available to correct for bias due to fatigue, sleep deprivation, acute anxiety, and so on, they might be less less able to correct for that bias when they're making these shoot, don't shoot decisions, right? And as we talked about, right, like in the Corel study, that was the last one that we talked about. They're looking at shooting behavior under ideal circumstances, right? Quiet, distraction-free, safe lab environment. They'RE not anxious, all that.

0:44:05 Thomas: Yeah.

0:44:42 Thomas: It's great day for a shooting. Ideal shooting circumstances.

0:44:46 Lindsey: Yeah. So this study that I'm about to talk about, they looked at the same first-person shooter task, but they examined one potentially important and common way in which officers might be cognitively depleted, which was sleep deprivation. So this paper's by Ma and colleagues. It's from 2013. And the paper's entitled, When Fatigue Turns Deadly, the Effects of Cognitive Depletion and Sleep Deprivation on the Decision to Shoot. And they examine the role of fatigue in inhibiting police officers' stereotypic responses in a first-person shooter task. And so the idea behind their study is that in order to prevent these automatic associations from resulting in racially biased shooting decisions, they need to exert cognitive control. If their ability to do that is impaired because they're fatigued, they should exhibit more racial bias in their shooting decisions and less accuracy as a function of race.

0:45:35 Lindsey: Okay, so this was actually a two-study paper, but the first one was on college students. I feel like the study two on officers is a little more relevant to the point that we want to make, so I'm going to focus on that. So they recruited, or they got a sample of police recruits who completed the first-person shooter task on the very first day of their training, and they also self-reported the number of hours of sleep they'd had the night before. And bottom line, what they found in the study was that, racial bias in the decision to shoot across all participants, they found significant racial bias in the decision to shoot across all participants, such that black targets were shot significantly more often than white targets. And again, this was their first day of training. So that's consistent with the results from the last study, right? But critically, the racial bias and shooting behavior was significantly greater as sleep deprivation increased, right.

0:46:23 Thomas: And that's all just self-report. So they just take the people who have self- reported the least amount of sleep and they tend to have these outcomes. Okay.

0:46:29 Lindsey: Yes. Yeah, exactly. Right. They broke it up into like three categories of like, you know, least sleep, moderate sleep, and like most sleep. And essentially, you got the biggest result with the least sleep group, right,

0:46:35 Thomas: Yeah. Well, plus I feel like a huge problem is unless they're going to do this first person shooter game, Call of Duty, whatever, and then just open up a door and there's a live alligator there. And then that way they genuinely feel afraid for their lives. I don't know how you're going to replicate that part of it, you know, which seems to me to be a very tough part of this test. You know, if out in the field, they genuinely have to be on high alert, you know, anxiety wise for whether or not they're going to be killed that day.

And that's kind of what their training tells them or what their culture and, you Know, not to mention, they are in dangerous scenarios, like, you know, everyone could acknowledge that, like there's certain danger to being a police officer, at least you're around danger. If you can't, I don't see how you'd possibly replicate that in this study.

And I don't know, but it seems plausible that if certain effects like sleep or whatever, you know, sleep deprivation can remove that ability to fight these biases and we can never test that. It's going to be very hard to show this in a lab,

0:46:55 Lindsey: Uh huh.

0:47:43 Thomas: Whereas, you know, once you're out in the field, that might instantly change those results, don't you think?

0:47:48 Lindsey: Yeah, no, I absolutely do. And that's, that's why I think it's really important to have evidence that's coming from like these two very different sources, right? You need the field data, you need to show that like, nope, out in the field, you know, unarmed black men are being shot at disproportionately higher rates, right. Because that captures all that Messiness. This

0:47:54 Thomas: Yeah. Yes. You can't do what the right does and what the IDW does and say, well, because we have these lab tests that show more or less not much bias, for example, you know, for argument's sake, and we have this field data that does show a discrepancy that must mean that black people are just more dangerous, you know, like you can't do it the other way around. That's what the Sam Harris types do, is they say like, well, you know, they've actually tested that and there isn't bias, so therefore if black people are being shot more, they must just be that much more threatening that is an inappropriate conclusion to draw to say the least kind

0:48:06 Lindsey: Yes. Yes.

0:48:35 Lindsey: No. That's Right, yeah. It Absolutely. Is Right. Yeah, so if you, if you stopped with that first study, right, and then you said, see, so it can't be race, right, that would be an inappropriate inference, right? Um, but this study is showing that like, no, actually, you know, you, you are getting these sort of biased decisions, right, under these different circumstances, and you it's a little clean, it's a little controlled, but it also helps to to remove move these confounds that you're on.

0:48:55 Thomas: Of hints at what's happening

0:49:02 Lindsey: Yeah, exactly. Right. Yeah. And also, you know, so another limitation of the sleep deprivation study is that it's not randomly assigned, right? This is just people's reports of how much they're sleeping. But there was a study that actually came out last year where they experimentally manipulated cognitive load. So they had people do this, this shooter bias task. And at the same time, they had to do this second audio task. So, they were, they They were keeping – they had to keep track of this listening task that they were doing. And they found that when you – and also, right, they had both novices at the task and people that were trained in the task doing this. And they were able to make all those comparisons. And

0:49:02 Thomas: Yeah right yeah when you say the task do you mean the secondary task they had to do, or you mean, okay gotcha

0:49:42 Lindsey: – no, the first-person shooter task. And Yeah, they find that, like, essentially the people under cognitive load looked like the novices in terms of the racial bias in their decisions, right?

0:49:54 Thomas: Well and like you have to do more than one thing out in the field as an officer, you know, you have to do a lot of things oh god let liberal scientists won't do what we need to snowflakes

0:50:01 Lindsey: Exactly. Right. And I think that the cognitive, so even though you can't put an alligator in the lab, doing a cognitive, I'm sorry, IRBs, IRBS are the worst. But no, I mean, you can put an alligator in the lab, but you can do things like cognitive load, right? And the reason that that anxiety going to interfere with your accuracy in these circumstances is because it's increasing cognitive load. It's impairing your ability to think. So I don't know, experimental stuff does something, right?

0:50:14 Thomas: Yeah right yeah that makes sense, yeah oh for sure, yeah, I'm sorry, I didn't mean to, didn't mean to bash all experiments, you're right, you you scientists have found really creative ways to try to replicate real world stuff, and it's awesome right

0:50:37 Lindsey: But no, yeah, no, no. I mean, but the point that you're making is valid. You need both forms of evidence. You know, they're providing complementary information, right. Yeah. So anyway, just to sum up the shooter stuff, like studies like these, or like this one in particular that we just talked about, right, show that implicit bias doesn't need to translate into biased decision-making under ideal circumstances, but again, like in the field, right, altercations that lead to shootings are not always going to be ideal, right? And also in the sleep deprivation paper, the researchers noted some evidence that police officers are often like very, very sleep deprived, like to disturbing levels. So seems very reasonable to take these concerns seriously. Sleep the evidence that I wanted to talk about with the first person shooter tasks. And next we're going to talk about a couple of really interesting studies about the role of implicit racial bias in sentencing decisions. So, that'll be fun.

0:51:07 Thomas: Yeah we're right oh my gosh, there's so much there, yeah we are out of time for part one. I will say this, you know, again, I'm so happy to have Lindsey on the show. I'm still excited about that announcement. And we are going to try to do more, you know, given now that there's two of us, we could try to do a little more for patrons. We want to encourage you to support the show on patreon.com slash series pod. And I will get that part two out to patrons early. I promise. Scout's honor.

Non-patrons, sorry, you're gonna have to wait a little longer, but we do want to try to deliver stuff to the people who've been supporting the show, and along those lines, we're gonna do, you know, some more like Q and a stuff, um, possible bonus episodes. So keep an eye on it. Watch this space, everybody watch this space. I don't know how you would do that, but, um yeah, cause we're going to try to do that. It'll be a lot of fun, but um, thanks so much for taking us through that. Uh, the, those, those first set of studies, Lindsay, and, uh, there's plenty more systemic racism where that came from parts

0:51:43 Lindsey: Yeah. Yay.

0:52:37 Thomas: Two through 9,875. Well, we'll have to suffice for part two, coming, uh, very soon to patrons and next week for everybody else. So thank you for listening. Thanks, Lindsay.

0:52:36 Lindsey: Yes, there is.

0:52:48 Lindsey: Thank you.

0:52:57 Clip: Thank you so much for listening to Serious Inquiries Only. If you like the show, the absolute best thing you can do is support us at patreon.com slash SeriousPod. The second best thing you can do is please share it on social media or by word of mouth. Speaking of social media, follow me on Twitter at Serious Pod. If you would like to be Facebook friends, send me a request at facebook.com Thomas podcast. If not up for that level of intimacy, that's fine. Maybe you want to join the discussion at facebook.com/groups/SIO members. Thank you again for listening. We'll see you next time.