SIO289: Our Disgustingly Racist Death Penalty

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. And This Lindsay has already resigned as co-host, so it's just me again. God, that didn't...

0:00:34 Lindsey: Yeah, sorry.

0:00:35 Thomas: Oh, no, you're here! Okay, you coming back? No, I'm just kidding. Lindsay's here to stay. How are you doing, Lindsay? I, what studies are we going to talk about today? Okay.

0:00:36 Lindsey: Oh, yeah. Awesome. How Ooh, really good studies. Really good studies, so yeah, so last time we talked about some evidence, the implicit associations with race impact in these shooting tasks, right? And in this part, I wanted to talk about some of the evidence regarding regarding racial bias in sentencing decisions, and particularly in like death penalty cases? Yeah, yeah. So we're gonna start with some real world evidence that black defendants receive death sentences more often than white defendants, and that the degree, this is interesting, the degree to which a black defendant looks stereotypically black is also associated with the probability of receiving a death sentence.

0:00:45 Thomas: Yikes. Right.

0:01:22 Thomas: Away, I'm already mad that we have this much data about this, you know, like we can, we can do cross sections of all the people the state has put to death, probably falsely, you know, like that's that's not a good sign. Okay.

0:01:25 Lindsey: Know. Yeah. Yeah. No, it is not. Yeah. So that's the first thing we're going talk about. And then I wanted to talk about some experimental evidence showing that implicit activation of racial stereotypes may be a plausible mechanism for these deferential sentencing decisions. And as we emphasize in part one, I, you know, I think it's important to discuss the real world data, because this is the, you know, the stuff, the actual outcomes that we care about. But because those data can be messy in terms of figuring out actual causes, the controlled experimental evidence is really important as well, because that can help us make causal inferences. So I think this should be a nice complimentary set of studies.

0:02:10 Lindsey: Yeah. So we're going to start with some evidence presented in a Cornell Law Review article by Baldus and colleagues. It was published in 1998, and it was entitled, Racial Discrimination and the Death Penalty in the Post-Furman Era, an Empirical and Legal Overview with Recent Findings from Philadelphia. It's a hell of a title, I I don't know, I was kind of hoping, you know, because it sounds familiar, it sounds like, uh, yeah.

0:02:06 Thomas: Cool. What's the Post-Furman Era, that's Well, Furman must be a case. Is it like the case that decided, v. Georgia. Okay, don't worry. I'm just channeling. I'm doing a seance and channeling my inner Andrew here for a sec.

0:02:42 Lindsey: Yeah. Oh. In.

0:02:47 Thomas: Yeah, that was the 1972 case that banned. Okay, it said revolves around the Eighth Amendment, ban on unusual punishments and death penalty cases, uh, it, my memory, didn't they, like, turn it off for a minute and then Republicans were like, nah, and then they turned back on. I can't remember. In Phil, no, this is the Supreme court.

0:03:02 Lindsey: Philadelphia, or I mean sorry in Pennsylvania Oh.

0:03:05 Thomas: This. Yeah. This is the whole, but anyway, uh, it says, following this decision, the use of death penalty was put on hold while states revised criminal statutes to ensure death penalty was not applied arbitrarily or discriminatorily. The death penalty was then reinstated after the 1976 case. Oh, okay. I was right. Greg v. Georgia. Yeah. They turned it off for four years and like, well, let's get to the bottom of this. And then they're like, game on. And it was fine.

0:03:31 Thomas: They're like car, stop the death penalty for a second. Okay. Sorry. I did in wings, but it's very inappropriate bit, but I feel like this is a dark show. So, so

0:03:06 Lindsey: Oh, okay. Yeah. Yeah. Yeah. Yep, fun stuff. So Yeah, it is a pretty dark show. Um.

0:03:41 Thomas: Furman.

0:03:41 Lindsey: We need need a few moments of, yeah.

0:03:42 Thomas: Yeah, Furman would be that decision then. Okay, we we cracked it. Don't, don't, nevermind, Andrew. You

0:03:47 Lindsey: You know, this paper, as the title suggests, has an empirical piece, right? And that's what I'm going to talk about because I'm the empirical talking guy. And it's also got legal analysis, which I will not attempt to reiterate because I am not the legal talking guy." So, yeah, so Baldus and colleagues were comparing racial disparities in the outcomes of death penalty cases in which defendants had similar levels of criminal culpability. So in other words, they were trying to ascertain whether Black defendants were more likely to receive the death penalty after relevant case facts were essentially held constant statistically.

So they collected data from 524 randomly selected first-degree murder cases in Philadelphia between 1983 and 1993 that were death penalty eligible based on the facts of the case. And the cases were selected from these official lists of first-degree murder cases and murder convictions in Philadelphia. So their sample included 118 cases in which the defendant ultimately received the death sentence, 230 cases in which the defendant ultimately received a life sentence but the case was death eligible, and 176 cases in which the cases didn't end up proceeding to the penalty phase. So I think one example of when this would happen would be because of like a plea bargain. They talked a lot about prosecutorial discretion in the papers as a potential avenue for bias in

0:03:59 Thomas: Okay, hmm. Oh,

0:05:11 Lindsey: Addition to jury judgments.

0:05:11 Thomas: Yeah, that's a huge one. Yeah, and that's going to be really hard to measure. I don't know, would be a, because I don't how you'd have all that data of, like, what the bargains were and that kind of thing.

0:05:13 Lindsey: Yeah. Yeah, I mean, so they have some, there's a lot of data in this analysis. We're not going to be able to talk about all of it, but they actually do a really careful job of trying to look at bias at kind of like each decision point along the road to potentially receiving the death penalty. It's really really interesting. Yeah, so that last 176 cases didn't end up proceeding to the penalty phase, but they were death eligible based on the facts of the case. And so Baldus and colleagues compiled, coded, and analyzed a truly staggering amount of detail about the facts of each case from court documents. So they coded cases for the presence of specific aggravating circumstances and mitigating circumstances. They had a measure of the subjective severity of the murder, which was coded by blind raters who are all law students. And this was based on a narrative about the facts of the crime.

0:05:25 Thomas: Oh okay,

0:06:15 Lindsey: They also coded for socioeconomic status of both the defendant and the victim, the race of both the defendant in the victim. And then finally, whether or not the defendant ultimately received the death penalty. So this is an incredibly rich dataset and a labor-intensive dataset. I mean, yeah, I can't imagine how long it took to code all these documents and put together all this information. Right, so what did they find? So first, they examined the representation of Black versus non-Black defendants at each sort of legal decision point that they were looking at. And so these first numbers I'm going to report are unadjusted for other factors like SES and and facts of the case and stuff. So in terms of represent...

0:06:33 Thomas: Yeah.

0:06:55 Thomas: That's socio-economic status.

0:06:57 Lindsey: Yeah.

0:06:57 Thomas: I solved it.

0:06:58 Lindsey: Oh, yeah. Sorry. Yeah. Yeah. Sorry, I gotta be careful with abbreviations.

0:07:01 Thomas: Cool Puzzle, did it that's all right Yeah, So

0:07:05 Lindsey: So in terms of representation in death eligible cases overall, Black defendants accounted for 78% of all death eligible cases in the study. So that's out of like all 500 and some cases. Cases. They accounted for 81% of the subset of those cases that advanced to a penalty trial and they accounted for 85% of all defendants who were sentenced to death during the penalty trial. So these unadjusted estimates are high and as you proceed along the path to a death sentence, they become nominally higher, right? So 78% to 81% to 85%. But of course, what we're really interested in here is whether race accounts for differential outcomes even even when relevant facts of the case are held constant, right? And so in order to determine this, the researchers then performed logistic regression analyses.

0:07:55 Lindsey: And we've talked about these before. You do logistic aggression when you want to predict the likelihood of some binary outcome. So in this case, death penalty versus no death penalty on the basis of other factors like race and SES and mitigating and aggravating circumstances and so on. So they ran several different models to examine how probabilities of receiving the death penalty changed between different subsets of cases. So for example, the probability of receiving a death penalty out of all the cases, the probability of the death among only the cases that advanced to a penalty trial, the probability of a death in cases that included or excluded hung juries and things like that. So they did a lot of analyses to determine the racial bias looked like in each of these different kind of comparisons, right?

0:08:45 Lindsey: Massive number of analyses. And. And so without getting too into the weeds on all of these analyses, here's the bottom line. So the researchers found strong and consistent evidence of racial bias across all of these analyses. So, after adjusting for all those covariates we talked about, including the facts of case and socioeconomic status and all those things, both the race of the defendant and the race of the victim uniquely predicted the likelihood of the defendant receiving the death penalty.

0:09:13 Thomas: So, Before we talk about that, I was gonna ask though, when you talk about socioeconomic status, are there ways when we're controlling for these variables already that we could be maybe skewing things? Because if, because you of the racial wealth and income gap, like more Black people are lower on socioeconomic status, does that mean like when you quote unquote adjust for that, like is that appropriate or is that another place? Would you be adjusting out some more racism, like some more systemic racism by saying, well, we'll hold this variable equal?

0:09:33 Lindsey: So, well, yeah, I mean, in some sense you are, right? But I don't think it's inappropriate to do that. Like we'd still want to know, like, all else being equal, if we take a poor white person and a poor Black person, are the probabilities of receiving the death penalty equal, or is there something beyond just SES that's operating here

0:09:52 Thomas: Okay. Right. But I guess both things are information you want, though. Because like if the answer is hypothetically, the answer we're like, no, it's just it's actually just poor people are the ones that get screwed. There's no racism. That's like, well, OK, but that's still going to be a super racist outcome.

0:10:06 Lindsey: Yes, That's Still Going to be a lot of Black people. Yeah. Exactly. Right. No, that's, that's a great point. Yeah, so, yeah, so when, when other factors were held constant, Black defendants were more likely to receive the death, and exactly how much more likely varied across the analyses. So for example, in cases, if you were just looking at the subset of cases that advanced to a penalty trial, the risk of receiving the death penalty was 9.3 times higher if the defendant was Black, after controlling for other factors.

0:10:18 Thomas: Yeah. OK, that's a lot higher.

0:10:46 Lindsey: Yeah, right. And there were other analyses of different comparisons that found risk estimates for Black defendants as high as 29 times higher. Yeah. Or as little as 3.3 times higher, but in all cases, this was a strong, large, consistent effect.

0:11:01 Thomas: As little as three point three times

0:11:02 Lindsey: I know that's bad phrasing, but yeah, that seems to be the lower limit.

0:11:03 Thomas: Higher is not very. Yeah. Wow. Yeah. Those are huge numbers. I even think it would be that big. It's really bad.

0:11:08 Lindsey: Yeah, right. Yeah and also, so that was the risk, the odds ratios like based on the race of the defendant, but defendants were also more likely to receive the death penalty if the victim was not black. And again, the precise estimates changed across the models, but there was a fairly consistent effect, although it was smaller than the effect of the defendant's race. So one analysis estimated that you were 1.4 times at higher risk of receiving the death penalty if the victim was not black, but they were estimates that were higher and lower than this, right? So analyses policies like these provide, right, from actual death penalty cases of racial disparities in these case outcomes. And the fact that racial effects were detected as consistently and strongly as they were across these

0:11:27 Thomas: Right. Right. Yeah. I

0:11:59 Lindsey: Comparisons indicates a strong likelihood of racial bias affecting sentencing decisions.

0:12:04 Thomas: Mean, to say the least you like to understate it, but yeah. Well, and reminder, this is in 1998, is that which yeah and

0:12:05 Lindsey: Yeah, yeah absolutely. Yeah, that was published in 1998. it I think they sampled from 10 years of cases. So it was like 1983 to 1993.

0:12:20 Thomas: I'd be really curious if they've ever, if they like updated or kept this up, even though it sounds like a labor-intensive data set, as you said

0:12:27 Lindsey: Yeah, for sure, for sure. But I think that they did, because the next study that we're going to talk about, they pull on the same data set, but they cite a larger range of cases, and so I think that they were, I think that they didn't end up expanding this data set. That's my inference from that. So, um, I think I think this analysis is also valuable because they account for, again, like a variety of different factors that should influence judgments of defendant culpability, right, and that nonetheless, even after you account for all those things, you still get this effect of race, right, um but right, it's difficult on the basis of results like this to make any inferences about causality, right, about the actual mechanisms that are producing this disparity, because we don't have any direct indicators of why a defendant's race is associated with a higher risk of receiving the death penalty. So implicit riot...

0:12:28 Thomas: Yeah I mean if you were born yesterday, yeah well

0:13:18 Lindsey: Yeah, I'm with you, right? But people who are looking to be particularly critical of these things could say that, and they wouldn't be wrong at some level. So implicit racial bias could be one of the things that's operating to produce this discrepancy, but this particular analysis can't directly speak to that, essentially.

0:13:37 Thomas: What is a single other hypothetical explanation? Well, but you know, because you, you held that consistent and you control for the level of crime, well that's why I'm I get that you're being a scientist here, but like, I can't, I'm trying to think of what even the Ben Shapiro hypothetical, like, other cause for this could be if you've controlled for things like, because it doesn't work to do the Sam Harris and right wing like, "Well, they just commit more crime," so it's like, "No, we held that constant, we, we, uh, we account for that." It's not that. What, what other explanation is it? I

0:13:41 Lindsey: You know, that black people are just more criminals, I guess.

0:13:48 Lindsey: You've, yeah, yeah. They controlled for all kinds of factors of case, but still, there could have been something that they didn't account for.

0:14:19 Lindsey: So it's a little bit abstract, right? But when you're doing a correlational study like this, there's always something that maybe you didn't count for. And so you just, you can't make a causal inference on the basis of, I mean, you can say that it's strongly suggestive of that since we've controlled for all these things, but you can't ever really make a causal claim. So I think that that's what they'd say. And I'm not sure what factors they would suggest we should have accounted for, but yeah.

0:14:46 Thomas: I don't know. That sounds weak, Lindsey. That sounds weak. I don't get it. I feel like if you've, process of elimination to a bunch of stuff, like what's left over, that's... I, whatever. Okay. I don't get it.

0:14:57 Lindsey: No, hey, look, I don't disagree with you. You know, All right. Right. So this next study actually examines the same data set that Baldus and colleagues constructed, but they add a variable that could provide some evidence regarding whether implicit bias is contributing to these racial disparities in death penalty sentencing. So, this paper is by Eberhard and colleagues. It was published in 2006, and the title is Looking Deathworthy, Perceived Stereotypicality of Black Defendants Predicts Capital Sentencing Outcomes.

So as the title suggests, they do something really interesting in the study. They don't use race as their predictor of death penalty outcomes. They actually limit their analysis only to cases with Black defendants. And what they assess instead is whether perceptions of stereotypicality of Black defendant's facial features, so the degree to which their faces are judged to match a stereotype of a Black face, and they use those stereotypical ratings to predict death penalty sentences. So that's kind of a, it's an interesting variable to use here, I right? Think.

So Eberhard and colleagues analyze a subset of the Philadelphia death eligible cases from the Baldus and colleagues paper, and they focus only on those cases in which the defendant was Black, and in which the case advanced to the penalty phase, meaning, based on my understanding of the Baldus paper, that all of the individuals were convicted of first degree murder, and that that there was at least one aggravating and one mitigating circumstance, and then the jury was in the position of having to weigh the aggravating and mitigating circumstances to decide whether the defendant received the death penalty or life imprisonment. That's my understanding, anyway. Hopefully. But so, to create a measure of facial stereotypicality,

0:16:40 Lindsey: The researchers presented 32 undergraduate raters with standardized pictures of the defendants from the cases, and they didn't tell the raters that these were defendants in death penalty cases, so they were blind to that aspect of the study, and then they had these undergraduates rate how stereotypically Black each of the phases were, yeah, and they told what

0:17:00 Thomas: I'd be like, feels like a trap. Is this a trap? I'm not doing it. This is going to get, I'm going to be canceled. I'm not doing this. I will not participate.

0:17:01 Lindsey: Yeah yeah mm-hmm yep. Nope, I think anonymity would have been particularly important, being they were told that they could use any number of features to make their judgments, and they, and they gave them examples of, like, uh, the shape of the lips and nose, hair texture, and skin tone, right, to arrive at their judgments, yeah. So, yeah, so it is a subjective measure, I don't particularly see a problem with that for a variable like this one, but

0:17:23 Thomas: That's Right. Well, yeah. I mean, you're trying to capture a subjective prejudice. So I guess you gotta, you gotta do what you gotta do. That's Just a, I can't, I was just putting myself in the place of like a student being asked to do that and being like, "I don't think this is okay. I don't know, I'm tapping out." Which

0:17:31 Lindsey: Exactly. Yeah. Yeah, I'm with you. No, I'd feel weird about it too. So 32 raters made that judgment, and then they took the aggregate rating for each face, and that's what they were using to predict the sentencing. Okay, so the researchers then examined whether these stereotypicality ratings uniquely predicted the likelihood of receiving the death penalty above and beyond the effects of a variety of covariates. And so they controlled for all the the same things that Baldus and colleagues did, but they also added a covariate that Baldis did not code for, which was the defendant's attractiveness. And that has been shown in previous studies to matter in sentencing decisions, and especially since their focal variable is about facial characteristics, right? I think it was smart for them to partial out effects of attractiveness, right?

0:18:28 Thomas: Again, could also have racism. It does also have racism reinforced into it. But

0:18:29 Lindsey: Hmm. Yeah. That's true, although they, they did note that the attractiveness ratings didn't, uh, vary by race. Yeah. So

0:18:39 Thomas: Interesting. Okay, but Attractiveness didn't vary by race when measured by a group of college students, right, but That wouldn't necessarily be what juries are going to think, possibly.

0:18:46 Lindsey: That's Truth. Yeah.

0:18:52 Thomas: Just Just putting it out there.

0:18:52 Lindsey: No. That's, that's very, very true. But right? Presumably, if you can partial out that effect here, that gives you a sense of whether race would have an Effect above and beyond that. Okay, so the key results in the study were as follows. So black defendants who... So they took that stereotypicality measure and they broke the defendants up into sort of two groups. Those that were below the center on stereotypicality and those that were above the center. They had like just two, you know, less stereotypical and more stereotypical groups. And they found that black defendants who were below that midpoint on stereotypicality were less likely to receive the death penalty than those who were above, but only if their victims were white.

0:18:56 Thomas: Mm-hmm. Sure. Huh. Yeah.

0:19:37 Lindsey: So in cases where there was a white victim, 57.5% of the high stereotypical defendants received the death penalty versus 24.4% of the low stereotypicality defendants. That's a stark difference. When we were looking at cases with a Black victim, on the other hand, there was no difference based on the stereotypicalty of the defendants' facial features. So 45% of high stereotypical defendants received the death penalty versus 46.6% of low stereotypicality defendants on that difference.

0:19:52 Thomas: Yeah.

0:20:08 Lindsey: Difference, um.

0:20:07 Thomas: So I have to imagine this is where because, you know, I'm sure a lot of our listeners like me have heard this thrown out there from time to time in different, you know, whatever, podcasts, documentaries, articles that like a black perpetrator and a white victim has like a way higher chance of getting the death penalty. I kind of have to assume that comes from this, right?

0:20:29 Lindsey: It may, this uh, I mean, I'm sure this is not the only study that's actually, the the baldest study, examined this too, right? Um, yeah.

0:20:30 Thomas: Hmm. Oh, that's I guess that's what I meant. Like this data set of whatever. Whatever. Anyway, at any rate, that's the same observation. Like if it's a white victim, specifically, if it doesn't get into like a white female victim versus like a black male victim.

0:20:41 Lindsey: No.

0:20:50 Lindsey: They don't, they don't have that fine grain of detail in it, yeah.

0:20:53 Thomas: OK, gotcha. Well, that's how I've always heard it, like said, you know. So who knows if that comes from another study or if that's, a you misreading of this or what? But either way, got you.

0:20:57 Lindsey: Yeah. I I am, I am positive that there have been other studies documenting these effects in other places, right? Um, yeah. Yeah, so so this is a stark difference. And it's adding something unique to the findings of Baldus and colleagues in the last study, right? If you're black, you're more likely to see the death penalty. But beyond that, you know, the more that you resemble the stereotype of a black man, the greater this risk appears to become. And this provides, I think, an interesting source of evidence, albeit still an indirect source of evidence, that the degree to which someone looks black may be activating stereotypes about criminality to different degrees, and that that may be a mechanism through which these sentencing decisions are being influenced. So from these first two studies, we have evidence, evidence of sentencing disparities on the basis of race, as well as on the basis of how well someone matches the stereotype of a black man, right?

0:21:55 Lindsey: So now I want to talk about this question of what the evidence is that these inequalities are, at least in part, because of implicit associations between race and criminality. And so this last paper experimentally investigates whether there are implicit automatic associations between black faces and crime, and whether perceived stereotypicality influences the strength of those associations. So they're explicitly performing tests to determine whether implicit bias is a plausible mechanism for these observed inequalities that we're talking about. Is another paper by Eberhard and colleagues. So the first author on this paper is the same as the last one. It was published in 2004. And the title of the paper is "Seeing Black, Race, Crime, and Visual Processing."

0:22:42 Lindsey: So there are five studies in this paper, not going to go through every detail of every one. That would make for a monster episode, but I'm going to focus on the details that I think are most relevant to understanding the overall results and why I think they're relevant to understand the effects from these other studies. So in study one, the authors examined whether exposing people to a black face would make them faster to identify crime-related objects, which you would predict if people associate black faces with the concept of crime, right? So this was a subliminal priming study, which is fun. They had participants first complete what they called an attentional vigilance task, in which participants were instructed to focus on a dot in the middle of a computer screen.

0:23:28 Lindsey: And as they were doing this, there would be these objects flashing in the peripheral vision, and participants were instructed to report in real time whether the objects that were appearing were on the left or on the right, while constantly maintaining focus on the central dot, okay? And unbeknownst to participants, the objects that were flashing in their peripheral vision were these blurred composite faces that were followed by a subliminal prime. So either a black face, a white face, or no face in the control condition. Yeah. Yeah. And so the blurred composite face, this is called a mask in these priming studies. It kind of helps to keep the stimuli subliminal.

0:24:06 Thomas: Sorry, the first Place I went is like, no face from Spirited Away. Way, I hope that's what you're talking about, but you actually mean just not a face, that'd be really funny. Like, ah, they actually associate No face with with crime, that's, I mean, no face is a criminal in Spirited Away, to be fair.

0:24:14 Lindsey: So, oh God, I hope so too. That would prime something completely different though. No face is a criminal. Yeah, man.

0:24:22 Thomas: Yeah. Okay, but not that, yeah.

0:24:25 Lindsey: No, not that, but we should do that study. That sounds really fun. Oh man. Imagine the title on that paper.

0:24:34 Lindsey: Yeah. So okay. So in in subliminal priming studies, you have to do this thing called masking, right, so that you can obscure the effects of the prime. And so the way that they presented these stimuli was as followed. So they had this blurred composite face, and that's up for 100 milliseconds or a tenth of a second. Then that's followed by the actual sublimital prime, which is the black face or the white face or no face from spirited away, for 30 milliseconds, which is 0.03 seconds, very fast. And then that's followed by another mask for 100 milliseconds, and so this is a standard procedure for subliminal priming, um, and past research indicates that while these black and white faces are being sensed by participants, the speed and the masking helps keep the faces from being consciously perceived, um.

0:24:27 Thomas: Yeah. I wonder If this kind of thing, so, you know, you're telling us that this is a common procedure, Whatever, something that works, I wonder if this kind of thing is where people get the impression, maybe from going back to our first episode, part one, about like, wow, these uncontrollable, inaccessible biases, you know, where it's like you're showing something you don't even perceive that you're seeing, no face from Spirit Away.

0:25:26 Lindsey: Yeah. Yeah.

0:25:46 Thomas: Way, and you already have all these associations with no face because, let's face it, the no face is a criminal, it wanders into the the the bathhouse, you know, gets to all kinds of trouble, yeah, so maybe that's where that impression comes from, like, you can just flash this, your conscious brain doesn't even see it, you, that does that Does, does that contradict what you were saying in that episode, or not?

0:25:50 Lindsey: Yeah. Eats all the food, yeah. Yeah. Exactly. No. I, because I I think, um, I think I was careful to qualify that like we can sometimes be unaware of influences, it's just it's not the same as saying that like all bias, all implicit bias is unconscious, like, Like, the content of our biases is accessible to us. But, yeah, no, they operate all the times in ways that we don't recognize.

0:26:23 Thomas: Hmm. Okay. All right.

0:26:31 Lindsey: Yeah, and one other thing, because I – there is sort of – there's some ideas out there about subliminal priming research that are a little distorted, too. So these sublimital primes, they have effects, but they tend to be very small and they tend to be short-lived. So, you know, there was like some scare decades ago where somebody was claiming that they subliminally primed movie goers with messages that said like eat popcorn and drink Coke.

0:26:57 Lindsey: And then it was influencing consumer behavior and that was bullshit. So they don't have powerful enough or long-lived effects enough to like justify this impression of like mind control. It just – you can get these like brief activation effects essentially. Essentially. All right, so they subliminally primed them with black or white faces. And they did confirm with a sort of manipulation check, they confirmed that participants were not consciously aware that they had seen black and white faces in this part of the experiment. Okay, so subliminely primed black or White faces, and then they had participants engage in a second, presumably unrelated task, in which participants had to identify objects in these really degraded images. So participants were presented with a series of these short movie-like segments of objects that started off really, really fuzzy.

0:26:57 Thomas: All right, I

0:27:45 Lindsey: So you should imagine visual white noise, like a TV with bad reception. So it starts off really really fuzzy, and you can kind of see some vague ambiguous outline of an object, but it's almost impossible, possible to identify what it is, but then gradually over the frames it becomes clearer and clearer until on the last frame it's just like a perfectly clear image of an object.

0:28:07 Thomas: Just realized we're old and that example doesn't work anymore. Like if I told my kid, you know, bad reception, they'd be like, I don't know what the hell you're, we're gonna have to replace it. We're gonna say, you Know, when like the internet stops working and so the TV goes to low definition for a bit, you Now, it'll be like that. That's what we'll have to tell my kids.

0:28:15 Lindsey: Uh, hmm, yeah.

0:28:26 Thomas: They'll still have the eye doctor, maybe it'll be like, you know, when the eye doctor like does the fuzzy one, and is like, which one's better you know, do that that's

0:28:25 Lindsey: Yeah, so oh there you go yeah yeah, yeah, that's that's a good comparison. Okay, all right, so whatever works, audience, yeah. So in frame one, you can barely detect the object, by frame 41, it's very clear. And so some of the objects that were presented in this task were were crime relevant, so they, they chose weapons in this study, so they had images of guns and knives, and then some of the objects were crime irrelevant. So they had pictures of like a pocket watch, a phone, keys, stapler, things like that.

And the question of interest was whether people would be able to correctly identify the crime relevant objects at an earlier, more degraded frame if they were primed with black faces as opposed to white faces or the control, right? And if they are, then this would be consistent with the idea that there is a stereotyped association between black faces in crime, and that priming people with black faces is also priming crime indirectly. So here's what they found, for the crime relevant objects only, participants identified the objects on a significantly earlier frame if they were primed black faces compared to the control group or the, or the white faces prime. And this is kind of interesting, they measured explicit racism in this study, so they had participants fill out an explicit racism scale, and that did not predict how early they were able to detect

0:29:45 Lindsey: The objects well

0:29:45 Thomas: What I'm saying that's where people get that in this, uh, impression from the from the earlier episode, where it's like, it doesn't really matter what your explicit bias might be, sure.

0:29:56 Lindsey: It's, um, so there's... there's one distinction I'd want to make here, though, because there's a difference, I think, between like consciously held racist ideas and what you're willing to say on an explicit racism scale. So, but yeah, so but from what they're capturing in that explicit racism skill, that's not predicting the the main outcome here, so this is not an effect of explicit racism, it seems more likely that it's a function of these automatic associations between the concept of black men and criminality. So the

0:30:21 Thomas: Yeah. Which just is so depressing that like, what does this say about, TV you shows, movies, culture, society, like what do we do about that?

0:30:32 Lindsey: Fact that black men do make up a disproportionate amount of our prison populations, right, which is because of all the systemic influences that we've talked about.

0:30:40 Thomas: Right. And then reform reinforces it. You know, it's like the cycle that we talked about earlier.

0:30:46 Lindsey: Absolutely, right? It's the problem. So that was study one. In study two, they kind of they – were examining the same thing but in reverse. So they examined whether criming people with images of crime-related objects would prompt an attentional bias toward photos of black men over white men. Yeah. Which, again, you'd expect if there's this implicit association between black men and crime. So they subliminally primed people with crime-relevant images. So guns, knives, police badges, handcuffs, fingerprints. Or with, like, the same images but all jumbled up so that they're uninterpretable. Which is kind of clever, right? It's like it's the same pixels and everything. They're just not forming an image, essentially. And then they had the participants complete a reaction time task called a dot probe task in which you have two images of faces that appear side by side on a screen briefly, and then they both disappear. And immediately after they disappear, a dot appears where one of the two faces was. And so the participant's task is to locate the dot as quickly as possible and press a key to indicate where it is.

0:31:02 Thomas: Huh? Okay.

0:31:50 Lindsey: And if the participant was paying more attention to one of the two faces when it was on the screen, then they should respond faster. They should be faster to find the dot when it replaces that face. So yeah, it's kind of clever, right? Yeah. So if they're looking at the black face, they should be fast to find the dot. They do.

0:31:58 Thomas: Yeah. Huh. Yeah, I I always love when you when you give us that that, Lindsay, because you're like, you know, scientists, they're, they do good work, right? All right. Trying to like sell everybody on your career. You know, we do some good stuff here now and then. Like, right. I totally agree. Like it's just funny, but

0:32:25 Lindsey: I got to do a little little salesmanship. Otherwise, it's just, you

0:32:27 Thomas: You're also really modest about it. Yeah. It's pretty, pretty good, huh? Not so, not so bad. Would you have come up with that non-scientist?

0:32:32 Lindsey: Know, I wouldn't have come up with that. I think it's great. Yeah. Yeah, so this is cool. This is like a subtle measure of attentional bias. It's nice.

0:32:37 Thomas: Yeah. Yeah, interesting.

0:32:43 Lindsey: Yeah? So here's the question. Are people more attentive to the black than the white faces after being subliminally primed with crime relevant images? Yes, they are. So the researchers, yeah, so the researchers

0:32:51 Thomas: Well, I assume they did like a control group that's just Just like, do people, what if, you with no priming of any kind, what if people, I don't know, for basic color perception reasons or something like tended to, you Know, their eyes went to one face or the other. Did they, what's the base rate of that? Is that not, well,

0:32:52 Lindsey: Uh yeah, no, so they didn't do, because they have to have two faces, right, they can't really have a, can, yeah

0:33:13 Thomas: I know, but they don't have to have the priming. Of the item or whatever. Wasn't that part of it?

0:33:18 Lindsey: So, I mean, so their control group was just like the jumbled image. I, the prime had a, had a picture of like handcuffs, they would take that picture of handcuffs and just like jumble up the image so that, you know, all the pixels are sort of

0:33:29 Thomas: Okay. So they do have a control of like, there's no priming whatsoever. And there's a base, because I was actually curious if just for like, I don't know, human color perception reasons, like you would tend to favor one face or the other. I can see it either way. Like, or if it has something to do with personal identity, like if you're black or if you're white, would your eyes tend to favor one face or the other? And they'd have to like account for that kind of base rate, right?

0:33:33 Lindsey: Yeah.

0:33:55 Lindsey: Yeah, they, they absolutely would want to do that, and they did, and they actually found that at baseline, when there wasn't that priming, that people had a bias to look at the white face, um.

0:33:59 Thomas: Right. Interesting.

0:34:06 Lindsey: Yeah, so, but, but when crime was primed, when they had been subliminally primed to the crime images, they showed an attentional bias toward the black face. So they were significantly faster to find the dot when it replaced the black face in that case. So it was a reversed effect, which is interesting.

0:34:15 Thomas: Okay. Right. Yeah. So it means it's kind of even stronger than you'd think maybe?

0:34:24 Lindsey: Yeah, yeah. Yeah, because I think that you're right. If there's some, like, if it's because of just the, you know, the darkness of the image or something, then you would expect, mm-hmm, mm‑hmm,

0:34:32 Thomas: Yeah. I could imagine a lot of reasons, you know, just that people's eyes, well, like I don't, I'm not a a color perception scientist, but I could imagine like, just, there could be some baseline reason either way.

0:34:44 Lindsey: No, no, absolutely. Yeah. And so, yeah, so you're absolutely right. That's why it's so important to have that control condition because then you can sort of see whether factors like that are really responsible or whether it's about race, right?

0:34:49 Thomas: Well, long story short is depressing that if we see like a knife or something, we look at the black face. That is a wow.

0:35:01 Lindsey: Yeah. Yeah so These results definitely converge with study one. It's indicating that there's an implicit association between these two categories, black and crime. Priming one makes the other concept more active and that this is influencing visual processing at this kind of automatic level. So study three, uh, they use a similar procedure, study two, but instead of priming people with crime-related objects, they instead primed them with objects relating to a positive stereotype about black men, which is that they're good at basketball ball. And yeah, I know.

0:35:03 Thomas: Yikes. Okay.

0:35:36 Lindsey: So they're trying to establish here that the effect is related to like the specific content of racial stereotypes rather than just like an overall negativity or something like that. Yeah. So the participants were either primed with basketball relevant words or jumbled versions of those words. So that's essentially, it's non-word string and It should not prime basketball, but it helps to standardize like the length of the words and stuff. So I, they prime them with those words, and then they have them do a dot probe task that's the same as the last study. So pictures of two faces, dot replaces it, and they have to find the dot. And they found that when basketball was primed, people were significantly faster to identify the location of the dot when it replaced the black face than when it replace the white face. And so this demonstrates that it's not just priming some, like, negative association that's explaining this attentional bias to black faces. It really does seem to be about stereotype-relevant content, which is interesting. Um, now these last two studies, this is, this is good stuff, right? Good in a manner of speaking, right? So

0:35:44 Thomas: Hmm. Okay, mm-hmm. Yeah. It's good science.

0:36:35 Lindsey: Yeah, yeah, it is. Um, so in study four, they essentially did a replication of study two, so they're priming crime and then looking at attentional bias toward black and white faces, but there's two important differences here. So the first one is that they used police officers as participants, and the second is that they gave a surprise memory test after the dot-probe task for the faces that appeared in the task at the end of the experiment to see if attentional biases affect memory for faces.

So yeah, so after they did the dot probe task with black and white faces being replaced by dots, each participant looked at two lineups, one that had five black faces in it and one that had five white faces in it, and in each lineup there was one face that had actually appeared during the dot probe task and four distractors, and the officers were asked to select which of the five faces had appeared in the dot-probe task. Right, so this is interesting, and also I think it's relevant to describe how they selected these faces. So they had one face that they selected from each of the dot-probe tasks, and then they also selected faces that were, were more and less stereotypically black or white than the target face. And so these errors, if the participants make errors in selecting somebody other than the target face, they can essentially show whether their errors are systematic in some direction. Like, are they systematically picking more stereotypical black faces, or is it just like random errors around the target face? So, yeah.

0:38:07 Thomas: Jeez.

0:38:08 Lindsey: Yeah. So they replicated the results of study two. After being subliminally primed with crime, officers were faster to find the dot when it replaced a black face compared to a white face. And in the surprise memory test, officers were more likely to falsely identify a black face that was more stereotypically black than the actual face that appeared in the task. Yeah.

0:38:28 Thomas: You knew it was coming. You know, it was kind of like, I don't know. I

0:38:31 Lindsey: Yeah. So that was true overall. And the effect got even stronger if they had been primed with crime, And they did not find the same effect for white faces. So, yeah. So in conclusion, officers, you know, they attend more to the black faces after being primed with the concept of crime, and their memory for the specific faces that they saw were not more accurate as a result of that prime. And you could have, that could have been the case, right. You could have predicted that because they spent more time looking at those faces, but that wasn't the case. Instead, they they identified faces that were better matches to sort of the stereotypic category of black faces after seeing that prime.

0:38:43 Thomas: Jeez.

0:39:08 Thomas: You know, one thing you haven't talked about at all, and I don't know that it matters, but have you talked at all about like the race of the perceiver here and whether or not that makes an impact? You know that's.

0:39:08 Lindsey: Yeah.

0:39:19 Lindsey: No, it's a good question. And I think, yeah, I don't want to speak off the cuff about this because I think it does matter, but I can't remember how consistently it matters or how strongly it matters. And it wasn't an analysis that they did in these studies.

0:39:21 Thomas: Oh, God. Oh OK. My, my guess was going to be it doesn't matter because I feel like there have been plenty of studies where it actually doesn't make a difference. But who knows? We can look into that later. And it just to address. I mean, this is a go to two argument for, you know, a go to bad argument for people on the right, which is like what it's black officers, too. So that means it's not racism. You know, like that's the essentially It can't be right. Which is a fallacious argument. You know, like black black people can have negative associations and biases toward black people, just the same way. It's probably the same exact mechanism that they absorb it through culture And through, you know, it's, it doesn't, there's no reason why anyone would be immune from that necessarily.

0:39:42 Lindsey: That's well, yeah.

0:39:56 Lindsey: Mm-hmm. Yes. Yeah. Exactly. No. Yeah, and I am confident that I've seen findings about, like, Black people showing overall negative biases on things like the IAT. Um, but I, again, like I, I doubt that it would look identical across all of these groups because, you know, you're, if you have a lot of like positive experiences with somebody from a particular racial group that has negative stereotypes about it, that's going to weaken the negative stereotype.

0:40:32 Thomas: Sure. Yeah.

0:40:42 Thomas: Plus, just while we're on the topic, know, that that isn't to say that more black representation wouldn't help because this is a vast systemic problem. And having I've heard studies reference that like having a critical mass of black people. So like if you have one black officer, like that doesn't make a difference in the bias of the department. But if you have, like, you know, four or five, some number that's enough to where they, they have more, I think, social influence within the department than that makes a difference in like the the disparate numbers. So like that isn't to say that it wouldn't be beneficial to have more black people in positions of power and leadership and in the police. But just in this one measure, I think it may, you know, maybe comparable from from what I remember. But I guess we can look at that data another time, maybe.

0:40:50 Lindsey: Yes. Right.

0:41:30 Lindsey: Yeah. Yeah. We should. Um, yeah, maybe we can, we can follow up about that. Yeah. Um, I think that's a great point. Okay. Okay, so last study, study five, they directly examined the study in assumption of study four, which was that police officers perceive more stereotypically black faces as being more criminal. They sort of, you know, they presented that as a speculative reason that they found the effect that they did, but they wanted to actually test it. So they presented another officer sample with black or white faces. So each officer only saw faces of one race. And they did this to prevent demand characteristics, because in this study, they're going to, you know, be looking at associations between the stereotypicality of black and white faces and perceptions of criminality, right? And so if they show all the officers a bunch of black-and-white faces and then ask them about stereotypicality and criminality, that's likely to tip them off at like, hey, I think that they're looking at racism here, and that might change their responses.

0:42:26 Thomas: Yeah. Yeah, interesting.

0:42:28 Lindsey: So they made this like a completely like between-subjects design, essentially, essentially. So, so each officer only saw faces of one race. And, uh, each officer answered one of three questions about the faces that they saw. Uh, so a third of the officers asked, answered the question, how stereotypically black or white, if they were seeing white faces, is this face on a scale of one to seven? Uh, a third of the officers were asked, does this person look criminal, guess or no? And then a third of the officers were asked, how much would others find this face attractive on a scale of one to seven? And again, black and white faces received equivalent average ratings on attractiveness. So what did they find? They found that black faces that were rated as being high in stereotypicality were perceived to look more criminal on average than those low in stereotypicality, and they saw the opposite trend for white faces.

0:43:21 Lindsey: So white faces that were rated as more highly stereotypical of a white face resulted in significantly lower judgments of criminality compared to the faces that were judged as low stereotypical. Yeah. And I think it's worth noting that overall, white faces were judged to look less criminal than the black faces. But there doesn't seem to be a significant difference between the low stereotypically white and the low stereotypically black faces, so both of them look to be about the same in terms of how criminal they were judged. So the difference here is really being driven by the high stereotypicality white and high stereotypicality black faces, so high stereotypicality white particularly unlikely to be judged as looking criminal and high stereotypically black faces judged as looking very criminal.

0:43:49 Thomas: Interesting. I wonder if that's how you might, or it partially explains how you might walk right by, say, a Kyle Rittenhouse who's holding a gun, who's about to turn himself in, walk or run right by him because he can't be the shooter kind of thing. Could have something to do with it.

0:44:19 Lindsey: Yep. Yep. I imagine that it makes a difference, right? Yeah. Yeah. So just overall conclusions from the study, right, so across five studies, these researchers are providing, I think, very compelling evidence of associations between black faces and this concept of criminality. They do it from different directions, right. They establish that the associations are present in both undergraduate samples and officer samples. And they indicated that these associations appear to operate at an automatic level and suggest that priming people with either the concept of criminal or black activates the other concept. So I think that these findings support a plausible mechanism for the racial bias that we observed in the sentencing disparity cases or the sentcing disparities studies. And I also think that they could provide a plausible mechanism for the findings we discussed in part one, right, the first person shooter task.

0:44:24 Thomas: Yeah.

0:45:13 Lindsey: If black targets are sort of priming the concept of crime, this could have been at least partially responsible for the fact that, you know, the people were faster to shoot when they were encountering black armed subjects as opposed to white armed subjects, right? Speaks to that that criming thing. So that's my data. And I hope that like this gives folks a sense of kind of like the richness of evidence that has been collected around these questions. They've been examined from sort of different levels of analysis and there's a mix of real world data and these controlled experimental methods and they all seem to converge on the same picture here, which is that implicit biases are an important mechanism by which these kinds of disparities get perpetuated.

0:45:24 Thomas: Yeah, oh, I thought you were gonna say it was that racism was solved in 1968 or whenever they passed the Civil Rights Act, so you weren't gonna say that.

0:46:03 Lindsey: Oh, that's that's what I meant. Did I misspeak? Okay. I'm glad she corrected that.

0:46:05 Thomas: Okay, this is a a huge problem I and it amazes me, you know, 1998 isn't by now a long time ago, especially when you're talking about the death penalty stuff. And, uh, while I think we've made progress, there's still the death felony, so like what the hell, this should have been like, holy crap, look at this. And then, every, like, in a just world we, well, we wouldn't have had to begin with, but this should have been like, wow, oh, immediately we need to, uh, uh, rule that the death penalties is unconstitutional because of the, which they apparently tried to do.

We don't have to grab Andrew for this, but apparently they did in 1972 for four years because it was disproportionate. And then I think maybe enough like Nixon appointees got into the court and they're like, let's undo that. I'm pretty sure. So, oh What's your sense of as a scientist for all, for the non-scientists out there, you what is your senseof like, how strong are these findings? This is something I still struggle with. And I ask you every time I know, but like, I have a hard time knowing when somebody says, or when you say, or when a study says, well, here's a finding. They're more likely to blah. I wish we had some consumer friendly

0:46:19 Lindsey: Yeah. Yep. Mm hmm. Yep,

0:47:14 Thomas: Alert color. Talk about dated references. Something that told me like, oh, this is a huge effect. Earlier you talked about the death penalty one. That was huge. But what's your sense of how strong these effects are? Yeah.

0:47:26 Lindsey: Think that these are real effects. I think that they are consistent real effects and I think that sort of if you're talking about like effect size, I think that that depends on the domain that we're talking Right. Convinced at this point after, you know, obviously, as always, I read a lot more widely than I ended up talking about for these episodes. And I don't think that these are just like fluke effects that have shown up sometimes and aren't replicable. These seem to be reliable, replicable findings, and I think that they're meaningful.

0:47:37 Thomas: Okay.

0:48:00 Lindsey: Very

0:48:24 Thomas: Yeah, it's hard to evaluate all that, I guess. It's hard to say what amount of it, yeah, would lead to what. So maybe you don't need much at all to start this process of a vicious cycle that reproduces itself, you know? Maybe you don't need much baseline rate at all, or maybe you would. I guess that's a tough thing to calculate.

0:48:27 Lindsey: Much.

0:48:42 Lindsey: Yeah, it is, right? I mean, that's the peril of human data in general, right. All these things are just hideously complex systems. There are tons of factors that are interacting to produce these outcomes at any point in time. But yeah, I mean when you look at these sort of highly controlled laboratory studies, it indicates that it does seem to be one factor that's part of that that mix, I guess. Yeah.

0:48:47 Thomas: Yeah.

0:49:02 Thomas: Well, I think we can say we've, you Know, firmly established here on Serious Inquiries Only that racism is a big problem in criminal justice. Have we made that scientific finding clear? All right. Well, thanks as always for the breakdown, Lindsay. I'm looking forward to more science, more episodes, a grim topic, but a really important one. So thanks for taking us through the science.

0:49:12 Lindsey: Hope so. Yeah, yeah, thank you. Yeah yeah thank you

0:49:33 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/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 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.