Fault Lines
28 April 2017

Dueling Judges: A Judge’s Duty To Deal With Implicit Bias

November 16, 2016 (Fault Lines) — Ed. Note: In light of the concern about the impact of racial bias in the criminal justice system, we have “so ordered” Senior Judge  Mark W. Bennett of the Northern District of Iowa and Senior Judge Richard G. Kopf of the District of Nebraska to debate: “Should judges address the issue of ‘implicit bias’ with jurors?” This is Judge Bennett’s argument:

I have immense admiration and respect for my dear friend, Judge Richard Kopf. He is on my top 5 list of best federal trial judges in the nation. That’s even after I “creep” him out, allegedly engage in “consciously manipulating” jurors, and elevate “authoritarianism” over “justice.” Judge Kopf is brilliant, hard-working, transparent, outspoken, and puts devotion to the rule of law over his personal legal views in his decisions. His ruling in the partial birth abortion case was likely at odds with his personal views, and knowingly cost him his probable elevation to the Eighth Circuit Court of Appeals—a position he naturally aspired to from his clerking days. He is that principled and has that kind of unsurpassed integrity.

So much for the love fest because, respectfully, Judge Kopf is seriously under-informed, misinformed, and misguided in his truncated views of implicit bias and my decision to both discuss it with, and instruct jurors about it.

He is right about a few things in his post. I am a “true believer,” and I have discussed with and instructed jurors about implicit bias for more than 8 years in every civil and criminal trial in five jurisdictions spanning from the Middle District of Florida to the District Court for the Northern Mariana Islands. I might add, all without a single objection by a party—including governmental entities, from municipalities to the federal, AUSA’s, and some of the most powerful corporate entities in the world.

Judge Kopf claims “… if we look at actual behaviors such as serving as an actual juror in a real case, there is precious little reason to believe that ‘implied bias’ is at work.” He is partially correct that, in Noble Prize winning Daniel Kahneman’s lingo from Thinking Fast and Slow, there is less concern about implicit bias in System II thinking (long deliberations) than in System I thinking (fast thinking, like the first impression of a criminal defendant), but he is incorrect in stating “there is precious little reason to believe that ‘implied bias’ is at work.”

Indeed, in the only two studies of mock jurors evaluating ambiguous evidence or in misremembering facts, implicit bias strongly and negatively influenced jurors on the outcome of the case for a minority defendant.* More importantly, studies have consistently shown that white jurors find minority defendants “more guilty” on the same evidence than white defendants, and are less able to give them the them the full benefit of the presumption of innocence.**

Judge Kopf writes:

Judge Bennett is convinced that white people, who comprise the great bulk of his jurors and mine, hold unconscious implicit biases against people of color.

Almost right. I would add “most” to the white people—hundreds of studies indicate over 80%, but many people of color also have race-based implicit biases. Many folks of every race, including females, have strong gender implicit biases. The same is true about religion and many other categories that give rise to significant cultural stereotypes—often salient features in cases before us.

Judge Kopf, engages in serious misdirection—shifty running quarterback, Tommy Armstrong, of his beloved hometown Nebraska Cornhuskers, could learn a thing or two. The first, most obvious misdirection, is a critical, and fatal reasoning flaw that comes very early in Judge Kopf’s analysis when he claims: “The academic concept of implicit bias flows from the Implicit Association Test (IAT).” Judge Kopf’s understanding of implicit bias is backward. The IAT purports to measure implicit bias. It is a tool, nothing more. To claim that the psychological cognitive bias, implicit bias, flows from the tool of the IAT is like claiming that planet Mars comes from the telescope that first sighted it, or that cancer cells come from the first microscope that identified them.

Setting aside this flaw, Judge Kopf’s primary, if not sole, criticism of implicit bias is his sincere concern about the validity and reliability of the IAT. He correctly cites two studies by Professor Philip E. Tedlock, and co-authors, that are critical of the IAT. I have read everything Professor Tedlock has written criticizing the IAT. I applaud Tedlock, and his co-authors, for adding their 2 cents worth to this important debate.

However, one should put Tedlock’s views in the context of the broader cognitive psychological community of scholars. The fact is, there are substantially more members of the Flat Earth Society than there are believers in Tedlock’s views on the IAT. The world may be flat and Tedlock may be right, but I’m not buying it. Tedlock may be the next Professor Elizabeth Loftus, who single handedly uncloaked the pseudo-science of “repressed memories” by exposing and proving it to be a complete fraud.  However, I would not bet 2 cents on that happening. Tedlock’s views are widely rejected by thousands of social and cognitive psychologists, neuro-psychologists, legal scholars, and other implicit bias scholars who do empirical work about implicit bias. Tedlock is in such a finite minority of implicit bias thinkers, I bet he sends Judge Kopf a birthday present for citing his work on this blog.

Each person should decide for themselves whether the IAT is a valid and reliable test. I have done my own ongoing due diligence starting nearly a decade ago: reading over 600 scientific journal articles; spending on average 5 hours per week (before and after business hours) reading new scholarly articles; conducting my own national empirical research of state and federal judges and lawyers; publishing four law review articles on the subject with three more in the works; and spending countless hours of in-depth discussions with a dozen or so of the leading implicit bias scholars in the world.

Based on these cumulative efforts, I respectfully disagree with both Tedlock and Judge Kopf. Explaining why would take an entire law review article and then some—this is a blog post. In sum, I firmly believe the IAT is both valid and reliable as a general tool. It is important to recognize that I entered the implicit bias discussion as a complete skeptic, even more skeptical than Judge Kopf. However, my due diligence—which pales in comparison to Judge Kopf’s (tongue firmly in cheek) —changed that.

Let’s assume that both Tedlock and Judge Kopf are right about the IAT. That fact is irrelevant to this discussion. Not only is it irrelevant, it is a world record red herring. I discuss implicit bias with jurors because, more than a century of social science research establishes, beyond all doubt: ingroup favoritism and outgroup hostility; deeply rooted stereotyping of Blacks and other minorities with dark skin with violence and dangerousness going back to slavery; studies demonstrating how stereotyping works and its prevalence; dozens of fMRI studies of facial recognition demonstrating dramatically increased amygdala activity when viewing black faces (our brain’s fear center); and unconscious bias by most whites against most people of color and vice versa, but in different ratios. These findings are what prompted a recent heading in an amicus brief filed in the Supreme Court earlier this year in Buck v. Davis: There Is Deeply Rooted In Our History, Indelibly Stamped in Our Psyche, And Stubbornly Present In Our Culture A Cruel And Dehumanizing Stereotype That Black People Are Uniquely Violent And Dangerous.***

I have never suggested that the IAT can pass a Daubert analysis, that jurors should be required to take one (as some scholars have), or that an IAT test can predict specific future or past behavior. The IAT test is simply what got me interested in implicit bias. I would instruct jurors about implicit bias even if the IAT had never been invented based on my current knowledge of the subject.

Interestingly, Judge Kopf fails to challenge the language I use to instruct jurors on implicit bias:

“As we discussed in jury selection, everyone, including me, has feelings, assumptions, perceptions, fears, and stereotypes, that is, “implicit biases,” that we may not be aware of.  These hidden thoughts can impact what we see and hear, how we remember what we see and hear, and how we make important decisions.  Because you are making very important decisions in this case, I strongly encourage you to evaluate the evidence carefully and to resist jumping to conclusions based on personal likes or dislikes, generalizations, gut feelings, prejudices, sympathies, stereotypes, or biases.”

I am unsure what is “authoritarian,” “consciously manipulating,” “creepy,” or even slightly objectionable about this language. Moreover, the State of California adopted my instruction with only slight variations to it years ago. The ABA is promoting the idea that judges should discuss and instruct on implicit bias with jurors. Despite my “creepiness” and “authoritarianism,” hundreds of state and federal judges have requested my instructions and the videos I show to introduce the discussion of implicit bias to potential jurors. The District of Washington is developing a training video on implicit bias for prospective jurors.

Judge Kopf writes, “In our hypothetical counterterrorism case, shouldn’t the active ‘implicit bias’ judge demand a certification from jurors that their verdict was not driven by a dislike of the government?” I would do that, but in the same way I always do it, by discussing with potential jurors if they have any negative feelings about the federal government. If some do, we would discuss those feeling and I would inquire if those feelings could possibly impact their verdict. If the government wanted a certification, I would likely give one.

The value of the certification shown to prospective jurors at the end of jury selection is that, when they learn they are required to sign their name to it, often one or two express difficulty. For example, I had potential juror a few months ago with a minority defendant on trial. The prospective juror said: “Now that I know I would have to sign my name to it, I don’t think I could. I know my mother would turn over in her grave but I would be more willing to convict the defendant simply because he is Hispanic. I know that’s not right but that’s how I feel.” I thanked him profusely for serving his country this way and politely excused him with the parties’ consent and thanks. The certification doesn’t root out implicit bias, but it does help with explicit biases. The line between the two is often blurry. It is an important part of my “authoritarian” toolkit and passion for rooting out biases of all kinds in potential jurors—which is was my job.

I was momentarily offended and bristled at Judge Kopf’s notion that, by discussing and instructing jurors on subconscious bias, I was authoritarian and subverting justice (I forgave Judge Kopf because that is what friends do, and I love a good pot stirrer). And I do apologize for “frightening” my dear friend. I suspect Judge Kopf should be more frightened by the fact that the only two empirical studies that have given judges IAT tests confirm that judges have greater implicit biases than members of the public. **** If the IAT was as invalid as Judge Kopf implies why would there be a statistical difference between judges and the general public – twice confirmed in unrelated and independent studies with different cohorts of judges?

We see authoritarianism as differently as we understand implicit bias. Intuitively, I suspect it is the authoritarian judge types versus the egalitarian judge types that generally oppose discussing and instructing on implicit bias. Specifically, if there is not a model jury instruction on it, an authoritarian judge isn’t going to use it given it lacks the official imprimatur of the powers that be.*****

Here’s what scares and creeps me out: judges who blindly follow DOJ sentencing recommendations: judges, even after Booker and Gaul, who are wedded to the Sentencing Guidelines (deeply flawed from their inception) and fail to fairly apply the §3553(a) sentencing factors; judges who sleep at night after imposing crack/powder sentences at the 100:1 ratio when there is not a shred of empirical evidence to support that grave injustice; judges, for example, who accept the 30+ specific offense characteristics in the fraud guideline without critical scrutiny, or fail to request the government to put on any evidence they can muster that any of the Guideline SOC’s scored are empirically based—because they are not.

I respectfully suggest that blind adherence to the Sentencing Guidelines, which are not empirically based, when it comes to the federal criminal justice system, is a much greater marker of authoritarianism than discussing implicit biases with jurors. Judges who engage in the above sentencing practices are neither better nor worse than judges who are more “progressive,” for lack of a better term, but they are more status quo oriented and, thus,  in my view, less likely to embrace implicit bias.

I suspect that my instructing on implicit bias has not caused a single offender to serve an extra day in prison. Judge Kopf, I doubt you can say the same thing about judges who engage in the above described sentencing practices. Isn’t it more likely true that judges with a heavy dose of authoritarianism are far less likely to discuss or instruct on implicit biases?

One final point. I recently conducted a national study of federal judges and their blind spot biases and published the results in an upcoming article in the Yale L. J. Forum. I noted that Justice Anthony Kennedy recently penned an excellent definition of the cognitive blind spot bias in judges, without naming it, when he wrote about the unconstitutional failure of a state supreme court justice to recuse himself in a criminal case: “Bias is easy to attribute to others and difficult to discern in oneself.”****** Senior federal district court judges have a statistically stronger blind spot bias than active district court judges and a much stronger blind spot bias than federal magistrate judges, bankruptcy judges and members of the general public. Now this is something to be fearful about.

I thank Judge Kopf for raising these important questions and for his unfailing commitment through his extraordinary openness to discuss difficult issues to enhance our system of justice.

* Justin D. Levinson, Forgotten Racial Equality: Implicit Bias, Decisionmaking, and Misremembering, 57 Duke L.J. 345 (2007); Justin D. Levinson & Danielle Young, Different Shades Of Bias: Skin Tone, Implicit Bias, and Judgments of Ambiguous Evidence, 112 W. Va. L. Rev. 307 (2010).

** Justin D. Levinson, Huajian Cai & Danielle Young, Guilty By Implicit Bias: The Guilty/Not Guilty Implicit Association Test, 8 Ohio St. J. Crim. L. 187 (2010). (The study established that there is implicit racial bias in being able to give black defendants the same benefit of the presumption of innocence that white defendants receive. Specifically, study participants held implicit associations between black and guilty as compared with white and not guilty.)

*** Brief for the National Black Students Assoc. as Amici Curiae In Support of Petitioner,  Buck v. Davis, 2016 WL 4073688, at *xx (No. 15-8049) (July 29, 2016 U.S.).

**** Jeffrey J. Rachlinski et al., Does Unconscious Racial Bias Affect Trial Judges?, 84 Notre Dame L. Rev. 1195 (2009) and Justin D. Levinson, Mark W. Bennett & Koichi Hioki, Judging Implicit Bias:  A National Empirical Study of Judicial Stereotypes  69 Fla. L. Rev. (forthcoming January 2017)

***** I served on the Eighth Circuit’s Model Jury Instruction committee for a decade, but resigned because I never used the model instructions and couldn’t move the committee in the direction of the plain English bullet point jury instructions I use.

****** Williams v. Pennsylvania, 136 S. Ct.1899, 1905 (2016). See my article Mark W. Bennett, Implicit Racial Bias in Sentencing: The Next Frontier, 126 Yale L.J. Forum (forthcoming 2017), you can find it pre-publication on SSRN.

12 Comments on this post.

Leave a Reply



By submitting a comment here you grant this site a perpetual license to reproduce your words and name/web site in attribution.

  • CLS
    16 November 2016 at 9:57 am - Reply

    Judge Bennett:

    This is the best Fault Lines debate so far.

    Thank you and Judge Kopf for slugging this one out.

  • Richard G. Kopf
    16 November 2016 at 12:05 pm - Reply

    From Kopf:

    Three responses to Mark’s persuasive post.

    First, I did not ignore Mark’s jury instruction, I just didn’t discuss it. I found it benign and not worth arguing about. That said, it is too long. The Eighth Circuit Model Instruction (3.02) on this point is moderately better because it is short and makes the injunction strongly: “Do not allow sympathy or prejudice to influence you. The law demands of you a just verdict, unaffected by anything except the evidence, your common sense, and the law as I give it to you.”

    Second, Mark’s suggestion that those who doubt “implicit bias” are members of the flat earth society is not much of an argument. (Take a look at the background of the authors cited in my post disputing the validity of IAT and application of the thesis in the courts.) Indeed, Mark seems even to imply that the IAT might not pass Daubert analysis.

    Third, the assertion that jurors act on “implicit biases” cannot be falsified—proven to be untrue. The inability to falsify the assertion makes the assertion unscientific (at least if you believe the preeminent philosopher of science, Karl Popper). For example, in the Levinson article that Mark cites, the author frankly admits (much to his credit) that mock juries and other such rough simulations cannot replicate what really goes on during a trial:

    “The methodology used in the study imperfectly matched true legal processes. For example, the study used stories that resembled some aspects of real-life cases, but did not use real cases. The study also tested university study participants, not real jurors (or judges). Rather than participating in a trial, participants simply filled out a multipart survey. This survey presented the facts from only one perspective, and
    was quite far from matching true trial processes, which allow for exposure to multiple witnesses’ perspectives as well as the opportunity for attorneys to cross-examine witnesses.”


    Finally, let me conclude my comment this way:

    (1) Mark Bennett is perhaps the best active or senior federal trial judges in the nation. He is certainly one of the finest human beings who presently serve on the federal bench. Additionally, he is a scholar of the first rank. I am lucky indeed that he is my friend.

    (2) As far as our posts are concerned on the subject of “implicit bias” and how trial judges should deal with jurors on that subject, the reader is advised to view our writings with the penetrating skepticism of the mean ass editor of Fault Lines.

    All the best.


  • Mark Bennett
    16 November 2016 at 2:50 pm - Reply

    Dear Rich,

    Let me clarify three points. I did not mean to imply that you or other doubters are members of the Flat Earth Society, although I would be happy to purchase a lifetime membership for you😄 What I thought I wrote was they have more members than there are scholars who doubt the IAT. As to the IAT and Daubert, it would pass on the general proposition of its reliability, IMHO, without more data than I am aware of regarding its validity. It would not pass to show a specific test result predicted a past or future decision. I could not disagree with you more about the model instruction you cite except I agree it is shorter. Unless there is extensive discussion of what it means in the context of priming and stereotypes, with respect, it is worthless. All my best.

  • Brian
    16 November 2016 at 6:33 pm - Reply

    I only have two questions – what happens to those certifications after voir dire? What about after the trial?

  • Mark Bennett
    16 November 2016 at 9:50 pm - Reply

    The certification is shown to the jurors during jury selection and is the last part of the verdict form that all jurors sign to show the verdict is unanimous and that they reached their verdict without discrimination. It is part of the public record.

  • Mark Bennett
    16 November 2016 at 9:52 pm - Reply

    Thanks, it’s always informative to learn of disagreements with someone I hold in such high regard.

  • Peter Gerdes
    16 November 2016 at 10:28 pm - Reply

    Two comments:

    First, regarding the IAT I don’t think the consensus is as strong as you believe but even if it is one has to be careful about what the consensus is claiming. In particular you need strong evidence that the IAT correlates with observed bias in deliberative decision making. But let’s take that for granted for the moment.

    If that is true the fact that IAT bias can’t be waved away by pointing out the danger of bias or any explicit warning of this kind means that the very evidence you cite undermines the efficacy of the kind of solutions you consider.

    Second, the instruction mentioned either collapses into uselessness (don’t use race/gender/etc.. except when it is objectively appropriate) or rules out critical reasoning the jury must be able to engage in.

    We need juries to reason as follows: That neighborhood is almost entirely black, witnesses saw a white man fleeing and a white man was picked up a couple blocks away matching the description. Since there are so few white men in the area it’s pretty likely they picked up the same guy witnesses saw. But logically speaking using the fraction of Quakers who carry guns or the fraction of blacks/whites who carry guns uses the exact same kind of reasoning.

    Maybe what you mean to exclude is using one’s prior probability that someone of race/religion/etc.. X committed the crime at issue in this courtroom. That is more defensible but still troubling. For instance surely juries should use their priors about the rarity of stranger sexually motivated female on male rape when evaluating such claims. If you say no you basically exclude all folk psychology and the jury can’t even evaluate motives.

    I guess I’m suggesting that maybe you should just say “please be aware of the possibility of racial/etc.. bias” rather than using anything like a certification.

  • traderprofit
    16 November 2016 at 10:41 pm - Reply

    “Moreover, the State of California adopted my instruction with only slight variations to it years ago”

    Oh, well that certainly proves…nothing

    • Mark W. Bennett
      17 November 2016 at 6:54 am - Reply

      Dear Tradeprofit,
      So tell us how many instructions have you had approved by other states, the ABA, and hundreds of other judges? Best

  • Mark Bennett
    17 November 2016 at 7:59 am - Reply

    Peter, thanks for your comments. I invite you and any one else to write an article establishing the consensus is not as strong and consistent as I suggest. Secondly, while debiasing techniques that have been empirically verified are few and far between there is virtually uniform consensus that awareness is a critical first step. My instruction was carefully vetted and revised by a dozen of the leading IB scholars in the world. So, with respect, I value their opinion more than yours but appreciate yours as well.

  • Keith
    21 November 2016 at 12:09 pm - Reply

    Judge Bennett,

    “I instruct before opening statements in all civil and criminal cases, and the certification is part of the verdict form at the conclusion of the instructions, and I get that to the lawyers usually 7-10 days before trial, there is ample opportunity to raise issues relating to the certification well in advance of trial. No lawyer has ever done so but I would welcome them to.”

    Is there a risk that, having knowledge that this is something you deeply believe in, the attorneys might be skittish about coming forward with objections to having the warning mentioned? Have you ever enlisted the attorneys in the process of deciding which biases should / shouldn’t be covered or do you typically send them a prepared statement and ask for comment? Do you treat sample jury instructions in a similar way (giving them a sample and asking for comment vs eliciting samples from the parties)? Do you think counsel would be in a good (better) position to know which biases may affect their client more than others than the Judge (especially if it’s not you)?

    “Now that I know I would have to sign my name to it, I don’t think I could. I know my mother would turn over in her grave but I would be more willing to convict the defendant simply because he is Hispanic. I know that’s not right but that’s how I feel.”

    If people have an innate inherent bias they are aware of, have you considered that person may be able to fulfill the logical part of his / her duty in a better way than someone who is unable or worse, unwilling, to understand the role the bias plays in the process? In other words, are defendants actually better off by automatically striking such jurors? Do you give them the opportunity to keep them?

    If someone refuses to sign the certification, are they always struck for cause?

    Judge Kopf: “By the way, “it can’t hurt” is not an adequate answer because it just might hurt. That’s why doctors are so conservative when treating people–first do harm they are taught. We should never forget that caution.”

    Judge Bennett: “I suspect that my instructing on implicit bias has not caused a single offender to serve an extra day in prison.”

    Judge Kopf,

    Can you articulate what the harm is here and who is being harmed? If they make it less likely to convict someone, is there harm? Is there a chance that bringing up race can make it more likely someone would convict? In other words, can the act of measuring, affect the thing being measured?

    • Mark W. Bennett
      22 November 2016 at 11:42 am - Reply

      Dear Keith,

      Great questions. sorry for my delay in answering. Here are my responses:

      “Is there a risk that, having knowledge that this is something you deeply believe in, the attorneys might be skittish about coming forward with objections to having the warning mentioned? Have you ever enlisted the attorneys in the process of deciding which biases should / shouldn’t be covered or do you typically send them prepared statement and ask for comment? ”

      There is always some risk but in my experience good trial lawyers are never, nor should they be, shy about making a record. If you are shy you should have gone to dental school. I think good trial lawyers are always in abetter position to make strategic judgments about their cases than I am. I welcome their input.

      “If people have an innate inherent bias they are aware of, have you considered that person may be able to fulfill the logical part of his / her duty in a better way than someone who is unable or worse, unwilling, to understand the role the bias plays in the process? In other words, are defendants actually better off by automatically striking such jurors? Do you give them the opportunity to keep them?”

      Yes, I have considered this and I agree with you in a close case the lawyers should make the call. In the example I gave I did some follow up questioning of the potential juror and encouraged the lawyers to do so before I struck the juror. If I were a criminal defense lawyer I might want a juror like this. Judge Kopf and I were chatting about this yesterday and we both agreed this juror might be terrific and better than jurors who don’t confront their biases.

      “If someone refuses to sign the certification, are they always struck for cause?”

      Yes. The more interesting question would be if they are selected and then refuse to sign it on the verdict form. I hope I don’t face that issues because I can see excellent arguments on both sides of it.