In the U.S. criminal justice system, a court’s discussion of mental health diagnoses can perpetuate harmful stereotypes about such conditions. Recently, in Glossip v. Oklahoma,1 the Supreme Court ordered a new trial in a death penalty case featuring a prosecution that Oklahoma admitted was flawed.2 The Court agreed that the prosecution failed to correct a material misrepresentation that, in part, concealed how Justin Sneed’s bipolar disorder diagnosis, combined with his history of drug use, demonstrated a proclivity for violence and impulsivity.3 That justification may have adverse consequences for criminal defendants and witnesses with mental illnesses by suggesting the diagnosis of a mental health condition is sufficient to demonstrate violent proclivities, even though individuals with the same diagnosis experience different symptoms.
In 1997, Sneed killed Barry Van Treese with a baseball bat in a hotel Van Treese owned.4 Richard Glossip lived in and managed that hotel.5 Glossip eventually confessed to police that Sneed had admitted to killing Van Treese, despite prior inconsistent statements about Sneed’s role in the murder.6 Sneed flipped the script, asserting that he committed the murder at the behest of Glossip, who had “offered him thousands of dollars to” kill Van Treese.7 Based on Sneed’s account, Oklahoma brought charges against Glossip for capital murder.8 Glossip, however, “maintain[ed] his innocence,”9 refusing a plea deal that would have allowed him to avoid the death penalty if he had testified against Sneed.10 Prosecutors then offered Sneed the same deal to testify against Glossip, which Sneed accepted.11 “Sneed’s testimony was the only direct evidence” tying Glossip to the murder.12 At trial, the jury convicted and sentenced Glossip to death.13 The Oklahoma Court of Criminal Appeals (OCCA) reversed unanimously, holding that because any “evidence at trial . . . corroborat[ing] Sneed’s testimony was extremely weak,” Glossip’s “counsel’s failure to cross-examine Sneed on his . . . inconsistent statements” constituted ineffective assistance of counsel.14
Later, “[i]n 2004, . . . Glossip rejected another plea offer,” and Oklahoma tried him again.15 At the second trial, multiple witnesses confirmed Glossip’s story, and Glossip’s counsel proved “that Van Treese had been attacked with a knife . . . [in addition to] a baseball bat.”16 “The prosecution also asked Sneed whether anyone had prescribed him any medication” after his arrest.17 Sneed stated: “When I was arrested I asked for some Sudafed because I had a cold, but then shortly after that somehow they ended up giving me [l]ithium for some reason, I don’t know why. I never seen no psychiatrist or anything.”18 He also testified “that he used illegal drugs including marijuana and . . . [methamphetamine]” twice weekly before he got arrested.19 The prosecution’s closing argument portrayed Sneed as having “no propensity to violence except at Glossip’s direction.”20 Glossip was again convicted and sentenced to death.21 This time, the OCCA affirmed and found that circumstantial evidence about Glossip’s mismanagement of the hotel, together with Glossip’s initial dishonesty after the murder, corroborated Sneed’s account.22
As “concerns over the integrity of Glossip’s conviction” mounted, a group of Oklahoma legislators hired a law firm to independently investigate Glossip’s case.23 In 2022, the firm “reported . . . ‘grave doubt’” about Glossip’s conviction,24 highlighting that “the prosecution had . . . destroyed ‘key physical evidence’ before Glossip’s” second trial.25 After this report, and almost twenty years after Glossip’s second conviction, Oklahoma disclosed seven boxes of documents from Glossip’s trials that had previously been withheld.26
Among the documents in these boxes was a note from Connie Smothermon, the head prosecutor. Glossip alleged the note provided evidence of witness tampering.27 Smothermon sent this note, discussing Sneed’s testimony, “to Sneed’s lawyer before Sneed testified at the second trial.”28 The note complained that the knife remained the “biggest problem” as Sneed had told “the police that the knife fell out of his pocket and that he didn’t stab the victim with it.”29 Yet the victim nonetheless had knife-inflicted lacerations.30 Smothermon expressed confusion with how Sneed on his own could have simultaneously controlled both the knife and the bat.31 She noted that they “should get to [Sneed] [that] afternoon.”32 Glossip filed a motion for postconviction relief, arguing that Smothermon had violated the rule of sequestration by interfering with Sneed’s testimony.33 The OCCA held that “Glossip’s claims were procedurally barred [and] meritless.”34
Soon thereafter, Oklahoma discovered “an eighth box of . . . documents,”35 which included evidence Glossip alleged showed the prosecution “fail[ed] to correct Sneed’s false . . . testimony.”36 Box eight contained “notes handwritten by Smothermon during a pretrial interview with Sneed” in which Sneed told Smothermon that he was on lithium not by accident,37 as he had claimed during Glossip’s second trial,38 but “in connection with a ‘Dr. Trumpet.’”39 At the time Sneed was incarcerated, “[o]nly a single psychiatrist,” Dr. Larry Trombka, worked in Sneed’s facility.40 A now-disclosed summary of Sneed’s medical records also revealed that Sneed had been diagnosed with bipolar disorder and prescribed lithium to treat the disorder.41 He also “had a history of . . . angry outbursts[] and substance abuse.”42
On review of the evidence, Oklahoma’s Attorney General determined that Smothermon had “elicited false testimony from Sneed,”43 so he “disclosed Box 8 to Glossip and retained an independent counsel to . . . review . . . Glossip’s conviction.”44 That investigation “concluded that . . . [Smothermon’s] failure to turn over Sneed’s statements about his mental health treatment violated Brady v. Maryland,”45 and Smothermon’s “failure to correct Sneed’s false . . . testimony” about the lithium violated Napue v. Illinois.46 The independent counsel also recommended that the OCCA overturn Glossip’s conviction.47 Glossip then filed another petition for postconviction relief, asserting claims under Brady and Napue, along with cumulative error and actual innocence claims.48 However, “[t]he OCCA denied Glossip’s . . . petition” — though unopposed — “without a hearing.”49 “Th[e Supreme] Court the[n] stayed Glossip’s execution,” on a motion supported by Oklahoma, and granted a writ of certiorari to review the OCCA’s decision.50 Oklahoma did not dispute Glossip’s claim, so “the Court appointed . . . [an] amicus curiae to” argue on behalf of the OCCA’s decision.51
In February 2025, the Supreme Court reversed the OCCA.52 Writing for the majority, Justice Sotomayor53 first concluded that the “Court . . . ha[d] jurisdiction to review the [OCCA’s] judgment.”54
The Court then held that the State had “violated its constitutional obligation” under Napue “to correct false testimony.”55 In Napue, the Court held that a conviction violates the Due Process Clause of the Fourteenth Amendment when it is “knowingly ‘obtained through use of false evidence.’”56 Relevant here, “a new trial is warranted” under Napue if the defendant can show that the prosecution allowed material false testimony “to go uncorrected when it appear[ed].”57 The burden is on the government in this case to show, beyond a reasonable doubt, that the misstatements are not material; that is, that “the error complained of did not contribute to the verdict obtained.”58 The Court noted that “Oklahoma’s attorney general join[ed] Glossip in asserting a Napue error,” before establishing “that Sneed’s testimony was false” because it was inconsistent with his medical records.59 The Court also found sufficient evidence to establish “that the prosecution knew Sneed’s statements were false” at the time of his testimony.60 To reach that conclusion, the Court relied on the prosecution’s access to Sneed’s medical file, which referenced Sneed’s “lithium prescription and . . . bipolar diagnosis”; Smothermon’s “pre-trial conversation with Sneed at which he mentioned ‘lithium’ and ‘Dr. Trumpet’”; and Sneed’s testimony at trial, which was inconsistent with that evidence.61
After concluding that there was false testimony and that the prosecution knew about it, the Court determined that the testimony was material such that a Napue violation had occurred.62 The Court concluded that Sneed’s false testimony was material for two separate reasons: (1) The fact that Sneed lied could have undermined his credibility, and (2) the fact that he was diagnosed with bipolar disorder could have frustrated the prosecution’s theory that he was nonviolent without Glossip’s encouragement.63 First, the Court emphasized that Sneed’s “credibility plainly would have suffered” if the prosecution had corrected him.64 The jurors could have convicted Glossip only if they believed Sneed’s story, suggesting that his credibility was material.65 Indeed, Justice Sotomayor clarified that “[e]vidence can be material even if it ‘goes only to the credibility of the witness.’”66 But, second, Justice Sotomayor went a step further by suggesting that “Sneed’s false testimony also bore on Glossip’s guilt in a more direct way.”67 She emphasized that Sneed’s bipolar disorder diagnosis and lithium prescription “would have been . . . important . . . for the defense to know,” since some of the symptoms of bipolar disorder “can be exacerbated by illicit drug use, such as methamphetamine.”68 Because the prosecution was portraying Sneed as a harmless puppet controlled entirely by Glossip, “there [was] a reasonable likelihood that correcting Sneed’s testimony would have affected the judgment of the jury.”69
Justice Barrett concurred in part and dissented in part.70 She agreed that the Court had jurisdiction to review the case and that the OCCA misstated Napue as a matter of law.71 However, in disagreement with the Court, she noted that she would not have ordered the OCCA to vacate Glossip’s conviction.72 Instead, she argued that the case should have been remanded for further proceedings.73 Traditionally, the Court’s appellate function is limited to matters of law, but, Justice Barrett argued, here the majority improperly drew its own factual determinations based on the record.74
Justice Thomas filed a dissenting opinion.75 He first disagreed with the Court’s jurisdictional conclusion.76 He argued that the OCCA had an “independent state ground” for its decision, which served as a “‘jurisdictional’ limitation” for the Court.77 Despite the State’s confession to a Napue violation, the OCCA found that Glossip had failed to satisfy other prerequisites for postconviction relief under Oklahoma’s Uniform Post-Conviction Procedure Act.78 Justice Thomas would have concluded that “independent state ground bar[red]” relief from the Court.79
Next, Justice Thomas disagreed with the Court’s Napue analysis, arguing that Sneed’s false statements about lithium were not material.80 He reasoned that those misstatements could not have changed the jury’s verdict.81 To Justice Thomas, the misstatements “did not bear on any contested issue.”82 In fact, “the jury already knew that Sneed had been prescribed lithium, used illegal drugs, and behaved impulsively.”83 The jury also knew that Sneed had been violent: “[H]e admitted that he beat a man to death with a baseball bat in the middle of the night with no advanced planning.”84 In other words, even when the prosecution failed to correct the misrepresentation, the jury already knew what the Court claimed the bipolar disorder and drug use could show. With existing evidence that Sneed had behaved impulsively and violently, other evidence of impulsivity and violence seemed unlikely to affect the jury’s conclusions.85 Justice Thomas also emphasized that defense counsel “knew that Sneed . . . suffered from” mood disorders and bursts of anger that were treated by lithium “[a]s early as 1997,” yet “elected not to raise Sneed’s mental [health] condition at [Glossip’s] second trial.”86 A pretrial competency evaluation by the forensic psychologist Dr. Edith King resulted in a report in which Dr. King determined that Sneed qualified as a “mentally ill person or a person requiring treatment” with “an atypical mood swing disorder in his past characterized by ‘ups and downs’ including anger outburst[s].”87 In other words, any evidence related to Sneed’s bipolar disorder did not introduce anything new that related to “any contested issue”;88 thus, the concealed evidence could not have been material to the conviction.
The majority’s conclusion that Sneed’s bipolar disorder diagnosis was material because it “would have undermined the prosecution’s theory that Sneed was harmless”89 was incorrect or misleading. Evidence of Sneed’s dangerousness and impulsivity was already in the record, so the majority’s conclusion requires one of two possible explanations. Either the majority was wrong, and the misrepresentation was not material at all under Napue since it could not have impacted the jury’s conclusions. Or the bipolar disorder diagnosis and illegal drug use were material for a reason different from the one that the majority fronted: Perhaps they could have impacted the jury’s assessment for reasons related to harmful stereotypes that a person may be dangerous or unreliable solely because of their mental health disorder and substance use.
By determining that a Napue violation occurred, the Court proclaimed that the evidence about Sneed’s bipolar disorder and methamphetamine use was so important that it may have impacted the jury’s conclusions. The Attorney General of Oklahoma conceded that Sneed’s testimony that he had been given lithium for a cold and had “never seen no psychiatrist or anything” was untrue.90 The record showed that Sneed was diagnosed with bipolar disorder and prescribed lithium,91 and Smothermon’s notes arguably suggested that Sneed told her these facts.92 But a Napue violation requires something more than the prosecution’s awareness of false testimony.93 The evidence must also be material.94 The party benefiting from an alleged material misstatement has a high bar for defending the misstatement under Napue: To prevent the defendant from obtaining a new trial, the party must “prove beyond a reasonable doubt that the error . . . did not contribute to the verdict obtained.”95 Thus, the defendant need only show that there’s a reasonable doubt the misstatement “may have had an effect on the outcome of the trial.”96 The Court has emphasized that the determinative question is whether the false testimony “in any reasonable likelihood [could] have affected the judgment of the jury.”97
The Glossip majority’s first reason for the materiality of Sneed’s statement is well founded; however, the second lies on a shakier ground of stereotypes about mental illness. To be sure, Justice Sotomayor rightly emphasized how Sneed’s credibility would have suffered if the prosecution had corrected his false testimony, since “[a] lie is a lie.”98 Thus, the Court’s first materiality assessment made sense insofar as Sneed’s lie on the stand could have demonstrated to the jury his propensity to lie about other topics on the stand — like Glossip’s role in the killing. However, Justice Sotomayor’s separate basis for concluding that the testimony was material outside of its bearing on Sneed’s credibility reflected inaccurate assumptions about how mental health conditions necessarily impact an individual and his behavior. By finding that Sneed’s “diagnosis with a disorder that could trigger impulsive violence when combined with drug use . . . would have undermined the prosecution’s theory that Sneed was harmless on his own,”99 Justice Sotomayor at least claimed that Sneed’s bipolar disorder diagnosis and illicit drug use could have impacted the jury’s conclusions because of their tendency to show some sort of dangerousness, impulsivity, or violent behavior.
This second basis for materiality exposes Justice Sotomayor’s improper assumption that those with mental health diagnoses and a history of drug use would be more likely to commit violence than those without them. This conclusion is supported by Justice Sotomayor’s questions in oral argument. She noted:
[T]he issue wasn’t about him taking lithium. The issue was about why he was taking the lithium. . . . And so the fact that the jury knew he had taken lithium during incarceration doesn’t tell them anything about whether he had . . . a bipolar condition, that his use of drugs would have led to impulsive and violent behavior . . . .100
By suggesting that the jury had to know if Sneed had “a bipolar condition,” Justice Sotomayor suggested that Sneed’s mental health condition may have impacted the jury’s conclusion. This status could be relevant only if it could lead the jury to conclude that Sneed was even more dangerous than they would have already concluded. Moreover, Justice Sotomayor’s language that illicit drug use “would” have resulted in violent behavior suggests that all individuals with bipolar disorder diagnoses would have behaved violently having taken methamphetamine. She could have used the word “may,” indicating a possibility of more violent behavior. She didn’t, highlighting how she incorporated a monolithic view of mental health symptoms into her materiality consideration.
This reasoning could lead to pernicious arguments that individuals living with addictions or mental health disorders are more violent or impulsive solely because of their conditions and substance use: claims that are empirically untrue.101 Though one recent study showed that up to seventy-five percent of the U.S. population believes “that people with mental illness are violent,” this is a false assumption.102 At the very least, the claim that those with mental illnesses are more likely to be violent than their neurotypical counterparts “remains unclear.”103 For instance, though many previous studies have claimed “that bipolar disorder . . . in adults is associated with aggressive behaviors,”104 many are empirically limited as they studied only inpatients and thus may neither reflect the experience of bipolar disorder outside medical facilities nor account for “possible confounding factors.”105 On the other hand, more recent research has established that individuals with bipolar disorder are not inherently violent.106 If the Court, then, found Sneed’s misrepresentation to be material based on such a premise, that premise would be faulty and would perpetuate the false and widespread belief that mental illness, such as bipolar disorder, is a sufficient condition for violence.
If that same faulty premise were to infect other legal conclusions, it would have serious impacts on people with mental health conditions more generally. Other courts have reached similar faulty conclusions in the past about diagnoses of mental health conditions being sufficient for dangerousness — and those conclusions almost always have led to harmful outcomes for those with the mental health disorders. For instance, courts have argued that “[i]t is not unreasonable for the State to believe that a person with a mental disorder of a sexual nature is qualitatively more dangerous than another mental patient who nonetheless threatens danger to himself or others.”107 This conclusion has made treating individuals with certain sexual mental health disorders “differently from other civilly committed persons” permissible.108 In other words, certain mental health conditions justify a treatment of an individual as “dangerous,” even absent evidence of such behavior — and in spite of the common knowledge that different individuals with “the same diagnosis [will] experience different combinations of . . . symptoms” and behave differently in response to those symptoms.109 Further, after individuals with mental health conditions are convicted, there are clear pathways to prison sentences that are upward departures from the federal sentencing guidelines.110 Thus, if Glossip expands the false association between mental health conditions and violence in the public perception, it can legitimize further harmful archetypes.111
In Glossip, the Court threatened to expand this dangerous precedent. The majority clarified that it is permissible to categorize someone as more dangerous solely because of their mental health condition and history of drug use. In the criminal justice system, a significant portion of defendants have a mental health condition: Official government statistics found that “[a]bout [forty-three percent] of state . . . prisoners had a history of a mental health problem,”112 compared to 18.3% in the general population.113 And just over half “of inmates meet the criteria for drug abuse or dependence.”114 With such a large proportion of those impacted by the criminal justice system thus vulnerable to status-based stereotypes, the Glossip Court’s reasoning perpetuates the harmful and inaccurate stereotype that people with mental health or substance use disorders are always more “dangerous” than the rest of the population.