No. 24-3086

Appeal from the United States District Court for the Southern District of Ohio at Columbus. No. 2:20-cv-06038—Algenon L. Marbley, District Judge.

Argued: December 11, 2024

Decided and Filed: November 20, 2025

Before: GILMAN, READLER, and BLOOMEKATZ, Circuit Judges.

ARGUED: Laren E. Knoll, THE KNOLL LAW FIRM LLC, Dublin, Ohio, for Appellant. Jeffrey S. Hiller, LITTLER MENDELSON, P.C., Columbus, Ohio, for Appellees. ON BRIEF: Laren E. Knoll, THE KNOLL LAW FIRM LLC, Dublin, Ohio, Daniel I. Bryant, BRYANT LEGAL, LLC, Columbus, Ohio, for Appellant. Jeffrey S. Hiller, Andrew Klaben-Finegold, LITTLER MENDELSON, P.C., Columbus, Ohio, for Appellees. Georgina C. Yeomans, EQUAL EMPLOYMENT OPPORTUNITY COMMISSION, Washington, D.C., for Amicus Curiae. ON PETITION FOR REHEARING: Jeffrey S. Hiller, Andrew Klaben-Finegold, LITTLER MENDELSON, P.C., Columbus, Ohio, for Appellees. ON RESPONSE: Laren E. Knoll, THE KNOLL LAW FIRM LLC, Dublin, Ohio, Daniel I. Bryant, BRYANT LEGAL, LLC, Columbus, Ohio, for Appellant. ON AMICUS BRIEF IN LIGHT OF THE PETITION FOR REHEARING: Benjamin M. Flowers, Shams H. Hirji, ASHBROOK BYRNE KRESGE FLOWERS LLC, Cincinnati, Ohio, for Amicus Curiae.

BLOOMEKATZ, J., delivered the amended opinion of the court in which GILMAN, J., concurred. READLER, J. redelivered his original dissenting opinion (pp. 18–43) and delivered an addendum (pp. 44–56) in light of the amended majority opinion.

Page 2BLOOMEKATZ, Circuit Judge. Monica Gray helped a colleague secure an accommodation under the Americans with Disabilities Act (ADA). The colleague’s supervisor opposed the accommodation and, a few months later, reported Gray for timecard falsification. State Farm investigated the report and fired Gray. Gray then sued for retaliation under the ADA and Ohio law, claiming that the supervisor singled her out for conduct widespread in the agency because she had helped her colleague advocate for an accommodation. But the district court granted State Farm summary judgment, reasoning that the company had an honest belief that Gray had engaged in misconduct. Because Gray can proceed on a theory of vicarious liability based on the supervisor’s alleged bias, we reverse.

Monica Gray worked at State Farm for fifteen years until she was fired. At the time of her termination, Gray was a claim specialist on a team managed by Chris Martin. Her friend, Sonya Mauter, had also been with State Farm for many years. But she worked on a different team managed by Joe Kyle.

Mauter had an ADA accommodation that exempted her from overtime work. In August 2017, Kyle told Mauter that State Farm would no longer accommodate her work schedule. He placed Mauter on leave until she agreed to work overtime. And if she did not agree to overtime, Kyle threatened, she could eventually be terminated. When Mauter asked to use her leave under the Family and Medical Leave Act, Kyle said she had none left. Mauter later discovered that was not true.

Mauter turned to Gray for help. Gray researched ADA law and State Farm policies, contacted human resources for information, and even lodged an internal complaint against Kyle. Gray coached Mauter on how to advocate for herself. She also advised Mauter to seek legal Page 3counsel and file an EEOC discrimination charge, which Mauter did. Throughout this process, Mauter repeatedly informed Kyle that Gray was assisting her.

Meanwhile, Kyle intensified his scrutiny of Mauter. One time, he issued her a warning for discussing accommodations with colleagues, threatening her with “action up to and including termination.' Warning, R. 30-21, PageID 2101. Mauter's lawyer sent State Farm a letter alleging that Kyle engaged in illegal retaliation. State Farm investigated Mauter's allegations and agreed to transition her to another team.

In November 2017, Kyle substituted for Martin—the supervisor of Gray’s team—while Martin went on a short vacation. At the time, Martin took a relaxed approach to State Farm’s timecard policies. State Farm’s written policy required employees to adhere closely to scheduled shifts and allotted meal breaks. The policy also required employees to record time worked to the minute. But Martin was not a stickler for those rules. He did not monitor or verify his employees’ timesheets, and at least some on his team rounded time rather than reporting it to the minute.

Notwithstanding Martin's relaxed practice, Kyle pored over Gray's timesheets in Martin's absence. He compared her time entries to her computer activity and noticed three instances when she reported time while logged off her computer. Kyle raised the issue with his and Martin’s direct supervisor, Denise Hensley. And as required by company policy, Hensley told Kyle to report the discrepancies to HR. Policy, R. 30-2, PageID 631 (requiring all employees to “immediately” report timekeeping violations that they learn about). Kyle did so, telling HR that Gray had manually adjusted her time to hide long lunches and early departures. He incorrectly claimed that Gray had previously been coached for such behavior. And he suggested that HR would find more discrepancies if it investigated Gray’s records.

When Martin returned from vacation, Kyle met with him, Hensley, and Geri Keeling from HR. Kyle shared the evidence that he had gathered on Gray, prompting Keeling to launch an investigation. Keeling reviewed Gray's timesheets, computer activity, and building-entry records. She found more errors, including seven instances when Gray reported returning from lunch before she had re-entered the building.

Page 4Keeling and Martin interviewed Gray about these discrepancies. This was the first time Gray learned that she was being investigated. Gray denied any wrongdoing and insisted that Kyle had targeted her for helping Mauter secure an accommodation. She also claimed that the timing was suspicious, and she asked if State Farm was reviewing other employees’ timesheets. Keeling relayed Gray’s allegation of retaliation to Hensley, but no one at State Farm pursued the issue further.

A week later, Martin and Hensley recommended Gray’s termination to upper management. Gray did not know about the pending recommendation. In the meantime, she filed a retaliation charge with the EEOC. A few days later, State Farm fired her for falsifying her timesheets.

Gray sued Kyle and State Farm for retaliation under the ADA and its Ohio counterpart. See 42 U.S.C. § 12203; Ohio Rev. Code § 4112.02. She alleged that Kyle selectively reported her for otherwise common behavior, prompting State Farm to investigate and ultimately fire her. After discovery, the district court granted State Farm’s motion for summary judgment. The court reasoned that State Farm held an “honest belief” that Gray falsified her time and fired her for that reason. Op., R. 52, PageID 3743. It thus concluded that Gray could not show pretext, as required for her retaliation claims. Gray timely appealed.

We review the district court’s grant of summary judgment de novo. See King v. Steward Trumbull Mem'l Hosp. , Inc. , 30 F.4th 551, 559 (6th Cir. 2022). We affirm if, viewing the evidence and drawing all reasonable inferences in Gray's favor, Gray has failed to create a genuine dispute of material fact. See id. We reverse if Gray has presented evidence from which a jury could reasonably find in her favor. See id. Contrary to the district court’s opinion, Gray claims that she had such evidence for her retaliation claims.

We analyze retaliation claims under the ADA and Ohio law identically. Id. at 560. Because Gray relies on indirect evidence of retaliation, we apply the McDonnell Douglas burden-shifting framework. Kirilenko-Ison v. Bd. of Educ. , 974 F.3d 652, 661 (6th Cir. 2020). Under that framework, Gray must establish a prima facie case of retaliation. Id. If she meets that threshold, then State Farm must articulate a legitimate, non-retaliatory reason for its adverse Page 5action. Id. If State Farm provides such a reason, then Gray must show that the reason is a pretext designed to mask retaliation. Id. At summary judgment, we examine whether a genuine factual dispute exists at each of these steps. Id.

In this case, Gray invokes both direct and vicarious theories of liability. And that fact affects our analysis to some extent. In the discrimination context, an employer is directly liable for the unlawful motives of its decisionmakers-that is, those who made the challenged employment decision. See Sloat v. Hewlett-Packard Enter. Co. , 18 F.4th 204, 210 (6th Cir. 2021). So for direct liability claims, we focus our McDonnell Douglas analysis on the relevant decisionmakers.

Vicarious liability, commonly known as 'cat's paw' liability, hinges on a neutral decisionmaker acting on a biased subordinate’s input. See Marshall v. Rawlings Co. LLC , 854 F.3d 368, 377 (6th Cir. 2017). The theory thus shifts the motive analysis from the decisionmaker to the subordinate. See Sloat , 18 F.4th at 210-13. And since Kyle is the allegedly biased subordinate here, we use the McDonnell Douglas framework to assess his motives. Id. In this case, we consider Gray’s direct and vicarious allegations together, evaluating her evidence for both theories at each step of our analysis.²

Page 6To establish her prima facie case, Gray must present evidence that (1) she engaged in protected activity, (2) State Farm knew about her activity, (3) she suffered an adverse action, and (4) the adverse action was causally connected to her protected activity. Kirilenko-Ison , 974 F.3d at 661. For the third element, Gray relies on her termination, which unquestionably qualifies as an adverse action. See Burlington N. & Santa Fe Ry. Co. v. White , 548 U.S. 53, 62-64, 68 (2006). We therefore focus on the other elements.

We start by addressing the first two elements: Gray’s protected activity and State Farm’s knowledge of it. Gray says that she engaged in protected activity by helping Mauter “reinstate and secure” her ADA accommodation despite Kyle’s opposition. Mauter Decl., R. 42-2, PageID 3223. We agree. The ADA's anti-retaliation provision protects employees who aid others' exercise of their statutory rights. Rorrer v. City of Stow , 743 F.3d 1025, 1046 (6th Cir. 2014); see also White , 548 U.S. at 63. Gray supported Mauter throughout the ADA process, actively pushing back against Kyle's denial of Mauter's accommodation. Those efforts qualify as protected activity.³

Gray also presents evidence that Kyle and others knew about her involvement. Mauter recalls informing Kyle and Hensley of Gray’s assistance. And Gray says that she told Keeling and Martin that Kyle had targeted her for helping Mauter. Gray thus has evidence that Kyle and at least two of the relevant decisionmakers (Hensley and Martin) knew about her protected activity. State Farm disagrees, but its argument relies on credibility determinations that belong Page 7to the jury. See Kirilenko-Ison , 974 F.3d at 660. We therefore hold that Gray has raised genuine disputes over the protected activity and knowledge elements of her prima facie case.

We next turn to the causation element of Gray's prima facie case. At the prima facie stage, the burden of showing causation is “minimal.” Upshaw v. Ford Motor Co. , 576 F.3d 576, 588 (6th Cir. 2009). The plaintiff must merely 'put forth some evidence to deduce a causal connection” between the adverse action and protected activity. A.C. ex rel. J.C. v. Shelby Cnty. Bd. of Educ. , 711 F.3d 687, 699 (6th Cir. 2013) (citation omitted). Because retaliation takes many forms, our inquiry depends on the facts of each case. Dixon v. Gonzales , 481 F.3d 324, 335 (6th Cir. 2007). Gray alleges a single adverse action: her termination. So her evidence must support “a causal connection” between her termination and her assistance to Mauter. A.C. , 711 F.3d at 699.

We have held many times that employees can establish prima facie causation by showing that their employer began scrutinizing them more heavily shortly after they engaged in protected activity, and then used its findings to justify termination. E.g. , Hamilton v. Gen. Elec. Co. , 556 F.3d 428, 435-36 (6th Cir. 2009); Upshaw , 576 F.3d at 588-89; Adamov v. U.S. Bank Nat'l Ass’n , 681 F. App’x 473, 479 (6th Cir. 2017); E.E.O.C. v. New Breed Logistics , 783 F.3d 1057, 1070–71 (6th Cir. 2015). Gray relies on this theory, arguing that shortly after she helped Mauter, Kyle scrutinized her for timekeeping practices that no one had questioned before. And she says that scrutiny provided fodder for a broader HR inquiry that ultimately led to her firing.

Gray's evidence aligns with our precedent. To start, no one disputes that Gray faced more scrutiny after she helped Mauter advocate for an accommodation. Martin supervised Gray for two years. Until December 2017, he 'admittedly' did not inspect his team's timesheets. Appellees Br. at 10. So before Gray helped Mauter, her time entries went unexamined, but afterward, they received close review and ultimately resulted in her discharge.

State Farm responds that this increased scrutiny resulted not from Gray's protected activity, but from a change in supervisors. According to State Farm, Kyle, unlike Martin, always verified manual time entries and simply followed that practice when he covered for Martin.

Page 8But as we explain below, Gray has evidence that Kyle did not scrutinize other employees on Martin's team in the same way as her. So the differences between Kyle's and Martin's supervisory styles do not explain the particular scrutiny that Gray encountered.

Gray also has evidence that this heightened scrutiny occurred soon after her protected activity. State Farm investigated Gray only two or three months after she helped Mauter with her accommodation. So Gray not only experienced scrutiny for previously ignored conduct, but she did so shortly after she engaged in protected activity. That evidence is enough to raise a triable question on her direct liability claim. See Upshaw , 576 F.3d at 588–89.

When it comes to Gray’s cat’s paw claim, the temporal proximity is even closer. That is because when assessing temporal proximity, we consider whether the biased actor took the first opportunity that he had to retaliate. See Kirilenko-Ison , 974 F.3d at 665. Kyle’s first opportunity to retaliate came in November 2017, when Martin went on vacation and Kyle filled in for him. Kyle then reviewed Gray’s timesheets and immediately reported her to HR without consulting Martin. That timing by itself shows prima facie causation between Gray’s protected activity and Kyle’s actions. Id. at 665–66.

We have held similar evidence sufficient to show causation at this stage. E.g. , Hamilton , 556 F.3d at 432, 435-36; Upshaw , 576 F.3d at 588-89; Adamov , 681 F. App'x at 479. We therefore hold that Gray has established prima facie causation, and with that, her prima facie case of retaliation.

We now consider State Farm's stated non-retaliatory reason and Gray's evidence of pretext. State Farm says that it terminated Gray for timecard falsification. That is a legitimate and nondiscriminatory reason, so we turn to Gray’s evidence of pretext. See Espitia v. Procter & Gamble Co. , 93 F. App’x 707, 710 (6th Cir. 2004).

Our caselaw does not prescribe any particular method for showing pretext. See Miles v. S. Cent. Hum. Res. Agency, Inc. , 946 F.3d 883, 888 (6th Cir. 2020). Usually, plaintiffs show pretext by arguing that the employer’s proffered reason lacked a factual basis, did not actually Page 9motivate the adverse action, or was insufficient to motivate the adverse action. Id. But at bottom, pretext is about causation: was the defendant’s act motivated by the proffered reason or retaliation? Sloat , 18 F.4th at 210. And since Gray relies on both direct and vicarious theories of liability, that question applies to both Kyle’s and the ultimate decisionmakers’ motives.

We start with Gray’s direct liability claim, and like the dissent, we quickly dispose of it. Recall that in investigating Gray's records, State Farm found several instances when her timesheets did not match her computer activity. When asked about those discrepancies, Gray said she could not recall and gave mostly vague answers. So even if Gray had not falsified her time, her records—and her failure to explain them—pointed in that direction. See Michael v. Caterpillar Fin. Serv. Corp. , 496 F.3d 584, 598-99 (6th Cir. 2007). Gray responds that her records were no worse than those of other employees. And she emphasizes that State Farm never investigated anyone else, even after she identified potential comparators and alleged that Kyle had singled her out for retaliation. But Gray made that allegation when State Farm was already “contemplat[ing]” firing her. See Clark Cnty. Sch. Dist. v. Breeden , 532 U.S. 268, 272 (2001). Gray, moreover, does not have evidence of other employees who, like her, reported working when they were out of the building. In short, Gray has failed to call into question State Farm’s stated reason for firing her. Her direct liability claim thus fails at this stage.

But Gray also seeks to hold State Farm vicariously liable for Kyle’s actions. She argues that Kyle reported her for retaliatory reasons and that his report influenced State Farm’s decision to fire her. State Farm counters that Kyle's decision to report Gray was based solely on her timekeeping violations, not on any retaliatory animus. According to State Farm, Kyle inspected timesheets as part of his “normal” managerial practice; reviewed Gray’s timesheets in the same way he reviewed others'; and reported Gray for 'well-founded' discrepancies in her records. Appellees Br. at 27, 50, 68. These explanations are legitimate and non-retaliatory, so we turn to Gray’s evidence of pretext.

In arguing pretext, Gray insists that she never misreported her time. Gray does not dispute that her records showed certain discrepancies between her time and computer usage. She argues instead that her computer activity failed to capture tasks like answering phone calls or helping other employees. But this argument goes toward explaining the discrepancies; it does Page 10not render the discrepancies factually baseless in the first place. Gray has therefore failed to raise a dispute over the factual basis of Kyle’s report to HR about her timesheets.

But Gray’s alleged timekeeping improprieties do not end our analysis. That is because Kyle could have used those improprieties as “a legal, legitimate reason” to “cover up his true, [retaliatory] motivations.' See Hamilton , 556 F.3d at 436 (citation omitted). So we look at Gray’s other evidence of pretext. On that front, Gray argues that even if she overstated her time, others did the same but were not reported or disciplined. We have upheld retaliation claims when the plaintiff “was singled out for adverse treatment.” Hubbell v. FedEx SmartPost, Inc. , 933 F.3d 558, 570 (6th Cir. 2019) (citation omitted); see also Amos v. McNairy County , 622 F. App’x 529, 540 (6th Cir. 2015). Gray claims that happened here. And we agree that she has enough evidence of differential scrutiny to raise a material dispute over Kyle’s motives.

To start, consider the discrepancies for which Kyle reported Gray. On November 22, Gray manually ended her lunch at 3:10 p.m. but did not resume her computer activity until 3:16 p.m. On November 27, she manually ended her lunch at 3:13 p.m. but did not resume her computer activity until 3:30 p.m. And on November 29, Gray said she worked until 4:30 p.m., but her computer activity stopped at 4:21 p.m.

As Gray laid out in bullet points before the district court, Diane Parker, who was also on Martin’s team, had nearly identical discrepancies in her records. See MSJ Opp'n, R. 42, PageID 3192–94. On November 22, Parker manually ended her lunch at 2:14 p.m. but did not resume her computer activity until around 2:30 p.m. On November 24, Parker manually ended her lunch at 4:00 p.m. but did not resume her computer activity until 4:15 p.m. And on November 29, Parker manually ended her lunch at 1:45 p.m. but resumed her computer activity at 2:00 p.m. Parker’s 15-to-16-minute discrepancies tracked the 6-to-17-minute discrepancies that Kyle found in Gray’s records. Both Gray and Parker manually changed their lunch breaks so that they were at or just under the 50-minute limit. And Kyle undisputedly knew about Parker's timecard discrepancies at the time he reported Gray.

Despite knowing about their similar conduct, Kyle did not treat Parker and Gray in the same way. He viewed Gray’s actions as an “integrity” issue and reported her to HR. Yet he Page 11considered Parker’s conduct a simple “performance” problem and informally “coached” her for her policy violations. Appellees Br. at 48–49. Gray therefore has evidence that Kyle scrutinized her more than Parker for virtually identical conduct. See Hubbell , 933 F.3d at 570. That is sufficient for her to withstand summary judgment. See Strickland v. City of Detroit , 995 F.3d 495, 514 (6th Cir. 2021).

State Farm’s explanation does not justify this differential treatment, at least not beyond a genuine dispute. State Farm says Kyle reported Gray because, unlike Parker, Gray claimed to be working “when she was not even in the building.” Appellees Br. at 49 (citation omitted). But Kyle did not report Gray for that reason. He reported her based on discrepancies between her time entries and computer activity. It is undisputed that Gray's alleged physical absences surfaced only after Kyle reported her. In fact, State Farm says Kyle lacked access to employees’ building-entry records, so he could not have reported Gray based on those records. We cannot discern how Kyle could have distinguished between Gray and Parker based on information that he seemingly did not possess.

Apart from Parker, Gray also points to other employees who clocked into work but spent time socializing or visiting the cafeteria. State Farm concedes this happened, but frames it as a 'performance' rather than a 'falsification' issue. Id. at 46. But employees who claim to be working when they are not misreport their time. And State Farm's written policy treats both 'inaccurately report[ing] time' and '[f]alsify[ing] a timesheet' as policy violations, without distinguishing between their severity. Policy, R. 30-2, PageID 630-31. To be sure, the jury could conclude that Gray engaged in more serious misconduct than these other employees. But because State Farm’s policy does not draw that distinction, the question of whether Kyle singled Gray out for otherwise common behavior is a matter for the jury. See Strickland , 995 F.3d at 514.

State Farm identifies no other reason that could justify Kyle’s unique scrutiny of Gray. Gray was the second highest performing member of her team in 2017. Kyle acknowledged that she was considered 'a higher-level performer with more knowledge and technical skills than most.” Kyle Dep., R. 30-21, PageID 2009. And Martin viewed her as “intelligent and capable of very high performance.” Email, R. 30-5, PageID 969. Gray also served as “a lead resource Page 12on policy, workflow, and process' for her colleagues. Martin Dep., R. 30-12, PageID 1321. This is therefore not a case in which performance issues precede and then prompt special scrutiny. See, e.g. , Kenney v. Aspen Techs., Inc. , 965 F.3d 443, 450 (6th Cir. 2020); Hall v. Ohio Bell Tel. Co. , 529 F. App’x 434, 440–41 (6th Cir. 2013). Rather, heightened scrutiny seemed to come only after Gray engaged in protected activity, with no other prevailing reason to justify it.

In sum, Gray has enough evidence to show that Kyle’s stated reason for reporting her was a pretext designed to mask retaliation. We therefore hold that she has raised a genuine dispute over Kyle’s motives, as required for her cat’s paw claim.

Even if Kyle harbored retaliatory animus toward Gray, he did not participate in the decision to terminate her. Gray therefore seeks to hold State Farm vicariously liable for Kyle’s retaliation, arguing that his motives can be imputed to the actual decisionmakers. See Sloat , 18 F.4th at 210. To succeed on that theory, Gray must show (1) that Kyle intended to cause an adverse action against her for retaliatory reasons, and (2) that his actions proximately caused her termination. See Staub v. Proctor Hosp. , 562 U.S. 411, 422 (2011); E.E.O.C. v. Ford Motor Co. , 782 F.3d 753, 768 (6th Cir. 2015) (en banc). We have already determined that a factual dispute exists over Kyle's motives for reporting Gray, so we need not analyze the intent prong. See Bishop v. Ohio Dep’t of Rehab. & Corr. , 529 F. App’x 685, 696 (6th Cir. 2013). We therefore focus our analysis on proximate causation, as the parties do here.

Proximate causation requires “only some direct relation” between a biased supervisor’s action and the final adverse decision. Staub , 562 U.S. at 419 (cleaned up). Because injuries often have “multiple proximate causes,” a biased supervisor does not have to be the sole driver behind the employer’s decision. Id. at 419-20. The employer remains liable even if it relied only partially on the supervisor’s “biased report.” Id. at 419–21. And it escapes liability only if the supervisor’s influence was “too remote, purely contingent, or indirect.” Id. at 419 (citation omitted).

Gray has enough evidence to show that Kyle proximately caused her termination. Kyle reported Gray for conduct that was 'virtually identical' to Parker's. Madden v. Chattanooga Page 13City Wide Serv. Dep’t , 549 F.3d 666, 678 (6th Cir. 2008). He falsely claimed that Gray had been previously coached for taking long lunches. He also suggested that Gray likely misreported time on more than the three occasions that he discovered. A jury could conclude that Kyle caused State Farm to investigate Gray while anticipating other discrepancies that could lead to her termination. And it could conclude that State Farm relied on Kyle's 'biased report' when it chose to investigate and ultimately terminate Gray. Staub , 562 U.S. at 421.

State Farm responds that Kyle’s report was not biased because he honestly believed in the truth of his allegations. But a supervisor does not have to lie in order to be biased. As we have repeatedly recognized, a supervisor can cause an employee’s termination by reporting true yet selective information. Marshall , 854 F.3d at 378; Madden , 549 F.3d at 677–78. A report is true yet selective if it singles out an employee for conduct that others also engage in. See Madden , 549 F.3d at 677–78; Chattman v. Toho Tenax Am., Inc. , 686 F.3d 339, 353 (6th Cir. 2012). Gray therefore does not have to show that Kyle fabricated information to prevail.

State Farm next argues that even if Kyle made a biased report, State Farm can’t be liable because it independently confirmed Kyle's allegations. We have held that an employer can escape liability by conducting 'an in-depth and truly independent investigation' into an otherwise biased report. Marshall , 854 F.3d at 380. But when a supervisor reports true but selective information, an investigation will always confirm the supervisor's allegation. State Farm’s argument would thus foreclose liability even for selective reports that are “ designed and intended to produce the adverse action.' Staub , 562 U.S. at 420. The Supreme Court has rejected such an 'implausible' reading of antidiscrimination provisions. Id.

A simple example illustrates why. Imagine a workplace where five employees were engaged in the same pattern of misconduct. If a supervisor made a true but selective report of wrongdoing against one of the five employees because of that employee’s race, they would be attempting to use the company’s human resources as the “conduit” for their bias. See Romans v. Michigan Dep't of Hum. Services , 668 F.3d 826, 835 (6th Cir. 2012). If human resources then opened an independent investigation into that one employee, it would confirm the biased report, but it would not necessarily negate the supervisor's bias in singling out one employee based on race. That’s why the Supreme Court “declined to adopt [] a hard-and-fast rule” thatPage 14 “an independent investigation has a claim-preclusive effect” or “somehow relieves the employer of ‘fault.’” Staub , 562 U.S. at 420–21.

Our precedents likewise demonstrate that employers cannot avoid liability simply because they independently confirm the substance of a selective report. Consider Madden , where a supervisor reported an employee for setting off firecrackers while ignoring identical behavior by others. 549 F.3d at 678. The supervisor's report prompted an investigation by senior managers, during which the employee admitted to the supervisor’s allegations. Id. at 670. The investigation thus left no question about the truth of the supervisor's report. But we nonetheless held the employer liable because 'there was evidence that [the supervisor] discriminated in the information that he provided about employee misconduct to senior managers.” Id. at 678. We followed the same logic in Chattman , where an allegedly biased supervisor reported an employee for otherwise common behavior. 686 F.3d at 350, 353. There too, higher-ups independently confirmed the supervisor’s report, directly interviewing witnesses rather than relying on the supervisor’s account. Id. at 344. We still held that senior managers could serve “as the conduit of the supervisor’s prejudice” if they investigated the employee based on the supervisor’s selective report. Id. at 350–51 (cleaned up). Evidence of such a report, we concluded, created a genuine dispute over proximate causation. Id. at 352–53.

Gray presents similar evidence. She claims that Kyle selectively reported her to HR, and that his report led State Farm to investigate and terminate her. What’s more, Gray alerted State Farm to potential retaliation, but State Farm failed to take her allegation seriously. Gray told Martin and Keeling that Kyle targeted her because she had helped Mauter with her accommodation. And she identified specific colleagues who also rounded their time entries but were not reported. Yet State Farm made no effort to determine whether Kyle had singled Gray out for retaliatory reasons. From these factual allegations, a jury could conclude that State Farm acted as a quintessential 'conduit' of Kyle's bias. Even under a narrower view of vicarious liability, a jury could conclude that by ignoring Gray's allegation, State Farm effectively 'delegated part of [its] decisionmaking power' to Kyle. Staub , 562 U.S. at 425 (Alito, J., concurring in the judgment).

Page 15Our decision today does not impose a bright-line rule that a supervisor's true-butselective report will always be the proximate cause of any subsequent adverse employment action. Rather, we simply echo Staub ’s holding that a subsequent investigation that does nothing more than confirm a supervisor’s true-but-selective report is by itself insufficient to break the chain of proximate causation. Staub , 562 U.S. at 420-21. An employer can still negate causation by establishing that “the employer’s investigation result[ed] in an adverse action for reasons unrelated to the supervisor’s original biased action.” Id. at 421. Put another way, an employer will not be liable if its investigation uncovers a superseding 'cause of independent origin that was not foreseeable” from the supervisor’s biased action. Id. at 419 (quoting Exxon Co., U.S.A. v. Sofec, Inc. , 517 U.S. 830, 837 (1996)). We have previously held that the existence of a superseding cause is a question of fact. See Chattman , 686 F.3d at 353 (holding that whether an employer’s adverse action was “unrelated” to a supervisor’s biased action is an “issue of material fact”).

Here, the undisputed record shows that Kyle reported Gray for “manually changing” her time entries and identified three discrepancies as proof of that fact. State Farm’s HR employee, Keeling, thereafter conducted her own investigation that revealed additional discrepancies beyond the three identified by Kyle, including instances where Gray reported working while she was not in the building. State Farm claims that it fired Gray based on these additional out-ofbuilding discrepancies. On summary judgment, the question that we must answer is whether Gray’s termination based on these additional discrepancies was so “unrelated” to Kyle’s original report and so 'not foreseeable' by him that no reasonable factfinder could find proximate causation.

We hold that this question cannot be answered as a matter of law. Kyle reported generally that Gray was “manually changing” her time, not that she had done so only in the three instances that he identified. He also falsely informed Keeling that Gray had previously been reprimanded for similar conduct, suggesting that his report concerned a broad pattern of timekeeping issues. And he suggested in his report to HR that an investigation of Gray would uncover additional timekeeping errors. Given that Kyle's report and Gray's termination both Page 16related to Gray's timekeeping entries, a reasonable jury could find that the former improperly influenced—and was a proximate cause of—the latter.

Our decision in Romans does not counsel otherwise. Romans does not stand for the proposition that an independent investigation always breaks the chain of causation. Nor could it after the Supreme Court's rejection of that argument in Staub . It instead held that the independent investigation broke the causal chain because of the particular facts at issue in that case. There, the decisionmaker expressly disclaimed reliance on an allegedly biased report and conducted a separate investigation into the plaintiff’s alleged misconduct. 668 F.3d at 836. The separate investigation found that the plaintiff had “violated four work rules, only one of which was related to [the allegedly biased] report, and each of which would have individually supported a termination.” Id. at 837. Therefore, Romans falls comfortably within the scenario delineated in Staub where an independent investigation “results in an adverse action for reasons unrelated to the supervisor’s original biased action.” 562 U.S. at 421.

By contrast, here, State Farm relied on Kyle's report and opened an investigation that confirmed his allegations. State Farm then took the adverse action that Kyle's report was “designed and intended” to produce by firing Gray for timekeeping discrepancies. Staub , 562 U.S. at 420. So a jury could conclude that State Farm’s investigation took into account Kyle’s biased report and failed to determine that the adverse action was justified apart from his recommendation. See id. at 421. Romans supports rather than undermines this analysis.

State Farm nonetheless asserts that the additional discrepancies identified by Keeling are a distinct cause of Gray’s termination because they occurred when Gray was out of the building, indicating an “integrity” issue as opposed to a “performance” issue. But whether that distinction is sufficient to break the chain of proximate causation is a question for the jury. This is not a case where State Farm's further investigation revealed, for example, that Gray was a wanted fugitive or had embezzled funds from the company. Any such offenses, which have no connection whatsoever to timekeeping, would clearly be so “unrelated” to Kyle’s original report and so “not foreseeable” by him that we could find proximate causation lacking as a matter of Page 17law.⁴ Here, by contrast, the discrepancies for which Gray was fired are part of the same category of false timekeeping entries that Kyle reported initially. And Kyle certainly 'could [] have foreseen” the discovery of more timekeeping errors since he himself suggested to HR that an investigation would uncover additional discrepancies. Dis. Op. at 41. Furthermore, supervisor Martin’s termination recommendation for Gray cited only that she “over-reported time worked,” without distinguishing between discrepancies occurring in the building and those outside of it. A jury should therefore decide whether to credit State Farm’s in-building versus out-of-building distinction. We should not make that determination at the summary-judgment stage.

Finally, State Farm contends that even if Gray can show proximate causation, she cannot demonstrate that Kyle’s report was a but-for cause of her discharge, as required by the ADA. See Ford Motor Co. , 782 F.3d at 767. State Farm reasons that its investigation revealed additional misconduct not identified by Kyle, and that misconduct alone justified Gray's termination. We reject this crammed view of but-for causation. State Farm concedes that Martin never questioned his team’s time entries. And it admits that it investigated Gray only after Kyle reported her. Put differently, State Farm’s proffered reason for firing Gray resulted from Kyle’s alleged “antecedent” retaliation. Sloat , 18 F.4th at 213. A jury could conclude that State Farm would not have investigated and eventually fired Gray had Kyle not reported her.

For all of the foregoing reasons, we reverse the district court's grant of summary judgment and remand for further proceedings consistent with this opinion.

Page 18READLER, Circuit Judge, dissenting. “In our adversarial system of adjudication,” we trust the parties to “advanc[e] the facts and argument entitling them to relief.” United States v. Sineneng-Smith , 140 S. Ct. 1575, 1579 (2020) (citation omitted). That familiar understanding, otherwise known as the party presentation principle, reflects an important restraint on jurists: “The parties,” not judges, “frame the issues for decision.” Id. (citation modified). We, in turn, “decide only questions presented by the parties.” Id. (citation omitted).

Doing so makes this an easy case. On eight occasions in a single month, Monica Gray made manual fraudulent time entries, falsifying that she had returned to work from her lunch break when, in reality, she had not even swiped her badge to reenter the building. Each time, Gray reported a phony break of either 49 or 50 minutes, knowing that a 51-minute entry would have netted her a disciplinary point for exceeding her allotted break time. As Judge Marbley rightly concluded in granting State Farm’s motion for summary judgment, no reasonable juror could view this evidence as anything but a calculated pattern of intentionally submitting doctored timesheets to avoid discipline.

In seeing things otherwise, the majority opinion fails to heed the settled limits on the judicial power. Laboring to rehabilitate Gray's flawed appeal, the majority opinion resolves issue after issue in Gray's favor on grounds she never bothered to raise, with reference to evidence she never bothered to cite. Rewriting the arguments presented to us is troubling enough. Berry v. Experian Info. Sols., Inc. , 115 F.4th 528, 545 (6th Cir. 2024) (Readler, J., concurring in part and dissenting in part). But the fallout from doing so should make the practice unbearable. Embracing positions Gray never advanced plainly prejudices State Farm, who was not on notice that it needed to address those points with the Court. And then consider Judge Marbley. Reversing the district court based on arguments never made to it deeply disrespects that court’s weighty efforts, which, in Gray’s case, spanned more than three years, including a year dedicated to resolving State Farm’s dispositive motion, a decision we now undo. Taking the case as it was presented to us, the district court’s judgment should be affirmed.

Page 19As Gray pursued this case in district court and presented the case to us on appeal, her efforts were not enough to overcome State Farm's summary judgment motion. In a nutshell, Gray’s theory was that State Farm retaliated against her by terminating her employment several months after she helped a disabled coworker oppose perceived ADA violations by that employee's manager, Joe Kyle. Her analysis, in other words focused on her employer, State Farm, and its direct liability tied to its purported retaliatory efforts.

Viewing Gray’s case through the familiar lens of the McDonnell Douglas burden shifting framework, we can accept her prima facie case of retaliation as well as State Farm’s response, namely, that Gray was fired due to her scheme to falsify her time entries. That takes us to step three of McDonnell Douglas . At that point, Gray had to 'produce sufficient evidence from which the jury' could 'reasonably reject' State Farm's basis for terminating her as a mere “coverup” for retaliation. Manzer v. Diamond Shamrock Chems. Co. , 29 F.3d 1078, 1083-84 (6th Cir. 1994). She did not. In Judge Marbley’s words, “even if Ms. Gray can establish a prima facie case for retaliation, she cannot prove that State Farm's proffered reason for terminating her”—specifically, her “falsification of timecard entries”—“was pretextual.” Gray v. State Farm Mut. Auto. Ins. Co. , No. 20-cv-6038, 2024 WL 419001, at *6 (S.D. Ohio Feb. 5, 2024).

State Farm provided indisputable evidence documenting Gray's misconduct, justifying the company’s decision to terminate her. Based on discoveries initially made by Kyle, who was temporarily assigned to oversee Gray's work, and as later confirmed by State Farm in much greater detail, Gray reported working when she was not even badged into the building on each of November 3, 8, 9, 15, 17, 22, 24, and 27. On all but one of those occasions, she claimed a false lunch break of exactly 49 or 50 minutes—immediately under the 51-minute disciplinary cutoff. If she had reported her actual lunch breaks, she would have received enough attendance points to be terminated. And her putative explanation for these events, which addressed only one of these many suspect entries, falls well short of providing “sufficient basis in the evidence” for the “jury [to] reasonably reject [State Farm’s] explanation.” Manzer , 29 F.3d at 1083 (emphasis omitted).

Page 20Gray fails to counter this straightforward conclusion. She contends that her time theft 'was insufficient to warrant' termination, pointing to other employees she says committed similar misconduct, yet were not terminated. Jackson v. VHS Detroit Receiving Hosp., Inc. , 814 F.3d 769, 779 (6th Cir. 2016) (citation omitted). But Gray was obligated to produce evidence that the comparator employees-Jill Dillon and Diane Parker-fell 'outside [her] protected class.' Noble v. Brinker Int'l, Inc. , 391 F.3d 715, 728 (6th Cir. 2004). Her failure to demonstrate that Dillon and Parker did not engage in any ADA-protected activity, see 42 U.S.C. § 12203(a), is thus fatal to her comparator argument, see Noble , 391 F.3d at 728; Walden v. Gen. Elec. Int'l, Inc. , 119 F.4th 1049, 1056-57 (6th Cir. 2024) (explaining that 'conjecture' and “conclusory accusations” on a necessary aspect of a plaintiff’s claim are insufficient). Nor did either coworker engage in “substantially identical conduct” to that committed by Gray. Jackson , 814 F.3d at 779. Perhaps, as Gray emphasizes, all three made numerous manual entries. But compare that single shared characteristic with the other aspects of Gray’s misconduct. Gray does not claim, much less provide “significant probative evidence,” Walden , 119 F.4th at 1057, that either Dillon or Parker, like Gray, (1) made entries inconsistent with their computer-activity software, (2) clocked in when they were out of the building, or (3) systematically changed their lunch breaks to the maximum allowed. Those realities distinguish the “severity of [Dillon’s and Parker’s] actions” from Gray’s, making the pair poor analogies for proving that Gray’s course of conduct did not amount to a fireable offense. Jackson , 814 F.3d at 780.

No more persuasive is Gray’s assertion that her time theft did not “actually motivate the termination.” Smith v. Leggett Wire Co. , 220 F.3d 752, 759 (6th Cir. 2000). To prove as much, she must first “admit[],” for argument’s sake, both “the factual basis underlying the discharge” and 'that such conduct could motivate the dismissal.' Id. Once we 'acknowledge[]' these premises, id. , Gray's purported 'significant evidence' of State Farm's illicit motivation all but melts away, Br. Appellant 54. Two of her contentions-that State Farm and manager Chris Martin respectively rejected her attempted explanations for her unusual timekeeping practices— directly contest, rather than 'admit[],' as she must, the 'factual basis underlying' her firing. Smith , 220 F.3d at 759. Nor, if we accept that Gray committed fireable time theft, is there anything suspicious about State Farm acting on that information without launching further investigations into Kyle's actions or the behavior of other employees. So too as to Martin's Page 21apparent lack of candor to Gray during the investigation. There is nothing nefarious in wanting to get all HR-related ducks in a row before notifying an employee that she will be fired. Finally, that Kyle may have provided the HR investigator false information does not show that the State Farm decisionmakers (of whom Kyle was not one) used time theft as a coverup for their real motives. In short, Gray has not provided evidence to demonstrate that State Farm acted pretextually. That should end our analysis.

Resisting this straightforward conclusion, the majority opinion casts Gray's case in an entirely new light. Ignoring the fact that Gray’s legal arguments focus solely on State Farm’s decision making, the majority opinion instead discerns two separate claims woven together “inartfully” in Gray’s briefing: one seeking to hold State Farm directly liable for its own actions and a distinct one seeking to hold it vicariously liable for Kyle's. Maj. Op. 6-7 & n.2. Conceding that the first theory, as shown above, is a non-starter, id. at 11, the majority opinion rests Gray's case on a theory tied to Kyle's motives rather than State Farm's. Having recast Gray’s case in this manner, the majority opinion then concludes that Gray’s claims could prevail under a cat’s paw theory of liability, which, in essence, would penalize State Farm—even if it had no discriminatory motives—for illicit motivations held by Kyle. See Staub v. Proctor Hosp. , 562 U.S. 411, 422 (2011).

The problem is that this now-all-important cat's paw claim is neither the case nor the strategy Gray pursued. Rather, she framed her case as one presenting a theory of direct liability based on State Farm's own actions. In response to State Farm's summary judgment motion, Gray did not raise the cat’s paw notion as her theory of liability, let alone provide evidence on each essential element of the cat’s paw claim on which she bore the burden of proof at trial. Viet v. Le , 951 F.3d 818, 823 (6th Cir. 2020); see also 10A Charles Alan Wright et al., Federal Practice and Procedure § 2727.2 (4th ed. 2016). Nor did she do so on appeal. In neither setting did she mention the cat’s paw theory to inform her prima facie case of retaliation. And save for rebutting an alternative argument made by State Farm regarding pretext, one tied to the honest belief doctrine, Gray’s pretext arguments likewise forwent any reliance on a cat’s paw claim. In other words, the cat’s paw theory has played at most a fleeting role in Gray’s case.

Page 22Unsatisfied with Gray’s litigating choices, the majority opinion unilaterally remolds the afterthought cat's paw argument as the key legal theory that allows Gray to avoid summary judgment. In so doing, the majority opinion “shifts” the entire case’s focus from the “motive[s]” of the 'decisionmaker' (State Farm) to those of the 'subordinate' (Kyle). Maj. Op. 6. This could not be further from the way in which Gray asked us to view her lawsuit.

A. To see why, first consider how Gray might have—but did not —present her lawsuit. Under our precedents, the cat's paw theory can play a very robust role in a retaliation case, providing a “theory of liability” that centers the entire analysis around the actions of the “biased subordinate” rather than the employer. Marshall v. Rawlings Co. , 854 F.3d 368, 377 (6th Cir. 2017). The 'cat's paw theory,' we have explained, ultimately represents an 'application of respondeat superior principles.' Bose v. Bea , 947 F.3d 983, 991 (6th Cir. 2020). Today's majority opinion, which utilizes the cat’s paw doctrine in this broader role, is a good example. In sending Gray’s claim forward to the jury, the majority opinion does so not to hold State Farm directly liable for its own actions but to potentially hold the company 'vicarious[ly] liab[le]' based on retaliation by Kyle-a non-decisionmaker in Gray's termination. Maj. Op. 1. Yet Gray never asked us (or the district court) to do so. In fact, she never mentioned the supposedly key notion of 'vicarious liability' either in the district court or on appeal, let alone made the arguments needed to support such a theory.

Simply put, Gray never made an affirmative case on any of the cat’s paw elements. By way of background, the Supreme Court has authorized liability in this context only when (1) “a supervisor” (2) “performs an act motivated by [retaliatory] animus” that is both (3) “ intended by the supervisor to cause an adverse employment action' and (4) the 'proximate cause of the ultimate employment action.” Staub , 562 U.S. at 422. Gray barely hints at any argument on these elements.

Take the last element—proximate cause—first. This requirement is key in any cat’s paw case, because it demonstrates which biased employee actions can fairly serve as the basis of vicarious liability for the unbiased company. Woods v. City of Berwyn , 803 F.3d 865, 870 (7th Cir. 2015). Yet Gray never made the case to Judge Marbley that there was a “direct relation” between Kyle’s actions and her termination. Staub , 562 U.S. at 419-20 (quotations omitted).

Page 23And the only references to “proximate cause” in her appellate briefing are tucked inside a long block quote, otherwise going undiscussed in the surrounding text. Br. Appellant 51. At best, the two paragraphs that follow this block quote are obliquely aimed at rebutting one of State Farm’s arguments against proximate cause-namely that its 'independent investigation cut[] off any retaliatory animus.” Id. at 52. But Gray never makes any argument for proximate cause as part of a broader cat’s paw case. As this critical aspect of cat’s paw liability confirms, Gray did not frame her case around an affirmative cat’s paw theory, as the majority opinion does.

Doing so, as would the majority opinion, raises a host of other unbriefed questions as to the remaining cat’s paw elements. We must ask, for instance, was Kyle “a supervisor”? Staub , 562 U.S. at 422 & n.4 (citing Burlington Indus., Inc. v. Ellerth , 524 U.S. 742, 758 (1998)). And did he have the “specific intent to cause” Gray’s termination, either by “desir[ing] to cause” that “consequence[]” or by being “substantially certain” it would “result from” his actions? Id. at 422 & n.3 (quoting Restatement (Second) of Torts § 8A (A.L.I. 1965)). Gray addresses neither point, again belying any notion that she pursued a case of vicarious liability. Nor, understandably, did Judge Marbley, with Gray never raising those issues in district court (or here).

Cementing the point is Gray’s framing of the threshold McDonnell Douglas inquiry that leads off her case. Kirkland v. City of Maryville , 54 F.4th 901, 910 (6th Cir. 2022). At that stage, to make the broad cat’s paw argument envisioned by the majority opinion, Gray would use the burden-shifting framework to satisfy the remaining requirement from Staub —i.e., to show that Kyle “perform[ed] an act motivated by [retaliatory] animus.” Staub , 562 U.S. at 422. In other words, she would need to first 'appl[y] . . . the McDonnell Douglas framework to [her] allegations against [Kyle] .' Marshall , 854 F.3d at 381 (emphasis added). Only then, after determining that Gray can create a jury issue on “each step of the McDonnell Douglas burdenshifting test” to support her “allegation that [Kyle] was biased against her,” would we move on to analyze State Farm's vicarious liability for Kyle's illegal motives by analyzing proximate cause, supervisor status, and intent. Id. at 383; see also Staub , 562 U.S. at 422. Yet Gray's McDonnell Douglas arguments uniformly attempt to cast State Farm's actions as retaliatorynot Kyle’s.

Page 24Turn to the first McDonnell Douglas threshold, where Gray’s prima facie case required her to establish a causal link between her 'protected activity' and some subsequent 'adverse action.” Rorrer v. City of Stow , 743 F.3d 1025, 1046 (6th Cir. 2014). Whose “action” did she choose? Id. State Farm's, not Kyle's. The purportedly retaliatory act she identified was her termination-an action taken exclusively by State Farm. See Br. Appellant 33 (arguing that “State Farm took an adverse employment action against her when it terminated her employment”); see also id. at 60 (Gray admitting Kyle was not one of the 'decisionmakers' who terminated her). In tying her prima facie case to State Farm’s allegedly retaliatory response to Gray’s protected activity, in other words, Gray, to prevail, must demonstrate that State Farm (not Kyle) retaliated against her. If Gray had instead preferred to rest her case on Kyle “perform[ing] an act motivated by” retaliation, as the majority opinion now does, it was her duty to identify that “act” and provide prima facie evidence that it was “motivated by” retaliation. Staub , 562 U.S. at 422; Marshall , 854 F.3d at 379–80. She did not do so.

Gray’s remaining prima facie arguments sing the same tune. Take, for example, the lead heading in Gray's appellate brief. It asserts that 'State Farm Retaliated against Gray' and, tellingly, never mentions Kyle. Br. Appellant 21 (bold omitted). Or ponder the fact that two of Gray’s claimed “protected activities” occurred on December 12 and 24, with any role played by Kyle having ended, at the latest, on December 4. To the same end, in arguing that “State Farm was aware of [her] protected activity,” id. at 27 (underline omitted), Gray identified four State Farm employees (in addition to Kyle) with the requisite knowledge. These are odd choices if Gray’s case turned on Kyle’s retaliatory mental state rather than that of the company in general. And, again, Gray argues only that 'State Farm'-not Kyle-'took an adverse employment action against [her] when it terminated her employment.” Id. at 33 (underline omitted). Fairly read, Gray’s prima facie briefing seeks only to hold State Farm directly liable for its own actions rather than vicariously so for Kyle’s.

Her pretext arguments echo this theme. Over and over, they focus on State Farm's justification for its actions rather than Kyle’s justification for his. Br. Appellant 46 (querying whether “State Farm fire[d] Gray for the stated reasons or not”); id. at 49–50 (disputing whether “State Farm had an honest belief”); id. at 54–55 (arguing pretext based on circumstances “State Page 25Farm never investigated' and facts 'State Farm knew'); id. at 58 (arguing that 'State Farm treated Gray's coworkers more favorably'). Demonstrating that 'State Farm['s] . . . stated reasons” for firing Gray were pretextual, as Gray seeks to do, would support an effort to hold State Farm liable for its own actions. Id. at 46. But it does nothing to show that Kyle was “motivated by . . . animus” or, accordingly, that State Farm can be held vicariously liable based on his distinct actions. Staub , 562 U.S. at 422.

Were there any doubt on that score, look back to Gray’s summary judgment response in district court. There, Gray similarly painted her prima facie case focused on the theory that 'State Farm retaliated' based on evidence that 'State Farm knew about Ms. Gray's protected activity' and that 'State Farm took an adverse employment action against Ms. Gray.' Pl.'s Mem. Opp'n Defs.' Mot. Summ. J., R. 42, PageID#3174, 3181, 3186 [hereinafter Summ. J. Opp’n] (formatting omitted). As here, she did not purport to elevate the cat’s paw argument as informing her overriding theory of retaliation. Nor, by and large, did Gray emphasize Kyle’s actions as part of her broader pretext analysis in the district court; she made two different pretext arguments before even thinking to mention the cat's paw theory and Kyle's corresponding relevance. And, as in her appellate brief, Gray’s pretext arguments before Judge Marbley almost without exception asked the district court to evaluate State Farm's motives, not Kyle's. Id. , PageID#3201 (targeting “State Farm’s proffered reasons” and asking whether “State Farm fire[d] Ms. Gray for the stated reasons”); id. , PageID#3203 (citing evidence that “State Farm’s offered reason . . . was pretextual'); id. , PageID#3204 (discussing whether 'State Farm . . . honestly believed that Ms. Gray committed time theft'). At every turn, in other words, Gray targeted State Farm's retaliatory conduct and forwent any attempt to 'appl[y] the McDonnell Douglas framework to [Gray]’s allegations against [Kyle].” Marshall , 854 F.3d at 381.

Gray’s understanding of her case is confirmed yet again in the separate state claim she brought against Kyle for aiding and abetting retaliation. See Ohio. Rev. Code Ann. § 4112.02(J) (2025). In arguing that Kyle can be held liable on that basis, Gray characterizes her previous argument under the ADA as demonstrating that ' State Farm retaliated against Gray,' a fact which allows her to further conclude that “ Kyle is liable for aiding and abetting retaliation.' Br. Appellant 59 (emphasis added). These statements would be perplexing if they came from a Page 26litigant who purportedly sought to hold State Farm vicariously liable for Kyle’s retaliation rather than directly liable for its own.

B. By comparison, Gray's mentions of the cat's paw theory in this litigation were fleeting. Unearthing her actual use of the cat’s paw theory requires significant spadework. Both in district court and on appeal, Gray and State Farm primarily quarreled over the nature and severity of Gray’s misconduct and whether State Farm was justified in terminating her on that basis, or whether State Farm instead was motivated by a retaliatory animus due to Gray’s earlier support for a disabled employee. At one point deep into this back and forth on appeal, Gray claimed that State Farm’s “assertion that [she] falsified her timecards” was “factually false.” Br. Appellant 47 (underline omitted). State Farm, of course, firmly believed that its allegation was true, and explained as much. But even if it was not, State Farm added, the company had an 'honest belief'' that the misconduct occurred, an alternative basis for overcoming Gray's assertion of pretext. Br. Appellees 54.

A few words, then, on the honest belief doctrine. The concept applies in a targeted way to a small set of employment disputes. Sometimes, an employee may attempt to show that her employer acted in a pretextual manner by disciplining the employee for conduct that 'never happened.” Miles v. S. Cent. Hum. Res. Agency, Inc. , 946 F.3d 883, 889 (6th Cir. 2020). If an employee “provide[s] evidence that the employer’s allegations” are in fact false, the employer, “as a defense to [the] lack of a factual basis,” may still “defeat [the] pretext claim by showing that it had an ‘honest belief’ in the” reason it asserted for disciplining the employee. Id. at 889, 890 n.5 (citation omitted). State Farm invoked the doctrine here, explaining that even if Gray were correct that her misconduct never happened, the company honestly believed that it did, shielding it from liability.

Only at this late stage in the parties’ respective arguments do we reach what the majority opinion now features as the main event. In a defensive posture, Gray injects the cat's paw theory-and narrowly at that. Turn to page 50 of her 60-page opening brief on appeal. Anticipating and responding to State Farm’s reliance on the honest belief rule, Gray mentions “the cat’s paw theory of liability” because “the honesty of [State Farm’s] belief does not matter” if the cat's paw theory applies. Br. Appellant 50-51 (quoting Marshall v. Rawlings Co. , 854Page 27 F.3d 368, 380–81 (6th Cir. 2017)). This cat’s paw point, in fact, was not even Gray’s leading challenge to State Farm’s honest belief assertion—it was the “[s]econd” of her two reasons that “the District Court erred” in holding that “State Farm had an ‘honest belief’” that she falsified her timecards. Id. at 50. And Gray presented the argument as tied merely to a factual dispute, burying it under a heading that read “State Farm’s assertion that Gray falsified her timecards is factually false.” Id. at 47, 49–50 (underline omitted). In short, Gray raised the cat’s paw theory merely as a rebuttal point to arguments regarding the last step of the McDonnell Douglas threestep analysis. We should respect that decision.

Given the limited way in which Gray utilized the cat’s paw theory, it makes sense that she put no legwork into proving its requirements for demonstrating vicarious liability. See Staub , 562 U.S. at 422. For instance, she neither describes what she must prove to succeed in a cat’s paw case nor links those requirements to the record in this case. See Br. Appellant 50–52. Instead, she “discusse[s] the relationship between the cat’s paw theory and honest belief” only to challenge State Farm’s counterargument about its “independent investigation.” See id.

Gray’s confined reliance on the cat’s paw theory is confirmed by her briefing in district court. There, as here, Gray’s reliance on the cat’s paw notion was far from a leading light, first surfacing 72 pages into her 78-page summary judgment response brief, long after the parties conducted their debate over Gray’s asserted prima facie case of retaliation and had moved on to Gray’s response to State Farm’s asserted non-retaliatory basis for her termination. And when she finally introduced the idea, she did so only to rebut State Farm’s assertion of honest belief, dedicating just one paragraph to describing the logic behind the theory before spending the next page and a half arguing that it precludes reliance on honest belief. As already explained, had Gray instead opted to make wider use of a cat’s paw theory, as the majority opinion now does, she would have needed to, among other things, list the elements of cat's paw liability and connect those elements to record evidence, as was her duty at summary judgment. See Viet , 951 F.3d at 823; see also Rozumalski v. W.F. Baird & Assocs., Ltd. , 937 F.3d 919, 925 (7th Cir. 2019) (Wood, J.) (plaintiffs seeking to avoid summary judgment in a cat’s paw case must “make a specific argument” “inform[ing] the trial judge of the reasons, legal or factual, why summary Page 28judgment should not be entered,' rather than merely stating their 'intent to use a cat's paw theory” in “general terms”).

C. Finally, consider the most troubling aspect of reframing the case in this manner—that Gray’s briefing never gave State Farm or Judge Marbley a fair chance to respond to the Kylecentric case the majority opinion now embraces. Start with State Farm. Because Gray did not brief her theory of retaliation as resting on Kyle and his role as the shadow director of the cat’s paw, State Farm similarly did not brief the case in that fashion. Again, Gray invoked a cat’s paw argument only in response to State Farm’s assertion of the honest belief doctrine. Taking Gray at her word, State Farm similarly relegated its own short cat’s paw discussion to a subheading under its “Honest Belief” argument. Br. Appellees 54, 60. And it did so, it bears noting, only as a fallback position—its honest belief “defense” was an alternative argument to its lead theory, again, that Gray failed to make the initial showing that the allegations against her had “no basis in fact.” Br. Appellees 54; see also Miles , 946 F.3d at 890 n.5. That makes Gray's cat's paw point merely a rebuttal to a secondary argument.

The remainder of State Farm’s briefing confirms that it understandably took Gray at her word in limiting her cat’s paw argument to the issue of honest belief. State Farm’s McDonnell Douglas arguments, like Gray's, focus on the company's actions and motives-debating, for example, whether 'State Farm's decisionmakers knew about' Gray's protected activity and whether 'State Farm's termination decision was a pretext.' Br. Appellees 32, 52 (formatting omitted). Proof positive that State Farm fairly detected no overarching invocation of cat’s paw liability comes in its prima facie case argument, where it asserts that 'Kyle was wholly uninvolved in . . . the decision to terminate Gray’s employment.” Id. at 34. The point addresses the case Gray argued—one based on holding State Farm directly liable for its own actions. But it would be a nonsensical response to a case premised entirely on the cat's paw theory and vicarious liability.

And let us not forget Judge Marbley. By failing to make a broad, threshold cat’s paw argument in district court, Gray never asked Judge Marbley to find that 'each step of the McDonnell Douglas burden-shifting test' supported her 'allegation that [Kyle] w[as] biased against her.' Marshall , 854 F.3d at 383. Accordingly, Judge Marbley's opinion, like State Page 29Farm’s briefing, understandably responds to Gray’s State Farm–framed arguments. See Gray , 2024 WL 419001, at *3 (describing Gray’s “circumstantial evidence that State Farm discriminated against her'); id. at *4 (discussing whether 'State Farm knew about [Gray]'s protected activity'); id. at *6 (noting Gray's inability to 'prove that State Farm's proffered reason for terminating her was pretextual”). All of this well illustrates why points not made in district court are ordinarily foreclosed on appeal. See Glennborough Homeowners Ass'n v. USPS , 21 F.4th 410, 414 (6th Cir. 2021).

D. How does the majority opinion explain all of its norm-breaking? It shrugs these widespread concerns aside in a single footnote. Maj. Op. 6–7 n.2. Of the very few points made there, one is the entirely unhelpful admission that Gray never used critical phrases like “vicarious liability' or-except in the limited context noted above-'cat's paw,' notions that, we are nonetheless told, Gray 'invokes' as the key facet of her case. Id. at 6 & n.2. Despite these blaring omissions, the majority opinion portrays Gray as having raised the cat's paw theory “again and again” in her appellate and district court briefing. Id. at 6 n.2. Again and again, by my count, amounts to just five occasions, scattered across 175 pages of briefing between the two courts (in other words, once every 35 pages of briefing). Two are the exact passages discussed above, again, tied narrowly to rebutting the honest belief rule. One is Gray’s repetition of that argument in her reply brief. And the final two are simply the same point, regurgitated in the “Summary of the Argument” sections of Gray’s principal briefs here and in the district court. In short, Gray’s briefing, fairly portrayed, alludes to the cat’s paw issue hardly at all, and only in the context of her honest belief argument.

True, as the majority opinion emphasizes, Gray does mention Kyle's role in her termination at various points in her papers. Taking an inch to leap a mile, the majority opinion understands those references as meaning only one thing: that Gray's 'inartful[]' briefing includes an unlabeled, unexplained affirmative cat's paw claim. Id. at 6-7 n.2 (quotation omitted). I can think of a far more plausible explanation: Kyle was Gray’s fill-in supervisor when her time theft was noticed and dealt with. It is thus hardly surprising that Gray’s briefing, while singularly focusing its affirmative legal theories on State Farm’s decision to terminate her, would mention facts involving Kyle that led up to that action. But Gray’s recitation of Kyle’s Page 30name does not mean that she otherwise silently tied her overarching legal theory to his acts. Equally true, that some of Kyle’s actions could be packaged to potentially support a distinct cat’s paw claim does not license us to moonlight as quasi-detectives, quasi-prosecutors, investigating the record for any clues that might fill out an unargued legal theory for us to assert on Gray’s behalf.

On this point, it bears emphasizing that the dispute here is not merely one over whether Gray “forfeit[ed]” an argument, as the majority opinion declares. Id. at 6 n.2. Rather, the issue is far more fundamental: Gray, in responding to State Farm’s summary judgment motion, never “invoke[d]” the overarching legal theory her case purportedly depends on. Id. at 6. Keep in mind the blackletter understanding that as the 'nonmoving party,' Gray, to avoid summary judgment, was “require[d] . . . to do [her] own work.” Guarino v. Brookfield Twp. Trs. , 980 F.2d 399, 406 (6th Cir. 1992). So had Gray in fact sought to avoid summary judgment by anchoring her case to the cat’s paw theory, she had to identify the theory and then “demonstrate” each of its elements by 'set[ting] forth specific facts showing a triable issue,' conditions we ordinarily enforce without hesitation. Milczak v. Gen. Motors, LLC , 102 F.4th 772, 783 (6th Cir. 2024) (Bloomekatz, J.); accord. Franklin v. Franklin County , 115 F.4th 461, 470 (6th Cir. 2024) (Gilman, J.).

Gray and her counsel opted not to do so. Instead, she confined her cat’s paw argument to a rebuttal point within a case she framed around State Farm’s (not Kyle’s) actions. Unlike the majority opinion, I would honor that litigating choice. See Sineneng-Smith , 140 S. Ct. at 1579. “[W]e [should] take [Gray’s] case as it is, not as we might think it could be.” Berry , 115 F.4th at 544 (Readler, J., concurring in part and dissenting in part).

Nor can State Farm shoulder Gray’s summary judgment burden for her, as the majority opinion oddly seems to suggest. With little to point to in Gray’s summary judgment briefing, the majority opinion instead seizes on State Farm's citation of Staub in a single paragraph of its summary judgment motion. Maj. Op. 6 n.2. But State Farm cited the case merely to counter one of many possible causation arguments Gray might raise in opposing summary judgment. That is all. State Farm never mentioned the cat’s paw theory, much less suggested that it might control the entire case. In any event, whatever State Farm did (or did not) say in its brief could not Page 31absolve Gray of her duty to articulate the theories she relied on to avoid summary judgment, and to lay out the elements of those claims. Guarino , 980 F.3d at 406.

The point here is thus not about magic words. Maj. Op. 7 n.2 (quoting Coffee v. Carroll , 933 F.3d 577, 586 (6th Cir. 2019)). Instead, it is about articulating basic elements and claims so that the parties and the court have fair notice of what issues underlie the litigation, and are thus ripe for resolution. That is the least we ask of parties. Not surprisingly, the majority opinion’s citations do not say otherwise. Indeed, even if one were to view the issue here as merely a matter of forfeiture, as does the majority opinion, Coffee did not conclusively decide whether oversights like Gray's required a finding of forfeiture because, giving those defendants the benefit of the doubt on the question, the panel rejected their arguments on the merits. Coffee , 933 F.3d at 586. I would be happy to follow a similar practice here. But waving aside presentation concerns only to find the possibly forfeited theory successful is quite another matter. Alternatively, if Gray in fact asserted a cat’s paw claim in district court, one that inadvertently went unruled upon, our standard course would be to remand for the district court to “resolve[]” the matter “in the first instance.” Stanek v. Greco , 323 F.3d 476, 480 (6th Cir. 2003); see also United States v. Kimbrough , 138 F.4th 473, 480 (6th Cir. 2025) (“We are a court of review, not first view.” (citation modified)). But even that customary practice is jettisoned by the majority opinion, all in the name of advancing Gray’s case to trial.

In the end, the majority opinion finds this mountain of concerns “puzzling.” Maj. Op. 6 n.2. Of course, the majority opinion must be equally puzzled by Judge Marbley, who similarly detected no hint of a cat’s paw claim in his handling of the case, which spanned more than three years, far longer than we have spent with the matter. See generally Op. & Order, R. 52 (never mentioning “cat’s paw” or “vicarious liability”). Yet in reversing the district court, the majority opinion curiously fails to describe what error Judge Marbley in fact committed. This too is far from regular order. Customarily, we defer to a district court’s assessment of what issues were raised before it, and when. See Evanston Ins. Co. v. Cogswell Props., LLC , 683 F.3d 684, 692 (6th Cir. 2012). An ounce of that same deference would make sense here as well, where Judge Marbley’s summary judgment opinion necessarily determined that Gray never raised a cat’s paw claim. Yet far from receiving the benefit of the doubt, Judge Marbley instead is now being Page 32reversed for an error the contours of which the majority opinion never bothers to explain. Puzzling indeed.

The decision to rewrite Gray’s lawsuit is all the more curious when one realizes that, had Gray in fact pursued her retaliation claim under a cat's paw theory, with Kyle's purported animus imputed to State Farm, as the majority opinion now puts forward, she still could not prevail. Turn from Gray’s case as pleaded to the one depicted in the majority opinion—namely, one in which the cat’s paw doctrine frames the analysis from stem to stern around the motives of “the subordinate,” Kyle, rather than “the decisionmaker,” State Farm. Maj. Op. 9. Even in this artificial construct, Gray could not demonstrate two of the requirements for cat’s paw liability: retaliatory motive and proximate cause. Staub , 562 U.S. at 422.

A. Consider first whether Kyle was “motivated by [retaliatory] animus” when he first brought Gray’s timekeeping issues to light. Id. As noted above, our cat's paw case law requires Gray to first marshal her circumstantial evidence of Kyle's retaliation through the steps of McDonnell Douglas . Marshall , 854 F.3d at 379–81. In other words, only if Gray (1) presents a prima facie case of retaliation and (2) shows that Kyle’s offered reasons are pretextual may she move on separately to trying to prove the cat’s-paw-specific requirements of proximate cause, intent, and supervisor status (the latter two, although not contested here, likely could be, as explained later). Id. ; see also Staub , 562 U.S. at 422. Gray's case, as now constructed by the majority opinion, stumbles in multiple respects.

1. 1. Beginning with Gray’s burden to establish a prima facie case of retaliation, reframed under the cat’s paw theory to focus on Kyle rather than State Farm, Gray would have to show that her (1) protected activity, of which (2) Kyle was aware, was (3) the but-for cause of (4) the action he took against her. A.C. ex rel. J.C. v. Shelby Cnty. Bd. of Educ. , 711 F.3d 687, 698–99 (6th Cir. 2013). Steps one and two may be conceded, as Gray took protected activity by opposing perceived ADA violations, facts of which Kyle was seemingly aware.

Things get more difficult from there. What “act” did Kyle “perform[] . . . motivated by” retaliation? Staub , 562 U.S. at 422. That is, what was the relevant act on which to focus this Page 33prima facie showing of intentional retaliation? A.C. ex rel. J.C. , 711 F.3d at 698–99. As noted by the majority opinion, Kyle was not “the decisionmaker” who fired Gray. Maj. Op. 6. True, he did “report” “Gray to Human Resources.” Br. Appellant 4. Yet Gray has acknowledged that Kyle did so at the behest of his supervisor, who both “advised” Kyle to investigate Gray’s time entries beyond the initial three he flagged, and ultimately “told [him] to contact HR.” Summ. J. Opp’n, supra , R. 42, PageID#3153–54. Gray’s failure to point to any other action by Kyle alone likely sinks her hypothetical cat’s paw claim, as Kyle’s report to HR was indisputably motivated by obedience to a superior’s direct order, and thus not evidence of retaliation.

We could generously elevate an action that Kyle did take on his own prerogative (and thus that might have been motivated by personal animus): the earlier decision to flag three of Gray’s time entries that he “was not quickly able to resolve in [his] mind,” by taking them to his supervisor and saying: “[T]his is what I found. What would you like me to do?” Kyle Dep., R. 30-21, PageID#1953; see also Summ. J. Opp'n, supra , R. 42, PageID#3153 (Gray describing similar facts in district court). Even then, that leaves us to ask whether Gray’s protected activity was the but-for cause of Kyle flagging these time entries. The majority opinion believes it was, characterizing Kyle's actions as 'heightened scrutiny' of previously ignored conduct, which, coming in the wake of Gray's protected activity, allows for an inference of causation. See Hamilton v. Gen. Elec. Co. , 556 F.3d 428, 435 (6th Cir. 2009). Concluding as much is problematic for two reasons.

For one, Gray made this argument only in her reply brief. Her opening brief never mentions “heightened scrutiny,” never cites any of the majority opinion’s case law on this point (save for one case, in an unrelated context), and never develops any argument under this theory. And in district court, she never mentioned the point whatsoever. Rather, Gray's causation arguments, both in her opening appellate brief and before Judge Marbley, focused solely on comparator evidence and temporal proximity. This double 'omission' results in double 'forfeiture' of the heightened scrutiny argument, see Glennborough Homeowners Ass'n , 21 F.4th at 414, a reality the majority opinion ignores.

For another, even had the issue been preserved, a heightened scrutiny theory does not apply here. To demonstrate as much, Gray needs evidence that Kyle’s review of her timesheets Page 34departed from his previous practices. See, e.g. , Hamilton , 556 F.3d at 432 (finding causation because the plaintiff's supervisors 'harassed him more than they ever had before' (emphasis added)). She has not presented any. Gray’s heightened scrutiny argument to us, referenced only in reply, does not mention Kyle's previous practice in reviewing timesheets. Even more problematic is her district court briefing, which in essence concedes the issue, emphasizing that Kyle consistently took a strict approach to employee time entries, 'expect[ing] his Claims Specialists to let him know any time they manually enter[ed] time.' Summ. J. Opp'n, supra , R. 42, PageID#3152. Kyle’s testimony confirmed as much. Unlike Chris Martin, Gray’s normal manager, Kyle made a practice of reviewing the computer activity software that revealed Gray’s falsifications. In Kyle’s words, “[T]his is what I do, you know. This is normal for me.” Kyle Dep., R. 30-21, PageID#2035. Without evidence that Kyle's efforts marked 'intensified' analysis, Gray cannot succeed on this theory of causation. Hamilton , 556 F.3d at 432.

The majority opinion tries to sidestep this problem by asking us to direct our attention to Martin's managerial practices rather than Kyle's. The move is understandable, given that Martin’s “very laid back” managerial philosophy, as Gray put it, contrasted sharply with Kyle’s. Summ. J. Opp’n, supra , R. 42, PageID#3152. But mere interpersonal contrast is not the stuff of heightened scrutiny. Typically, a heightened scrutiny–based causal argument takes the following form: The manager did not care about the behavior before the protected activity. He did care after. The inference? The protected activity is what caused him to change his approach. See, e.g. , EEOC v. New Breed Logistics , 783 F.3d 1057, 1070 (6th Cir. 2015) (citing evidence that “prior to [the plaintiff’s] rejection of [her manager’s] sexual advances, [he] had no problems with [her] tardiness”). Note the very different story here: A lazy manager (Martin) did not care about checking timesheets. He was replaced by a strict one (Kyle) who did. What inference does this allow? Only that Kyle took this managerial duty more seriously than Martin did.

Just ask Gray. She conceded that Kyle 'never asked Mr. Martin about his policy on manual entries' and 'was not aware of Mr. Martin's expectations of his team' on that front. Summ. J. Opp’n, supra , R. 42, PageID#3152. With this in mind, Gray's complaints about the “common practice for Martin’s team,” Reply Br. Appellant 17, 21, do not help discern Kyle’s motivations. Kyle’s attention to detail may have been upsetting to Gray, as it may be for any Page 35employee suddenly overseen by a by-the-book manager. But, again, these distinctive management styles are not evidence that Kyle's 'scrutiny' ever 'intensified,' Hamilton , 556 F.3d at 432, or, it follows, that Gray’s protected activity caused him to flag her time entries.

Consistent with our cat’s paw precedent, the majority opinion’s effort to save Gray’s case falters at the pretext stage if Kyle had an “honest belief” in this pattern of deceptive entries that “justified adverse action against” her. Marshall , 854 F.3d at 380 n.3. Conversely, it can succeed here by showing (1) that Kyle lacked an honest belief that Gray engaged in the conduct noted, or (2) that this observation did not justify referral to Kyle’s supervisor. Id.

a. Beginning with the first consideration, Kyle’s honest belief that Gray misreported her

lunchbreaks is above reproof. See Maj. Op. 11–12. So is his related belief that these entries

showed a pattern of ducking in just under the disciplinary cutoff. As Kyle has repeatedly

explained, it was this entire pattern of events, not Gray merely “overstat[ing] her time,” , that

caught his attention, see Kyle Dep., id. R. 30-21, PageID#1984 (“Why put 50 minutes is the first

thing that comes to my mind. . . . 50 is significant, because . . . if she would have manually input

[one minute later], she would have been assessed points. . . .”); id. , PageID#2015 (“[S]eeing the

multiple 50 minutes manual adjustments, . . . [i]t's an indicator that potentially we have a

problem.”); Kyle Decl., R. 43-2, PageID#3712 (“[W]hat caught my attention was the pattern of Page 36Plaintiff changing meal times . . . with manual entry of 49[-] or 50-minute meals that did not align with her computer activity.”).

State Farm's computer and timekeeping records confirm the point. As those records demonstrate, Kyle saw Gray manually enter inaccurate lunch breaks of exactly 50 minutes on each of November 22, 24, and 27. All things considered, it was more than “reasonabl[e]” for Kyle to 'honestly rel[y]' on these 'particularized facts' in concluding that Gray habitually changed her lunch breaks to fall right under the time limit. Chen v. Dow Chem. Co. , 580 F.3d 394, 401 (6th Cir. 2009).

b. That leaves whether Kyle’s “honest belief . . . justified” his actions, see Marshall , 854 F.3d at 380 n.3—that is, whether the pattern he noted was “[]sufficient to warrant” consulting his supervisor, Jackson , 814 F.3d at 779 (citation omitted). To show it was not, Gray would have to provide “evidence that” Kyle did not flag the time entries of “other employees,” who were (1) 'not in [Gray's] protected class,' even though (2) 'they engaged in substantially identical conduct.” Id. at 779. The majority opinion cites just one such employee-Parker. But as a comparator, Parker falls short in both respects.

To start, Gray, as already discussed, failed to “produce[] . . . evidence as to . . . whether” Parker “was outside the . . . class” of people who take “protected” action under the ADA. Noble , 391 F.3d at 731. Nor, for similar reasons, did Gray show that Parker’s time entries, as reviewed by Kyle, were “substantially identical” to hers. Jackson , 814 F.3d at 779. Recall that Gray's evidence failed to prove as much as to State Farm's decision. The same applies to Kyle's decision. Focusing only on 'manual entries for meals,' as does Gray, Br. Appellant 40, disregards two key facets of Kyle's explanation that distinguish 'the severity of' Gray's and Parker’s “actions,” Jackson , 814 F.3d at 780—namely that Gray’s entries did not align with her computer activity, and that they demonstrated a pattern of 50-minute breaks.

Enter, once again, the majority opinion. Unlike Gray, the majority opinion detects “enough evidence of differential scrutiny,” Maj. Op. 12, by changing the focus from the “manual entries' Gray pointed out, Br. Appellant 40, to the 'discrepancies' between the time each employee entered and the time reflected on the computer-activity software, Maj. Op. 12. Gray’s Page 37appellate briefing, it bears emphasizing, never mentioned discrepancies in Parker’s timekeeping records, let alone argued that Parker was comparable on that basis. Ordinarily, we accept the fundamental 'premise' that a party 'know[s] what is best for [her]' and is 'responsible for advancing the . . . argument entitling [her] to relief.' Sineneng-Smith , 140 S. Ct. at 1579 (citation omitted). We likewise routinely consider arguments made in district court but abandoned in a party’s appellate briefing forfeited. Doe v. Mich. State Univ. , 989 F.3d 418, 425 (6th Cir. 2021). Yet the majority opinion jettisons this foundational practice too, instead suggesting that Gray’s listing of Parker’s discrepancies “in bullet points before the district court” properly presents to us a discrepancies-based argument for review. Maj. Op. 13. Here too, our departure from longstanding precedent is left unexplained.

Indeed, not only has Gray not made the legal argument forged by the majority opinion, but she has also failed to offer evidence to prevail on the point. On the comparator issue, Gray cited only two documents-Parker's complete timesheet audit for 2017 and a chart Gray prepared for litigation, summarizing all of Parker’s manual entries for the year. Conspicuously absent from Gray’s presentation is any mention of the computer-activity software, the item the majority opinion now embraces to discern 'discrepancies' in Parker's timesheets. Utilizing software printouts neither cited nor discussed by Gray, the majority opinion tediously works up calculations it believes support its own theory of retaliation. Compare Maj. Op. 13 (citing Summ. J. Opp'n, supra , Ex. 11, R. 42-11), with Br. Appellant (not doing so), and Reply Br. Appellant (same). Disrespecting Gray’s “strategic litigation choice,” which focused on manual entries, is unusual enough. Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll. , 143 S. Ct. 2141, 2239 n.21 (2023) (Sotomayor, J., dissenting) (citation omitted). Compiling data Gray never deployed only compounds the matter, wholly absolving Gray of her 'responsib[ility] for advancing the facts' needed to support this now-rewritten argument. Sineneng-Smith , 140 S. Ct. at 1579.

Ironically, even the majority opinion’s manufactured focus on “discrepancies” does not establish that Kyle observed Parker and Gray committing “substantially identical” misconduct. Jackson , 814 F.3d at 779. Through its calculations, the majority opinion derives one metric of arguable comparability: 'Parker's 15-to-16-minute discrepancies tracked the 6-to-17-minute Page 38discrepancies that Kyle found in Gray’s records.” Maj. Op. 13. Fair enough. But focusing on this factor alone ignores Kyle's key reason for reporting Gray: her pattern of 50-minute lunchbreaks. In that way, the majority opinion's observations about 'discrepancies' are as unresponsive as Gray’s reflections on “manual entries.”

Neither metric addresses the greater 'relative severity' marked by Gray's pattern of entering the maximum time allowed. Jackson , 814 F.3d at 780. This practice, recall, was the “indicator” of “potential misrepresentation” “that got [Kyle’s] attention.” Kyle Dep., R. 30-21, PageID#1984–85. Parker, by comparison, never made a single 50-minute entry during the week Kyle reviewed her timesheets, much less a pattern of such entries, as did Gray.

Sweating none of the details, the majority opinion sees Gray’s and Parker’s lunch breaks as one in the same, as both reported breaks that “were at or just under the 50-minute limit.” Maj. Op. 13. But details matter. Gray's altered lunch breaks during the relevant week were 50 minutes, 50 minutes, and 50 minutes. Parker’s, on the other hand, were 40 minutes, 45 minutes, and 45 minutes. From Kyle’s vantage point, Parker’s entries displayed no discernible pattern— and certainly not the pattern of meeting the exact daily maximum time allowed, as did Gray’s entries. Accordingly, Kyle did not ignore “virtually identical conduct” by Parker. Id.

B. The majority opinion similarly fails to demonstrate on Gray's behalf that Kyle's actions proximately caused her termination. Here, it needs proof of 'some direct relation' between Kyle's flagging of her time entries and Gray's later firing. Staub , 562 U.S. at 419 (citation omitted). And any causal link, it bears noting, is cut off if Gray was ultimately fired because of information 'of independent origin that was not foreseeable' to Kyle. Id. at 420 (citation omitted).

The majority opinion, to its credit, seems to acknowledge that the “three occasions” Kyle “discovered” were not the basis for Gray’s termination. Maj. Op. 16. In other words, Gray was fired due to information that emerged later, that is, “other discrepancies” in her timekeeping as well as evidence of her “physical absences.” Id. at 14, 16. As Kyle played no role in providing this later information, it is 'of independent origin,' dooming Gray's effort to demonstrate a causal link to Kyle’s actions. Staub , 562 U.S. at 420 (citation omitted).

Page 39Resisting this conclusion, the majority opinion deems the information's emergence 'foreseeable' to Kyle, id. , because he 'anticipat[ed]' the discovery of 'other discrepancies,' Maj. Op. 16. This conclusion is yet one more Gray never asked us to draw. Neither in district court nor on appeal did she claim that Kyle foresaw HR discovering more time discrepancies than he reported. Having not made that argument, it is thus perhaps no surprise that she likewise did not cite the evidence the majority opinion relies on to make this point—a statement in a form Kyle sent to HR noting that he 'noticed other errors' on '2 random days in October.' Kyle Dep., R. 30-21, PageID#2104.

In this instance too, the argument the majority opinion crafts is as flawed as the (albeit different) one made by Gray. All things considered, the majority opinion fails to demonstrate that Kyle foresaw the investigation uncovering sufficient evidence to terminate Gray. Indeed, the majority opinion’s cursory analysis here seems to push Staub ’s proximate cause requirement towards a simple 'but for' standard, under which an employer is liable 'any time a biased employee . . . sets in motion the process that leads to an adverse employment action.” Poland v. Chertoff , 494 F.3d 1174, 1181 (9th Cir. 2007). Kyle, it is true, “set[] in motion” the events that led to Gray’s firing. Id. But what evidence do we have of a “direct relation” between his actions and Gray’s termination? Maj. Op. 16 (quoting Staub , 562 U.S. at 419).

The majority opinion gives three examples. First, that Kyle selectively reported Gray for conduct 'virtually identical' to Parker's. Id. This is a strained notion on its own, as shown above. But it is also inapt. That fact, if true, would show Kyle’s bias. Yet it does nothing to show whether he foresaw the investigation uncovering additional evidence against Gray. Second, Kyle apparently conveyed a false claim about Gray’s disciplinary history. Id. That too demonstrates very little about what he foresaw.

That leaves a third tenuous thread linking Kyle's actions and Gray's termination: the “other discrepancies” he noticed and reported to HR. Id. As an initial matter, Kyle detected these errors only after his supervisor instructed him to investigate, making their usefulness in showing what Kyle foresaw when he initially flagged Gray's entries doubly dubious. But more to the point, focusing on these errors elides Martin's role in this series of events. Martin, the manager who first decided to terminate Gray, testified that he did so because he learned that she Page 40had reported working when she was not in the building. In Martin’s words, “telling me you’re here when you're not' indicates an 'integrity issue' that justifies 'termination,' not merely a “productivity” or “performance issue,” which merits “counsel[ing] or coach[ing].” Martin Dep., R. 30-12, PageID#1391, #1403. Yet evidence of Gray's 'physical absences,' the majority opinion reminds us, 'surfaced only after Kyle reported her,' at which point HR pulled the “building-entry records” to which Kyle “lacked access.” Maj. Op. 12. Without evidence that Kyle anticipated HR discovering that Gray was out of the building on these instances, that “[un]foreseeable” “cause of an independent origin” is also a “superseding” one. Staub , 562 U.S. at 420 (citation omitted).

Having introduced an argument not made by Gray, the majority opinion then expands the underlying legal doctrine on proximate cause, seemingly crafting a bright-line rule that 'report[ing] true but selective information' will 'always' be the proximate cause of any subsequent employment action—no matter what information a later investigation uncovers. Maj. Op. 17. This case is a poor vehicle for doing so, as Kyle’s report, again, was not selective—none of Gray’s coworkers engaged in “virtually identical misconduct.” Madden v. Chattanooga City Wide Serv. Dep’t , 549 F.3d 666, 678 (6th Cir. 2008). And even if they had, the selective reports that served as “conduit[s]” for supervisor bias in past cases, id. (citations omitted), are different in kind from Kyle’s. In those instances, it bears emphasizing, the selective report provided the entire basis for the ultimate employment action, circumstances which the subsequent investigation “foreseeabl[y]” confirmed. Staub , 562 U.S. at 420 (citation omitted).

Take Madden , for example. There, a biased supervisor selectively reported that an employee had set off fireworks at work—the exact conduct for which the employee was later fired, even though it turned out to be a common form of delinquency. Madden , 549 F.3d at 677. The company’s ensuing investigation merely confirmed the supervisor’s allegations. Id. at 678. Given as much, the result of that investigation was obviously “foreseeable” to the supervisor—it was coterminous with his report. Staub , 562 U.S. at 420 (citation omitted). Likewise for Chattman v. Toho-Tenax, America, Inc. , 686 F.3d 339 (6th Cir. 2012), where a biased supervisor selectively reported the exact incident of horseplay for which the employee was disciplined. Id. at 352-53. The facts of that incident, as the supervisor could have foreseen, were later Page 41confirmed following an investigation. Id. at 344, 352. In short, these cases do not displace the ordinary rule that an intervening cause cuts off liability when that cause was “not foreseeable.” Staub , 562 U.S. at 420 (citation omitted). They merely apply it to the easy situation where the investigation reveals the same facts as reported.

Employing that same rule here yields a different result, given the intervening basis for Gray’s termination. To that end, the conduct Kyle reported was not enough to terminate Gray, nor could he have foreseen the emergence of the evidence that did ultimately justify that decision. The investigation, in other words, did not merely “confirm [Kyle]’s allegation.” Maj. Op. 17. Instead, the HR investigator uncovered additional, indisputable evidence that turned the discrepancies Kyle flagged from what he called a “concern,” Kyle Dep., R. 30-21, PageID#1984, into, in Martin's independent evaluation, an 'integrity issue[]' meriting 'termination,' Martin Dep., R. 30-12, PageID#1391.

As a final matter, it is difficult to accept the majority opinion’s suggestion that State Farm needed to disprove Gray’s allegation of retaliation before it could proceed to discipline her. See Maj. Op. 18. The unworkability of such a rule should give us pause from the outset: Can employees across our circuit now temporarily immunize themselves from workplace discipline merely by spitting out a claim of retaliation upon being informed of the allegations against them? Given this far-fetched result, it is no surprise that this rule finds no support in the cited passage from Justice Alito’s Staub concurrence.

True, the concurring opinion called for 'an independent investigation of the matter ' when a decisionmaker 'is put on notice that adverse information may be based on [illegal] animus.” Staub , 562 U.S. at 425 (Alito, J., concurring in the judgment) (emphasis added). But context makes clear that the referenced “matter” to be investigated is not the alleged retaliation itself, but rather the “accuracy of th[e] information” the supervisor provided. Id. By the time Gray 'alerted' State Farm 'to the possibility that [the] adverse information' implicating her “may be tainted,” State Farm had already conducted its “reasonable investigation,” one that not only confirmed “the accuracy of that information,” but also found separate evidence supporting Gray’s termination. Id. In short, under any duty to investigate to be gleaned from Justice Alito's opinion, State Farm did not “delegate[] . . . decisionmaking power to” Kyle. Id.

Page 42* * * * *

For jurists, it can sometimes be tempting to offer “solutions that the parties do not seek.” Zakora v. Chrisman , 44 F.4th 452, 486 (6th Cir. 2022) (Sutton, C.J., concurring in part and dissenting in part). But at day’s end, each of us must remain vigilant in honoring our modest role as “neutral arbiter of matters the parties present.” Sineneng-Smith , 140 S. Ct. at 1579. As has long been the case, it is “utterly inappropriate for the court to abandon its position of neutrality in favor of a role equivalent to champion for the non-moving party: seeking out facts, developing legal theories, and finding ways to defeat the motion.” Guarino , 980 F.2d at 406. Today’s case amply demonstrates how casting aside these party presentation principles to instead frame a party’s case in whatever way we see fit yields a wave of unending problems.

For starters, if one were in the business of crafting arguments the parties never made, that leaves available other unbriefed grounds on which we might also decide the case. Take, for example, the other aspects of cat’s paw liability described by the Supreme Court in Staub . In applying 'traditional agency principles' to base this theory on the actions of 'a supervisor,' Staub signaled that the agency-based framework from Burlington , 524 U.S. at 758, applies here too. Staub , 562 U.S. at 422 n.4 (citing Burlington and 'express[ing] no view as to whether the employer would be liable if a co-worker, rather than a supervisor, committed a discriminatory act that influenced the ultimate employment decision”). With that in mind, I doubt whether Kyle— who was not Gray's manager and seems to have lacked any 'hiring' or 'firing' power over her-was Gray's 'supervisor.' Vance v. Ball State Univ. , 570 U.S. 421, 424 (2013). Staub likewise calls on “general tort law” to demand proof of Kyle’s “specific intent to cause” Gray’s termination. Staub , 562 U.S. at 419, 423. On this point too, I see no 'significant probative evidence,” Walden , 119 F.4th at 1057, that Kyle either “desire[d] to cause” Gray’s termination or 'believe[d]' it 'substantially certain to result from' his actions, Staub , 462 U.S. at 422 n.3 (citation omitted). (The majority opinion, it also bears adding, confusingly collapses this “intent prong,' which focuses on the desired result, with the separate requirement of a retaliatory “motive[],” which hinges on the reasons for desiring that result. Maj. Op. 15. The Supreme Court clearly framed “motivat[ion]” and “inten[t]” as distinct requirements in a cat’s paw claim. Staub , 562 U.S. at 422.)

Page 43And then reflect on the ramifications on the litigation's other participants. How disappointed State Farm must be to find out it lost a case it was never asked to defend. Judge Marbley too, who has now been reversed for errors he never had the chance to commit. While there are no true winners in these extra-judicial circumstances, Gray, on the other hand, must feel like she reaped a windfall. After all, she seemingly found more capable advocates on the appellate bench than she did at counsel table. Better to be lucky, it turns out, than good.

As the case was presented to us by the parties, State Farm was entitled to summary judgment. On that basis, I would affirm the district court.

Page 44After we published our opinion in this case, see Gray v. State Farm Mut. Auto. Ins. Co. , 145 F.4th 630 (6th Cir. 2025), the majority amended its opinion in light of briefing at the panel rehearing stage. See Amended Maj. Op. at 5 n.2, 15-17 (adding, among other things, new language adopting a foreseeability analysis). These developments highlight yet again the erroneous path taken here from the outset.

A.1. To understand why, let me briefly recap the two reasons I dissented from the panel opinion. First, the theory of the case crafted by the majority opinion, in my view, completely rewrote Monica Gray’s briefing by expanding the cat’s paw issue from an afterthought rebuttal point to the purported central premise of Gray’s affirmative claim against State Farm. Second, even if Gray had in fact presented the cat's paw issue as her theory of liability, the majority opinion applied circuit law in an incorrect way to reach an incorrect decision. One reason I raised the threshold party presentation point was to emphasize the risky nature of an appeals court tackling on its own an issue no one has briefed. Why? Because without the benefit of briefing, we are likely to miss something the parties would have caught. See Gray , 145 F.4th at 661–62 (Readler, J., dissenting); United States v. McReynolds , 964 F.3d 555, 571 (6th Cir. 2020) (Griffin, J., concurring in part and dissenting in part) ('Acting . . . contrary [to the partypresentation principle] . . . increases the chances that we may decide an issue erroneously.' (citing Elonis v. United States , 575 U.S. 723, 742 (2015))); United States v. Campbell , 26 F.4th 860, 895 (11th Cir. 2022) (en banc) (Newsom, J., dissenting) (explaining that the party presentation principle 'enhances the prospect that courts will render correct judgments' and 'minimize[s] the risk of error' (citation modified)). After all, it is well understood that “[c]ounsel almost always know a great deal more about their cases than we do.” Greenlaw v. United States , 554 U.S. 237, 244 (2008) (citation modified).

That warning was prescient. Upon receiving our original panel opinion, the parties discovered that the majority opinion treated this case as one raising the cat’s paw theory as a basis for relief, contrary to how the parties and Judge Marbley had understood the case, and Page 45contrary to how Gray had presented her arguments in district court and on appeal. The response was foreseeable. Not long after our decision issued, the United States Chamber of Commerce, a familiar amicus brief filer across the federal courts, along with State Farm, brought to our attention the fact that we overlooked an on-point case, Romans v. Michigan Department of Human Services , 668 F.3d 826 (6th Cir. 2012), which, if considered, would have led to a different resolution here. Br. Amicus Curiae Chamber of Commerce of the United States of America Supporting Pet. Reh’g or Reh’g En Banc 8; see also Pet. Reh’g or Reh’g En Banc 6. Romans was our first decision interpreting Staub v. Proctor Hospital , 562 U.S. 411 (2011). In applying that precedent, Romans imposed the following rule for determining the key element of proximate cause in cat's paw cases: An employer is absolved of liability for disciplining an employee if the employer, upon conducting an investigation independent of the initial biased tip, discovers additional evidence supporting the adverse action. Romans , 668 F.3d at 836–37.

To illustrate the rule’s force, consider the facts in Romans . An employee was fired based off an independent investigation headed by a non-biased investigator, following a tip from an allegedly biased actor. While the subject matter of the initial biased investigation and subsequent investigation overlapped (both concerned the plaintiff’s treatment of a co-worker and how the plaintiff's conduct affected the workplace), the independent investigation revealed additional rule violations beyond what the initial report had discovered. A unanimous panel held that, in a setting like this, the supervisor’s actions were not the proximate cause of the subsequent firing. By independently confirming and discovering additional untainted grounds justifying the employee’s termination, the company’s investigation “determined” that the termination “was . . . entirely justified” “apart from” any influence from the biased supervisor. Id. at 836; see also Staub , 562 U.S. at 421 (explaining that a 'supervisor's biased report' is not a 'causal factor' when an 'independent investigation . . . determin[es] that the adverse action was, apart from the supervisor’s recommendation, entirely justified”); EEOC v. Ford Motor Co. , 782 F.3d 753, 768–69 (6th Cir. 2015) (en banc) (applying Romans and holding no proximate cause under cat’s paw theory in light of an independent investigation preceding plaintiff’s termination, even when a biased supervisor had an “effect” on plaintiff’s termination by “overs[eeing plaintiff’s] performance and report[ing] her failures” to decisionmakers). Put another way, it was irrelevant that the employer relied on the information reported by the biased actor in terminating the Page 46employee. Why? Because an independent investigation both confirmed the information and unearthed new information that justified the firing, removing the causal link between the biased actor and the employment action.

This straightforward rule should have controlled Gray’s case as well. Joe Kyle, Gray’s allegedly biased supervisor, reported discrepancies between when Gray claimed she was at work and inactivity on her computer on three occasions in November 2017. The company's subsequent investigation, in which Kyle played no role, discovered eight additional violations of State Farm’s pay policy by Gray. And it learned that her conduct was especially egregious, as the evidence indicated that Gray was not even present in the building on several of the occasions for which she falsely clocked in, a fact Kyle could not have known as he did not have access to these records. This conduct, separate and apart from what Kyle reported, was a sufficient basis for the company to terminate Gray. See Gray , 145 F.4th at 645 (Readler, J., dissenting). Under Romans , it follows, Kyle's report was not the proximate cause of Gray's termination. Instead, that action was “entirely justified” “apart from” the allegations Kyle relayed. Romans , 668 F.3d at 836. The company, in short, was neither Kyle’s cat’s paw nor a “conduit” for Kyle’s bias, but rather its own, uninfluenced decisionmaker. See Amended Maj. Op. at 13.

Had the cat’s paw issue been briefed as a core premise of Gray’s case, one (if not both) of the parties would surely have brought Romans to our attention. But as the cat's paw notion was hardly mentioned by the parties and never formed as an affirmative theory of relief, no party bothered to brief the issue in any detail (including by citing Romans ) at any point before we issued our initial opinion. So fleeting were the parties’ references to cat’s paw liability before the district court that Judge Marbley understandably did not include that theory in his wellreasoned opinion resolving State Farm’s summary judgment motion. Those of us on the court of appeals were in the same boat, receiving only scant references to the cat’s paw issue, one, again, never framed by Gray as an affirmative theory of recovery. Admittedly, in response to the majority opinion raising the cat’s paw theory, I missed Romans in my research. The same must be true for the majority opinion, which likewise did not cite, let alone address Romans in holding that Kyle's report about three violations could be the proximate cause of Gray's ensuing termination, even where additional violations untainted by Kyle's allegedly biased report Page 47justified that decision. See Gray , 145 F.4th at 643–44 . I do not fault any of us for overlooking Romans under the circumstances. But I continue to believe the majority opinion was unwise to invent Gray’s cat’s paw theory for her. Compare Ford Motor Co. , 782 F.3d at 768–69 (holding that the cat's paw theory was 'doubly forfeited'), with id. at 785 (Moore, J., dissenting) (interpreting a genuine dispute of material fact as to a judicially created cat’s paw theory). These later developments only cement that conclusion. Rather than the panel opinion responding to the parties’ arguments, as is our traditional practice, the backwards process here had the parties at the rehearing stage responding to our arguments resolving previously unbriefed issues.

Unwilling to accept Gray’s decision not to advance an affirmative cat’s paw theory at summary judgment, the amended majority opinion now suggests that State Farm forfeited its opportunity to raise Gray's failures in district court by not asserting the point at oral argument. See Amended Maj. Op. at 5 n.2. At the outset, it is difficult to ignore the irony in the majority raising this forfeiture point for the first time in its amended opinion, not its original one. It is equally difficult to accept this reframing, which focuses on the wrong party. It bears repeating that Gray, in response to State Farm’s summary judgment motion, had the obligation to support each element of an affirmative cat's paw claim before the district court. She never did. See Gray , 145 F.4th at 647-48, 653 (Readler, J., dissenting). Whether State Farm emphasized that shortcoming or emphasized another argument in response to the Court's questions is entirely beside the point.

Despite these oversights, Romans nonetheless controls here. And going forward, we must apply Romans as the law of the circuit. See United States v. Ferguson , 868 F.3d 514, 515 (6th Cir. 2017) (“One panel of this court may not overrule the decision of another panel . . . .”).

1. 2. Now on notice of Romans, the amended majority opinion attempts to reconcile that decision with the decision here. The amended majority opinion begins by reading Romans as holding that a supervisor who makes a “true but selective report of wrongdoing” against one of a handful of employees found to be engaged in the same pattern of misconduct 'would be attempting to use the company’s human resources as the ‘conduit’ for [the supervisor’s] bias.” Amended Maj. Op. at 13 (quoting Romans , 668 F.3d at 835). This statement, while superficially legally sound, is entirely irrelevant to Gray’s case. As explained in my original dissent, GrayPage 48 failed to provide comparators who engaged in her same pattern of misconduct and who were members of the same protected class. See Gray , 145 F.4th at 657-58 (Readler, J., dissenting). Thus, while Kyle’s report turned out to be true, Gray failed to provide evidence that the report was selective. And even if Gray had provided adequate comparators, State Farm was not a “conduit” for Kyle’s bias. Amended Maj. Op. at 13. True, an ultimate decisionmaker can be considered a 'conduit' of the biased supervisor when the grounds for termination repeat the company policy violations listed in the report. See, e.g. , Madden v. Chattanooga City Wide Serv. Dep’t , 549 F.3d 666, 678 (6th Cir. 2008). Indeed, that is the reason Staub declined to adopt a 'hard-and-fast rule' that all subsequent investigations by an employer necessarily break the chain of causation. 562 U.S. at 420-21. But here, as already explained, State Farm's independent investigation unearthed eight additional violations—none of which were included in Kyle’s initial biased report. See Gray , 145 F.4th at 661 (Readler, J., dissenting).

Nor is the amended majority opinion’s alternative path, namely, to distinguish Romans , any more persuasive. See Amended Maj. Op. at 16–17. According to the amended majority opinion, Romans held “that the independent investigation broke the causal chain because of the particular facts at issue in that case.” Id. at 16. But the 'particular facts' of Romans are not materially distinguishable from the facts at issue here. Just as the decisionmaker in Romans 'disclaimed' reliance on the biased report, State Farm's HR investigator here conducted a de novo review of Gray's timesheets and did not include Kyle in any conversation during that investigation. Likewise, while the plaintiff in Romans was terminated for violating four work rules, only one of which was related to the biased report, all violations stemmed from the behavior complained of in the report. Here, the eight additional timesheet discrepancies unearthed by the independent investigation were separate grounds for termination outside of the behavior reported by Kyle. In other words, State Farm’s independent investigation resulted “in an adverse action for reasons unrelated to [Kyle’s] original biased action,” thereby eliminating any proximate causation. See Staub , 562 U.S. at 421. Compare that conclusion to the unusual approach advocated by the amended majority opinion. It takes Staub and Romans as together creating a new 'hard-and-fast rule' in the other direction: No subsequent investigation-no matter how independent from the biased report—can ever break the chain of causation in a cat’s paw case involving a tip that turns out to be true.

Page 49Again, Romans held that an employer is absolved of liability if, upon conducting an investigation independent of the initial biased tip, the employer discovers additional evidence supporting the adverse action. If the amended majority opinion believes that Romans conflicts with Staub , see Amended Maj. Op. at 16 (noting that Romans could not hold 'that an independent investigation always breaks the chain of causation” in light of Staub ), those in the majority should call for an en banc panel to overrule Romans . But see Ford Motor Co. , 782 F.3d at 768–69 (en banc court citing Romans favorably for the proposition that there must be a “direct relationship between the injury asserted [termination] and the injurious conduct alleged” (citation modified)). But we may not simply ignore our binding precedent, as regrettably turns out to be the case once again today. See also Nash v. Bryce , -- F.4th --, No. 24-1263, 2025 WL 2778548, at *28 (6th Cir. Sept. 30, 2025) (Readler, J., dissenting) (“Precedent, in the end, is precedent, no matter how much we may dislike where it leads.”); United States v. Florence , 150 F.4th 773, 778 n.1 (6th Cir. 2025) ('As Justice Gorsuch explained, we don't get to 'pick and choose' the passages from the case ‘we happen to like.’” (quoting Loper Bright Enters. v. Raimondo , 144 S. Ct. 2244, 2292 (2024) (Gorsuch, J., concurring))).

B. Instead of honoring , the amended majority opinion sends this case to a jury.

That necessary, the Romans amended is majority opinion now explains, in light of Staub ’s

“foreseeability” element, a term never used in the original majority opinion. According to the

amended majority opinion, its decision 'simply echo[es] Staub ’s holding that a subsequent

investigation that does nothing more than confirm a supervisor’s true-but-selective report is

itself by insufficient to break the chain of proximate causation.' Amended Maj. Op. at 15 (citing

Staub , 562 U.S. at 420). By the same token, says the amended majority opinion, “an employer

will not be liable if its investigation uncovers a superseding 'cause of independent origin that

was not foreseeable’ from the supervisor’s biased action.” Id. (quoting Staub , 562 U.S. at 419

(citation modified)). But here, the amended majority opinion concludes, State Farm has not overcome that hurdle.

1. The amended majority opinion's foreseeability test manipulates Supreme Court

precedent and entirely ignores our own. Beginning with the Supreme Court, while Staub

references 'foreseeability,' 562 U.S. at 420, it says nothing about a 'supervisor's true-but- Page 50selective report” being “ itself insufficient to break the chain of proximate causation.” Amended Maj. Op. at 15. Nor can we pluck the word “foreseeable” from the remainder of the Supreme Court’s opinion, as does the amended majority opinion, to elevate it above Staub ’s holding: An employer is not liable where the employee’s termination is the result of an action—such as an independent investigation—that is “too remote, purely contingent, or indirect” from any biased conduct. Staub , 562 U.S. at 419. Like every decision, Staub must be read in context . And as with most cases in a cat’s paw setting, Staub turned on whether the hostile supervisors' action was the 'causal factor' underlying the termination decision. Id. at 421. In that case, the employment decision at issue was based exclusively on the report from the hostile actors. No independent investigation into that report took place, nor were additional facts uncovered to justify the termination. In thus concluding that the supervisors’ biased action caused the adverse employment action, Staub did not even undertake a foreseeability inquiry. Which makes sense. Foreseeability, remember, is merely a consideration inherent in a broader proximate causation analysis. See id. at 420-22. In practice, assessing foreseeability in this context entails asking whether an employer's adverse action was based on 'reasons unrelated to the supervisor's original biased action.' See id. at 421. So in cases where an HR department does launch an independent investigation following a hostile actor’s report, we must consider the causal nexus and determine whether that investigation uncovered unknown facts that, standing alone, justify termination. An independent investigation, in other words, resolves the foreseeability inquiry.

This is how we have interpreted Staub . Turn back to Romans , which held that there is no causal nexus between a hostile supervisor’s report and the ultimate adverse action—even if the employer relied on information in the hostile supervisor's report-where the employer also relied on additional information uncovered in an independent investigation. 668 F.3d at 836–37. Romans could not have been more clear on that point: “Defendant conducted an independent investigation at [an independent decisionmaker’s] direction that breaks the causal chain between [the biased employee’s] alleged animus and [the decisionmaker’s] action.” Id. at 836. So long as additional reasons for termination are unearthed during the independent investigation, an employer is not the 'cat's paw' of the bad actor, even if the reasons for termination include information from the bad actor’s report. See Romans , 668 F.3d at 836–37; Goodsite v. Norfolk S. Ry. Co. , 573 F. App'x 572, 587 (6th Cir. 2014) (applying Staub to uphold district court's Page 51determination on summary judgment that the company's independent investigation broke the causal nexus); Sami v. Detroit Med. Ctr. , 591 F. App'x 419, 426-27 (6th Cir. 2014) (applying Romans to conclude, on summary judgment, that the company’s independent investigation broke the causal nexus). In other words, liability does not attach to an employer’s actions so long as it does not “rubber[] stamp” a biased report. Bose v. Bea , 947 F.3d 983, 991 (6th Cir. 2020).

Romans demonstrates the flaw in the amended majority opinion's invocation of a freestanding foreseeability inquiry. If that were the proper approach, Romans would have required a jury to consider whether there was any relationship between the initial purported biased report and the subsequent employment decision, even after an independent investigation. It did not. Indeed, until today, we have uniformly resolved questions of liability on summary judgment in the employer's favor where 'the employer's investigation' uncovered additional grounds justifying the employer’s actions. Romans , 668 F.3d at 836 (quoting Staub , 562 U.S. at 421); see, e.g. , Goodsite , 573 F. App’x at 587; Sami , 591 F. App’x at 426–27. By unearthing an independent foreseeability requirement and then directing the issue as a question for the jury, the amended majority opinion disturbs settled ground and stands alone.

Equally groundbreaking is the amended majority opinion’s decision to disqualify as valid grounds for an employer’s post-investigative decision any consideration of offenses that might be related in some sense to the originally reported offense. See Amended Maj. Op. at 16. That approach, again, is inconsistent with our case law, which asks whether an independent investigation occurred and, if so, whether it uncovered new violations of company policy. See Romans , 668 F.3d at 836-37 (recognizing that violations independently uncovered by the company’s investigation were still related to the incident described in the biased actor’s report). We have never required that there be no relationship whatsoever between a newly discovered violation and an earlier reported one before an employer can act on the later-discovered violation. See Ford Motor Co. , 782 F.3d 768–69 (holding that a biased supervisor’s tip that had an “ effect on her termination” does not by itself establish proximate cause for a cat’s paw claim). And for good reason. Otherwise, employees would be shielded from any job-related consequences for massive workplace misconduct where it just so happens that the initial discovery of that wrongdoing was made by an allegedly biased supervisor. See Woods v. City of Page 52Berwyn , 803 F.3d 865, 870 (7th Cir. 2015) ('If the ultimate decision-maker does determine whether the adverse action is entirely justified apart from the supervisor’s recommendation, then the subordinate's purported bias might not subject the employer to liability. . . . To hold otherwise would be to rule that whenever a discriminatory subordinate makes an allegation or institutes a charge and the plaintiff-employee is fired, there are no steps the ultimate decisionmaker could ever take to break that chain of proximate causation. That cannot be so.'). Yet under the amended majority opinion’s rule, an employee reported for embezzling $50 would be untouchable even where a later independent investigation uncovers embezzlement of $50,000, or even $5 million. So too for an employee alleged to have sexually harassed a co-worker, where a subsequent review reveals the employee has been serially harassing employees across the workplace. Why the amended majority opinion opts to contort our precedent to protect wrongdoers like these is difficult to understand.

The amended majority opinion’s citations do not change things. See Amended Maj. Op. at 14. Both cases were addressed in my original dissent. Madden , a preStaub case, involved an employee who was fired for the “exact conduct” that was reported by the biased supervisor with no independent investigation. Gray , 145 F.4th at 660 (Readler, J., dissenting). What is more, Madden carries little weight postStaub . Likewise, in Chattman v. Toho Tenax America, Inc. , 686 F.3d 339 (6th Cir. 2012), there was no daylight between what the biased supervisor reported and the reason the employee was fired. See Gray , 145 F.4th at 660 (Readler, J., dissenting). More broadly, Chattman does not stand for the proposition the amended majority opinion suggests: that questions about whether an employer’s adverse action was related to the biased actor’s tip are for the jury. True, Chattman found a genuine dispute of fact over the issue of causation. But that was the case because the disputed evidence when viewed in a light most favorable to the plaintiff showed that the investigation at issue was not independent, as our cat’s paw precedent requires. According to that evidence, the biased supervisor was no ordinary employee-he was the HR supervisor. Chattman , 686 F.3d at 353. And unlike Kyle, the supervisor there “misinformed and selectively informed” corporate officials about the incident at issue, and later 'actively inserted himself into the decisionmaking process,' tainting that investigation to the point that it was no longer independent. Id. But nothing in Chattman suggests that the question of relatedness is one the jury should determine no matter the facts Page 53presented, in particular where, as here, no one disputes that State Farm undertook an investigation independent of Kyle. See id. (“Thus, we cannot say that the investigation conducted by Verbruggen and Lane was ‘unrelated’ to Tullock’s actions.” (emphasis added)). And more to the point, Chattman does not concern, nor does it undermine, the rule that an independent investigation that reveals additional grounds for termination absolves an employer of liability. Romans , 668 F.3d at 836–37. In any event, Chattman , which cited Romans, cannot be read to overturn that earlier decision. See Chattman , 686 F.3d at 351 n.10; Ferguson , 868 F.3d at 515.

Beyond not having any basis in law, the amended majority opinion's newly created relatedness test is also arbitrary in its application. According to the amended majority opinion, all 11 of Gray's company policy violations fall under the relatedness umbrella of 'false timekeeping” and, therefore, are a subject for the jury to sort out. See Amended Maj. Op. at 17. It is difficult to rationalize why that level of generality-categorizing every violation as a “timekeeping” matter—is appropriate. It is equally confounding why a court cannot distinguish, as a matter of law, the violations at issue here, since the facts are undisputed. Other than generically having to do with 'time,' any reasonable person would find a stark distinction between the subject matter of Kyle’s tip (that Gray was not in front of her computer when she said she was working) and what the subsequent investigation revealed (that Gray’s conduct went beyond mere timekeeping errors-she was a repeated truant from the workplace). And then consider the amended majority opinion’s concession that we can sometimes resolve questions of relatedness at summary judgment. For this case, the amended majority opinion reasons, had State Farm's subsequent investigation revealed 'that Gray was a wanted fugitive or had embezzled funds from the company,” such violations would be, as a matter of law, “unrelated” and “not foreseeable.” See Amended Maj. Op. at 16. But why those respective violations could not similarly be tucked under a similarly generic umbrella of “theft” or “integrity” is anyone’s guess (after all, is stealing time all that different than stealing money?). Put another way, if it was somehow foreseeable to Kyle that his tip about Gray's computer time would reveal that Gray repeatedly was not even in the office while being paid, it seems equally foreseeable that Gray could have been engaging in other integrity-based misconduct.

Page 54Falling short in all other respects, the amended majority opinion turns to newly embraced evidence in an effort to bolster its foreseeability analysis. In its amended opinion, the majority opinion now claims, without citation, that Kyle 'suggested in his report to HR that an investigation of Gray would uncover additional timekeeping errors,” making it foreseeable that HR would unearth additional violations, which, the amended majority opinion surmises, was Kyle’s intent all along. See Amended Maj. Op. at 15. But the evidence apparently relied upon by the amended majority opinion reveals no such thing. The opinion appears to invoke Kyle’s “Performance Scoping Document,” an exhibit from his deposition. Kyle Dep., R. 30-21, PageID 2103. Yet the exhibit contains neither a suggestion that State Farm would uncover additional violations nor a roadmap for HR to follow in finding more violations. At most, Kyle states that '[h]e looked at 2 random days in October and noticed other errors' in addition to the three November violations delineated in his report to HR. Id. at PageID 2104. Those observations regarding October errors are hardly “suggestions” to HR when the independent HR investigation focused exclusively on November violations. That investigation, again, uncovered eight additional violations in November alone. And, it bears repeating, those violations were different in kind from the three violations reported by Kyle. Gray , 145 F.4th at 645 (Readler, J., dissenting) (noting that the HR investigation uncovered that Gray “was not even badged into the building” during instances where she reported working). There is no indication that HR even assessed Kyle’s purported “suggestion,” let alone acted on it. In the end, the amended majority opinion's last-minute effort to portray Kyle as having implicitly directed State Farm's HR investigation is entirely at odds with the undisputed record.

2. Even if one accepted the amended majority opinion’s numerous misreadings of Circuit precedent, deferring the proximate cause question to the jury here, as the amended majority opinion orders, makes little sense in light of the undisputed record confirming that Gray was fired due to information that emerged from the independent HR investigation. Appellees Br. 18– 19; R.30-32, PageID 2690–94. This is not a case where State Farm “unthinkingly adopt[ed] the recommendations of [its] biased lower-level supervisors.” Marshall v. Rawlings Co. , 854 F.3d 368, 378 (6th Cir. 2017). Nor does State Farm seek to evade liability 'through willful blindness.” Id . As all acknowledge, State Farm did not merely rubber stamp Kyle’s complaint. See Bose , 947 F.3d at 991. Rather, the company initiated an independent investigation into Page 55whether Gray violated State Farm's policies. Appellees Br. 11; R.30-5, PageID 958. The independent investigator retained by State Farm’s HR department requested and reviewed Gray’s workstation and time sheet audit records for November 2017, which uncovered 11 time-entry discrepancies. R. 30-5, PageID 959; R. 30-32, PageID 2693–94. Some discrepancies revealed that Gray was not even in the building at the time she logged in. R. 30-9, PageID 1232–33. The investigator then interviewed Gray, where she failed to explain a single time-entry discrepancy. R. 30-2, PageID 372; R. 30-5, PageID 960. After the interview and a subsequent discussion with Gray’s former supervisor (Martin), the investigator prepared an Executive Summary. And it was that document which provided the basis for Gray's termination. R. 30-32, PageID 2693-94. Kyle’s original report, which included only three instances of discrepancies in Gray’s timekeeping, is nowhere mentioned in the Executive Summary or the related internal communications concerning the decision to fire Gray. R. 30-32, PageID 2690–94. Remember, Kyle's suggestion that Gray's timekeeping entries did not match her computer records was simply a tip and made no mention of the fact that those entries encompassed periods when she had not even swiped her badge to reenter the building. Indeed, as the amended majority opinion acknowledges, Kyle would have had no way of knowing as much, as he 'lacked access' to building entry records. Gray , 145 F.4th at 641.

With these facts not in dispute, whether Kyle's 'true-but-selective report [was] the proximate cause of [the] subsequent adverse employment action” taken against Gray, Amended Maj. Op. at 18, is a legal question for the court to decide at summary judgment. Eisenhour v. Weber County , 897 F.3d 1272, 1280 (10th Cir. 2018) ('[P]roximate cause . . . becomes a question of law for a court to decide when no evidence supports proximate causation.”); Lahar v. Oakland County , 304 F. App’x 354, 355 (6th Cir. 2008) (affirming grant of summary judgement in ADEA claim because plaintiff “failed to demonstrate a genuine issue of fact with respect to . . . causation”); Devonshire v. Johnston Grp. First Advisors , 166 F. App’x 811, 814 (6th Cir. 2006) (“Federal Rule of Civil Procedure 56(c) . . . instructs federal courts to grant summary judgment on substantive legal issues, proximate cause included, when there are no genuinely disputed issues of material fact and the moving party is entitled to judgment as a matter of law.'); Corrigan v. E. W. Bohren Transp. Co. , 408 F.2d 301, 303 (6th Cir. 1968) ('The question of proximate cause is ordinarily one of fact, but, where there is no conflict in the evidence, such Page 56question becomes one of law.” (citation modified)). Put differently, “whether Gray’s termination” was “based on . . . additional discrepancies” discovered by State Farm, Amended Maj. Op. at 19, is a question of law given that the underlying record of events is undisputed. Bluecross Blueshield of Tenn., Inc. v. Nicolopoulos , 136 F.4th 681, 687 (6th Cir. 2025) (explaining that Fed. R. Civ. P. 56 entitles a party to judgment as a matter of law when no genuine issue of material fact exists). So it is our obligation, indeed duty, to resolve this case now.

Adopting the amended majority opinion's contrary approach in essence nullifies the summary judgment stage. If questions like Gray’s may not be resolved as a matter of law when the underlying facts are not in dispute, no cat's paw case will ever be suitable for summary judgment. Not only is that result contrary to the Rules of Civil Procedure, but it also runs counter to how we routinely resolved cat’s paw cases before now. See, e.g. , Romans , 668 F.3d at 836–37, 843 (affirming district court grant of summary judgment of cat’s paw liability claim); Davis v. Omni-Care Inc. , 482 F. App’x 102, 111 (6th Cir. 2012) (per curiam) (same); Sami , 591 F. App'x at 427, 429 (same); Min Li v. Qi Jiang , 673 F. App'x 470, 474-76 (6th Cir. 2016) (same). As in these cases, we should affirm Judge Marbley's grant of summary judgment to State Farm.