SANDERS v. United States of America, Intervenor. (2026)

OPINION

Petitioner David Lee Sanders was sentenced to death for the murder and robbery of two men inside a Kentucky convenience store in 1987. A jury rejected Sanders's only defense: that he was not guilty by reason of insanity. After Sanders's initial sentencing, a protracted series of proceedings followed, which yielded four Kentucky Supreme Court decisions, three denials of certiorari from the U.S. Supreme Court, and a federal district court decision (and reconsideration). Before us now, Sanders makes two main arguments. First, he argues that the Antiterrorism and Effective Death Penalty Act (“AEDPA”), which narrows our ability to grant habeas relief when a state court decision has already decided a petitioner's claim on the merits, is unconstitutional. Second, he argues that ambiguous jury instructions regarding his insanity defense, deficient performance by trial counsel, and cumulative error violated his constitutional rights.

We hold that AEDPA is constitutional and, applying it to Sanders's claims, the Kentucky Supreme Court's decisions on jury instructions and several of Sanders's ineffective assistance of counsel claims were not contrary to law and did not involve an unreasonable application of federal law. As for Sanders's claims not accorded AEDPA deference, we hold that they still fail under de novo review because Sanders fails to show prejudice. Finally, Sanders's cumulative-error claim was procedurally defaulted and otherwise lacks merit. Thus, we affirm the district court's denial of Sanders's habeas petition in its entirety.

I.

On January 28, 1987, a customer stopped by the Boone Variety Store near Richmond, Kentucky. Upon entering the store, she was confronted with a horrific scene. Behind the counter, Jim Brandenburg lay slumped over. Near the door, Wayne Hatch was face down, gurgling in a pool of his blood. Both men had bullet holes in the back of their heads. Two bank bags had been taken from the store, and the men's wallets had been stolen. Less than an hour earlier, Sanders had stopped by the store. After paying for his purchases, Sanders asked Brandenburg, who was behind the counter, if he could use the phone. Brandenburg directed Sanders to the payphone. Sanders then called his mother and confirmed he would come over on Thursday evening. Then he called his wife to let her know he would not come home until Friday.

Sanders then went into his truck and retrieved two loaded guns. He reentered the store, where he saw Brandenburg behind the counter with his back to him. Sanders walked to the counter, placed the barrel of one of his guns against the back of Brandenburg's head, and pulled the trigger, killing him. Sanders then took Brandenburg's wallet, as well as a bag of money sitting on the counter. Hatch then entered the store through the front door while Sanders still had his gun out. Although it is uncertain how Hatch got on the floor, Sanders shot Hatch in the back of the head while he was lying face down on the floor and took his wallet. Sanders then drove home, tossing some items he had stolen out the window. 1

A.

Based on the timing of the murders and outgoing calls from the store's telephone, it did not take long for officers to suspect Sanders was responsible. When police first interviewed Sanders, he denied committing the murders and told various stories supporting his innocence. However, as the evidence accumulated, Sanders eventually confessed. After Sanders's confession, concerns about his mental health developed. On the day of his confession, the jail psychologist, Dr. John Moffit, wrote a letter to the trial judge stating that he felt that Sanders “[was] suicidal at this time, and hence, all precaution should be taken to prevent him from acting on these impulses.” 2 DE 165-1, Moffit Ltr, Page ID 2813. A few weeks later, Sanders's attorney, J. Kevin Charters, moved to commit Sanders to a state institution for a mental health examination, believing that he might be mentally ill. In response, the Kentucky state trial court ordered Sanders to be moved to the Kentucky Correctional Psychiatric Center (“KCPC”), where he was to be evaluated for competence to stand trial, competency at the time the crimes were committed, and for any other mental issues.

A month and a half later, Dr. Candace Walker from KCPC sent her report to the trial court. The report concluded that Sanders suffered from no mental illness that could absolve him of responsibility for his crime. According to the report, Sanders understood the proceedings against him, was of at least average intelligence, and could cooperate with his attorney in his defense. The report also concluded that Sanders did not suffer from any mental disease or defect that could prevent him from conforming his conduct to the law and that there was no evidence that he was suffering from a mental defect when he committed the alleged crimes. Finally, the report also noted that Sanders described his background, particularly his childhood, differently to various KCPC members.

One of the KCPC team members involved in the report was Dr. Frank Flenning. Flenning wrote an independent report that was not included in the ultimate Walker report. Flenning's report was more mixed and concluded that there was some evidence to suggest that Sanders suffered from a mental deficiency resulting from traumatic experiences in his childhood. But notably, the existence of this report was never made known to any of the parties or presented to the jury. This is likely due to Walker's report that they had “found no medical evidence to indicate that he was suffering from such a disorder at the time of the alleged crime.” DE 165-4, Walker KCPC Pretrial Competency Evaluation Rep., Page ID 2819. Such a statement indicated that Walker's team was unanimous in its conclusion.

Sanders's attorney also did not interview Walker or any member of her team. Instead, Charters opted to retain Dr. Stuart Cooke, a clinical psychologist, to testify on Sanders's behalf. Cooke interviewed Sanders and administered various tests—all within two and a half hours. Cooke also interviewed Sanders's brother and read the Walker report. With this information, Cooke concluded that Sanders suffered from a depersonalization disorder during the time of the crime and could not control his actions.

B.

At trial, Sanders asserted an insanity defense. The defense called two witnesses: Sanders himself and Cooke. Sanders admitted to shooting and killing Brandenburg and Hatch but insisted he could not control himself. Sanders testified that as he went to his truck, he felt “funny” and remembered telling himself: “Oh God no. Not – not again. You are not going to do it again.” DE 166-3, Trial Tr., Page ID 3781. Sanders testified that he tried to stop himself but could not. Sanders further told the jury: “I watched my left hand raise, and it had a loaded gun in it. I could not stop it․ I didn't watch that man fall, I didn't want that to happen. I didn't want any of that to happen.” Id. at 3782–83.

Next, the defense called Cooke to the stand. Cooke corroborated Sanders's testimony and told the jury that Sanders suffered from a depersonalization disorder, causing him to lose control of his body and his awareness of reality. In response, the prosecution called Walker to the stand. Walker described the full battery of psychological tests performed on Sanders. After a month and a half of testing, Walker testified that, while Sanders likely had a personality disorder, he did not suffer from any disease that would impair his competence. She noted that Sanders had never presented any evidence of depersonalization disorder. Instead, Walker suggested that Sanders's shifting stories and attempts to manipulate staff members suggested that he may have been attempting to fabricate a mental illness to avoid responsibility.

According to Sanders, Cooke's credibility was significantly undermined in the prosecutor's closing argument in several ways that support ineffective assistance of counsel argument. First, Cooke mispronounced his last name as “Saunders,” a fact the prosecution pointed out to the jury. Second, Cooke made a few factual mistakes such as testifying that he saw Sanders on a date and time when Sanders was unquestionably in court. Third, Cooke only interviewed and tested Sanders for two and a half hours, far less than the one and a half months of extensive testing Walker and her team had conducted.

After all the testimony was completed, the court gave the jury instructions on four offenses: two counts of murder and two counts of first-degree robbery. Sanders did not object to any of these instructions. The court listed each count and told the jury it could only find Sanders guilty “if they believe[d] from the evidence beyond a reasonable doubt [that he committed the crimes.]” DE 166-4, Trial Tr., Page ID 3914–16. After listing these offenses, the court further told the jury that they “shall find [Sanders] not guilty” for reasons of insanity if they “believe[d] from the evidence” that Sanders could not control his actions. Id. 3917. This is a different standard than the beyond a reasonable doubt standard. Under Kentucky law, trial courts instruct juries that they can find insanity under the lower standard of “from the evidence.” Gall v. Commonwealth, 607 S.W.2d 97, 110 (Ky. 1980), overruled on other grounds by Payne v. Commonwealth, 623 S.W.2d 867 (Ky. 1981). The jury instructions, as the jury heard and read them, were as follows:

Even though you might otherwise find the defendant guilty of the offenses mentioned in instructions number 2, 3, 4 and 5, [(murder and robbery instructions)] if you believe from the evidence that at the time David Lee Sanders committed these offenses, if he did so, he was of unsound mind, you shall find him not guilty and say in your verdict that you find him not guilty on the ground of insanity. Before the defendant can be excused from the ground of insanity, you must believe from the evidence that at the time of the act in question the defendant as a result of mental disease or defect, A, did not have substantial capacity to appreciate the criminal nature of the act or, B, if he did have such capacity he did not have substantial capacity to conform his conduct to the requirements of the law. The term “mental disease” or “defect” does not include an abnormality manifested only by repeated criminal or otherwise anti-social conduct.

DE 166-4, Trial Tr., Page ID 3916–17 (emphases added).

The jury found Sanders guilty on all counts.

C.

Having been found guilty of the murders, Sanders's trial proceeded to the penalty phase. The entire penalty proceeding, including the closing argument by Charters, lasted about an hour. Sanders first presented testimony from his employer, his neighbor, the assistant principal of his high school, and his sister. All expressed shock that Sanders had committed these crimes and told the jury that he was a good person. After these four witnesses, Sanders himself testified. Sanders testified that he could not control his actions and did not even need the money he stole. Sanders also told the jury that the evaluators at KCPC did not appropriately understand him or his mental issues. Sanders did not testify to, and was not asked by his counsel, about any childhood abuse that he had suffered. The jury recommended, and the court imposed a sentence of death for Sanders's killing of Brandenburg and Hatch. Sanders v. Commonwealth, 801 S.W.2d 665, 668 (Ky. 1990) (“Sanders I”).

D.

After Sanders's conviction and sentence, a series of appeals and collateral attacks followed. First, Sanders directly appealed to the Kentucky Supreme Court, which affirmed his sentence. Sanders I, 801 S.W.2d at 684. The Kentucky Supreme Court rejected his argument that the trial court had failed to specify the burden of proof, noting that in Kentucky, the common insanity instruction is “believed ‘from the evidence.’ ” Id. at 679 (citing Gall, 607 S.W.2d at 110). Sanders subsequently petitioned for a writ of certiorari, which the U.S. Supreme Court denied. Sanders v. Kentucky, 502 U.S. 831, 112 S.Ct. 107, 116 L.Ed.2d 76 (1991).

Sanders then pursued several postconviction proceedings in Kentucky state court, seeking to vacate his death sentence. Sanders v. Commonwealth, 89 S.W.3d 380 (Ky. 2002) (“Sanders II”), overruled on other grounds by Leonard v. Commonwealth, 279 S.W.3d 151 (Ky. 2009); see also Ky. Rule of Crim. Pro. 11.42 (Kentucky criminal procedure rule for modifying a defendant's sentence). In Sanders II, the Kentucky Supreme Court considered and rejected all of Sanders's arguments that his counsel was constitutionally defective. Sanders II, 89 S.W.3d at 394. Sanders filed another petition for a writ of certiorari, which the U.S. Supreme Court again denied. Sanders v. Kentucky, 540 U.S. 838, 124 S.Ct. 96, 157 L.Ed.2d 70 (2003).

Sanders then petitioned for a writ of habeas corpus in federal court in 2003. The district court held Sanders's petition in abeyance, so Sanders could exhaust his state court claims. In 2011, the Kentucky Supreme Court again rejected all of Sanders's claims for post-conviction relief. Sanders v. Commonwealth, 339 S.W.3d 427 (Ky. 2011) (“Sanders III”). Sanders filed a third petition for a writ of certiorari, which the U.S. Supreme Court denied once again. Sanders v. Kentucky, 566 U.S. 907, 132 S.Ct. 1792, 182 L.Ed.2d 620 (2012).

Having exhausted all his state claims, Sanders returned to federal court to file a supplemental petition for writ of habeas corpus. In February 2015, the district court denied Sanders's petition on all grounds. In June 2016, the district court first granted Sanders a Certificate of Appealability (“COA”) for his argument that Charters had provided constitutionally ineffective assistance of counsel by failing to appropriately prepare and present mitigating evidence at the trial's penalty phase. Then, in May 2017, the district court granted Sanders a COA for his claims that the jury instructions on insanity had violated due process, that the Kentucky Supreme Court had unreasonably applied Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), and that Sanders had sustained cumulative prejudice due to his counsel's deficiencies. We later granted additional COAs concerning whether Charters had rendered ineffective assistance of counsel for (1) “failing to retain an adequate mental-health expert”; (2) “not pursuing leads relating to the evaluation of a mental-health expert”; (3) “inadequately preparing Sanders to testify during the trial's sentencing phase”; and (4) “repeatedly failing to move for additional evaluation of Sanders's competency.” CA6 R.34-1, Order of the Court, at 3–4.

At this point, we again held Sanders's case in abeyance while he pursued a new claim in Kentucky state court arising from a recent Kentucky Supreme Court decision, which Sanders argued called into question the correctness of the state trial court's previous ruling denying his motion to reopen. Ultimately, the Kentucky Supreme Court rejected Sanders's request to reopen his case. Sanders v. Commonwealth, 2022 WL 3640912, at *4 (Ky. Aug. 18, 2022) (“Sanders IV”).

Now back before us, Sanders argues in his supplemental briefing that § 2254(d) of AEDPA is unconstitutional. See 28 U.S.C. § 2254(d). Because the constitutionality of a federal statute was called into question, the United States exercised its statutory right to intervene as an independent party. See 28 U.S.C. § 2403(a). Thus, we are now presented with two general claims: (i) whether AEDPA is unconstitutional and (ii) whether Sanders's constitutional rights were violated during his state trial proceedings under AEDPA review.

II.

Sanders brings a habeas petition under § 2254(d) of AEDPA, which limits our ability to grant remedial relief to a state prisoner whose claims were previously adjudicated on the merits in the state court. Reiner v. Woods, 955 F.3d 549, 556 (6th Cir. 2020). Under AEDPA, we may only grant a writ of habeas corpus if the claim disputed in the underlying state court proceeding “(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States” or “(2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(1), (2). The Supreme Court has regularly “reminded” courts that this test is “difficult to meet.” White v. Woodall, 572 U.S. 415, 419, 134 S.Ct. 1697, 188 L.Ed.2d 698 (2014) (citation omitted).

Under § 2254(d)(1), or the “contrary to” clause, “a federal habeas court may grant the writ if the state court arrives at a conclusion opposite to that reached by [the Supreme] Court on a question of law or if the state court decides a case differently than [the Supreme] Court has on a set of materially indistinguishable facts.” Williams v. Taylor, 529 U.S. 362, 412–13, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000). State prisoners must make two basic showings. See Bergman v. Howard, 54 F.4th 950, 957 (6th Cir. 2022), cert. denied, ––– U.S. ––––, 143 S. Ct. 2445, 216 L.Ed.2d 423 (2023). First, they must identify a “clearly established” principle of “Federal law” that the “Supreme Court” has pronounced. 28 U.S.C. § 2254(d)(1). This language allows prisoners to seek relief based on just one source: Supreme Court decisions. Id. Even more narrowly, prisoners must rely on the “holdings” of those decisions; the Court's dicta does not clearly establish binding law. Woodall, 572 U.S. at 419, 134 S.Ct. 1697 (citation omitted). Second, once a prisoner has identified a clearly established holding with specificity, the prisoner must show that a state court's denial of relief was “contrary to” or an “unreasonable application” of this holding. 28 U.S.C. § 2254(d)(1); Bergman, 54 F.4th at 961. “A state decision cannot be ‘contrary to’ a Supreme Court holding unless it adopts a conflicting legal rule or reaches the opposite result in a case with materially identical facts.” Fields v. Jordan, 86 F.4th 218, 232 (6th Cir. 2023) (en banc) (citation omitted), cert. denied, Fields v. Plappert, ––– U.S. ––––, 144 S. Ct. 2635, 219 L.Ed.2d 1270 (2024).

Under § 2254(d)(2), or the “unreasonable application” clause, “a federal habeas court may grant the writ if the state court identifies the correct governing legal principle from [the Supreme] Court's decisions but unreasonably applies that principle to the facts of the [petitioner's] case.” Williams v. Taylor, 529 U.S. at 413, 120 S.Ct. 1495. “[T]he state court's factual findings are presumed correct unless rebutted by the habeas petitioner by clear and convincing evidence.” Moore v. Mitchell, 708 F.3d 760, 775 (6th Cir. 2013) (citation omitted). In short, federal habeas relief is available only if the state court's decision was “objectively unreasonable.” Lockyer v. Andrade, 538 U.S. 63, 75, 123 S.Ct. 1166, 155 L.Ed.2d 144 (2003); Joseph v. Coyle, 469 F.3d 441, 468 (6th Cir. 2006). A federal court sitting in habeas review may not supersede a state court's factual determination just because “[r]easonable minds reviewing the record might disagree about the finding in question.” Brumfield v. Cain, 576 U.S. 305, 314, 135 S.Ct. 2269, 192 L.Ed.2d 356 (2015) (citation omitted). In fact, under the “unreasonable determination of the facts” prong, the burden is higher than “clear error.” Davis v. Ayala, 576 U.S. 257, 271, 135 S.Ct. 2187, 192 L.Ed.2d 323 (2015).

Under either clause, the petitioner must show that the state court's ruling was “so lacking in justification that there was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Harrington v. Richter, 562 U.S. 86, 103, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011). Thus, the state court's determination also cannot be unreasonable if “fairminded jurists” could disagree on the decision. See id. at 101, 131 S.Ct. 770. Section 2254(d) is a “purposefully demanding standard,” and it requires that state court determinations “be given the benefit of the doubt.” Haight v. Jordan, 59 F.4th 817, 831 (6th Cir. 2023) (quotations omitted), cert. denied, ––– U.S. ––––, 144 S. Ct. 578, 217 L.Ed.2d 307 (2024). That is because habeas under AEDPA is not for ordinary error correction; it is a way for federal courts to police the edges of the Constitution and “guard against extreme malfunctions in the state criminal justice systems[.]” Harrington, 562 U.S. at 102, 131 S.Ct. 770. Ultimately, we are prohibited from issuing a remedy merely because we conclude in our independent judgment that the state court reached a mistaken, or even clearly erroneous, result. Price v. Vincent, 538 U.S. 634, 641, 123 S.Ct. 1848, 155 L.Ed.2d 877 (2003); Fields, 86 F.4th at 232.

Because the district court did not conduct an evidentiary hearing and relied exclusively on the state court record in its habeas decision, we review both the district court's legal and factual findings de novo. Barton v. Warden, S. Ohio Corr. Facility, 786 F.3d 450, 460 (6th Cir. 2015).

III.

Before applying AEDPA, we must address its constitutionality. Sanders contends that the deference provision of AEDPA, see 28 U.S.C. § 2254(d), is unconstitutional for two distinctive, but overlapping reasons. First, Sanders argues that in eliminating “Chevron deference,” Loper Bright Enterprises v. Raimondo, 603 U.S. 369, 144 S.Ct. 2244, 219 L.Ed.2d 832 (2024) also renders “AEDPA deference” unconstitutional. See 603 U.S. at 413, 144 S.Ct. 2244. In particular, Sanders directs us to the concurrences of Justices Thomas and Gorsuch, which claim that Chevron deference violates our Constitution's separation of powers. See id. at 413–16, 144 S.Ct. 2244 (Thomas, J., concurring); id. at 416–48, 144 S.Ct. 2244 (Gorsuch, J., concurring). Second, Sanders argues that even putting Loper Bright aside, AEDPA intrudes upon our Article III powers by forcing us to abdicate our constitutional responsibility to “say what the law is.” Marbury v. Madison, 1 Cranch 137, 177, 2 L.Ed. 60 (1803). We disagree and join every court that has addressed this issue and affirm the constitutionality of AEDPA.

A.

AEDPA deference is constitutional. The Supreme Court has dealt with AEDPA several times and has never cast doubt on its constitutionality. In Williams v. Taylor, the Supreme Court explicitly noted the constraints AEDPA places on the power of a federal habeas court, holding that “the writ may issue only if” the decision “was contrary to clearly established Federal law, as determined by the Supreme Court of the United States” or “involved an unreasonable application of clearly established Federal law, as determined by the Supreme Court of the United States.” 529 U.S. 362, 412, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (citation modified) (citing § 2254(d)(1)). After saying so, the Supreme Court applied AEDPA—and has continued to do so for decades. Just a few years ago, the Supreme Court explicitly stated that “[w]hen Congress supplies a constitutionally valid rule of decision, federal courts must follow it. In AEDPA, Congress announced such a rule.” Brown v. Davenport, 596 U.S. 118, 127, 142 S.Ct. 1510, 212 L.Ed.2d 463 (2022); see also Renico v. Lett, 559 U.S. 766, 769, 130 S.Ct. 1855, 176 L.Ed.2d 678 (2010) (applying AEDPA deference and holding “that the Michigan Supreme Court's application of federal law was not unreasonable” so no habeas relief was available).

The argument that AEDPA deference is unconstitutional has been rejected in every circuit to consider the question. 3 Backing up our precedent are many cases in the district courts of our circuit upholding AEDPA deference. 4 And the Constitution commands us “to construe the law with ‘[c]lear heads ․ and honest hearts,’ not with an eye to policy preferences that ha[ve] not made it into the statute.” Loper Bright, 603 U.S. at 403–04, 144 S.Ct. 2244 (quoting 1 Works of James Wilson 363 (J. Andrews ed. 1896) (first and second alterations in original)).

B.

“The writ of habeas corpus indisputably holds an honored position in our jurisprudence.” Engle v. Isaac, 456 U.S. 107, 126, 102 S.Ct. 1558, 71 L.Ed.2d 783 (1982); see also Slack v. McDaniel, 529 U.S. 473, 483, 120 S.Ct. 1595, 146 L.Ed.2d 542 (2000) (“The writ of habeas corpus plays a vital role in protecting constitutional rights.”). But the scope of the writ of habeas corpus has been limited by Congress. The Supreme Court has “long recognized that ‘the power to award the writ by any of the courts of the United States, must be given by written law,’ and we have likewise recognized that judgments about the proper scope of the writ are ‘normally for Congress to make.’ ” Felker v. Turpin, 518 U.S. 651, 664, 116 S.Ct. 2333, 135 L.Ed.2d 827 (1996) (quoting Bollman, 8 U.S. at 94 and Lonchar v. Thomas, 517 U.S. 314, 323, 116 S.Ct. 1293, 134 L.Ed.2d 440 (1996)) (citation omitted). Of course there are still some constraints. See Boumediene v. Bush, 553 U.S. 723, 745, 128 S.Ct. 2229, 171 L.Ed.2d 41 (2008). Therefore, by no means should our opinion today be read to say that Congress has complete control over habeas. Like the Supreme Court, we acknowledge that in a few instances there may be limits.

We cannot ignore the Constitution's longstanding history of allowing significant—but not plenary—limiting factors on habeas relief in our constitutional analysis. See Zivotofsky ex rel. Zivotofsky v. Kerry, 576 U.S. 1, 23, 135 S.Ct. 2076, 192 L.Ed.2d 83 (2015) (“In separation-of-powers cases this Court has often ‘put significant weight upon historical practice.’ ”) (citation omitted). It should not escape our attention that the ebb and flow—restrictions and extensions—of habeas can be conceptualized as both judicially created and congressionally created depending on which ebb and which flow we are examining. Here, we focus on Congress's role in extending or restricting habeas, not the courts. In so doing, we see no indication that Congress has ever been constitutionally required to extend habeas beyond AEDPA (although they are certainly allowed to).

C.

Against this backdrop, Sanders claims that the history of the Madisonian Compromise indicates that AEDPA cannot stand. In so doing, he misapplies the Madisonian Compromise, which, properly understood, provides significant support for the idea that Congress can regulate habeas with respect to jurisdiction and remedies. We draw three important principles from the Madisonian Compromise, and all three of them support AEDPA. One, the Constitution presumes state courts are competent to adjudicate federal constitutional rights. Two, Congress has broad powers to regulate the jurisdiction of inferior federal courts. Three, Congress has broad powers to regulate our ability to grant remedies.

1.

Underlying the Madisonian Compromise was an explicit understanding that state courts had a role in interpreting the federal Constitution. This was the understanding at the Founding: “When ․ we consider the State governments and the national governments, as they truly are, in the light of kindred systems, and as parts of ONE WHOLE, the inference seems to be conclusive that the State courts would have a concurrent jurisdiction in all cases arising under the laws of the Union, where it was not expressly prohibited.” The Federalist No. 82, at 403 (Alexander Hamilton) (Dover Thrift Edition 2014). And this is the understanding today: “[S]tate courts have inherent authority, and are thus presumptively competent, to adjudicate claims arising under the laws of the United States.” Tafflin v. Levitt, 493 U.S. 455, 458–59, 110 S.Ct. 792, 107 L.Ed.2d 887 (1990).

Sanders interjects and raises the specter of state courts acting inappropriately, citing Federalist No. 22 and the risk of “bias[ed] local views and prejudices.” The Federalist No. 22, at 105 (Alexander Hamilton) (Dover Thrift Edition 2014). There are a few things to note about this passage. First, it is included in the discussion of the Supreme Court of the United States, not in that of inferior federal courts. Id. at 105. Additionally, Hamilton's example related to the interpretation of federal treaties, which could be subject to many different interpretations and local biases. The fear was that (as was a problem with the old Articles of Confederation, see Smith v. Turner, 48 U.S. 283, 569, 7 How. 283, 12 L.Ed. 702 (1849); Medellín v. Texas, 552 U.S. 491, 543–44, 128 S.Ct. 1346, 170 L.Ed.2d 190 (2008) (Breyer, J., dissenting)) there would be no uniform treaty interpretation. Without a single Supreme Court engaged in uniform interpretation, Hamilton rhetorically asked: “[i]s it possible that foreign nations can either respect or confide in such a government?” The Federalist No. 22 at 106 (Alexander Hamilton).

It is incorrect to view the federal writ of habeas corpus as a constitutionally required instrument by which the lower federal courts serve as courts of special review for every state court criminal judgment pursuant to adequate jurisdiction. Not only does such a view conflict with explicit Supreme Court precedent, but it also undermines our federalist system. Intervention under habeas is “an affront to the State and its citizens who returned a verdict of guilt after considering the evidence before them.” Shinn v. Ramirez, 596 U.S. 366, 390, 142 S.Ct. 1718, 212 L.Ed.2d 713 (2022). To continue, an understanding of habeas unable to be regulated in any way by Congress does not accord with the purpose of habeas as articulated by our history or the Supreme Court; habeas relief is not intended for “ordinary error correction,” but to police the bounds of constitutionality—i.e., to ensure clear egregious constitutional violations do not stand. Harrington, 562 U.S. at 102, 131 S.Ct. 770 (citation omitted). Not to grapple with this conception of AEDPA is to upset the delicate federal system created by our Constitution and the Supreme Court. See Calderon v. Thompson, 523 U.S. 538, 555–56, 118 S.Ct. 1489, 140 L.Ed.2d 728 (1998). Congress's decision in limiting habeas remedies serves the “well-settled meaning and function of habeas corpus in the federal system.” Harrington, 562 U.S. at 104, 131 S.Ct. 770. Sanders had three opportunities to petition the U.S. Supreme Court for direct review of his federal constitutional issues, and his petitions were denied each time. See, e.g., Sanders, 502 U.S. at 831, 112 S.Ct. 107; Sanders, 540 U.S. at 838, 124 S.Ct. 96; Sanders, 566 U.S. at 907, 132 S.Ct. 1792. Congress is not constitutionally required to give him de novo review in the federal district court and this court of appeals on top of the de novo review he already received in the Supreme Court.

The Madisonian Compromise left the decision of the reach of federal courts to Congress, not to the judiciary and no matter how much we may desire the writ of habeas corpus to be unlimited, its scope is not solely within our control. And while Article III grants power to the federal courts, it also serves as a check on our power. See Kaplan v. Cent. Bank of the Islamic Republic of Iran, 896 F.3d 501, 517 (D.C. Cir. 2018) (Edwards, J., concurring) (“Under Article III, jurisdiction is limited both by the bounds of the ‘judicial power’ as articulated in Article III, § 2, and by the extent to which Congress has vested that power in the lower courts, see U.S. Const. art. III, § 1.”).

2.

Lastly, Congress can regulate a federal court's ability to grant remedies. For example, the Supreme Court case of Lauf v. E.G. Shinner & Co. concerned a federal law under which no federal court “shall have jurisdiction to issue a temporary or permanent injunction in any case involving or growing out of a labor dispute[.]” 303 U.S. 323, 329, 58 S.Ct. 578, 82 L.Ed. 872 (1938). The Supreme Court held that this remedy-stripping provision of the law comported with Article III, specifically stating that “[t]here can be no question of the power of Congress thus to define and limit the jurisdiction of the inferior courts of the United States ․ It follows that in issuing the injunction [the district court] exceeded its jurisdiction.” Id. at 330, 58 S.Ct. 578. So, we know that Congress can limit remedies.

Within habeas, even valid constitutional rules do not necessarily have retroactive effect. Teague v. Lane, 489 U.S. 288, 310, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). Relief can be restricted for successive habeas petitions. McCleskey v. Zant, 499 U.S. 467, 494–95, 111 S.Ct. 1454, 113 L.Ed.2d 517 (1991). Habeas review can be limited for issues not fully preserved. Wainwright v. Sykes, 433 U.S. 72, 87, 97, 97 S.Ct. 2497, 53 L.Ed.2d 594 (1977). Thus, the proposition that habeas remedies for adjudged violations can be restricted is unsurprising and fairly uncontroversial. The Supreme Court has explicitly told us so. See Felker, 518 U.S. at 658, 116 S.Ct. 2333 (upholding the constitutionality of habeas restrictions even “[ ]though the Act impose[s] new conditions on our authority to grant relief[.]”); see also Lonchar, 517 U.S. at 322, 116 S.Ct. 1293 (explicitly noting that restrictions on the writ of habeas corpus can be made by Congress).

AEDPA is a statute that limits relief and jurisdiction except when the underlying decision was contrary to Supreme Court precedent or involved an unreasonable application of its precedent. This represents a constitutionally sound and permissible policy decision taken by Congress that (1) adheres to the structure of Article III and (2) balances the importance of the finality of state judgments with Congress's belief that federal courts should still police the outward bounds. Cf. Dretke v. Haley, 541 U.S. 386, 393, 124 S.Ct. 1847, 158 L.Ed.2d 659 (2004) (noting the importance of “States’ finality and comity interests” in dealing with habeas).

Sanders seems to misunderstand what federal courts do in AEDPA review. According to him, we are forced to take in the state court's constitutional interpretation as our own. That is not true. We undertake our own legal analysis under AEDPA. Then we ask whether the Constitutional violation was so egregiously wrong that habeas relief is merited under AEDPA. If not, we exercise restraint. That end result—judicial restraint in the face of reasonable competing views—is not a subversion of the Constitution; instead, it upholds core constitutional principles: Congressional control over jurisdiction and remedies and leeway for state courts to analyze their own criminal law, Shinn, 596 U.S. at 390, 142 S.Ct. 1718, particularly so if that is what Congress desires.⁵ Congress has the power to delineate our lines of habeas review. The existence of AEDPA deference is in the hands of Congress, not the judiciary.

D.

Next, Sanders points to United States v. Klein to argue that AEDPA violates our independent adjudicatory powers. 80 U.S. 128, 13 Wall. 128, 20 L.Ed. 519 (1871). But if anything, Klein and its progeny help further bolster our conclusion that AEDPA is constitutional. In Robertson v. Seattle Audubon Society, the Supreme Court held that Klein’s holding did not invalidate a Congressional law that targeted a specific case by name and substituted in new environmental standards to replace old ones. 503 U.S. 429, 434–35, 441, 112 S.Ct. 1407, 118 L.Ed.2d 73 (1992).

Next came Bank Markazi v. Peterson, in which Congress passed a specific statute designating a particular set of assets owned by Iran at a New York bank that would be available to satisfy specific judgments for victims of terrorism sponsored by Iran. 578 U.S. 212, 136 S.Ct. 1310, 194 L.Ed.2d 463 (2016). The Supreme Court read Klein narrowly and held that this statute was fine because Congress can direct courts to “apply a new legal standard to undisputed facts.” Id. at 230, 136 S.Ct. 1310. One sees the comparison between AEDPA habeas relief and that “new legal standard” to “undisputed facts” immediately. AEDPA is merely Congress's articulation of a “new legal standard” for habeas. Going further, the Supreme Court in Bank Markazi even noted that this “new legal standard” created by Congress could leave “only one possible outcome” to the case and still be permissible under Klein and the Constitution. See id. Thus, Klein and its progeny, if anything, support the constitutionality of AEDPA. Congress has the power to articulate new legal standards even if the articulation of these new legal standards changes the odds of victory. And Congressional grants of jurisdiction are a prerequisite to judicial power.

E.

The caselaw and the history show that AEDPA is constitutional. In response, Sanders contends that the Supreme Court's holding in Loper Bright wipes the slate clean and eliminates all the aforementioned precedent and history. We disagree.

Loper Bright concerned an interpretation of the APA and did not hold that all sorts of deference are unconstitutional. See 603 U.S. at 398, 144 S.Ct. 2244 (“Chevron defies the command of the APA that ‘the reviewing court’—not the agency whose action it reviews—is to ‘decide all relevant questions of law’ and ‘interpret ․ statutory provisions.’ ” (quoting 5 U.S.C. § 706) (alterations in original)). Moreover, AEDPA is a poor comparator for agency deference—AEDPA does not mandate deference to a state court's constitutional determination; it just states that a court cannot grant habeas relief merely because its own independent interpretation differs.

The issue with Chevron deference was that it mandated the Court to accept an interpretation of the law that was not its own. See id. at 427–28, 430–31, 144 S.Ct. 2244 (Gorsuch, J., concurring). AEDPA does not mandate that federal courts accept a state court's federal constitutional interpretation as their own, it only prevents those federal courts from intervening when the state courts make reasonable decisions as opposed to Chevron, which forced a court to defer to an agency decision so long as it is reasonable before even being allowed to engage in its own independent statutory analysis. See id. at 397, 144 S.Ct. 2244 (noting that so long as the agency's interpretation of the statute is reasonable, the court “had to set aside the traditional interpretive tools and defer to the agency”); see also Chevron, 467 U.S. at 840, 104 S.Ct. 2778. With AEDPA, even if a circuit court decides a state court's interpretation of the Constitution is “reasonable” that state supreme court's interpretation will not be considered binding law in the circuit. For AEDPA to be unconstitutional, the law would have to state that if a state court's interpretation of a federal constitutional right is reasonable, the federal court must adopt it as its own. But AEDPA does not say that.

The Supreme Court has repeatedly warned circuit courts against such behavior: “[i]f a precedent of this Court has direct application in a case, yet appears to rest on reasons rejected in some other line of decisions, the Court of Appeals should follow the case which directly controls, leaving to this Court the prerogative of overruling its own decisions.” Rodriguez de Quijas v. Shearson/Am. Exp., Inc., 490 U.S. 477, 484, 109 S.Ct. 1917, 104 L.Ed.2d 526 (1989). For these reasons, we join every other circuit to address this issue and reaffirm AEDPA's constitutionality. 6

IV.

Moving to the application of AEDPA, Sanders's various claims can be consolidated into three issues: (1) the jury instructions incorrectly defined the burden of proof for insanity defenses and violated Sanders's due process rights; (2) Sanders's trial counsel was ineffective in failing to (a) obtain an adequate mental health expert; (b) present mental health testimony from a psychologist and guard who observed Sanders after his arrest; (c) investigate and present mitigating evidence about Sanders's childhood abuse; (d) prepare for the penalty phase testimony; and (e) request more competency evaluations as Sanders's health deteriorated; and (3) the cumulative prejudice of trial counsel's ineffectiveness prejudiced him at trial. None of these arguments has merit.

A.

As discussed, Sanders's insanity was his sole defense at trial. In giving its instructions to the jury, the trial court repeatedly reiterated that Sanders's crime had to be proven “beyond a reasonable doubt.” DE 166-4, Trial Tr., Page ID 3914–15. But in dealing with the insanity defense, the judge told the jury that the defense would be established if the jury “believe[d it had] from the evidence.” Id. 3914–17. While the words of the two jury instructions are different and create different standards, Sanders contends that they were improperly blended so as to imply that his insanity defense—like his criminal offenses—must be established beyond a reasonable doubt. Because of these ambiguous jury instructions, Sanders argues that his Fourteenth Amendment due process rights were violated, and that the Kentucky Supreme Court clearly erred in holding otherwise. We disagree.

1.

In Kentucky, while the prosecution must prove the crime beyond a reasonable doubt, the defendant's insanity defense need only be proven by a preponderance of the evidence. Ball v. Commonwealth, 81 Ky. 662, 664 (1884). But Kentucky does not use the phrase “preponderance of the evidence” and instead prefers the phrase “[believe] from the evidence.” Gall, 607 S.W.2d at 110; see, e.g., Hall v. Commonwealth, 645 S.W.3d 383, 404 (Ky. 2022) (noting that a “believe from the evidence” instruction “appropriately incorporated the preponderance standard”). The question we must decide is whether this Kentucky rule, which was used to affirm the jury instructions that led to Sanders's conviction, was unconstitutional as applied to Sanders and whether it was contrary to or involved an unreasonable application of clearly established federal law as articulated by the Supreme Court. See 28 U.S.C. § 2254(d)(1).

There is no due process requirement with respect to a state's choice of a burden of proof for establishing an insanity defense. See Leland v. Oregon, 343 U.S. 790, 798–99, 72 S.Ct. 1002, 96 L.Ed. 1302 (1952). That decision is left up to the states. See Engle v. Isaac, 456 U.S. 107, 120, 102 S.Ct. 1558, 71 L.Ed.2d 783 (1982). However, once the state has settled on the burden of proof, the defendant has a due-process interest in having the appropriate burden of proof applied. See Hicks v. Oklahoma, 447 U.S. 343, 346, 100 S.Ct. 2227, 65 L.Ed.2d 175 (1980).

2.

The Kentucky Supreme Court's ultimate disposition of Sanders's claims regarding the jury instruction issue was short. In its entirety, the court reasoned as follows:

It is further submitted that the instructions failed to specify the burden of proof regarding the defense of insanity. The jury was instructed that it might find the defendant not guilty by reason of insanity if it believed “from the evidence” that he was insane at the time of the offenses. We find no error in this instruction.

Sanders I, 801 S.W.2d at 679 (citing Gall, 607 S.W.2d at 110).

A relevant and on-point case is Boyde v. California, 494 U.S. 370, 110 S.Ct. 1190, 108 L.Ed.2d 316 (1990). There, the U.S. Supreme Court held that the proper inquiry when determining whether ambiguous jury instructions were unconstitutional is “whether there is a reasonable likelihood that the jury has applied the challenged instruction in a way that prevents the consideration of constitutionally relevant evidence.” Id. at 380, 110 S.Ct. 1190. With this case available (although not cited), the Kentucky Supreme Court held that the jury instructions were not erroneous. Sanders I, 801 S.W.2d at 679. To be sure, the court did not elaborate on its reasoning. Nonetheless, “[w]hen a state court rejects a federal claim without expressly addressing that claim, a federal habeas court must presume that the federal claim was adjudicated on the merits[.]” 7 Johnson v. Williams, 568 U.S. 289, 301, 133 S.Ct. 1088, 185 L.Ed.2d 105 (2013). And because the claim was adjudicated on the merits, AEDPA deference applies. Robinson v. Howes, 663 F.3d 819, 822–23 (6th Cir. 2011). Sanders fails to show that the Kentucky Supreme Court's decision “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1).

As the district court noted, fair-minded jurists can conclude that the jury instructions were not confusing in this instance. In our system of justice, we presume that jurors are fitted for their duty “by their natural intelligence and their practical knowledge[.]” Aetna Life Ins. v. Ward, 140 U.S. 76, 88, 11 S.Ct. 720, 35 L.Ed. 371 (1891). This is particularly true in criminal trials. Francis v. Franklin, 471 U.S. 307, 324 n.9, 105 S.Ct. 1965, 85 L.Ed.2d 344 (1985). The rule in Kentucky applies a similar presumption and thread of logic. The two standards of “beyond a reasonable doubt” and “believe from the evidence” are arguably materially different. Kentucky, trusting the natural abilities of its citizens—an undeniably fair choice—has crafted a rule that replaces “preponderance from the evidence” with “believe from the evidence.” In so doing, Kentucky made the reasonable decision that it would be unnecessarily redundant to include the phrase “preponderance of.” See also Pryor v. Rose, 724 F.2d 525, 530 n.3 (6th Cir. 1984). Ultimately, the trial court accurately stated the law, and a defendant is not entitled to his specific choice of words. United States v. Frederick, 406 F.3d 754, 765 (6th Cir. 2005).

Thus, the question in applying Boyde and similar cases here is whether it is unreasonable for the Kentucky Supreme Court to conclude that the jury instructions, which had no conflict on their face and expressly conformed to its longstanding practice, prevented the jurors from considering constitutionally relevant evidence at Sanders's guilt or mitigation proceedings. We hold that a fair-minded jurist could take the position that the jury instructions were constitutionally valid.

B.

We turn to Sanders's claims of ineffective assistance of counsel. Under the Sixth Amendment, criminal defendants have a “right to the effective assistance of counsel.” Strickland v. Washington, 466 U.S. 668, 686, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984) (citation omitted). To establish a violation of this right, Sanders must prove two things. First, he must show that his counsel “made errors so serious that counsel was not functioning as the ‘counsel’ guaranteed [to] the defendant by the Sixth Amendment.” Id. at 687, 104 S.Ct. 2052. “Judicial scrutiny of counsel's performance must be highly deferential.” Id. at 689, 104 S.Ct. 2052. We must be careful to avoid needlessly second-guessing strategic decisions made by counsel on the ground. Id. Second, Sanders must show that his counsel's deficient performance prejudiced his defense. Id. at 687, 104 S.Ct. 2052.

In confronting most of Sanders's Strickland claims, we apply AEDPA review. Even without AEDPA, we are already “highly deferential” to the performance of the underlying counsel. Id. at 689, 104 S.Ct. 2052. Habeas review, itself highly deferential, adds another layer of deference. So our review of Sanders's ineffective assistance of counsel habeas claims under AEDPA must be “doubly deferential.” Knowles v. Mirzayance, 556 U.S. 111, 123, 129 S.Ct. 1411, 173 L.Ed.2d 251 (2009). Thus, “it is no longer enough for [Sanders] to show that his attorney made a decision that no reasonable lawyer would have made. He now must show that ‘every fairminded jurist would agree that every reasonable lawyer would have made a different decision.’ ” Fields, 86 F.4th at 240 (quoting Dunn v. Reeves, 594 U.S. 731, 739–40, 141 S.Ct. 2405, 210 L.Ed.2d 812 (2021)) (citation modified).

Sanders makes five ineffective assistance of counsel claims. He asserts that Charters failed to (1) obtain an adequate mental health expert; (2) present mental health testimony from a psychologist and guard who observed him after his arrest; (3) investigate and present mitigating evidence about his childhood abuse; (4) prepare for the penalty phase testimony; and (5) request additional competency evaluations as his health deteriorated. We ultimately find that the Kentucky Supreme Court did not contravene clearly established federal law in rejecting ineffective assistance of counsel arguments in claims one, three, and four. As to claim two, while the Kentucky Supreme Court may have contravened clearly established law in its analysis, we need not address that issue as Sanders's claim would still fail under de novo review because he cannot show prejudice. And although the Kentucky Supreme Court did not reach claim five, he also fails to show prejudice under de novo review. We address each claim in turn.

1.

When insanity is a defense, a mental health expert must be provided to an indigent defendant. 8 See Ake v. Oklahoma, 470 U.S. 68, 83, 105 S.Ct. 1087, 84 L.Ed.2d 53 (1985). We give trial counsel broad discretion in selecting a mental health expert and determining how much time to dedicate to the expert's preparation. See Dunn, 594 U.S. at 739, 141 S.Ct. 2405. This is because we recognize that there are many ways to try a case and counsel may sometimes choose to prioritize resources differently depending on the specifics of a given case. See Harrington, 562 U.S. at 106, 131 S.Ct. 770; Yarborough v. Gentry, 540 U.S. 1, 5–6, 124 S.Ct. 1, 157 L.Ed.2d 1 (2003). Ultimately, “counsel [is] entitled to formulate a strategy that was reasonable at the time and to balance limited resources in accord with effective trial tactics and strategies.” Harrington, 562 U.S. at 107, 131 S.Ct. 770 (citation modified). Sanders's first claim—failure to obtain an adequate mental health expert—has two parts. The first part concerns Charters's failure to interview or retain Flenning. The second part concerns Charters's failure to adequately prepare Cooke for his testimony.

a.

As a reminder, Sanders was evaluated for one and a half months by Walker and her team, who concluded that Sanders suffered from no mental defect that would prevent him from being responsible for the murders. Unbeknownst to Sanders or his counsel, one of the members of Walker's team—Flenning—produced a personal sub-report that was more sympathetic to Sanders's insanity defense. Flenning concluded that the evidence showed that Sanders suffered from a mental deficiency. Sanders argued to the Kentucky Supreme Court in Sanders II that Charters's failure to interview and retain Flenning constituted ineffective assistance of counsel. 89 S.W.3d at 385–86. The Kentucky Supreme Court disagreed, noting that Charters had access to the full report and there may have been reasons he did not pursue the leads in Walker's report because the “report did not support the defense of insanity and could actually damage Sanders[.]” Id. And relatedly, the Kentucky Supreme Court held that there was likely no prejudice because the Flenning report would have been likely used by the prosecution to convincingly argue to the jury that Sanders changed his story depending on who he was talking to, as stated in the Walker report.

Under the “doubly deferential” standard, Knowles, 556 U.S. at 123, 129 S.Ct. 1411, under which we evaluate AEDPA Strickland claims, we cannot ignore the Kentucky Supreme Court's legal reasoning on this issue. The Walker report gave Charters no indication that any evidence existed that might contradict it. If anything, the report's use of the pronoun “we” in setting out its conclusion would indicate to a reasonable attorney that further digging into the members of Walker's team would only lead to more information that would be prejudicial and not helpful to Sanders's defense. In hindsight, Charters may have been mistaken not to inquire further into any potential disagreement among team members. “But hindsight bias is the very thing that Strickland’s deferential standard counsels against[.]” Hale v. Cool, 122 F.4th 637, 649 (6th Cir. 2024). An attorney in Charters's position had limited time and resources and was presented with a report that, by all indications, was prejudicial to his client. Rather than interviewing individual team members, Charters opted to retain a counter-expert. We do not believe that every fairminded jurist would agree that no reasonable lawyer would have done something differently. For that reason, we deny habeas relief on this claim. See Fields, 86 F.4th at 240.

b.

Ultimately, rather than pursue the Walker report, Charters opted to retain Cooke, who offered to provide his services pro bono. Cooke evaluated Sanders only a little over a week before trial and only for around two and a half hours. Based on the information obtained from this short examination, Cooke testified during trial that Sanders could not control his actions. The contrast between Cooke's short, cursory evaluation and Walker's long, multi-member evaluation was highlighted by the prosecution and likely significantly hurt Cooke's credibility with the jury.

According to Sanders, Charters's extremely limited preparation constituted ineffective assistance of counsel. The Kentucky Supreme Court disagreed. See Sanders II, 89 S.W.2d at 387–88. The court noted that Cooke's testimony had the intended effect: Cooke reduced the negative impact of the Walker report by disagreeing with it. Id. at 387. Analyzing the U.S. Supreme Court case of Ake v. Oklahoma, the Kentucky Supreme Court noted that Sanders was merely entitled to a state-provided psychiatrist, not to multiple examinations merely because he did not like the results of the state-provided psychiatrist. Id. at 387–88. Ultimately, Cooke testified that, in his opinion, Sanders could not control his actions, and that his work—along with Walker's comprehensive team—satisfied Sanders's right to have access to qualified mental health experts. Id.

In Ake, the U.S. Supreme Court was presented with the question “whether the Constitution requires that an indigent defendant have access to the psychiatric examination and assistance necessary to prepare an effective defense based on his mental condition, when his sanity at the time of the offense is seriously in question.” 470 U.S. at 70, 105 S.Ct. 1087. The Court answered that a psychiatrist must be provided. Id. at 80, 105 S.Ct. 1087. In particular, “the State must, at a minimum, assure the [indigent] defendant access to a competent psychiatrist who will conduct an appropriate examination and assist in evaluation, preparation, and presentation of the defense.” Id. at 83, 105 S.Ct. 1087; see also McWilliams v. Dunn, 582 U.S. 183, 198, 137 S.Ct. 1790, 198 L.Ed.2d 341 (2017).

From the outset, Sanders was not an indigent defendant, thus making it unclear if Ake’s command applies here. See 470 U.S. at 83, 105 S.Ct. 1087; see also Durr v. Mitchell, 487 F.3d 423, 433 (6th Cir. 2007) (stating Ake’s holding as applying to indigent defendants). We need not decide Ake’s reach, however, as Sanders's claim can be resolved even if Ake applies. Cooke was a licensed psychologist with extensive educational and clinical experience. He testified in Sanders's defense, using his expert knowledge to conclude that Sanders was not responsible for his actions.

Given the significance of the proceedings against Sanders, it is understandable that he would believe that if Cooke had spent more time on his case, the result might have been more favorable to him. But Cooke was apparently given enough time to come to a clear conclusion. Considering our doubly deferential standard of review under AEDPA and Strickland, we cannot conclude that every fair-minded jurist would conclude that every reasonable lawyer would have acted differently from Charters. Therefore, we cannot grant habeas relief on Sanders's habeas claim that Charters was constitutionally defective in his preparation of Cooke.

c.

Sanders further draws our attention to Rogers v. Dzurenda, 25 F.4th 1171 (9th Cir. 2022). In Rogers, the Ninth Circuit was not reviewing Rogers's Strickland claim under AEDPA. Instead, it was reviewing his Strickland claim de novo. See Rogers, 25 F.4th at 1181 (“On appeal, the State does not challenge the district court's decision not to apply AEDPA deference to the state court's adjudication of Rogers's ineffective assistance of counsel claim. Because the ineffective assistance of counsel claim before the district court was never adjudicated on the merits by the Supreme Court of Nevada, we review Rogers's [Strickland claim] de novo.” (footnote omitted)). From our perspective, this greatly affects the relevance of Rogers to Sanders as AEDPA deference is among the narrowest forms of review compared with de novo—the most lenient. See Burt v. Titlow, 571 U.S. 12, 19, 134 S.Ct. 10, 187 L.Ed.2d 348 (2013) (“AEDPA erects a formidable barrier to federal habeas relief[.]”). Thus, we are not persuaded by Rogers.

2.

Second, we address Sanders's ineffective assistance of counsel claim arising from Charters's failure to present testimony from the jail psychologist and two prison guards regarding Sanders's behavior after his arrest. When Sanders was initially arrested, the jail psychologist wrote a letter to the trial judge requesting that Sanders be taken to a psychiatric hospital because he was showing signs that he could be suicidal. According to Sanders, three witnesses during his time in his initial jail cell could have testified to his disturbed mental state. The deputy jailers could have testified that Sanders was in tears and banging his head against the wall and that the jail psychologist, Moffit, could have testified that Sanders was suffering from major psychological problems. Charters did not follow up, and Sanders alleges that such actions constituted ineffective assistance of counsel. The Kentucky Supreme Court disagreed, reasoning that this evidence was needlessly cumulative given that Sanders had testified to his mental state and that counsel's decision not to present cumulative testimony was not ineffective assistance of counsel. See Sanders II, 89 S.W.3d at 391.

Even if the Kentucky Supreme Court erred on this point, Sanders's claim still fails under de novo review because he has not provided any evidence of what the two jailers or psychologists would have said and thus, cannot prove prejudice under Strickland. To establish prejudice, a defendant must introduce evidence establishing what would have been said (such as an affidavit). See Tinsley v. Million, 399 F.3d 796, 810 (6th Cir. 2005). This is the defendant's burden. See Hale, 122 F.4th at 646. Sanders claims there is evidence in the form of a video recording of his behavior immediately after his arrest. But it is still unclear whether the jailers saw the contents of this videotape. The question is not what happened, but what the jailers’ testimony would be. Without this evidence, Sanders fails to satisfy his burden.

Sanders's allegations regarding Moffit are supported by some evidence. There is the February 3, 1987, letter Moffit wrote asking that Sanders be placed in psychiatric care. There is also a February 4, 1987, notarized letter reiterating concerns raised in the prior letter and noting that Sanders might be dissociative. However, while this is some evidence, Sanders still fails to show how it would have affected the proceedings’ outcome. Sanders claims that calling Moffit would have strengthened Cooke's testimony and his diminished capacity defense, but the letter makes no such assertions. Instead, the letter merely states that Moffit had reason to believe that Sanders suffered from mental deficiency; the letter does not suggest that Moffit would have supported Sanders had he testified. Because we do not know, Sanders has not satisfied his burden. See Baze v. Parker, 371 F.3d 310, 322 (6th Cir. 2004).

Nor was Charter's failure to use Moffit as an expert witness ineffective assistance of counsel. Moffit's letters merely recommend that Sanders be examined for mental deficiencies. The letters do not represent a diagnosis—just an initial concern. Charters chose instead to use his own independent expert, given his limited resources. See Clardy v. Pounds, 126 F.4th 1201, 1210 (6th Cir. 2025) (noting the latitude we give counsel in how they secure expert witnesses with limited resources); see also Mammone v. Jenkins, 49 F.4th 1026, 1052 (6th Cir. 2022) (“[S]electing an expert is the classic example of a strategic choice made by counsel.”). In sum, even putting aside AEDPA deference, Sanders still cannot show that failure to introduce corroborating testimony from the two jailers or psychologist prejudiced him or that Charters's decision not to use Moffit as a witness was ineffective assistance of counsel.

3.

During the penalty phase of his trial, Sanders's employer, neighbor, assistant principal of his high school, and his sister testified. Sanders himself also testified. He claims that had Charters engaged in any real preparation as required under Strickland, he would have interviewed his family and friends and discovered an abundance of additional mitigating evidence. In so doing, Sanders asks us to consider evidence not in the state court record. We are barred from doing so and find that the Kentucky Supreme Court's decision on this issue cannot be disturbed under AEDPA.

a.

The Kentucky Supreme Court rejected Sanders's claim on what, at first blush, appeared to be procedural grounds. See Sanders II, 89 S.W.3d at 390–91. The court noted that Sanders failed to offer the content of any testimony from the additional witnesses and thus failed under Kentucky state procedural rules. See id. at 390 9). But then the Kentucky Supreme Court went on to discuss Strickland and held that Charter's decision not to call these additional witnesses could have been a strategic one that should not be second-guessed. See id. at 390. The district court initially held that this was a procedural decision by the Kentucky Supreme Court that relied exclusively on state procedural law and that Sanders failed to attach any affidavits to show why the state court's procedural bar should be ignored. Therefore, the district court dismissed Sanders's habeas claim because the question of his procedural default was an independent decision that could support the state judgment. See Lovins v. Parker, 712 F.3d 283, 295–96 (6th Cir. 2013) (quoting Coleman v. Thompson, 501 U.S. 722, 729, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991)).

But after reflecting on its decision, the district court sua sponte ordered both parties to brief the issue. Both parties agreed that the Kentucky Supreme Court's decision was on the merits. Therefore, the district court re-analyzed its decision regarding the mitigation evidence under AEDPA deference and, after applying AEDPA, held that the state court's application of Strickland did not overcome AEDPA deference.

Generally, a federal habeas court is precluded from hearing claims that were “procedurally defaulted” on state procedural grounds. Shinn, 596 U.S. at 371, 142 S.Ct. 1718. But procedural default can be overcome if the prisoner can “demonstrate ‘cause’ to excuse the procedural defect and ‘actual prejudice’ if the federal court were to decline to hear his claim.” Id. (citation omitted). In so doing, “[o]ften a prisoner with a defaulted claim will ask a federal habeas court not only to consider his claim but also to permit him to introduce new evidence to support it.” Id. But under AEDPA, we are prohibited from expanding the factual record developed in the state court. Id.; see Frazier v. Huffman, 343 F.3d 780, 797 (6th Cir. 2003); see also 28 U.S.C. § 2254(e)(2). That is unless the petitioner's “claim relies on” “(i) a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court[ ] that was previously unavailable” or “(ii) a factual predicate that could not have been previously discovered through the exercise of due diligence” and this new fact must “be sufficient to establish by clear and convincing evidence that but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” 28 U.S.C. § 2254(e)(2)(A)–(B); see also Shinn, 596 U.S. at 371, 142 S.Ct. 1718.

Under procedural default, in determining whether an exception applies, a federal court considers what new evidence would be presented. So, before the district court ordered more briefing on whether the Kentucky Supreme Court's decision was on the merits, it also slightly changed its position on the procedural default issue and allowed Sanders to file various affidavits so it could decide whether there was an exception to procedural default. As a result, before the district court decided that the Kentucky Supreme Court decision was on the merits, Sanders filed six affidavits. 10

These affidavits were not before the Kentucky Supreme Court. They could be relevant if we were considering procedural default. The district court initially analyzed the Kentucky Supreme Court's analysis on mitigation evidence and held that the claim was procedurally defaulted, and Sanders did not present sufficient affidavits to overcome procedural default. The district court then changed its mind on the affidavits and allowed Sanders to submit new affidavits to consider whether this new evidence could be looked at by the habeas court. But then the district court had second thoughts about whether this was even a procedural default case and asked for further briefing. After briefing, the district court reversed itself and held that the Kentucky Supreme Court's decision was on the merits.

b.

Because the Kentucky Supreme Court's decision was on the merits, we cannot consider the new affidavits. But even if it were not, these new affidavits cannot overcome the barrier from Shinn. The district court correctly held that the Kentucky Supreme Court's decision on Sanders's Strickland claim was a merits analysis.

When a state court is presented with a federal claim, “it may be presumed that the state court adjudicated the claim on the merits in the absence of any indication or state-law procedural principles to the contrary.” Harrington, 562 U.S. at 99, 131 S.Ct. 770. To overcome this presumption the state court must “clearly and expressly” state in its decision that the judgment is because of a state procedural bar. Harris v. Reed, 489 U.S. 255, 263, 109 S.Ct. 1038, 103 L.Ed.2d 308 (1989) (quotation omitted). But the court analyzed Sanders's claim under Strickland without any indication that his claim could solely be decided on state procedural grounds. See Sanders II, 89 S.W.3d at 390. Given the presumption that the state court decided the merits unless it states otherwise, the Kentucky Supreme Court decided Sanders's Strickland claim on the merits.

c.

Sanders suggests that we should consider these six affidavits in our analysis, regardless of whether the ineffective assistance of counsel claim was a merits or procedural decision, because his new evidence is so substantial that it has created a “new” claim not yet adjudicated by the Kentucky Supreme Court.

Sanders lost in the Kentucky Supreme Court, in part, because there was insufficient evidence in the Kentucky 11.42 motion at the trial level. See Sanders II, 89 S.W.3d at 390. If that is true, according to Sanders, then the new affidavits so changed the facts that a new claim before the federal district court is created under habeas and there is no AEDPA deference to apply. Essentially, Sanders's claim is a new one not considered by the Kentucky Supreme Court, which had an incomplete record. Therefore, we need not apply AEDPA deference as the principles of “comity, finality, and federalism” are not implicated. See Gordon v. Braxton, 780 F.3d 196, 202 (4th Cir. 2015) (quotation omitted). Alternatively, Sanders asserts that he is bringing a separate ineffective assistance of counsel claim based on evidence not raised in the prior ineffective assistance of counsel claim because of that prior post-conviction counsel's incompetence. Because this separate collateral attack is a habeas claim not attacking a state supreme court's decision on the merits, AEDPA deference does not apply.

To further clarify, a state prisoner challenging the merits of a state court's constitutional decision cannot rely on new evidence to assess the state court's claim under AEDPA. 11 Upshaw v. Stephenson, 97 F.4th 365, 372 (6th Cir. 2024). Therefore, we cannot consider these affidavits to challenge the Kentucky Supreme Court's merits decision. Sanders argues that if the Kentucky Supreme Court's decision was on the merits with respect to the evidence before it, then the new evidence in federal court substantially improved his Strickland allegations to create a new claim. Thus, Sanders argues, because his claim was not technically considered on the merits in the state courts, AEDPA deference does not apply.

There are two problems with this. First, “AEDPA requires habeas petitioners to exhaust their claims in state court before turning to a federal court for relief.” Stermer v. Warren, 959 F.3d 704, 720 (6th Cir. 2020). Therefore, if it is true that the new evidence has transformed Sanders's ineffective assistance of counsel claim on mitigation into a new claim, then Sanders has failed to exhaust that new habeas claim anyway. Thus, his claim is now procedurally defaulted. See Shinn, 596 U.S. at 371, 142 S.Ct. 1718.

Second, Sanders's new evidence does not fundamentally alter his ineffective assistance of counsel claim. New evidence does not transform a claim into a new one just by strengthening the thrust of the original argument. See Franklin v. Jenkins, 839 F.3d 465, 474 (6th Cir. 2016). “Simply put, for purposes of AEDPA, ‘[n]ew evidence does not usually give rise to a new claim; it merely provides additional proof of a claim already adjudicated on the merits.’ ” Id. at 474–75 (citation modified). But here, the “new” evidence is simply more evidence for the same ineffective assistance of counsel claim he already brought to the Kentucky Supreme Court—that Charters failed to do appropriate research and obtain mitigating testimony. For these two reasons, there is no “new” habeas claim outside the scope of AEDPA.

d.

Having established that we cannot consider these six affidavits, we now analyze Sanders's ineffective assistance of counsel claim that Charters did not develop appropriate mitigation evidence, based solely on evidence from the state record and through the doubly deferential lens of AEDPA and Strickland.

i.

There are three reports to consider on this issue. The first is Walker's KCPC report. The report notes Sanders's inconsistent statements. At times, Sanders described his life as very good; other times, he said otherwise particularly with respect to his childhood. To Flenning, Sanders said he was forced to work at age four and did not get to play with other children. Sanders also described conflict with his parents. He said they beat him with a “tobacco stick” or “barbed wire.” DE 165-4, Walker Rep., Page ID 2823. Sanders also said his father was a “heavy drinker” and that he had been charged in the recent past with “Assault 3rd degree and several other charges.” Id. at 2824. From this, the report ultimately concluded that Sanders may have suffered some degree of abuse as a child. This report was available to Charters at the time of the trial.

The second report was produced by Dr. Roger H. Fisher in 1994, during state post-conviction proceedings. Fisher, along with a few other experts, had evaluated Sanders in another case. Fisher illuminates Sanders's background even more. Fisher notes that, according to Sanders, his father essentially used him and his siblings as a captive work force while he was drinking. Sanders's father hit him constantly—often when he was drunk, but also sometimes when Sanders did not complete a job in the way his father wanted. The third report is the Flenning report. Flenning reiterates Sanders's abusive childhood. The report also states that Sanders's mother abused him as well. Essentially, Sanders grew up alienated from his family and spoke only to his dog and himself.

ii.

Before Sanders's trial and penalty phase, Charters had access to the Walker report that indicated the potential of abuse. Charters did not investigate further or use this evidence at the mitigation phase of the trial. The Kentucky Supreme Court applied Strickland and found that this did not constitute ineffective assistance of counsel. See Sanders II, 89 S.W.3d at 390–91. Sanders cites several Supreme Court cases concerning the duty to conduct thorough investigations of the evidence to argue that the Kentucky Supreme Court's holding contradicted clearly established law. But most of these were decided after Sanders II and therefore, cannot be used to show that the Kentucky Supreme Court's decision was contrary to clearly established federal law. See Stewart v. Wolfenbarger, 468 F.3d 338, 346 (6th Cir. 2006). Only one— Williams v. Taylor—was decided before Sanders II. 529 U.S. at 362, 120 S.Ct. 1495 (plurality opinion). In Williams, the Supreme Court held that counsel was ineffective under Strickland when counsel “failed to introduce available evidence” that the defendant had a horrible upbringing, was intellectually disabled, and was non-violent in prison. Id. at 396, 120 S.Ct. 1495.

Williams also elaborated on the proper standards for evaluating claims under AEDPA. First, interpreting the “the contrary to” prong of AEDPA, the Supreme Court held that a state court decision conflicts with clearly established law if it applies a rule that contradicts the governing law in a previous decision. 28 U.S.C. § 2254(d)(1); see 529 U.S. at 405, 120 S.Ct. 1495 (plurality opinion). Or it is contrary if it applies the correct law to “materially indistinguishable” facts from a prior Supreme Court decision but arrives at a contrary result. Id. at 406, 120 S.Ct. 1495. However, Williams notes that a state court decision that identifies the correct Strickland rule and applies it to reject the prisoner's claim is not “diametrically different,” or “opposite in character or nature,” or “mutually opposed” to Strickland. Id.

Second, interpreting the “unreasonable application” prong, Williams concluded that a state court's decision is contrary to federal law when it identifies the correct legal principle but unreasonably applies it to the facts. 28 U.S.C. § 2254(d)(1); see Williams, 529 U.S. at 413, 120 S.Ct. 1495. It was not until 2011 that the Supreme Court established the rule that federal habeas relief is precluded so long as “fairminded jurists could disagree” on whether the law was unreasonably applied. 12 See Harrington, 562 U.S. at 101, 131 S.Ct. 770 (quotation omitted).

Per Williams, a state court decision can also conflict with federal law if it reaches a different conclusion despite confronting “materially indistinguishable facts” from a prior Supreme Court decision. See 529 U.S. at 405, 120 S.Ct. 1495. In Williams, the Virginia Supreme Court erred in misreading Strickland as being supplemented by Lockhart v. Fretwell, 506 U.S. 364, 113 S.Ct. 838, 122 L.Ed.2d 180 (1993) to require a separate inquiry into fundamental fairness even after a Strickland claim is satisfied. See Williams, 529 U.S. at 393, 120 S.Ct. 1495 (plurality opinion). By contrast, the Kentucky Supreme Court did not misapply Strickland here. The Virginia Supreme Court also had access to a wealth of mitigating evidence that should have been presented. Id. at 370, 120 S.Ct. 1495. Here, the mitigation evidence consisted of just three reports—one of which Charters could not have known existed; another created after the conviction; and the third potentially prejudicial to his own client. And all three contain second-hand reports of Sanders's abuse with no verification in the state court record. Cf. Mitchell v. The Hartford, No. 3:05CV-432-H, 2006 WL 1548956, at *5 (W.D. Ky. June 2, 2006) (“These statements are second hand and completely unverified.”) There are no direct reports in the record now before us. Accordingly, we cannot grant habeas relief for Charters's failure to offer more mitigation evidence at the penalty stage of Sanders's trial.

4.

Sanders next argues that Charters rendered ineffective assistance of counsel because he failed to appropriately advise and prepare him for his penalty phase testimony. The Kentucky Supreme Court rejected this claim because Sanders presented nothing but speculation as to why he was inadequately prepared to testify. See Sanders II, 89 S.W.3d at 391. The district court denied Sanders's habeas petition because he failed to show prejudice under Strickland and the Supreme Court of Kentucky reasonably denied him relief. We agree.

After Sanders's conviction, there was little time to prepare for the penalty stage. In fact, the first thing the state trial court said after Sanders's conviction was, “I didn't have any idea we would get to this stage this quickly.” DE 166-4, Trial Tr., Page ID 3956. Because Charters was unprepared to begin the penalty phase testimony, he asked the court to recess that day (Friday) so he could prepare over the weekend. The trial court hesitated, and Charters stated that he already “put on what would normally be the mitigation stuff,” but just wanted to check with a few family members if they wanted to testify about the personal effects the death penalty would have on them. Id. 3958. In response, the trial court told Charters that this type of mitigation is likely inadmissible. After a brief recess, Charters reiterated that he wanted the court recessed until Monday and that he could bring in several people from the community who knew Sanders growing up to testify about his childhood. The court obliged.

On Monday, Charters presented four witnesses in Sanders's defense (his employer, neighbor, assistant principal of his high school, and sister). The entire questioning took less than 20 minutes. Because Sanders had not testified, the court wanted the record to reflect that he knew he had the right to testify but did not want to. So, the trial court conducted a colloquy, reminding Sanders of his rights. The court then asked Sanders whether he wanted to testify, to which he responded, “I don't know.” DE 166-5, Trial Tr., Page ID 3978. The court then recessed so Sanders and Charters could discuss whether Sanders would testify. After four minutes, the court readjourned, and Sanders testified.

Sanders's testimony on the stand was scattered. He testified that he was not guilty of the crimes and that his mental health professionals failed to help him. Charters asked Sanders if he would cooperate with mental health professionals if offered treatment, but Sanders equivocated and insisted he had already cooperated. Sanders ended his testimony stating, “I just – I just don't understand what happened to me. Those people at KCPC told me I might not ever understand what happened. I asked those people to help me, and they didn't have time. They had a lot of people there, more people coming in all the time, and they didn't have time to talk to me.” DE 166-5, Trial Tr. Page ID 3982–83.

The prosecution seized on Sanders's testimony at closing. It noted that Sanders continued to blame others. The prosecution also stated that Sanders did not say he would cooperate with future treatments and asked the jury, “[d]oesn't that scare you[?]” Id. 3991. The prosecution also pointed out that Sanders never once expressed any remorse for his crime: “No remorse at all. He told you today he might be a little bit guilty, but he was mostly not guilty, much of him was not guilty. How scar[ ]y is that, to know that that man would be allowed to live?” Id. 3997.

Sanders argues that Charters's preparation fell short under the Strickland standard and that it prejudiced him. The Kentucky Supreme Court disagreed, holding that Sanders presented nothing but speculation that he was not fully informed of his rights and had not been adequately prepared to testify. See Sanders II, 89 S.W.3d at 391. Strickland has two prongs: (1) ineffective assistance of counsel and (2) prejudice, which Sanders must satisfy. See 466 U.S. at 687, 104 S.Ct. 2052. In Sanders's postconviction proceedings, the Kentucky Supreme Court reasonably concluded that Sanders had failed to indicate what Charters should have done and what would have been different. See Sanders II, 89 S.W.3d at 391.

The same holds true today. Sanders presents no evidence about what would have gone differently had Charters served as adequate counsel. 13 Contra Rogers v. Mays, 69 F.4th 381, 390 (6th Cir. 2023) (to show Strickland prejudice the petitioner must “show a reasonable probability that the trial would have gone differently but for counsel's errors”). Instead, Sanders only theorizes as to what would have gone differently, such as he would have shown more remorse during his testimony or would not have testified at all, which presumably would have made the jury not recommend the death penalty. But this only leaves us “with pure speculation on whether the outcome of the trial or the penalty phase could have been any different, [which is] an insufficient basis for a successful claim of prejudice.” Baze, 371 F.3d at 322. This is not to say that a defendant can never show prejudice when his counsel fails to prepare him to testify, and he then testifies. We conclude only that the Kentucky Supreme Court's conclusion that Sanders did not show prejudice based on the record before it, which included overwhelming aggravating evidence against Sanders, was reasonable and not contrary to federal law. Accordingly, we cannot grant Sanders habeas relief on this claim.

5.

Sanders argues that Charters's failure to call for a second competency evaluation for Sanders constituted a Strickland violation. Although we review this claim de novo, and not under AEDPA, we disagree.

After confessing to the crime and being arrested, Sanders began displaying signs of mental incompetence. This led Charters to move to commit Sanders to a state institution and have his mental health evaluated. The Kentucky state court granted the motion and ordered Sanders moved to KCPC. There, after one and half months of study, Walker produced her report finding Sanders competent. Despite being found competent, Charters continued to have concerns. At the beginning of Sanders's trial, Charters asked the judge if he could get a doctor visit for Sanders, stating that Sanders had “gone two nights now unable to sleep and he's – I'm losing him. I'm losing his attention. He can't concentrate.” DE 166-2, Trial Tr., Page ID 3725. The state court permitted Charters to make “arrangements” with the jail, but Charters did not request a new competency evaluation. After Sanders was found guilty, Charters told the court that Sanders was physically unable to testify that day, although after the weekend Sanders could testify.

After the jury's recommendation of death, the state court asked Charters and Sanders whether there was any reason that it should not impose judgment at that time. In response Charters stated: “I do not believe that Mr. Sanders is competent at this stage of the proceedings. I don't believe that he understands the nature of what is happening right now. I think he understands what you are doing, but he does not understand why. And I question whether he has the ability to go through this stage.” DE 166-6, Trial Tr., Page ID 4007. So, the trial court conducted its own colloquy and asked Sanders whether he understood what was happening. Sanders said that he did, but that he did not “understand how it's come this far.” Id. 4008. The trial court continued, asking Sanders if he understood that he had been found guilty of the crimes of murder and robbery and that it was the state court's job to sentence him. Sanders said that he understood. Satisfied, the court found him competent.

In Sanders I, the Kentucky Supreme Court held that, under these facts, the trial court “was not required to order a competency hearing pursuant to RCr 8.06.” 801 S.W.2d at 682 (citing Ky. Rule. Crim. Pro. 8.06)). In Sanders II, the Kentucky Supreme Court held that because Sanders had already brought his claim that Charters should have requested another competency evaluation in Sanders I, Sanders could not re-raise it. See 89 S.W.3d at 391. In so doing, the Kentucky Supreme Court misread Sanders I, which concerned the trial court's independent responsibility to order another competency evaluation; in Sanders II, Sanders brought a separate claim that Charters should have requested one. Therefore, the Kentucky Supreme Court in Sanders II did not reach the merits on this issue and AEDPA does not apply, and we review Sanders's Strickland claim de novo.

Constitutionally, trying an incompetent defendant violates due process. See Godinez v. Moran, 509 U.S. 389, 396, 113 S.Ct. 2680, 125 L.Ed.2d 321 (1993)․ The test is whether the defendant has “sufficient present ability to consult with his lawyer with a reasonable degree of rational understanding—and whether he has a rational as well as factual understanding[s] of the proceedings against him.” Drope v. Missouri, 420 U.S. 162, 170 n.7, 95 S.Ct. 896, 43 L.Ed.2d 103 (1975) (quotation omitted). Otherwise, someone who became “ ‘mad’ after the commission of an offense ․ should not be tried, ‘for how can he make his defense?’ ” Id. at 171, 95 S.Ct. 896 (quoting 4 W. Blackstone Commentaries, *24)). Charter's failure to seek another competency hearing is governed by Strickland. Thus, Sanders must show both deficient performance and prejudice. See United States v. Dubrule, 822 F.3d 866, 881 (6th Cir. 2016). To demonstrate prejudice, Sanders must show that if Charters had sought a new competency hearing, there is a reasonable probability that the outcome of the proceeding would have been different. Id. at 881–82.

We agree with the district court's finding that Sanders has failed to establish a reasonable probability that he would have obtained a second competency hearing if Charters requested one. Sanders had already received a comprehensive, six-week evaluation before trial. From this evaluation, he was found competent to stand trial. “There is no right to a continual succession of competency hearings in the absence of some new factor.” Pate v. Commonwealth, 769 S.W.2d 46, 47 (Ky. 1989). To obtain another competency evaluation, Sanders would have had to convince the judge that a new factor or change in his condition warranted one. See id. The only new factors were a statement that Sanders was having trouble paying attention and that, before sentencing, Charters believed Sanders understood what was happening but not why. But it is understandable that Sanders may have been stressed and indecisive about whether to testify. See Cox v. Compton, No. 98-6209, 2000 WL 84458, at *3 (6th Cir. Jan. 12, 2000) (“[C]onfusion and indecision do not seem to us inconsistent with legal competence, which requires only an ability to understand the proceedings and to consult rationally with defense counsel.”); United States v. Van Dyke, 605 F.2d 220, 226 (6th Cir. 1979) (“[S]ome degree of emotional strain exists in every criminal proceeding[.]”).

In addition, before sentencing, when Charters raised the issue of competence again, the state court conducted its own colloquy and was satisfied that Sanders was competent. Thus, these three situations together with the fact that Sanders already had an intensive six-week mental health evaluation and was found by the trial court to be competent, indicate that there is no reasonable possibility that, if Charters had asked for another mental competency evaluation, it would have been granted. Therefore, under de novo review, Sanders cannot satisfy the prejudice prong of Strickland. Accordingly, no Strickland violation arises from Charters's failure to ask for a second competency evaluation.

C.

Finally, Sanders argues that the cumulative effect of Charter's errors prejudiced him. According to Supreme Court precedent, while we analyze each Strickland claim individually, we should aggregate each alleged ineffective assistance of counsel violation and consider, as a whole, whether there was prejudice. See Rompilla, 545 U.S. at 393, 125 S.Ct. 2456; Williams, 529 U.S. at 399, 120 S.Ct. 1495; see also Mackey v. Russell, 148 F. App'x 355, 368–69 (6th Cir. 2005). Sanders did not initially raise his cumulative prejudice argument in state court. He later brought it before the Kentucky Supreme Court in Sanders III. See 339 S.W.3d at 436 (“(36) cumulative prejudice”). There, the Kentucky Supreme Court held that this claim, among many others, was barred for Sanders's failure to bring it in either Sanders I (direct review) or Sanders II (the initial collateral attack). See id. at 437. Therefore, the Kentucky Supreme Court relied “on a procedural bar to dispose of the claim at hand,” and Sanders's cumulative prejudice claim was not adjudicated on the merits. See Barton, 786 F.3d at 464.

As a result, Sanders's claim for cumulative prejudice was procedurally defaulted, and we cannot hear his claim unless he shows cause. Haliym v. Mitchell, 492 F.3d 680, 690 (6th Cir. 2007). To show cause, Sanders must establish “some objective factor external to the defense impeded counsel's efforts to comply with the State's procedural rule.” Murray v. Carrier, 477 U.S. 478, 488, 106 S.Ct. 2639, 91 L.Ed.2d 397 (1986). No such cause exists in this record. Sanders's cumulative prejudice claim thus fails.

V.

Accordingly, we uphold the constitutionality of AEDPA. Applying AEDPA to some claims and reviewing other claims de novo, we deny Sanders's habeas petition.

DISSENT

In my view, the majority opinion misinterprets Supreme Court caselaw regarding the appropriate role of federal courts and thereby errs in holding constitutional a law that requires courts to abdicate the authority the founders vested in the federal courts. Longstanding precedent demonstrates that AEDPA is unconstitutional. Further, regardless of the standard of review issue in this case, Sanders is entitled to habeas relief. Therefore, I respectfully dissent.

I. BACKGROUND

A. State Conviction

On January 28, 1997, David Lee Sanders, by his own admission, shot two men at the Boone Variety Store in Madison County, Kentucky. He walked into the store and purchased orange juice and candy. He used the store's pay phone. He walked out to his car, got his gun from the car, and returned to the store. He approached the store's counter where the store's owner was standing. The owner was writing something with his back towards Sanders. Sanders shot him in the back of the head. When another customer entered the store, Sanders shot that man too. He took money from both men.

Shortly after Sanders was arrested, he began showing disturbing mental health symptoms. The psychiatrist at the jail where Sanders was initially held after his arrest wrote a letter to the court expressing concern for Sanders's emotional state, worrying that Sanders was suicidal. Sanders's retained counsel moved for Sanders to be committed to a state institution for evaluation of competence and “neurological health” because he was in such a state of mental anguish that he was having difficulty assisting with his own defense and displaying symptoms including pain, dizziness, and sensitivity to light.

The court ordered that Sanders be transferred to the Kentucky Correctional Psychiatric Center (KCPC) where he would be evaluated for competency to stand trial, competency at the time of the crime, and any physio-neurological defects or deficiencies. After observing Sanders for a month and a half, a team of mental health professionals from KCPC produced a report. The KCPC team of psychiatrists concluded that Sanders did not suffer from a psychotic illness or major affective disorder, that he did not suffer from hallucinations, delusions, or paranoia, and that there was no evidence of central nervous system dysfunction. As a result, the KCPC Report indicated that Sanders had no condition that would render him not criminally responsible. The Report further concluded that Sanders understood the nature of the proceedings against him and was competent to stand trial.

In the lead up to trial, Sanders's counsel failed to interview any of the professionals who had treated or evaluated Sanders either at the jail or at KCPC. Instead, during the week before trial, counsel contacted Dr. Cooke, an independent expert who met with Sanders for three hours just days before testifying, to do an evaluation. Dr. Cooke reviewed the KCPC Report and police reports and interviewed one of Sanders's brothers but was not provided with any other psychiatric information about Sanders. Dr. Cooke concluded that Sanders suffered from “depersonalization disorder” at the time of the crime.

At trial, the defense presented an insanity defense based on the testimony of exactly two witnesses: Dr. Cooke and Sanders himself. Sanders admitted to having shot both victims as well as to having shot another woman in an earlier incident. He described significant amnesia regarding the incident and testified that, at the time he was unsure why he was acting the way he was and was unable to stop himself. Dr. Cooke then testified briefly. He was accepted as an expert and testified that Sanders suffered from depersonalization disorder at the time of the crime. He explained that a person with such a disorder will “feel[ ] that they are not themselves any longer, that they are somehow outside of their body, watching themselves do something ․ [and] that they do not have any control any longer over their body.” R. 166-3, Trial Tr., PageID 3857-58. Such symptoms, Dr. Cooke explained, can strike suddenly, without any warning. He concluded that, at the time of the crime, Sanders “did not have any control over what was happening to him.” Id. at PageID 3858.

Several key facts undermined Dr. Cooke's credibility. First, he had only evaluated Sanders for three hours (as compared to the KCPC doctors who had evaluated him over a month and a half). Second, perhaps because of his rushed preparation, he was caught in several errors at trial, which the prosecutor made a point of highlighting to the jury at closing argument. He misstated the date on which he had evaluated Sanders, testifying to having evaluated Sanders at a time when the jury knew Sanders had been in trial. He even misstated Sanders's name, calling him “Saunders.”

After testimony, the jury was instructed. The court began with a series of instructions setting out the elements of the State's burden for each crime with which Sanders was charged. Each began with uniform language instructing the jury that “You will find the defendant, David Lee Sanders, guilty under this instruction if and only if you believe from the evidence beyond a reasonable doubt all of the following[.]” The court then instructed the jury on the insanity defense:

Instruction number seven, Insanity. Even though you might otherwise find the defendant guilty of the offenses mentioned in instructions number 2, 3, 4 and 5, if you believe from the evidence that at the time David Lee Sanders committed these offenses, if he did so, he was of unsound mind, you shall find him not guilty and say in your verdict that you find him not guilty on the ground of insanity. Before the defendant can be excused on the ground of insanity, you must believe from the evidence that at the time of the act in question the defendant as a result of mental disease or defect, A, did not have substantial capacity to appreciate the criminal nature of the act or, B, if he did have such capacity he did not have substantial capacity to conform his conduct to the requirements of the law. The term “mental disease” or “defect” does not include an abnormality manifested only by repeated criminal or otherwise anti-social conduct.

R. 166-4, Trial Tr., PageID 3916-17. Sanders was found guilty and sentenced to death. Sanders v. Commonwealth, 801 S.W.2d 665, 668 (Ky. 1990).

B. Post-Trial Proceedings

Sanders appealed his convictions, raising a host of issues. Id. Relevant here, Sanders argued that the jury instructions given at his trial were misleading as to the burden of proof for insanity because the instructions failed to explain a burden other than “beyond a reasonable doubt.” Id. at 679. The Kentucky court analyzed the issue in its entirety as follows:

It is further submitted that the instructions failed to specify the burden of proof regarding the defense of insanity. The jury was instructed that it might find the defendant not guilty by reason of insanity if it believed “from the evidence” that he was insane at the time of the offenses. We find no error in this instruction. See Gall v. Commonwealth, Ky., 607 S.W.2d 97 at 110.

Id. The court affirmed Sanders's conviction. Id. at 684.

Sanders then initiated postconviction proceedings in the Kentucky courts pursuant to Kentucky Rule of Criminal Procedure 11.42. He raised multiple claims of ineffective assistance of counsel, including the claims that counsel was ineffective for failing to acquire competent psychiatric testimony either from a Dr. Flenning who had worked on the KCPC team of psychiatrists, or by better preparing Dr. Cooke. Sanders v. Commonwealth, 89 S.W.3d 380, 385-88 (Ky. 2002).

During his post-conviction proceedings, Sanders introduced evidence that Dr. Flenning had prepared a separate report from the one produced by the KCPC team as a whole. This report had never been acquired by the State and was not turned over to the defense. Id. at 385-86. It was not referenced in the KCPC Report. In fact, the KCPC Report represented itself as summarizing all of the findings of the team of psychologists including Dr. Flenning. The Flenning Report came to conclusions more favorable to Sanders than the KCPC Report, including that there was “strong evidence for an atypical psychosis” such as “pseudo-neurotic schizophrenia.” R. 165-16, PageID 3346-47. Dr. Flenning also concluded that “[u]nder unusual stresses and circumstances he has the potential for brief psychotic episodes wherein he would experience a loss of contact with reality and possibly delusional thinking.” Id. at PageID 3347.

Sanders also introduced evidence that he had been evaluated for a different murder case by a team of independent experts who, unlike Dr. Cooke, had time to adequately evaluate Sanders and prepare reports with conclusions. Two experts examined Sanders over the course of several sessions, and both concluded, as Dr. Cooke did, that Sanders suffered dissociative episodes characterized by depersonalization in which his ability to control himself was impaired. Both concluded that Sanders lacked the capacity to conform his conduct to the requirements of the law at the time of Sanders's other crimes. The state trial court denied Sanders's postconviction claims. Sanders, 89 S.W.3d at 384.

Sanders's appeal from his post-conviction proceedings again reached the Kentucky Supreme Court. Id. The Kentucky court reasoned that, even if Sanders's counsel should have acquired Dr. Flenning's report, there was no prejudice because “[t]he use of the Flenning report to cross-examine Dr. Walker, or by calling Dr. Flenning as a witness would have resulted in a reinforcement of the argument by the prosecutor about inconsistencies in Sanders’[s] statements and illustrate that Sanders repeatedly changed his version of various events as he was interviewed by different evaluators.” Id. at 386.

As for Sanders's claim that his counsel had failed to provide him with adequate expert representation in the form of Dr. Cooke, the court reasoned:

Ake v. Oklahoma, 470 U.S. 68, 105 S.Ct. 1087, 84 L.Ed.2d 53 (1985), commented that “Psychiatry is not, however, an exact science, and psychiatrists disagree widely and frequently on what constitutes mental illness, on the appropriate diagnosis to be attached to given behavior and symptoms, on cure and treatment, and on the likelihood of future dangerousness.” Sanders was sent on his own motion to KCPC for a six-week psychiatric evaluation by a team of social workers, psychologists, a psychiatrist and a neurologist․ The fact that an additional evaluation might be beneficial to the defense does not add credibility to his claim. The trial judge stated that Sanders submitted reports from a psychologist, a psychiatrist and a Ph.D. These experts were retained in connection with a defense of charges resulting from a 1986 shooting in Lincoln County. As correctly observed by the trial judge, the reports and the testimony would have made no difference in the outcome of the trial. The jury rejected the opinion offered by Dr. Cooke and there is no reason to believe they would have accepted a similar opinion simply because it came from a different defense expert. Dr. Cooke indicated that he was willing to testify that in his opinion Sanders satisfied the legal standard for insanity. Thus, Sanders had access to qualified mental health experts to establish his insanity defense. His complaints about ineffectiveness are without merit.

Id. at 387-88. The Kentucky Supreme Court affirmed the district court's decision to deny Sanders's postconviction claim. Id. at 394.

Sanders filed a petition for a writ of habeas corpus in federal district court on December 18, 2003, alleging that the Kentucky Supreme Court had acted unreasonably under or contrary to federal law. The district court denied the petition.

II. STANDARD OF REVIEW

Because the district court did not hold an evidentiary hearing, we review both the district court's legal conclusions and the district court's factual finding's de novo. Barton v. Warden, S. Ohio Corr. Facility, 786 F.3d 450, 460 (6th Cir. 2015). The majority opinion concludes that, because this is a habeas action, this court's ability to grant relief is governed by the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA). See generally 28 U.S.C. § 2254. AEDPA prohibits a federal habeas court from granting relief on “any claim that was adjudicated on the merits in State court proceedings unless” the state court decision “was contrary to, or involved an unreasonable application of, clearly established Federal law,” as determined by the Supreme Court of the United States, id. § 2254(d)(1), or the state court's decision “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding,” id. § 2254(d)(2).

The first fundamental question we must answer is whether this so called “AEDPA deference” is constitutionally permissible. The majority opinion holds that it is. I would hold that the § 2254(d) standard is unconstitutional. Analysis of this issue begins with the fact that Congress has granted federal courts jurisdiction to determine whether a prisoner is being held in violation of the Constitution. The fundamental issue is whether, when a federal court exercises that grant and finds the imprisonment unconstitutional, Congress can make that court's ability to grant relief dependent on deference to a state interpretation of the Constitution. As explained below, I would hold that such a rule is not constitutional and would, therefore, review Sanders's constitutional claims de novo without deference to the state courts.

A. Unconstitutional Deference

In The Federalist Papers No. 78, Hamilton recognized that the judiciary must be “truly distinct from both the legislative and executive. For I agree that there is no liberty, if the power of judging be not separated from the legislative and executive powers.” The Federalist No. 78 (quotation omitted). He noted that “[t]he interpretation of the laws is the proper and peculiar province of the courts. A constitution is, in fact, and must be regarded by the judges, as a fundamental law. It therefore belongs to them to ascertain its meaning.” Id. The historical development of that stark separation of powers sheds light on the force it should have.

The current text of Article III of the Constitution is the result of a compromise between two factions at the constitutional convention. James Madison originally proposed a federal judiciary that would exclusively resolve all questions of federal law at least at the appellate level. Letter from James Madison to George Washington (Apr. 16, 1787), in 9 The Papers of James Madison 382, 384 (Robert A. Rutland, et al. eds., 1975) [hereinafter Madison Papers]. His opponents wanted to leave federal question jurisdiction to the state courts, giving them a duty to apply national law, and to exclude lower federal courts entirely. 1 The Records of the Federal Convention of 1787, at 244-45 (Max Farrand ed. 1911) [hereinafter 1 Farrand]. At the same time, there was significant debate as to whether there should be a national legislative veto for state laws not in the national interest. Compare Letter from James Madison to Thomas Jefferson (March 19, 1787), in Madison Papers 317, 318-19 (setting forth a case for a national negative), and 2 The Records of the Federal Convention of 1787, at 27 (Max Farrand ed. 1911) [hereinafter 2 Farrand] (arguing that the negative is “essential to the efficacy [and] security of the” government), with id. at 28 (arguing that the negative would “disgust” all states and noting that it failed three states to seven). The compromise between these positions was to delegate to judges (including federal judges) the task of voiding unconstitutional state law. See 2 Farrand 28-29 (proposing a prototype of the Supremacy Clause immediately after the failure of the national negative); id at 38-39 (unanimously approving the addition of inferior federal courts the day after the failure of the national negative to review questions of federal law).

The express purpose of allowing federal judges this power was to avoid situations in which state judges, loyal to state law, would misapply national law to preserve their own state law. Letter from James Madison to George Washington (Apr. 16, 1787) in Madison Papers 382, 384. Even those opposed to Madison's positions acknowledged that there was a need for federal courts to protect “national rights.” 1 Farrand 124 (noting that Rutlidge and Sherman, staunch opponents of Madison's schemes, agreed that there was a need for a federal judiciary to protect national rights). Madison argued that the purpose of lower federal courts was to ensure there was a remedy “after improper [v]erdicts in State tribunals obtained under the [biased] directions of a dependent Judge” and to step in when the states persisted in behavior violative of the Constitution. Id. This history indicates that the Founders would have objected to a rule that required federal courts to subject their interpretation of the Constitution, even in a single case, to that of the state courts. Doing so would undermine the very purpose of lower federal courts—stopping biased state judges from circumventing the Constitution.

The Court in Loper Bright Enterprises v. Raimondo 1 recognized the problems with a federal court deferring to non-Article III actors:

The Framers also envisioned that the final “interpretation of the laws” would be “the proper and peculiar province of the courts.” ․ To ensure the “steady, upright and impartial administration of the laws,” the Framers structured the Constitution to allow judges to exercise that judgment independent of influence from the political branches․ This Court embraced the Framers’ understanding of the judicial function early on. In the foundational decision of Marbury v. Madison, Chief Justice Marshall famously declared that “[i]t is emphatically the province and duty of the judicial department to say what the law is.”

603 U.S. 369, 385, 144 S.Ct. 2244, 219 L.Ed.2d 832 (2024) (citations omitted). But the Court in Loper Bright was by no means setting out a new argument regarding the special role of Article III courts. Stern v. Marshall teaches that “[A]rticle III could neither serve its purpose in the system of checks and balances nor preserve the integrity of judicial decisionmaking if the other branches of the Federal Government could confer the Government's ‘judicial Power’ on entities outside Article III. That is why we have long recognized that, in general, Congress may not ‘withdraw from judicial cognizance any matter which, from its nature, is the subject of a suit at the common law, or in equity, or admiralty.’ ” 564 U.S. 462, 484, 131 S.Ct. 2594, 180 L.Ed.2d 475 (2011). As far back as 1824, the Supreme Court in Cohens v. Virginia held that it would be “treason[ous] to the [C]onstitution” for Article III courts to defer to a state reading of the Constitution and decline to provide an independent reading simply because the matters were doubtful. 19 U.S. (6. Wheat.) 264, 404, 5 L.Ed. 257 (1821). Presumably, Congress cannot mandate that courts commit treason against the Constitution.

Of those courts that have upheld AEDPA, most have conceded that a rule requiring the federal courts to defer to a non-Article III actor as to the proper interpretation of the Constitution would be unconstitutional. In the words of the Seventh Circuit, “Congress lacks power to revise the meaning of the Constitution or to require federal judges to ‘defer’ to the interpretations reached by state courts. Once the judicial power is brought to bear by the presentation of a justiciable case or controversy within a statutory grant of jurisdiction, the federal courts’ independent interpretive authority cannot constitutionally be impaired.” Lindh v. Murphy, 96 F.3d 856, 872 (7th Cir. 1996) (en banc). And the majority opinion in this case does not contend that Congress can mandate that Article III courts accept state interpretations of the Constitution. Maj. Op. at ––––.

B. Unconstitutional Dictation of Remedies

The next question, then, is whether Congress can create an end run around the unconstitutionality of deference to the state courts by saying that Article III courts can recognize a state's unconstitutional actions but cannot do anything about them unless the state acted unreasonably. Congress can, of course, limit the types of cases in which courts can grant remedies, and what types of remedies courts may give, even for constitutional violations. Felker v. Turpin, 518 U.S. 651, 658, 116 S.Ct. 2333, 135 L.Ed.2d 827 (1996) (“We conclude that although the Act does impose new conditions on our authority to grant relief, it does not deprive this Court of jurisdiction to entertain original habeas petitions.”). Thus, it is possible that, in some cases, a state court could violate a defendant's rights and a reviewing federal court would be unable to grant a remedy. As multiple circuits reviewing this issue have noted, the idea that Article III courts would be confronted with a constitutional violation and unable to remedy the issue is not new:

Plain error doctrine authorizes courts to correct forfeited errors only in rare circumstances. The harmless error rule plays a similar, albeit less severe, function. Qualified immunity can prevent meritorious constitutional plaintiffs from recovering. In the habeas context, the Supreme Court's retroactivity doctrine creates a chasm between valid claims and the right to relief. Supreme Court precedent similarly forecloses federal habeas relief for state prisoners convicted on the basis of evidence obtained in an unconstitutional search and seizure. In short, federal courts routinely deny relief even for known constitutional violations. The distinction between rights and remedies is fundamental.

Cobb v. Thaler, 682 F.3d 364, 375 (5th Cir. 2012) (citations omitted).

What AEDPA does, however, is subtly different from what occurs in all those other types of cases. It does not just provide a limit on relief for the underlying constitutional violation; it makes relief dependent on the state law regarding that violation—which state law may, itself, be unconstitutional. In the situations discussed above, it is the nature of the violation (e.g., retroactivity, qualified immunity, Fourth Amendment violations on habeas) or some fact with no constitutional valence that determines whether relief may be granted (e.g., the strength of the evidence in harmless error cases, or the lack of an objection in plain error cases). AEDPA makes a federal court's decision to grant habeas relief dependent on the substance, not only of the underlying constitutional violation, but on the rule of decisional law the state announces when it decides the prisoner's case. As Judge Ripple noted when the Seventh Circuit considered this issue:

There is a qualitative difference—a constitutional difference—between fixing the time frame at which federal law will be applied and requiring that the federal court defer to any application of constitutional principle that cannot be characterized as unreasonable. Under the latter approach, mandated by the amended statute, the federal court must be content with a careful application of the Constitution, even if it is wrong. In essence, the federal court is free to have its own opinion of what federal law requires, but it must grant or deny the writ on the basis of another non-federal tribunal's view.

Lindh v. Murphy, 96 F.3d 856, 889 (7th Cir. 1996) (Ripple, J., dissenting).

To see this distinction, consider two hypothetical prisoners: Joe and Bob. In Joe's case, a constitutional violation, X, happens under circumstances A, B, and C. The state supreme court takes up Joe's case and decides that X was not a constitutional violation and that, in any case, it was harmless under circumstances A, B, and C. In Bob's case, the same constitutional violation, X, occurs, under the same circumstances, A, B, and C. The state supreme court takes up Bob's case and decides that X was not a constitutional violation and never reaches the harmlessness analysis. Imagine further, that, in fact, under the Constitution, X is a constitutional violation and is not harmless.

Despite the fact that the same constitutional violation, X, occurred under the same circumstances, A, B, and C, in Joe's and Bob's trials, a federal court's ability to grant relief is materially different in each case due to the different handling by the state courts. In Joe's case, even if the federal court finds that there was a constitutional violation and that the state court acted unreasonably in holding otherwise, that court cannot grant relief unless the state court was also unreasonable in analyzing harmlessness. Brown v. Davenport, 596 U.S. 118, 122, 142 S.Ct. 1510, 212 L.Ed.2d 463 (2022). In Bob's case, if the federal court reaches harmlessness, it can grant relief after mere de novo review. Rompilla v. Beard, 545 U.S. 374, 390, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005). The rule of decisional law announced by the state supreme court in Joe's case but not in Bob's—that violation X is not harmless under circumstances A, B, and C—is, therefore, playing a material role in the federal court's decisionmaking separate from the role played by the underlying violation.

The question, then, is not merely whether there can be a constitutional violation for which Congress prohibits Article III courts from granting relief—clearly there can be. It is whether Congress can make a court's ability to grant relief for a constitutional violation dependent on the existence of a reasonable state law saying that the act in question is not a violation. More broadly, can Congress say that Article III courts cannot grant relief for a constitutional violation for which they could otherwise grant relief if a particular non-Article III actor happens to have a different interpretation of the Constitution?

First of all, the idea that we can simply say “Congress has broad powers to regulate the jurisdiction of inferior federal courts ․ [and] to regulate our ability to grant remedies” and have done with the question would have absurd results. Maj. Op at ––––. Taken to its logical extreme, this reasoning would allow Congress to pass a law saying that, while federal courts have jurisdiction to decide whether the laws of Congress (or of some state) are constitutional, Article III courts can only grant relief where there is no reasonable interpretation of the Constitution that would permit such a law. After all, in cases where the law was unconstitutional but reasonable, courts would not be prohibited from saying whether the law was constitutional, only from granting relief. But Marbury v. Madison teaches that Congress cannot do that. 5 U.S. 137, 177-78, 1 Cranch 137, 2 L.Ed. 60 (1803). Congress cannot pass a law requiring courts to give effect to an unconstitutional law:

It is emphatically the province and duty of the judicial department to say what the law is. Those who apply the rule to particular cases, must of necessity expound and interpret that rule. If two laws conflict with each other, the courts must decide on the operation of each.

So if a law be in opposition to the constitution; if both the law and the constitution apply to a particular case, so that the court must either decide that case conformably to the law, disregarding the constitution; or conformably to the constitution, disregarding the law; the court must determine which of these conflicting rules governs the case. This is of the very essence of judicial duty.

If then the courts are to regard the constitution; and the constitution is superior to any ordinary act of the legislature; the constitution, and not such ordinary act, must govern the case to which they both apply.

Those then who controvert the principle that the constitution is to be considered, in court, as a paramount law, are reduced to the necessity of maintaining that courts must close their eyes on the constitution, and see only the law.

This doctrine would subvert the very foundation of all written constitutions. It would declare that an act, which, according to the principles and theory of our government, is entirely void; is yet, in practice, completely obligatory. It would declare, that if the legislature shall do what is expressly forbidden, such act, notwithstanding the express prohibition, is in reality effectual.

Id. In other words, Article III courts cannot recognize that a law is unconstitutional and then apply it anyway to a particular case, letting it—rather than the Constitution—dictate the outcome of the case. At the very least, this teaches that, while Congress can prohibit Article III courts from granting relief for a constitutional violation in some cases, the fact that Congress dictates relief rather than directly dictating a rule of constitutional decision does not automatically make permissible Congress's attempts to interfere with constitutional adjudication by Article III courts. So, there must be some intermediate rule governing when Congress can and cannot prohibit Article III courts from granting relief that distinguishes between cases like plain error review and cases like the hypothetical in which Congress simply prohibits courts from granting relief from any of its reasonable but unconstitutional laws.

Two decisions provide some guidance on the matter. First, in United States v. Klein, certain kinds of relief were due to individuals residing in the South who had not participated in the Civil War. 80 U.S. 128, 137, 13 Wall. 128, 20 L.Ed. 519 (1871). The Court ruled that, under the Constitution, those who had been pardoned were, likewise, entitled to relief. Id. at 141-42. Unhappy with that ruling, Congress tried to use its ability to strip courts of jurisdiction to deny jurisdiction where there was evidence of a pardon. Id. at 143-44. The Court reasoned:

Undoubtedly the legislature has complete control over the organization and existence of that court and may confer or withhold the right of appeal from its decisions. And if this act did nothing more, it would be our duty to give it effect. If it simply denied the right of appeal in a particular class of cases, there could be no doubt that it must be regarded as an exercise of the power of Congress to make “such exceptions from the appellate jurisdiction” as should seem to it expedient.

But the language of the proviso shows plainly that it does not intend to withhold appellate jurisdiction except as a means to an end. Its great and controlling purpose is to deny to pardons granted by the President the effect which this court had adjudged them to have. The proviso declares that pardons shall not be considered by this court on appeal. We had already decided that it was our duty to consider them and give them effect, in cases like the present, as equivalent to proof of loyalty. It provides that whenever it shall appear that any judgment of the Court of Claims shall have been founded on such pardons, without other proof of loyalty, the Supreme Court shall have no further jurisdiction of the case and shall dismiss the same for want of jurisdiction. The proviso further declares that every pardon granted to any suitor in the Court of Claims and reciting that the person pardoned has been guilty of any act of rebellion or disloyalty, shall, if accepted in writing without disclaimer of the fact recited, be taken as conclusive evidence in that court and on appeal, of the act recited; and on proof of pardon or acceptance, summarily made on motion or otherwise, the jurisdiction of the court shall cease and the suit shall be forthwith dismissed.

It is evident from this statement that the denial of jurisdiction to this court, as well as to the Court of Claims, is founded solely on the application of a rule of decision, in causes pending, prescribed by Congress. The court has jurisdiction of the cause to a given point; but when it ascertains that a certain state of things exists, its jurisdiction is to cease and it is required to dismiss the cause for want of jurisdiction.

It seems to us that this is not an exercise of the acknowledged power of Congress to make exceptions and prescribe regulations to the appellate power.

The court is required to ascertain the existence of certain facts and thereupon to declare that its jurisdiction on appeal has ceased, by dismissing the bill. What is this but to prescribe a rule for the decision of a cause in a particular way? In the case before us, the Court of Claims has rendered judgment for the claimant and an appeal has been taken to this court. We are directed to dismiss the appeal, if we find that the judgment must be affirmed, because of a pardon granted to the intestate of the claimants. Can we do so without allowing one party to the controversy to decide it in its own favor? Can we do so without allowing that the legislature may prescribe rules of decision to the Judicial Department of the government in cases pending before it?

Id. at 145-46. From Klein we can draw a simple principle: Congress cannot use its power to regulate procedure to functionally effectuate an unconstitutional rule of decision. Congress can regulate relief in some ways, just as it can regulate jurisdiction. But Congress cannot use its power to regulate relief to functionally create an erroneous rule of decision any more than it can tailor jurisdiction to do the same. Tracking the logic of Klein, if Congress is permitted to require courts to ascertain the existence of an incorrect but reasonable non-Article III interpretation of the Constitution prohibiting relief and, as a result, declare that no relief can be provided, then Congress is permitted to functionally create an unconstitutional rule of decision.

Congress can create rules of relief like plain error that are just that: rules of relief. Similarly, Congress can strip the courts of jurisdiction to hear certain types of cases as acknowledged in Klein. Id at 145. But Congress can gerrymander neither relief nor jurisdiction as a “means to an end” to force courts to functionally decide cases by an unconstitutional rule of decision put forward by a non-Article III actor (in Klein’s case Congress, in AEDPA's the state court). Id.

Second, in Reynoldsville Casket Co. v. Hyde, after the Supreme Court had already found an Ohio “tolling” provision of its statute of limitations unconstitutional, Court precedent indicated that such prior ruling applied to all pending cases. 514 U.S. 749, 750-71, 115 S.Ct. 1745, 131 L.Ed.2d 820 (1995). The Supreme Court of Ohio, however, held that the prior ruling “may not be retroactively applied to bar claims in state courts which had accrued prior to the announcement of that decision.” Id. at 751-72, 115 S.Ct. 1745. Hyde argued, as the defenders of AEDPA do today, that the Court should “look at what the Ohio Supreme Court has done, not through the lens of ‘retroactivity,’ but through that of ‘remedy.’ ” Id. at 752, 115 S.Ct. 1745. The plaintiff's stance was that the State could simply decline to provide to the defendant any remedy for the unconstitutionality of the tolling provisions. Id. As examples, she cited both qualified immunity and Teague. Id. at 757, 115 S.Ct. 1745. The Court reasoned, however,

Thus, Hyde asks, why not look at what the Ohio Supreme Court has done in this case as if it were simply an effort to fashion a remedy that takes into consideration her reliance on pre-Bendix law? Here, the remedy would actually consist of providing no remedy for the constitutional violation or, to put the matter more precisely, of continuing to toll the 2-year statute of limitations in pre-Bendix cases, such as hers[.] ․ [W]e do not see how, in the circumstances before us, the Ohio Supreme Court could change a legal outcome that federal law, applicable under the Supremacy Clause, would otherwise dictate simply by calling its refusal to apply that federal law an effort to create a remedy. The Ohio Supreme Court's justification for refusing to dismiss Hyde's suit is that she, and others like her, may have reasonably relied upon pre-Bendix law[.] ․ But, this type of justification—often present when prior law is overruled -- is the very sort that this Court, in Harper[ v. Virginia Dept. of Taxation, 509 U.S. 86, 113 S.Ct. 2510, 125 L.Ed.2d 74 (1993)], found insufficient to deny retroactive application of a new legal rule (that had been applied in the case that first announced it). If Harper has anything more than symbolic significance, how could virtually identical reliance, without more, prove sufficient to permit a virtually identical denial simply because it is characterized as a denial based on ‘remedy’ rather than ‘nonretroactivity’?

Id. at 753-54, 115 S.Ct. 1745 (emphasis added). The upshot of this case is similar to the upshot of Klein: courts cannot simply substitute an unconstitutional reading of the law for a constitutional one by couching what they are doing in the language of “remedy” or “jurisdiction.” Where the effect of the choice of “remedy” is functionally the application of a different rule of decision, the choice of “remedy” is as unconstitutional as the original rule of decision was.

The full impropriety of AEDPA deference under the reasoning of Klein and Reynoldsville Casket Co, is evidenced by an analogous hypothetical. Suppose Congress appointed a partisan, political body to decide constitutional questions (e.g., a committee of partisan constitutional scholars) and then empowered courts to grant relief only in cases where that body's interpretation of the Constitution would suggest that the court could grant relief or where that body's interpretation was unreasonable. True, under that framework courts could still “decide what the law is,” but in close questions they could grant relief only where the partisan political body interpreted the Constitution in the same way.

In this hypothetical, the “relief,” owed to a plaintiff whose rights were violated under an Article III court's reading of the Constitution but not under the partisan body's would be like the “remedy” in Reynoldsville Casket Co. For all practical purposes, the “remedy” would be whatever the plaintiff would have received under the partisan body's interpretation of the Constitution, just as Hyde's “remedy” was effectively the continued application of the unconstitutional tolling provision. Id. at 753-54, 115 S.Ct. 1745. As in Klein, our ability to grant relief would be “founded solely on ․ a rule of decision” prescribed by a non-Article III actor. 80 U.S. at 146. Neither situation could stand, and if common sense did not tell us that Congress's insertion of a partisan body of constitutional arbiters into the decisionmaking of the court system was unconstitutional, Reynoldsville Casket Co. and Klein would.

Crucially, this hypothetical is identical to AEDPA except that the non-Article III body to whose musings Article III courts must subject their ability to grant relief is different. That difference is immaterial. As the history described above demonstrates, the Continental Congress expressly rejected the idea that States should be treated as though their interpretations of constitutional law were equal to the federal courts’ interpretations in finality and independence. Letter from James Madison to George Washington (Apr. 16, 1787) in Madison Papers 382, 384. The framers considered state courts to be just as much political entities as federal legislative and executive bodies. Id. Thus, the fact that the hypothetical above involves a panel of partisan experts and AEDPA involves state courts makes no difference. The necessary conclusion, then, is that Congress cannot use its power to regulate relief to, for all practical intents and purposes, force Article III courts to defer to state courts’ interpretation of the Constitution any more than Congress could simply mandate directly that Article III courts defer to state law.

One final case draws together these principles and applies them to a closely analogous set of facts. In St. Joseph Stock Yards Co. v. United States, the Court considered the constitutionality of an order of the Secretary of Agriculture fixing maximum rates for the stock yard's services. 298 U.S. 38, 56 S.Ct. 720, 80 L.Ed. 1033 (1936). The stock yards alleged that the rates were unconstitutionally “confiscatory” under the Fifth Amendment, but a representative of the Secretary of Agriculture had already concluded that they were not after conducting his investigation and hearing. Id. at 46, 56 S.Ct. 720. The district court that initially reviewed the case thought that, under the Packers and Stockyards Act, it could not reject the findings of the Secretary of Agriculture regarding the constitutionality of the rates if those findings were “supported by substantial evidence.” The Court explained why the district court was wrong:

[T]he Constitution fixes limits to the rate-making power by prohibiting the deprivation of property without due process of law or the taking of private property for public use without just compensation. When the legislature acts directly, its action is subject to judicial scrutiny and determination in order to prevent the transgression of these limits of power. The legislature cannot preclude that scrutiny or determination by any declaration or legislative finding. Legislative declaration or finding is necessarily subject to independent judicial review upon the facts and the law by courts of competent jurisdiction to the end that the Constitution as the supreme law of the land may be maintained. Nor can the legislature escape the constitutional limitation by authorizing its agent to make findings that the agent has kept within that limitation. Legislative agencies, with varying qualifications, work in a field peculiarly exposed to political demands. Some may be expert and impartial, others subservient. It is not difficult for them to observe the requirements of law in giving a hearing and receiving evidence. But to say that their findings of fact may be made conclusive where constitutional rights of liberty and property are involved, although the evidence clearly establishes that the findings are wrong and constitutional rights have been invaded, is to place those rights at the mercy of administrative officials and seriously to impair the security inherent in our judicial safeguards. That prospect, with our multiplication of administrative agencies, is not one to be lightly regarded. It is said that we can retain judicial authority to examine the weight of evidence when the question concerns the right of personal liberty. But, if this be so, it is not because we are privileged to perform our judicial duty in that case and for reasons of convenience to disregard it in others. The principle applies when rights either of person or of property are protected by constitutional restrictions. Under our system there is no warrant for the view that the judicial power of a competent court can be circumscribed by any legislative arrangement designed to give effect to administrative action going beyond the limits of constitutional authority.

Id. at 52, 56 S.Ct. 720 (emphases added). Congress, with AEDPA, has attempted to do the same thing the Court found impermissible in St. Joseph Stockyards: using a “legislative arrangement” designed to give effect to an unconstitutional action of a non-Article III entity on grounds that the same non-Article III entity reviewed the action and said it was not unconstitutional. Such a law cannot stand.

AEDPA deference is unconstitutional. The fact that it uses a limitation on relief rather than an explicit decisional rule to force Article III courts to subordinate their understanding of the Constitution to that of state courts does not make it any less unconstitutional. Congress may place some limitations on Article III courts’ ability to grant relief in habeas cases, but it may not make the ability of those courts to grant relief for a constitutional violation dependent on the existence of a state law saying that the act in question is not a violation.

C. The Developing Law Regarding AEDPA's Constitutionality

As noted above, there exists a long history of historical and legal discussion regarding the separation of power between our judicial and legislative branches of government. The principles undergirding that discussion address the relationship between Article III courts and state law as well as the role of legislative enactments such as AEDPA deference and have given rise to much scholarly and judicial debate. See, e.g., Anthony G. Amsterdam & James S. Liebman, Loper Bright and the Great Writ, 56 Colum. Hum. Rts. L. Rev. 54 (2024); Transcript of Oral Argument at 80, 114, 127, Relentless, Inc. v. Dep't of Com. 603 U.S. 369 (2024) (No. 22-1219) (companion case to Loper Bright). Drawing from these debates, the majority opinion puts forward a series of arguments in favor of the constitutionality of AEDPA deference.

It first suggests that Supreme Court precedent dictates that AEDPA is constitutional. Maj. Op. at ––––. But the Supreme Court has never squarely considered whether AEDPA is constitutional. Indeed, it has declined to do so. Although the issue of AEDPA's constitutionality was presented in Williams v. Taylor, 529 U.S. 362, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000), the Court declined to grant certiorari on that issue. Compare Petition for a Writ of Certiorari, Williams, 529 U.S. 362 (No. 98-8384) (presenting four proposed questions for the Supreme Court's review, the fourth being the constitutionality of § 2254(d)), with Williams v. Taylor, 526 U.S. 1050, 119 S.Ct. 1355, 143 L.Ed.2d 516 (1999) (granting certiorari only as to the first three questions). While the Supreme Court has often applied AEDPA deference, it has merely assumed its constitutionality. See, e.g., Brown v. Davenport, 596 U.S. 118, 127, 142 S.Ct. 1510, 212 L.Ed.2d 463 (2022). But the Court has never held that it is constitutional. And while other circuits have addressed and rejected the argument that AEDPA deference is unconstitutional, our circuit has not done so in a published opinion. 2

The Second argument raised is that restrictions on the writ of habeas corpus are consistent with historical tradition. Maj. Op. at ––––. Because the scope of habeas review has historically been quite narrow, the argument goes, any newly gerrymandered limit on habeas relief that narrows the scope of the relief closer to historical levels must be acceptable. But to frame the historically narrow scope of habeas review as an indication that Congress can restrict habeas in novel ways now is to misread our nation's constitutional history. As Justice O'Connor made clear, there is a reason habeas review was historically narrow, and it has nothing to do with congressional narrowing of the scope of relief to vindicate unconstitutional state actions:

While it is true that [prior to the 1900s] a state prisoner could not obtain the writ if he had been provided a full and fair hearing in the state courts, this rule governed the merits of a claim under the Due Process Clause. It was not a threshold bar to the consideration of other federal claims, because, with rare exceptions, there were no other federal claims available at the time․ [T]he guarantees of the Bill of Rights were not yet understood to apply in state criminal prosecutions. The only protections the Constitution afforded to state prisoners were those for which the text of the Constitution explicitly limited the authority of the States, most notably the Due Process Clause of the Fourteenth Amendment. And in the area of criminal procedure, the Due Process Clause was understood to guarantee no more than a full and fair hearing in the state courts. See, e.g., Ponzi v. Fessenden, 258 U.S. 254, 260, 42 S. Ct. 309, 66 L. Ed. 607 (1922) (“One accused of crime has a right to a full and fair trial according to the law of the government whose sovereignty he is alleged to have offended, but he has no more than that”).

Thus, when the Court stated that a state prisoner who had been afforded a full and fair hearing could not obtain a writ of habeas corpus, the Court was propounding a rule of constitutional law, not a threshold requirement of habeas corpus. This is evident from the fact that the Court did not just apply this rule on habeas, but also in cases on direct review.

Wright v. West, 505 U.S. 277, 297-98, 112 S.Ct. 2482, 120 L.Ed.2d 225 (1992) (O'Connor, J., concurring). In other words, historical limits on habeas did not exist due to some congressional restriction. There existed no provision for courts to say “the state acted unconstitutionally, but our hands are tied because Congress says so.” It was because habeas then, as today, was only available for violations of federal law. The state could violate constitutional law only if it violated the Due Process Clause, and the Due Process Clause was violated only if the state failed to give the prisoner a full and fair hearing. Id. If habeas was unavailable, it was because the state was not understood to have acted unconstitutionally in the first place, not because the state acted unconstitutionally and the federal courts were prohibited from granting relief. Id. Thus, the historical tradition argument tells us nothing about what restrictions Congress can impose when there is genuine unconstitutional conduct.

The third argument reminds us that state courts have “a role in interpreting the federal Constitution.” Maj. Op. at –––– (quoting Tafflin v. Levitt, 493 U.S. 455, 458-59, 110 S.Ct. 792, 107 L.Ed.2d 887 (1990)). The majority opinion is quite right to note that state courts have traditionally been conceived of as competent evaluators of the Constitution. See The Federalist No. 82. But it is one thing to find that the state courts are competent to do so in the first instance, and another to say that a federal court must enforce an admittedly erroneous state court reading of the Constitution once a federal court is presented with that interpretation. The history demonstrates that the balance struck by the Madisonian compromise included the concession that state courts could interpret the Constitution, along with the creation of federal courts to correct improper applications of federal law by state courts. 2 Farrand 28-39; see also James S. Liebman & William F. Ryan, “Some Effectual Power”: The Quantity and Quality of Decisionmaking Required of Article III Courts, 98 Colum. L. Rev. 696, 730-33 (1998) (summarizing the conception of the Supremacy Clause by the convention as a substitute for a national negative to handle unconstitutional state actions).

The Supreme Court, moreover, rejected the argument that state court competence has any bearing on a federal court's responsibility to correct unconstitutional state actions in 1821. Cohens, 19 U.S. 264. In Cohens, the defendants had been convicted of selling lottery tickets in violation of Virginia law. Id. at 375. The defendants argued that their convictions were illegal under federal law, and the Virginia courts rejected their arguments. Id. Virginia argued that, because state courts have concurrent authority to interpret the Constitution, a reviewing federal court should not re-examine the state court's decisions and overturn them. Id. at 320-21. The Supreme Court rendered a full-throated rejection of this argument:

But a constitution is framed for ages to come, and is designed to approach immortality as nearly as human institutions can approach it. Its course cannot always be tranquil. It is exposed to storms and tempests, and its framers must be unwise statesmen indeed, if they have not provided it, as far as its nature will permit, with the means of self-preservation from the perils it may be destined to encounter. No government ought to be so defective in its organization, as not to contain within itself the means of securing the execution of its own laws against other dangers than those which occur every day. Courts of justice are the means most usually employed; and it is reasonable to expect that a government should repose on its own Courts, rather than on others. There is certainly nothing in the circumstances under which our constitution was formed; nothing in the history of the times, which would justify the opinion that the confidence reposed in the States was so implicit as to leave in them and their tribunals the power of resisting or defeating, in the form of law, the legitimate measures of the Union.

Id. at 387-88. In sum, while state courts have concurrent authority to interpret the Constitution, their authority is not so broad as to make it appropriate for a reviewing federal court to decline to disturb their decisions when those decisions are unconstitutional.

The final argument posited is that Congress can narrow the scope of an Article III court's jurisdiction to hear habeas cases and regulate the scope of the remedies Article III courts can give and the circumstances in which we can give them. But this inherently bows to the same remedies argument the Warden puts forward; it accepts that even if we cannot be commanded to defer to the states, Congress can use its power to strip jurisdiction or prescribe remedies to create a remedies rule that has the same effect. As discussed above, this runs directly afoul of the rule set forth in Klein: Once Congress has granted jurisdiction over a type of case, Congress cannot use its power to regulate procedure to functionally effectuate an unconstitutional rule of decision. 80 U.S. at 145-46.

Faced with the tension between the outcome it reaches and Klein’s holding—that, once Article III courts have been given jurisdiction, Congress cannot use jurisdiction or remedies to impose a rule of decision—the majority opinion attempts to narrow Klein. It cites later precedent which, it claims, demonstrates that Klein did not actually prohibit Congress from dictating a rule of decision to Article III courts. Maj. Op. at –––– – –––– (citing Bank Markazi v. Peterson, 578 U.S. 212, 230, 136 S.Ct. 1310, 194 L.Ed.2d 463 (2016)). But Bank Markazi does not prove the point.

Bank Markazi concerned Congress's ability to dictate how we interpret laws passed by Congress itself. It considered the Iran Threat Reduction and Syria Human Rights Act of 2012, 22 U.S.C § 8772, which made a set of assets held by a New York Bank for Bank Markazi—the central bank of Iran—available to satisfy judgments gained under existing law against Iran by victims of terror attacks sponsored by Iran. 578 U.S. at 215, 136 S.Ct. 1310. Under prior law victims had been able to sue foreign governments for “personal injury or death” caused by terrorism but could only recover damages in the form of property used for commercial activity in the United States. Id. at 216-17, 136 S.Ct. 1310. Section 8772 added the assets of Bank Markazi to the pot of money victims could recover from and made the money available to victims with pending cases. Id. at 218, 136 S.Ct. 1310.

Bank Markazi held that § 8772 had appropriately amended the prior laws and appropriately made those amendments applicable to pending cases. Id. at 226-28, 136 S.Ct. 1310. The Supreme Court determined that Congress was well within its rights to do so and that doing so was functionally different than attempting to dictate that courts apply an incorrect reading of the law. Id. Thus, it concluded that Klein was not implicated at all because Congress was simply creating new, narrow, and highly specific legislation about a particular subject matter, rather than trying to interfere with Article III adjudication. Id. at 228, 136 S.Ct. 1310. The Supreme Court made clear that its holding rested on the fact that Congress had the power to modify the underlying law. Id. It differentiated Klein as follows:

Lacking authority to impair the pardon power of the Executive, Congress could not direct a court to be instrumental to that end. In other words, the statute in Klein infringed the judicial power, not because it left too little for courts to do, but because it attempted to direct the result without altering the legal standards governing the effect of a pardon—standards Congress was powerless to prescribe.

Id. (citation modified).

Bank Markazi does not confer on Congress the power to use remedies or other procedural rules to effectuate a rule of decision that it could not otherwise impose. It stands for the simple proposition that Congress can change its own laws in narrow ways, and those changes may affect the outcome of cases adjudicated under that legislation. AEDPA represents a much more significant assertion of power than the law at issue in Bank Markazi. There, Congress was attempting to modify its own laws, which it has every right to do, to ensure that the adjudication of those laws will reach the intended outcome. In AEDPA, however, Congress attempted to modify the legal standards for the interpretation of the Constitution—standards which, in the words of Bank Markazi, it was “powerless to prescribe”—through improper remedies provisions. Id.

The Supreme Court, moreover, has explicitly reaffirmed the key holding of Klein, as described in section II.B, within the last decade. In Patchak v. Zinke, the Secretary of the Interior took federal lands into trust for the Match-E-Be-Nash-She-Wish Band of Pottawatomi Indians pursuant to the Indian Reorganization Act. 583 U.S. 244, 246-47, 138 S.Ct. 897, 200 L.Ed.2d 92 (2018). Patchak sued, claiming that the Indian Reorganization Act did not permit the land to be taken into trust, and while that suit was pending, Congress passed the Gun Lake Trust Land Reaffirmation Act. Id. at 248, 138 S.Ct. 897. The Gun Lake Act provided that the Secretary of the Interior had the right to take the land into trust, made that provision binding on all pending suits, and stripped courts of jurisdiction to hear any challenge relating to that land. Id.

The Supreme Court again held, as it had in Bank Markazi that “Congress violates Article III when it compels findings or results under old law. But Congress does not violate Article III when it changes the law.” Id. at 250, 138 S.Ct. 897 (citation modified). The Court noted that changing the law to strip jurisdiction over an entire class of cases regardless of the legal merits of those cases was well within Congress's power, as was changing the law so that the Secretary's actions were legal. Id. at 257, 138 S.Ct. 897. It reasoned that the key distinction between the Gun Lake Act and the law in Klein was that the jurisdiction stripping in the Gun Lake Act removed the court's power in all cases in that subject matter rather than simply cases where the court was inclined to interpret the law differently than Congress. Id. The Court explained:

[T]he statute in Klein infringed the judicial power, not because it left too little for courts to do, but because it attempted to direct the result without altering the legal standards governing the effect of a pardon—standards Congress was powerless to prescribe. Congress had no authority to declare that pardons are not evidence of loyalty, so it could not achieve the same result by stripping jurisdiction whenever claimants cited pardons as evidence of loyalty. Nor could Congress confer jurisdiction to a federal court but then strip jurisdiction from that same court once the court concluded that a pardoned claimant should prevail under the statute.

Patchak's attempts to compare § 2(b) to the statute in Klein are unpersuasive. Section 2(b) does not attempt to exercise a power that the Constitution vests in another branch. And unlike the selective jurisdiction-stripping statute in Klein, § 2(b) strips jurisdiction over every suit relating to the Bradley Property. Indeed, Klein itself explained that statutes that do “nothing more” than strip jurisdiction over “a particular class of cases” are constitutional. That is precisely what § 2(b) does.

Id. In other words, Congress cannot have it both ways. It can strip courts of jurisdiction to hear certain types of cases, but when Congress does so, it must do so for all cases in the relevant class. Congress cannot strip courts of jurisdiction only in cases where courts are going to do something Congress does not like while conferring courts with jurisdiction to do things Congress does like.

Far from establishing that Klein was cabined to its facts, modern doctrine confirms that the principles in Klein are alive and well. Congress can change the law that courts must interpret, and when it does so, there may be implications for pending cases, implications to which we must give effect. Bank Markazi, 578 U.S. at 228, 136 S.Ct. 1310. Congress can even do so in response to highly particularized situations. Id. at 226-228, 136 S.Ct. 1310. But where Congress cannot change the underlying law—as in constitutional cases—it cannot pass new laws forcing courts to interpret the old law in a particular way. Id. at 228, 136 S.Ct. 1310. And although Congress can strip jurisdiction over a whole class of cases regardless of outcome, Congress cannot use jurisdiction to effectuate its preferred interpretation. See Patchak, 583 U.S. at 257, 138 S.Ct. 897.

This leads directly back to the conclusion set out in Section II.B. Because Congress cannot unilaterally change the Constitution, Congress cannot prescribe rules of decision under the Constitution. Bank Markazi, 578 U.S. at 228, 136 S.Ct. 1310. Though Congress can strip courts of jurisdiction it cannot use jurisdictional rules to change things only when a court would reach a certain outcome. Patchak, 583 U.S. at 257, 138 S.Ct. 897. Reynoldsville Casket Co. teaches that these same restrictions apply to Congress's attempted use of remedies, 514 U.S. at 753-54, 115 S.Ct. 1745, and AEDPA runs afoul of those restrictions.

AEDPA deference is a wolf in sheep's clothing. It may be dressed up as a simple remedies provision, but in substance, AEDPA contains Congress's attempt to use remedies to do what it cannot do on its own: force Article III courts to enforce and give credence to unconstitutional actions by state courts that deprive prisoners of their constitutional rights. It disrupts the delicate balance of power between state courts and Article III federal courts established by the Founders in 1787. Congress asks that we look only to the sheep. Over a century of Supreme Court caselaw demands that we acknowledge the wolf. AEDPA is unconstitutional.

III. CONSTITUTIONAL CLAIMS

I turn now to a review of the constitutional issues inherent in Sanders's case. On each claim Sanders must prove both constitutional error and prejudice to be successful. Brecht v. Abrahamson, 507 U.S. 619, 622-23, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993). I begin by setting out two constitutional errors in Sanders’ trial 3 and then address the prejudice flowing from the two errors. For the reasons discussed above, I review all matters of law de novo—the proper analysis given the above conclusions about the constitutionality of AEDPA deference.

A. Constitutional Error

1. The Insanity Instruction

As the majority opinion notes, the Due Process Clause of the Fourteenth Amendment does not mandate that states apply a particular burden of proof for the insanity defense. Leland v. Oregon, 343 U.S. 790, 798-99, 72 S.Ct. 1002, 96 L.Ed. 1302 (1952). Some states require that the state prove, beyond a reasonable doubt, that the defendant was not insane. Id. Others require that the defendant prove that he was insane but disagree as to the quantum of proof required. Id. This variation is permissible and is not grounds for habeas relief. Id. Nor may federal courts second guess state court interpretation of state statutes regarding the burden of proof in habeas proceedings. Engle v. Isaac, 456 U.S. 107, 120, 102 S.Ct. 1558, 71 L.Ed.2d 783 (1982). The states get to decide what the burden of proof is, both as a legislative matter and as a matter of statutory interpretation.

Once a state has established the decisional rules by which a jury may convict a defendant, however, it is a violation of the Due Process Clause for the state to give an erroneous jury instruction that sets forth different decisional rules. Hicks v. Oklahoma, 447 U.S. 343, 346, 100 S.Ct. 2227, 65 L.Ed.2d 175 (1980). As the court explained in Hicks:

Where, however, a State has provided for the imposition of criminal punishment in the discretion of the trial jury, it is not correct to say that the defendant's interest in the exercise of that discretion is merely a matter of state procedural law. The defendant in such a case has a substantial and legitimate expectation that he will be deprived of his liberty only to the extent determined by the jury in the exercise of its statutory discretion, ․ and that liberty interest is one that the Fourteenth Amendment preserves against arbitrary deprivation by the State.

Id. (citations omitted). As a result, habeas relief may be granted where jury instructions erroneously define the burden of proof. Id.

Kentucky allows a defendant to prove he is not guilty by reason of insanity by a preponderance of the evidence. Gall v. Commonwealth, 607 S.W.2d 97, 110 (Ky. 1980) (describing the preponderance standard as correct although admonishing that other language is better for describing the standard to a jury); see also Ball v. Commonwealth, 81 Ky. 662, 664-65 (Ky. 1884) (approving the preponderance standard for the insanity defense). Neither party disputes in this case that insanity must be proved only by a preponderance of the evidence in Kentucky. It would, therefore, be a violation of Sanders's due process rights, cognizable on habeas review, for the Kentucky court to instruct the jury to apply any other standard of review.

In cases where the claim is that the instruction was ambiguous and subject to an erroneous construction, the question is whether there is a “reasonable likelihood” that the jury has applied the instruction in an unconstitutional way (i.e., a way that applies an erroneous burden of proof). Waddington v. Sarausad, 555 U.S. 179, 190-91, 129 S.Ct. 823, 172 L.Ed.2d 532 (2009) (quoting Estelle v. McGuire, 502 U.S. 62, 72, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991)). In making this determination, “a single instruction to a jury may not be judged in artificial isolation, but must be viewed in the context of the overall charge.” Cupp v. Naughten, 414 U.S. 141, 146-47, 94 S.Ct. 396, 38 L.Ed.2d 368 (1973). Further, in coming to an understanding of whether a jury might be misled, the court must keep in mind that “[j]urors do not sit in solitary isolation booths parsing instructions for subtle shades of meaning in the same way that lawyers might․ [C]ommonsense understanding of the instructions in the light of all that has taken place at the trial likely [is] to prevail over technical hairsplitting.” Boyde v. California, 494 U.S. 370, 380-81, 110 S.Ct. 1190, 108 L.Ed.2d 316 (1990).

In this case, immediately before the challenged instruction, the trial court had repeated the following instruction four times: that the jury should take an action “if, and only if you believe from the evidence, beyond a reasonable doubt” some fact or facts. R. 165-5, Jury Instr., PageID 2830-2833. After the challenged instruction there was an instruction “you shall find the defendant in this case not guilty unless you are satisfied from the evidence alone, and beyond a reasonable doubt, that he is guilty.” Id. at PageID 2837. In other words, there was variation between the different instructions, but they all associated the words “from the evidence” with the words “beyond a reasonable doubt.” The challenged instruction was near identical to that in the other instructions. It instructed the jury to find Sanders not guilty on the ground of insanity “if you believe from the evidence that ․ he was of unsound mind.” R. 166-4, PageID 3916-17. The instructions did not clarify that this was a lower burden or a different burden.

In Spisak v. Mitchell, our court considered a case where the jury instructions explicitly required unanimity for the jury to find aggravating circumstances but were silent about the rule that unanimity is not required to find mitigating circumstances. 465 F.3d 684, 710-11 (6th Cir. 2006), rev'd on other grounds 552 U.S. 945, 128 S.Ct. 373, 169 L.Ed.2d 257 (2007). The language used to describe the mitigating circumstances and the procedures the jury was instructed to use to find that the mitigating circumstances outweighed the aggravating circumstances mirrored the language and procedures used for aggravation instructions with the above exception. Id. We held that, under those circumstances, because “the jury was never told, either expressly or impliedly, that individual jurors may consider mitigating factors in the weighing process regardless of the lack of agreement with other jurors as to the presence of that factor,” there was a reasonable likelihood that the jury applied a unanimity standard to their analysis of mitigating circumstances. Id. (granting habeas relief).

A similar issue arises here. The jury had been repeatedly instructed that, to find something “from the evidence,” they needed to find it “beyond a reasonable doubt.” It was never clarified that “beyond a reasonable doubt” was an additional requirement for those instructions, rather than one associated with finding something “from the evidence.” The jury was then told it needed to find insanity “from the evidence”—the same language used in the beyond a reasonable doubt instruction. This mirroring is enough to raise a reasonable likelihood that the jury would apply the “beyond a reasonable doubt” standard to insanity too, even if it was not directly instructed to, just as the mirroring of the aggravation instructions could cause the jury to apply an unstated unanimity requirement in Spisak. 4

The Warden argues, quoting the district court, that

A jury could understand Kentucky's formulation as distinguishing between “from the evidence” and “from the evidence beyond a reasonable doubt.” Jurors can reasonably be expected to identify the significance of the explicit absence of “beyond a reasonable doubt” following “from the evidence” in the instructions on insanity. See, e.g., Gall, 607 S.W.2d at 110 (“There is no reason to suppose or suspect that the jurors were ignorant bumpkins.”). In each reference to the state's burden, the trial judge specifically included “beyond a reasonable doubt,” but omitted that phrase when discussing the burden for Sanders’ insanity defense.

Appellee's Br. 30. This argument fails for several reasons. First, if it was the case that courts could reasonably presume under Boyde that jurors would reason from the absence of a few words that different standards applied to identical procedures described similarly, Spisak could not have been decided as it was. After all, in Spisak, the jury was explicitly told it must be unanimous as to aggravation and was not told so about mitigation. Id. at 709-10. Yet, in that case, this court did not assume that the jurors would reason from absence. Id. at 711.

Second, applicable to both this case and Spisak, it would be irresponsible to presume that jurors parsed the instructions for subtle shades of meaning, identifying the significance of the absence of just a few words from one instruction. Boyde, 494 U.S. at 380-81, 110 S.Ct. 1190. Instead, this court must presume that the jurors used the common sense understanding of the instructions. Consider an example: I tell my clerk on Monday to “cite check a document, being careful to check all case citations as well as the record citations” then use that exact language to direct her on Tuesday and on Wednesday, but, on Thursday, I simply say conduct a “cite check.” The commonsense assumption is not that this one time she does not need to check the record cites—it is that I used a shorthand for the same instruction I had given three times before. Similarly, in this case, one reasonable interpretation of the instructions that the jury must “believe [the elements] from the evidence, beyond a reasonable doubt” is that assessing the evidence under a “beyond a reasonable doubt” standard is part and parcel of “believ[ing] from the evidence.” Under that understanding, the next time the judge instructed the jury that it must believe something “from the evidence,” the correct understanding of that instruction would be that the jury must make such a finding “beyond a reasonable doubt.”

Third, even if one were to assume that the jurors were being closely attentive to small changes in language, the instructions given in this case might stymie them. The final instruction on burdens used language to describe the State's burden that was different from all the other preceding instructions. It instructed that the jury must be “satisfied from the evidence alone, and beyond a reasonable doubt” as opposed to simply “believ[ing] from the evidence, beyond a reasonable doubt.” R. 166-4, PageID 3918. A close parse of this distinction, paying attention to omitted and changed words, might suggest that this was a different standard. For instance, being satisfied from the evidence “alone” as opposed to simply believing “from the evidence” suggests that, in the former case, the evidence must be considered in isolation, whereas in the latter case, one can consider the evidence and other things. But, confusingly, these two pieces of language were intended to describe the same standard—the State's burden of proof. If similar but not identical language with subtly different meanings can be used to describe the same standard in one instance, it is not clear how the jury was to parse out that the lack of identical language in the insanity instruction imposed an entirely new standard that had never been explained to them.

To be sure, the Supreme Court has emphasized that, to determine whether a given instruction was impermissibly ambiguous, courts should consider not only the whole jury charge, but also the full trial record. Estelle v. McGuire, 502 U.S. 62, 72, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991). Here, there was nothing in the broader instructions that would have mitigated the ambiguity of the insanity instruction. Sanders's counsel did nothing to clear up the confusion because he did not discuss the burden of proof in his closing argument at all. The State did not clear things up either—instead, it arguably made things worse. The State asserted in its closing argument that Sanders must “prove to your satisfaction that he was mentally ill.” R. 166-4, Trial Tr., PageID 3931. This does not specify a standard of proof. Nor did the State specify the standard of proof at any other time. Further, in explaining that the State had met its burden, the prosecutor asserted, “I submit to you that we have done that. We have gone even beyond that. We have shifted the burden. The burden lies on the defendant to prove his insanity.” Id. This suggests that the State's burden has been shifted to the defendant, and not that the defendant has a new, different type of burden. It reinforces the idea created by the jury instructions that the defendant's burden was the same as the State's: beyond a reasonable doubt. Therefore, the use of these jury instructions was constitutional error.

2. Failure to Retain and Prepare Experts

Sanders alleges that his counsel was deficient because he failed to timely retain and adequately prepare Dr. Cooke. This claim turns on whether counsel's failure to timely prepare an expert and to provide that expert with the necessary resources to conduct a proper analysis fell below an objective standard of reasonableness. Richey v. Mitchell, 395 F.3d 660, 683-84 (6th Cir. 2005), rev'd on other grounds Bradshaw v. Richey, 546 U.S. 74, 126 S.Ct. 602, 163 L.Ed.2d 407 (2005).

The Supreme Court in Ake v. Oklahoma, 470 U.S. 68, 105 S.Ct. 1087, 84 L.Ed.2d 53 (1985) set the threshold for what services must be procured from a psychiatrist for a defendant pleading insanity. Ake was decided in the context of an indigent defendant who pled insanity. Id. at 72, 105 S.Ct. 1087. His counsel requested that the court appoint a psychiatrist to perform an examination on behalf of the defendant or provide funds so that the defense could arrange one. Id. The trial court denied counsel's request. The Supreme Court held that, “when the State has made the defendant's mental condition relevant to his criminal culpability and to the punishment he might suffer, the assistance of a psychiatrist may well be crucial to the defendant's ability to marshal his defense.” Id. at 80, 105 S.Ct. 1087. The court reasoned that:

[W]ithout the assistance of a psychiatrist to conduct a professional examination on issues relevant to the defense, to help determine whether the insanity defense is viable, to present testimony, and to assist in preparing the cross-examination of a State's psychiatric witnesses, the risk of an inaccurate resolution of sanity issues is extremely high. With such assistance, the defendant is fairly able to present at least enough information to the jury, in a meaningful manner, as to permit it to make a sensible determination.

***

We therefore hold that when a defendant demonstrates to the trial judge that his sanity at the time of the offense is to be a significant factor at trial, the State must, at a minimum, assure the defendant access to a competent psychiatrist who will conduct an appropriate examination and assist in evaluation, preparation, and presentation of the defense.

Id. at 82-83, 105 S.Ct. 1087. The court reasoned that the requirement that indigent defendants have access to a competent psychiatrist to conduct an appropriate examination and assist in evaluation, preparation, and presentation was part and parcel of ensuring that the defendant received his right to due process under the Fourteenth Amendment, on par with the right to effective assistance of counsel. Id. at 76, 105 S.Ct. 1087.

In McWilliams v. Dunn, the Supreme Court applied the standard set forth in Ake and held that a defendant has not received an adequate defense if his psychiatrist, while competent and qualified, is given so little time to prepare that he cannot effectively assist counsel in reviewing information, preparing for trial, and effectively presenting a case. 582 U.S. 183, 199, 137 S.Ct. 1790, 198 L.Ed.2d 341 (2017). Further, the mental health expert must be sufficiently independent from the prosecution to effectively assist in those roles. Id. at 198, 137 S.Ct. 1790. Thus, a court ordered evaluation by state psychiatrists was not sufficient to replace a psychiatrist who would specifically assist the defense. Id.

Sanders was not indigent, so there was no requirement that the State provide him with funds for a psychiatrist. But it is hardly reasonable to say that there is a threshold of representation so “crucial to the defendant's ability to marshal his defense” that the State must provide it to indigent defendants because otherwise “the risk of an inaccurate resolution of sanity issues is extremely high,” but counsel is not ineffective when they fail for no strategic reason to ensure its provision. Ake, 470 U.S. at 80-82, 105 S.Ct. 1087; accord Skaggs v. Parker, 235 F.3d 261, 272-73 (6th Cir. 2000) (noting that counsel was ineffective because his incompetence led to a failure to present an expert who was even marginally competent under Ake). “The right to counsel prevents the States from conducting trials at which persons who face incarceration must defend themselves without adequate legal assistance,” and a defendant's rights are violated where counsel fails to provide such assistance. Cuyler v. Sullivan, 446 U.S. 335, 344, 100 S.Ct. 1708, 64 L.Ed.2d 333 (1980). A counsel's failure to ensure that a defendant receives even the baseline assistance required to satisfy the Fourteenth Amendment for indigent defendants is ineffective assistance.

The Ninth Circuit recently considered an analogous case, Rogers v. Dzurenda, 25 F.4th 1171 (9th Cir. 2022). Rogers's counsel pursued an insanity defense, but she did not begin preparing the only expert who could testify to insanity until a month before trial, leaving the expert unable to prepare an opinion. Id. at 1184-85. She did not provide that expert with all the psychiatric information available to her, nor did she provide him with a complete investigation of Rogers's background and childhood history. Id. at 1185-86. Counsel gave the expert the opportunity to conduct only a “brief evaluation” of the defendant. Id. at 1178. She failed to adequately discuss potential lines of cross examination with the expert and prepare potential responses to the testimony of State witnesses. Id. at 1185-86. The court concluded that Rogers's counsel had been ineffective for failing to give the expert adequate time, sufficient information to form a supported version of his conclusion, and enough preparation to withstand cross examination. Id. at 1184-86.

A more extreme version of the same facts occurred here. Sanders's counsel pursued an insanity defense, but he did not begin preparing Dr Cooke, the only expert who could testify to insanity, until a few days before trial—far less than the month in Rogers—leaving him unable to prepare a report. He did not provide Dr. Cooke with all the information available to him, or with a complete investigation of Sanders's background and childhood (having never done one). Counsel allowed Dr. Cooke time to do only a cursory evaluation of a few hours the day before jury selection. He failed to prepare Dr. Cooke to testify sufficiently such that the doctor would be able to adequately remember Sanders's name or even respond to questions about the differences between Cooke's conclusions and the KCPC Report. For the same reasons stated in Rogers, that was deficient performance.

B. Prejudice

Even outside the bounds of AEDPA, we may only grant relief in a habeas case if the constitutional errors present in a defendant's trial “had [a] substantial and injurious effect or influence in determining the jury's verdict.” Brecht, 507 U.S. at 638, 113 S.Ct. 1710; see also Miller v. Genovese, 994 F.3d 734, 744 (6th Cir. 2021) (“We find prejudice when we have ‘grave doubt about whether a trial error of federal law had substantial and injurious effect.’ ”).

Much depended in this trial on the expert, Dr. Cooke. The difference between the presence and the absence of competent expert testimony from one party can materially alter the outcome of a trial. Richey, 395 F.3d at 687-88. Where defense counsel has failed to present an expert or has presented only an incompetent expert as to a question crucial to the defense, federal courts have found such substantial prejudice that a state court decision to the contrary was unreasonable. Id. The necessity of competent expert testimony in an insanity defense is precisely the reason courts are required to ensure that indigent defendants are provided with it: “the testimony of psychiatrists can be crucial and a virtual necessity if an insanity plea is to have any chance of success.” Ake, 470 U.S. at 82, 105 S.Ct. 1087 (quotations omitted). The Supreme Court has recognized that “without the assistance of a psychiatrist to conduct a professional examination on issues relevant to the defense, to help determine whether the insanity defense is viable, to present testimony, and to assist in preparing the cross-examination of a State's psychiatric witnesses, the risk of an inaccurate resolution of sanity issues is extremely high.” Id. Without such effective assistance, “a defense may be devastated.” Id. at 83, 105 S.Ct. 1087.

One of two things must be true: (1) the jury found Dr. Cooke lacking in credibility, or (2) the jury found Dr. Cooke credible. Begin with the first (and more likely) scenario: Dr. Cooke was not credible. If Dr. Cooke was not credible, Sanders was functionally putting on a case with no expert testimony. Even worse, there were only two witnesses to Sanders's insanity defense: Dr. Cooke and Sanders himself. Sanders, as the defendant, was not likely to be taken seriously, and, in any case, while he could describe his symptoms, he was unable to explain their medical significance. So, Sanders was putting on an insanity case without a single credible witness who could explain Sanders's putative insanity to the jury: precisely the “devastating” situation envisioned by Ake. Id.

As discussed above, both our caselaw and the Supreme Court's caselaw squarely recognize that, where a defendant does not receive the benefit of competent psychiatric testimony, the addition to the defendant's case of an expert who can testify directly and credibly as to the defendant's insanity might well make a difference in the verdict in an insanity case. See, e.g., Ake, 470 U.S. at 83, 105 S.Ct. 1087 (“It is in [insanity] cases that a defense may be devastated by the absence of a psychiatric examination and testimony; with such assistance, the defendant might have a reasonable chance of success.”).

This case presents even more reason to believe that the difference between a credible expert and a non-credible expert might have had a “substantial and injurious effect” on the jury's verdict than the generic insanity case. After trial, several experts considered the same evidence Dr. Cooke would have been able to present if properly prepared and concluded that Sanders was insane. R. 165-14, PageID 3333-34; R. 165-15, PageID 3341. This suggests that Dr. Cooke's conclusion was not simply an aberration that the jury would be likely to disregard had it been presented by a well-prepared witness. It was a conclusion, supported by facts, that reasonable people, even people with expertise and the skepticism that accompanies it, could believe. It suggests that it was possible with sufficient preparation, to put on a compelling case for insanity. And yet, Sanders's trial counsel certainly did not, because the sole expert witness to insanity was not credible. Therefore, in the scenario where Dr. Cooke was simply not credible, there is a reasonable probability that his lack of credibility was a significant deciding factor in the jury's conclusion that Sanders was not insane. The fact that Dr. Cooke's lack of credibility was the result of counsel's unprofessional errors raises grave doubt as to whether those unprofessional errors had a substantial and injurious effect or influence on the jury's verdict.

The Warden contends, however, that Cooke was not so lacking in credibility as to render the use of his services prejudicial to Sanders. Taking the Warden at her word, consider the second scenario described above: Sanders was credible. In this scenario, Sanders had a credible expert who concluded in no uncertain terms that Sanders was insane, and yet, the jury found that he had not met his burden of proof of establishing insanity. Thus, we must consider whether the erroneous jury instructions given by the court had a substantial injurious effect in producing that outcome.

“This court is mindful that a habeas petitioner faces an uphill battle in establishing that an erroneous jury instruction is so prejudicial that he or she is entitled to habeas relief.” Barker v. Yukins, 199 F.3d 867, 876 (6th Cir. 1999). The improper instruction must have infected the entire trial. Id. But “[a]lthough the burden is undeniably heavy, of course, this does not mean that a jury instruction may never rise to such proportions.” Id. Barker is instructive. In Barker, the defendant argued that she had attacked the victim in of self-defense to prevent herself from being raped, but the trial court had erroneously refused to instruct the jury that, under Michigan law, Barker was entitled to use deadly force to resist an imminent rape. Id. at 870. We acknowledged that we had no way of knowing for sure whether the jury's conviction was a result of disbelieving the evidence Barker presented that the force she used was necessary to prevent rape or a result of the improper jury instruction. Id. at 873. But we noted that, where there was “sufficient evidence in the record to support a rational conclusion” that Barker had acted in self-defense, “the district court's failure to specifically instruct the jury that she was justified in using deadly force to resist a rape had a substantial and injurious effect or influence in determining the jury's verdict and resulted in actual prejudice.” Id. at 873-74.

A similar situation presents itself here. Sanders presented credible evidence that he was legally insane in the form of credible expert testimony from Dr. Cooke. This is the kind of legally sufficient evidence that gives a defendant a “reasonable chance at success.” Ake, 470 U.S. at 83, 105 S.Ct. 1087. As in Barker, it was the kind of evidence that could lead to the “rational conclusion” that Sanders had met his burden of proof. Barker, 199 F.3d at 874. Yet, as in Barker, the trial judge cut that reasonable chance off at the knees by giving instructions that foreclosed the possibility that a juror, believing by a preponderance of the evidence, but not beyond a reasonable doubt, that Sanders was insane, would find him so. The same result is demanded: there is grave doubt as to whether the erroneous instruction had a substantial and injurious effect on the verdict. Id. at 873.

Thus, either position taken as to Dr. Cooke's credibility leads to the same result: Sanders was prejudiced by the constitutional errors at his trial.

IV. CONSTITUTIONAL CLAIMS UNDER AEDPA

Even if AEDPA deference was constitutional (as the Majority opinion concludes), Sanders’ constitutional arguments would still succeed.

As to the insanity jury instructions, the Kentucky Supreme Court acted contrary to or unreasonably under the Boyde standard when it held that there was no error in the jury instructions. The state court's reasoning was limited to a citation to Gall v. Commonwealth, 607 S.W.2d 97 (Ky. 1980). In that case, the court considered whether the defendant had been prejudiced by an instruction that informed the jury that it must find that the defendant met his burden of proving insanity “by a preponderance of the evidence.” Id. at 110. The court reasoned:

Instruction No. 3 presented the defense of insanity, and was in correct form except for its use of the words, “by a preponderance of the evidence” in lieu of “from the evidence.” Again, this portion of the instruction was given exactly as requested by Gall. Now, as a result, we find his appellate counsel crying out that “preponderance of the evidence” should have been defined. This court disapproves the use of that terminology. “From the evidence” is sufficient in all civil cases, and with the qualifying phrase “beyond a reasonable doubt” it is sufficient in a criminal case. We would not for that reason, however, reverse this judgment. There is no reason to suppose or suspect that the jurors were ignorant bumpkins.

Id. Gall stands for the proposition that instructing the jury that the plaintiff must meet its burden “from the evidence” in a civil case is an acceptable (and even preferable) substitute for “by a preponderance of the evidence” in civil cases and “from the evidence beyond a reasonable doubt” is acceptable in criminal cases. It appears the court in Sanders reasoned that “from the evidence,” being sufficient for civil cases, was also sufficient for criminal cases in which the defendant had a preponderance burden.

That is the wrong legal rule. The question is not merely whether requiring the jury to find insanity “from the evidence” sufficiently explains the preponderance standard in a vacuum (as it would be in a civil case). It is whether in the context of the jury instructions in this case there is a “reasonable likelihood” that a juror could get confused and apply the wrong standard. Boyde, 494 U.S. at 380, 110 S.Ct. 1190 (quoting Estelle, 502 U.S. at 72, 112 S.Ct. 475). The trial court neither identified nor applied this test. Instead, the court relied on reasoning that looks at the phrase “from the evidence” in a vacuum rather than in the context of the overall charge, as required by Cupp. 414 U.S. at 146-47, 94 S.Ct. 396 (noting that a jury instruction may not be viewed “in isolation”). The state court applied the wrong test and thereby acted contrary to clearly established federal law. Williams v. Taylor, 529 U.S. 362, 405, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (O'Connor, J., concurring) (“A state-court decision will certainly be contrary to our clearly established precedent if the state court applies a rule that contradicts the governing law set forth in our cases.”).

Where the state court erred by applying a rule contrary to clearly established federal law or by unreasonably applying clearly established federal law, even under AEDPA, the next step is to consider the merits of the petitioner's constitutional claim de novo. Fulcher v. Motley, 444 F.3d 791, 799 (6th Cir. 2006). And, as set forth in Section III.A.1, Sanders has succeeded in demonstrating that his constitutional rights were violated. Thus, even under the AEDPA standard, the inappropriate jury instructions violated Sanders's due process rights.

Turn then to the failure to retain and prepare experts. The state court seemingly recognized that a counsel's failure to ensure that a defendant receives adequate assistance under Ake could constitute deficient performance. But the state court held that Ake was satisfied, first, because “Sanders was sent on his own motion to KCPC for a six-week psychiatric evaluation by a team of social workers, psychologists, a psychiatrist and a neurologist.” Sanders, 89 S.W.3d at 387. The court reasoned that Ake “does not hold that an indigent defendant has a constitutional right to choose the psychiatrist.” Id. Second, as to Dr. Cooke, the court reasoned “Dr. Cooke indicated that he was willing to testify that in his opinion Sanders satisfied the legal standard for insanity. Thus, Sanders had access to qualified mental health experts to establish his insanity defense.” Id. at 388.

Both of these justifications for counsel's failure to prepare Dr. Cooke are contrary to or unreasonable under Supreme Court precedent. McWilliams demonstrates the impropriety of both. There, the Supreme Court held that it was not merely wrong, but an unreasonable application of Ake to find that a defendant's rights had been satisfied by an evaluation done by individuals who are not “available to the defense and independent from the prosecution.” McWilliams, 582 U.S. at 188, 197-99, 137 S.Ct. 1790 (requiring an expert sufficiently independent that they may assist the defense and finding a violation of rights even though there had been a full evaluation by a court ordered “Lunacy Commission”). Because McWilliams made no new law and merely determined the bounds of “proper application” of Ake, its holdings are applicable to cases like this one that were decided after Ake, even if they were decided before McWilliams and must now be evaluated under AEDPA. Wiggins v. Smith, 539 U.S. 510, 522, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003). In this case, The KCPC evaluation was not an independent evaluation and, thus, to rely upon it in lieu of a competent defense mental health expert sufficiently independent to assist in the preparation of a defense is not reasonable any more than reliance on the Lunacy Commission rendered the state court decision in McWilliams reasonable.

Similarly, in McWilliams, the Court found a violation of clearly established rights even though the defense expert had reached conclusions favorable to the defense because the expert was not given time to adequately assist with preparation and presentation. Id. at 190, 198, 137 S.Ct. 1790. And the Court found that a state court decision—which reasoned that McWilliams's rights under Ake had been satisfied because McWilliams had been provided with a psychiatrist to examine him—was unreasonable because that psychiatrist had not been given time to adequately prepare and assist in the presentation of the defense. Id. at 198-99, 137 S.Ct. 1790. Thus, despite the state court's claim in this case, the fact that Dr. Cooke formed a conclusion favorable to Sanders is not the end of the inquiry. The question is whether Dr. Cooke was given sufficient time to adequately play the role that a defense mental health expert should play. As in McWilliams, the state's assumption in this case—that an expert who was unable to fully assist the defense in the preparation and presentation of a case is adequate under Ake merely because he provided a finding of insanity—rendered the state decision on the matter unreasonable. Thus, even under AEDPA, Sanders meets the high bar to demonstrate that the state court's finding of a lack of prejudice was contrary to or unreasonable under clearly established federal law.

Finally, consider the prejudice flowing from each error. The state never conducted a harmlessness analysis as to Sanders’ jury instruction claim. Thus, even under AEDPA, the prejudice flowing from the jury instruction claim would be analyzed under the Brecht test applied above rather than the § 2254(d) standard and the analysis and the outcome would be identical to that set forth in Section III.B: Sanders suffered prejudice from the constitutional error. Fry v. Pliler, 551 U.S. 112, 120-21, 127 S.Ct. 2321, 168 L.Ed.2d 16 (2007). The state court reached prejudice for Sanders's claim of ineffective assistance of counsel. Sanders, 89 S.W.3d at 388. Therefore, if AEDPA were constitutional, we would review the prejudice prong of that claim with AEDPA deference.

But, once again, the state court did not apply the correct test. It reasoned:

The jury rejected the opinion offered by Dr. Cooke and there is no reason to believe they would have accepted a similar opinion simply because it came from a different defense expert. Dr. Cooke indicated that he was willing to testify that in his opinion Sanders satisfied the legal standard for insanity.

Id. In other words, because Dr. Cooke reached the conclusion that Sanders wanted, there is no reason to believe a more prepared expert could have done anything more. This is unreasonable under federal law. At the time of the Kentucky Supreme Court opinion, there was already precedent that credibility matters, and that a change in the credibility of a witness may change the outcome of a trial. See, e.g., Strickler v. Greene, 527 U.S. 263, 289, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999) (holding that evidence which reduces the credibility of a state witness may change the outcome of a trial); Skipper v. South Carolina, 476 U.S. 1, 8-9, 106 S.Ct. 1669, 90 L.Ed.2d 1 (1986) (holding that testimony from a credible witness may change the outcome of a trial even where it asserts the same base facts as the testimony given by a witness who was inherently not credible). The state court disregarded its duty to consider whether counsel's errors robbed Sanders, not of an expert to testify to insanity, but of a credible expert capable of rendering the services necessary to the defense. Because the state court acted unreasonably under this precedent, Sanders's claim succeeds even under AEDPA.

V. CONCLUSION

For the foregoing reasons, I would reverse the district court's judgments and grant the writ of habeas corpus. I, therefore, respectfully dissent.

FOOTNOTES

1.   These facts, which were presented to the jury before it recommended a sentence of death, are undisputed.

2.   The next day, Moffit also wrote a notarized letter reiterating his concerns and describing Sanders as “possibly of the dissociative [d]isorder type.” DE 167-1, Moffit Ltr., Page ID 4020.

3.   For example, Lindh v. Murphy, 96 F.3d 856, 868–71 (7th Cir. 1996), rev'd on other grounds by Lindh v. Murphy, 521 U.S. 320, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997); Green v. French, 143 F.3d 865, 874–75 (4th Cir. 1998); Crater v. Galaza, 491 F.3d 1119, 1121–29 (9th Cir. 2007); Evans v. Thompson, 518 F.3d 1, 12 (1st Cir. 2008); Cobb v. Thaler, 682 F.3d 364, 376–77 (5th Cir. 2012). The Tenth Circuit has also rejected the claim that AEDPA is unconstitutional in two unpublished opinions. See Bonomelli v. Dinwiddie, 399 F. App'x 384, 387 (10th Cir. 2010); Olona v. Williams, 13 F. App'x 745, 747 (10th Cir. 2001). We recently held that AEDPA deference was constitutional, albeit in an unpublished decision with minimal briefing. Miles v. Floyd, No. 24-1096, 2025 WL 902800 (6th Cir. Mar. 25, 2025).

4.   For example, Bowling v. Parker, 882 F. Supp. 2d 891, 899 (E.D. Ky. 2012); Parrish v. Simpson, No. 3:09-CV-254-H, 2010 WL 750204, at *2 (W.D. Ky. Feb. 26, 2010); Byrd v. Trombley, 580 F. Supp. 2d 542, 551 (E.D. Mich. 2008); Dennis v. Mitchell, 68 F. Supp. 2d 863, 875 (N.D. Ohio 1999), aff'd on other grounds, Dennis v. Mitchell, 354 F.3d 511 (6th Cir. 2003); Norris v. Warden, NCI, No. 2:08-cv-732, 2010 WL 883847, at *1–2 (S.D. Ohio Mar. 9, 2010). And further still are mountains of caselaw in the district courts of the other circuits that have not yet ruled on this question, all adhering to the same analysis—too many to list in full. See, e.g., McCoy v. Smith, No. 17-2162, 2018 WL 3304343, at *4 (E.D. Pa. July 5, 2018); Atkins v. Pringle, No. 3:22-cv-110, 2023 WL 11996936, at *9 (D.N.D. Mar. 14, 2023); Nelson v. Williams, No. 20-cv-00757-CMA, 2021 WL 7161830, at *35 (D. Colo. Dec. 14, 2021); Ingram v. Stewart, No. 1:17-cv-01464-LSC, 2021 WL 1208867, at *7 (N.D. Ala. Mar. 31, 2021).

5.   In upholding AEDPA, the Seventh Circuit has said the following: “How much leeway does the “unreasonable application” language create? None on questions of interpretation. It does not authorize or permit state courts to deviate from the Constitution. Federal courts acting within their jurisdiction are always entitled to interpret the law independently. Section 2254(d)(1) as we read it does no more than regulate relief. It tells federal courts: Hands off, unless the judgment in place is based on an error grave enough to be called ‘unreasonable.’ Historical practice [ ] likewise counseled restraint in use of the writ. Other rules limiting the remedy abound.” Lindh, 96 F.3d at 870.

6.   And we reiterate that our analysis is solely confined within habeas constituting collateral attacks on state court judgments. Our opinion does not speak to the Supreme Court's direct review of state supreme court decisions on federal constitutional law. See Martin v. Hunter's Lessee, 14 U.S. 304, 323, 1 Wheat. 304, 4 L.Ed. 97 (1816).

7.   Although a defendant can rebut that presumption subject to certain requirements, Sanders does not attempt to do so here. See Johnson, 568 U.S. at 301, 133 S.Ct. 1088.

8.   Note, however, that Sanders is not an indigent defendant.

9.   The procedural rule refers to Ky. Rule of Crim. Pro. 11.42(2), which states that “[t]he motion shall be signed and verified by the movant and shall state specifically the grounds on which the sentence is being challenged and the facts on which the movant relies in support of such grounds. Failure to comply with this section shall warrant a summary dismissal of the motion.”

10.   These affidavits are from people who did not testify at Sanders’ penalty phase. In order, they are from his aunt, his sister, his younger brother, his older brother, his neighbor, and his maternal cousin.

11.   This is different from a procedural default, where new evidence can be considered subject to a few exceptions. See Shinn, 596 U.S. at 371, 142 S.Ct. 1718.

12.   Part II of Justice Stevens's opinion in Williams rejects that idea stating that a decision can be unreasonable even if jurists disagree. See 529 U.S. at 377, 120 S.Ct. 1495. But that part of the decision was not joined by Justices Kennedy and O'Connor, so it constituted only a plurality. See id. at 367, 120 S.Ct. 1495. Instead, Justice O'Connor’s concurrence became the majority opinion on the unreasonable interpretation prong. Id. at 399, 120 S.Ct. 1495. Justice O'Connor noted the difficulty in defining that term and narrowly held that a decision can be incorrect but not unreasonable. Id. at 411, 120 S.Ct. 1495.

13.   Inadequate preparation, by itself, does not show prejudice. See Wise v. Bowersox, 136 F.3d 1197, 1207 (8th Cir. 1998).

1.   Although Loper Bright’s explanation of the role of federal courts is instructive, I agree with the majority opinion that the holding of Loper Bright has no bearing on the constitutionality of AEDPA. Loper Bright did deal with deference, but in a different context. The Supreme Court's decision in Chevron, U.S.A., Inc. v Natural Resources Defenses Council, Inc. read the Administrative Procedure Act (APA) to permit Article III courts to defer to reasonable agency interpretations of statutes even where those statutes did not expressly delegate authority to the agency. 467 U.S. 837, 842-43, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). In other words, Chevron read the APA as creating a rule of decision for Article III courts interpreting other statutes. Loper Bright overturned Chevron on statutory grounds, holding that the APA creates no such rule of decision. 603 U.S. at 412, 144 S.Ct. 2244. That holding is not applicable to habeas cases, because habeas cases are not decided under the APA. Additionally, Justices Thomas and Gorsuch both wrote concurrences in which they expressed the belief that Chevron style deference—i.e., a rule of decision by which Article III courts must defer to a non-Article III entity as to the reading of a document that did not expressly delegate interpretive authority to the entity—would be unconstitutional even if Congress had actually dictated it in the APA. Id. at 413-15, 144 S.Ct. 2244 (Thomas, J., concurring); id. at 433, 144 S.Ct. 2244 (Gorsuch, J., concurring). But the Supreme Court majority did not reach a conclusion as to this reading of the Constitution.

2.   Nor are jurists quite as unanimous as is suggested. Most circuits that have found the deference provision of AEDPA constitutional had at least one voice raising the alarm of unconstitutionality, and often multiple voices. See, e.g., Evans v. Thompson, 524 F.3d 1 (1st Cir. 2008) (Lipez, J., dissenting from denial of rehearing en banc); Crater v. Galaza, 508 F.3d 1261, 1261 (9th Cir. 2007) (Reinhardt, J., dissenting from denial of rehearing en banc); Irons v. Carey, 505 F.3d 846, 854 (9th Cir 2007) (Noonan, J., concurring); Davis v. Straub, 445 F.3d 908, 908 (6th Cir. 2006) (Martin, J., dissenting from denial of rehearing en banc); Lindh, 96 F.3d at 885 (Ripple, J., dissenting). In circuits that have not decided the issue, there are district courts that have expressed grave reservations. See, e.g., Figueroa v. Walsh, No. 00-CV-1160, 2008 WL 1945350, at *6-8 (E.D.N.Y. May 1, 2008) (denying relief because the habeas petition was a successive motion but noting that AEDPA deference is probably unconstitutional). In the First Circuit, moreover, district courts have expressed concern that, while AEDPA deference has been held constitutional in the circuit, the way it is actually applied may not be. See Woods v. Medeiros, 465 F. Supp. 3d 1, 15-16 (D. Mass. 2020).

3.   Because Sanders is entitled to relief on at least one of the insanity jury instructions or his counsel's failure to retain and prepare an expert witness, I do not reach his remaining claims.

4.   The majority elides the issue by asking whether the jury would notice the difference between the phrase “beyond a reasonable doubt” and “believe from the evidence,” which, the majority argues, are “different standards.” Maj. Op. at ––––. But that overlooks the key problem—the “beyond a reasonable doubt” instructions were phrased as “believe from the evidence, beyond a reasonable doubt,” raising the substantial likelihood that the jury would associate the “believe from the evidence” language with the “beyond a reasonable doubt” standard and fail to recognize the difference between “believe from the evidence” alone and “believe from the evidence, beyond a reasonable doubt.”

JULIA SMITH GIBBONS, Circuit Judge.

GIBBONS, J., delivered the opinion of the court in which SILER, J., concurred. STRANCH, J. (pp. –––– – ––––), delivered a separate dissenting opinion.