In August 2019, just a week before his scheduled execution, Dexter Johnson asked the Fifth Circuit for permission to file a second federal habeas petition. His only claim was based on Atkins v. Virginia, the Supreme Court decision holding that the Eighth Amendment bars the execution of people with intellectual disabilities.[i] Yet Atkins offered Johnson no new legal grounds. It had been decided in 2002, four years before his crime, and was available when he filed his first habeas petition. What had changed since then, though, was the science that determines who qualifies as intellectually disabled. In 2013, the American Psychiatric Association published the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (“DSM-5”), which moved away from the strict IQ cutoffs used to diagnose intellectual disability and recognized that a person scoring above 70 can still be intellectually disabled.[ii] Based on that change, the Fifth Circuit allowed Johnson’s petition to go forward.[iii] Texas appealed, and the Supreme Court recently heard oral arguments on the matter of Guerrero v. Johnson, and is set to issue a ruling soon.[iv]

The question the Supreme Court will answer is a technical one: whether a claim “relies on a new rule of constitutional law . . . that was previously unavailable” when the prisoner could have raised a claim based on that rule in a previous petition.[v] However, a bigger issue sits underneath this analysis. Fifth Circuit Judge, James C. Ho, dissented from the court’s denial to rehear the issue en banc, and argued that treating updated scientific standards as new law, would be “tantamount to delegating the judicial power to interpret the law to scientists.”[vi] Judge Ho presents a significant perspective, as the Fifth Circuit’s interpretation of the statute is certainly difficult to defend. However, it is not scientists who define the contours of the Eighth Amendment. It was the Supreme Court that tied the Atkins standard to clinical science, and that choice is the real source of the problem in Guerrero.

To understand the dispute, one must start with the statute. The Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”),[vii] generally gives a state prisoner only one chance at federal habeas review. Thus, a new claim in a second petition must be dismissed unless it falls into one of two narrow exceptions: (1) the claim relies on a new, retroactive, and previously unavailable rule of constitutional law, or (2) the claim is based on newly discovered facts that would show by clear and convincing evidence that no reasonable factfinder would have found the prisoner guilty.[viii] The second exception does not help Johnson because he is only challenging his death sentence, not whether he committed the crime, which leaves him with the first exception.

The Fifth Circuit’s approach followed its precedent in In Re Cathey, which held that “a claim must have some possibility of merit to be considered available.”[ix] Meaning, if an Atkins claim would have failed under the clinical standards in place when the prisoner filed his first petition, the claim was never really “available”, even though Atkins itself was. Johnson’s case shows how this works. At trial, his own expert estimated his IQ to be between 74 and 88 and thus concluded that he was not intellectually disabled. Yet by 2019, Johnson was ultimately diagnosed with mild intellectual disability by two experts who applied the new DSM-5 standards.[x] Other circuits disagree with this approach. Notably, the Eleventh Circuit rejected Cathey, holding that whether there is a new rule “depends solely on Supreme Court decisions, not on the issuance of a new diagnostic manual.”[xi] The Fourth Circuit similarly read “previously unavailable” to refer to whether the rule existed, but not whether a particular prisoner could win under it.[xii]

Looking at the text alone, Texas has the better argument, because as Texas points out, “previously unavailable” grammatically modifies “rule”, not “claim”.[xiii] So the statute asks whether a rule was available, not whether a claim would have succeeded. Additionally, Cathey’s test blurs the line that Congress drew between the two exceptions.[xiv] If new evidence can make an old rule “unavailable”, then any new fact that turns a losing claim into a winning one could get around the innocence requirement that Congress placed on new-facts claims.

That said, the framing of Judge Ho’s dissent leaves out an important part of the story. A new diagnostic manual alone is not what made Johnson’s claim viable. In Atkins, the Supreme Court left it to the states to develop new ways to enforce its rule.[xv] Later, in Hall v. Florida, the Court held that the Atkins inquiry must be “informed by the medical community’s diagnostic framework”, it struck down Florida’s strict IQ cutoff, and stated that “intellectual disability is a condition, not a number.”[xvi] Then, in Moore v. Texas, the Court rejected Texas’s nonclinical factors set forth in Ex parte Briseño, and held that states cannot disregard current medical standards.[xvii] Only after Moore did the Texas Court of Criminal Appeals hold that the DSM-5 controls intellectual disability claims.[xviii]

In essence, scientists are not taking judicial power for themselves. Instead, it is the Supreme Court who has consistently allowed their influence in. Under the Court’s own cases, the line between who can and cannot be executed now shifts when clinical evaluations are updated. There are good reasons for this approach, one being that it prevents the Eighth Amendment from depending upon outdated stereotypes. That is precisely what the Supreme Court found wrong with the Briseño factors in Moore. However, this also creates a problem that the drafters of AEDPA did not expect. AEDPA assumes that constitutional rules change only when the Supreme Court changes them, and it draws a clear line between new law and new facts. Yet, an Atkins standard that depends partly on outside clinical criteria falls right on that line. As a result, a prisoner like Johnson, who may be constitutionally ineligible for execution under current standards, can fall through the gap. His claim looks factual for purposes of the new-facts exception, but those facts relate to his sentence instead of his guilt. In practice, this issue affects very few people, but for those few prisoners impacted, the outcome is precisely a matter of life and death.[xix]

So, how should the court handle the gap? Ultimately, Guerrero gives the Supreme Court a choice. It can enforce the text of AEDPA by rejecting the Fifth Circuit’s futility exception, which its recent habeas decisions suggest it is likely to do. If it does, however, it should recognize that its own decision to connect the Eighth Amendment to evolving clinical standards is the reason these claims will keep coming up. Reversing without addressing that conflict would leave some prisoners with no court in which to argue that they cannot constitutionally be executed. The better outcome would be for the Court to read the statute as written, while making clear that the DSM-5 did not change the constitutional standard, the Court did. It did so in Hall and Moore, and those are rules the Court can either declare retroactive or choose not to. That approach would ultimately respect both the limits that Congress set in the AEDPA and the Court’s own Eighth Amendment precedents, while also giving lower courts guidance moving forward. Most importantly, it would ensure that when the line between life and death shifts, the Court that drew it is the one to say so.

[i] See Atkins v. Virginia, 536 U.S. 304, 321 (2002) (holding that the Eighth Amendment bars execution of intellectually disabled offenders); In re Johnson, 935 F.3d 284, 287–88, 291–92 (5th Cir. 2019).

[ii] See In re Johnson, 935 F.3d at 292 (“Higher IQ scores no longer bar a diagnosis of an intellectual disability.”); Johnson v. Lumpkin, No. 4:19-cv-3047, slip op. at 6 (S.D. Tex. Oct. 28, 2022), reprinted in Petition for Writ of Certiorari app. at 15a, Guerrero v. Johnson, No. 25-1003 (U.S. Feb. 19, 2026); Am. Psychiatric Ass’n, Diagnostic and Stat. Manual of Mental Disorders 37 (5th ed. 2013).

[iii] In re Johnson, 935 F.3d at 293–94; see also Johnson v. Guerrero, No. 23-70002, 2025 WL 2060781, at *2 (5th Cir. July 23, 2025) (per curiam) (affirming the denial of the State’s motion to dismiss), reh’g en banc denied, 164 F.4th 398 (5th Cir. 2026), cert. granted sub nom. Guerrero v. Johnson, No. 25-1003 (U.S. June 15, 2026).

[iv] See Guerrero v. Johnson, SCOTUSblog, https://www.scotusblog.com/cases/guerrero-v-johnson/ [https://perma.cc/5YAM-MTUN] (last visited Nov. 13, 2026).

[v] 28 U.S.C. § 2244(b)(2)(A); see Petition for Writ of Certiorari, supra note ii, at i.

[vi]  Johnson v. Guerrero, 164 F.4th 398, 400 (5th Cir. 2026) (Ho, J., dissenting from the denial of rehearing en banc).

[vii] Antiterrorism and Effective Death Penalty Act of 1996, Pub. L. No. 104-132, 110 Stat. 1214 (codified as amended in scattered sections of the U.S.C.); see Banister v. Davis, 590 U.S. 504, 509 (2020) (“[A] state prisoner always gets one chance to bring a federal habeas challenge to his conviction.”).

[viii] 28 U.S.C. § 2244(b)(2)(A)–(B).

[ix] In re Cathey, 857 F.3d 221, 232 (5th Cir. 2017) (per curiam).

[x] See Johnson v. Lumpkin, No. 4:19-cv-3047, slip op. at 4–5 & n.2, 8–9 (S.D. Tex. Oct. 28, 2022), reprinted in Petition for Writ of Certiorari, supra note ii, app. at 13a–14a & n.2, 17a–18a.

[xi] In re Bowles, 935 F.3d 1210, 1219 (11th Cir. 2019).

[xii] See Slusser v. Vereen, 36 F.4th 590, 596 (4th Cir. 2022); see also In re Williams, 364 F.3d 235, 239 (4th Cir. 2004).

[xiii] See Petition for Writ of Certiorari, supra note ii, at 26–27 & n.7.

[xiv] See Bowles, 935 F.3d at 1217–18 (rejecting the argument that updated clinical standards support a new-rule claim, as it would confuse Congress’s intent when drafting the two exceptions); see also 28 U.S.C. § 2244(b)(2)(B)(ii) (limiting new-facts claims to those showing that “no reasonable factfinder would have found the applicant guilty of the underlying offense”); Petition for Writ of Certiorari, supra note ii, at 25–26.

[xv] See Atkins, 536 U.S. at 317.

[xvi] Hall v. Florida, 572 U.S. 701, 721, 723 (2014).

[xvii] Moore v. Texas, 581 U.S. 1, 19 (2017); see also Moore v. Texas, 586 U.S. 133, 142 (2019) (per curiam) (reversing again after the state court continued to rely on nonclinical factors on remand).

[xviii] See Ex parte Moore, 548 S.W.3d 552, 560 (Tex. Crim. App. 2018).

[xix] See Justin Wise, Justices Will Consider Dispute Over Successive Habeas Challenges, Bloomberg L. (June 15, 2026, at 9:51 EDT), https://news.bloomberglaw.com/us-law-week/justices-will-consider-dispute-over-successive-habeas-challenges [https://perma.cc/YW69-3NL8] (describing the dispute over when prisoners bring successive habeas petitions raising Atkins claims).