EX PARTE MICHAEL DEAN GONZALES v. << (2026)

DISSENTING OPINION

The ultimate question in this case is simple: whether Michael Dean Gonzales, Applicant, is intellectually disabled under the Eighth Amendment and cannot be executed. That much is easily said. What is difficult is figuring out how to go about answering that question mindful of both the controlling, substantive constitutional question and the procedural difficulties stemming from our review posture. I dissent because I disagree with the process we have used to answer that question. I write separately to explain my concerns.

First, I believe the Court reaches the correct disposition when it “denies” Gonzales relief, given its view of the merits, rather than “dismissing” it because dismissing it will likely prevent a person who has credibly shown he is intellectually disabled under the Eighth Amendment from asserting the claim in the federal system before being executed.

Second, I'm not so sure that the Section 5(a)(3) writ bar applies to intellectual disability claims notwithstanding our earlier holding to that effect in Ex parte Blue, 230 S.W.3d 151 (Tex. Crim. App. 2007). I suspect that the prohibition on the execution of the mentally disabled reflects a substantive limitation on the exercise of state power and is not properly resolved on procedural grounds like the elevated burden of proof the majority deploys here.

Third, I agree that this Court is the “ultimate” factfinder but am troubled by the invocation of that authority without showing any real deference to the findings of the “original” factfinder or developing a plausible mechanism for making what amounts to new and different findings without hearing from or observing any witness, including and especially the expert witnesses who have opined that Gonzales is intellectually disabled. I worry that the Court is either abandoning the review function inherent to our posture or pivoting to act as a plenary finder of fact without the means to find the plausibly contested (or complex) factual questions suggested in our remand order. Skeptics might develop the misunderstanding that either of these things amount to placing an outcome-oriented thumb on the scale. While the determination of whether someone is intellectually disabled under the Eighth Amendment ultimately is a legal question, its answer is dependent on critical factual antecedents.

While I have no doubt that we are free to reject “expert” declarations of the ultimate issue where there is a genuine, material question as to whether a person sentenced to death is “intellectually disabled”; parsing the bases and methodology underlying written opinions is not an acceptable substitute for live observation of testimony under cross-examination or, at a minimum, a record of that testimony and examination preserved for later review.¹

I. BACKGROUND

A jury convicted Gonzales in 1995 of capital murder and sentenced him to death. See Tex. Penal Code § 19.03(a). This Court affirmed his conviction and sentence. Gonzales v. State, No. AP–72,317 (Tex. Crim. App. June 3, 1998) (not designated for publication). We then denied relief on Gonzales’ initial habeas application. Ex parte Gonzales, No. WR–40,541–01 (Tex. Crim. App. Mar. 10, 1999) (not designated for publication). He later moved to stay his execution. His motion was denied without written order in 2001.

In 2002, the United States District Court for the Western District of Texas granted Gonzales a new punishment hearing, and he was again sentenced to death. Gonzales v. Cockrell, No. MO–99–CA–073 (W.D. Tex. Dec. 19, 2002) (not designated for publication), aff'd by Gonzales v. Quarterman, 458 F.3d 384 (5th Cir. 2006). We affirmed that sentence on direct appeal in 2011. Gonzales v. State, 353 S.W.3d 826 (Tex. Crim. App. 2011). When asked whether he wanted counsel appointed for writ application challenging his new death sentence, he refused. Ex parte Gonzales, No. WR–40,541–03 (Tex. Crim. App. Nov. 10, 2010) (not designated for publication). The trial court concluded that his “election not to accept an appointed attorney and proceed pro se was intelligent and voluntary.” It's unclear whether Gonzales was represented by counsel at the May 2009 hearing, but it appears he was not.

Then this Court, after Gonzales failed to file a timely writ application, accepted his waiver and said that any future application would be subject to the Section 5 writ bar in Article 11.071. Id. True to our word, when Gonzales filed another writ application after his federal proceedings were abated to exhaust state remedies, we dismissed his application under Section 5 as an abuse of the writ. Ex parte Gonzales, 463 S.W.3d 508, 509 (Tex. Crim. App. 2015) (per curiam).

The instant writ application followed. Now represented by counsel, he argues among other things that he is intellectually disabled. We remanded the case after concluding that his claim was sufficiently plausible to satisfy Section 5, but we said further development of the record by an “original” factfinder was necessary. Ex parte Gonzales, No. WR–40,541–07, 2022 WL 628505, at *1 (Tex. Crim. App. Mar. 3, 2022) (not designated for publication). While the State initially opposed relief, it changed its position when its own retained expert tendered a report agreeing with Gonzales’ experts that he is intellectually disabled. The trial court found the expert opinions sufficient to preclude the need for live testimony, it entered findings to that effect, and it recommended granting relief. Today, the Court discards those findings and denies Gonzales relief.

II. THE MAJORITY PROPERLY DENIES, RATHER THAN DISMISSES, GONZALES’ INTELLECUTAL DISABILITY CLAIM

By denying Gonzales’ intellectual disability claim on its merits, the Court properly permits at least limited federal review of an issue of immense importance: Whether the Eighth Amendment forbids Gonzales’ execution. I think that this Court should be cautious when dismissing subsequent applications raising intellectual disability under Section 5 when such an important constitutional issue is at stake, a concern we seemingly dismissed in Blue. In Blue, we said “application of Section 5 does not violate the federal constitution just because it might deny a particular applicant review of an allegation of facts that, if true, might impose a fundamental bar to execution.” Blue, 230 S.W.3d at 156. Thus, according to the Blue majority, even if the Eighth Amendment categorically prohibits a person from being executed, that constitutional right must give way to procedural default rules. See id.

III. DISMISSING GONZALES’ SUCCESSIVE APPLICATION WOULD BE PARTICULARLY IRONIC GIVEN THE FACTS HERE.

The majority applies the greatly heightened standard of proof under Article 11.071's successive writ bar because Gonzales’ first pro se application failed to include the argument and accompanying expert opinion testimony on which it relies. But faulting an applicant for not properly raising an intellectual disability claim in a pro se application when he (or any like inmate) has later credibly proven intellectually disability by a preponderance flies in the face of common sense. Indeed, had Gonzales been capable of submitting a complete and persuasive application, we might point to its eloquence as the basis for denying relief. Joseph Heller explored a similar concept in his fictional novel, Catch–22. See Joseph Heller, Catch–22 (1961).

IV. APPLICATION OF THE WRIT BAR TO INTELLECTUAL DISABILITY CLAIMS SHOULD BE REVISITED, PARTICULARLY WHEN AN EARLIER PRO SE FILING IS INVOLVED.

I'm also not convinced that Section 5(a)(3) applies to intellectual disability claims for several reasons, and my discussion begins with our seminal case in this area: Ex parte Blue. See Blue, 230 S.W.3d at 151.

A. Ex parte Blue

Blue argued that he should be able to raise an Atkins claim at any time without regard for abuse-of-writ principles because the Atkins prohibition is absolute and that, alternatively, he could raise his claim in a successive writ application under Section 5(a)(3). Id. at 153; see Atkins v. Virginia, 536 U.S. 304, 317 (2002). Without too much wrangling, the Court knocked down both arguments.

We explained that an Atkins claim could not be raised at any time because, even if the Eighth Amendment's categorical prohibition is a systemic prohibition of the type identified in Marin, Marin is different. It's different we said because it's about appeals, not postconviction writs, and the ability to raise a Category 1 Marin right for the first time on appeal is predicated on having the right to appeal in the first place. Id. at 155; see Evitts v. Lucey, 469 U.S. 387, 401 (1985) (holding that when a state provides an appellate process, the process must comport with due process). We then observed that the legislature has also given Texans the right to file writ applications but did not expound on the point. Blue, 230 S.W.3d at 155–56.

Rather, we concluded Section 5 is the only way an applicant can raise an Atkins claim in a successive habeas application, shoring up our conclusion by explaining that Section 5 is not all that different than federal habeas corpus jurisprudence, which we concluded would also probably bar an applicant from raising an intellectual disability claim in a successive federal writ application.

B. What Ex parte Blue Means

In an initial writ application, an applicant must show by a preponderance of the evidence that Atkins prohibits his execution. See Franklin v. State, 579 S.W.3d 382, 386 (Tex. Crim. App. 2019); see Ex parte Briseno, 135 S.W.3d 1, 12 (Tex. Crim. App. 2004), abrogated by Moore v. Texas, 581 U.S. 1, 17–18 (2017). But since Blue, an applicant in a successive writ application must prove by clear and convincing evidence that no rational juror would have answered any of the special issues in the State's favor in a successive writ application. The legal standard is elevated, to say the least, and serves as the basis for the majority's denial here.

C. Why Ex parte Blue Needs to be Revisited

1. The remedy of habeas corpus should comport with due process.

The Blue majority talked about how a state that provides for appellate review can withhold that right in part or altogether. Blue, 230 S.W.3d at 155. We said the same of habeas writs. Id. at 155–57. The Blue majority did not meaningfully address Evitts, and its possible application here. In Evitts, the United States Supreme Court held that states need not provide an avenue to appeal a criminal conviction or sentence but that, if they do, it must comport with due process. Evitts, 469 U.S. at 400–01.

If Texas provides for appellate review that must comply with due process, and it provides for the habeas corpus review, shouldn't the postconviction process also have to comport with due process? The United States Supreme Court almost answered this question in Finley. Pennsylvania v. Finley, 481 U.S. 551, 558 (1987). It said that, even if Pennsylvania's postconviction writ process, which was at issue in that case, had to adhere to principles of due process, the petitioner “suffered no deprivation ․” Id.

2. Potential Texas Constitutional Problems

The Blue majority also did not grapple with possible conflicts between Section 5 and the Texas Constitution. While the legislature has the right to restrict access to the Great Writ, the Blue majority said the only thing the legislature cannot do when restricting access to it is to suspend it. See Tex. Const. art. I, § 12 (“The writ of Habeas Corpus is a writ of right and shall never be suspended. The Legislature shall enact laws to render the remedy speedy and effectual.”). True enough. And as I see it, there is an argument that the burden on an applicant in this context might be so high as to effectively suspend the writ (or deny due process).² And it is not clear to me whether a procedural default statute that bars some people who are categorically exempt from execution by resort to a nearly insurmountable legal standard effectively violates Texans’ constitutional right to open courts because the courts are effectively closed to them. Id. art. I, § 13 (“All courts shall be open ․”).

I have previously expressed the view that our Texas Constitution, including its open courts and anti-suspension provisions, would prohibit the execution of a person who has produced probative new evidence of substantive innocence proving that a jury would most likely not convict him. See Ex parte Wood, No. WR–45,746–04, 2025 WL 2148862 (Tex. Crim. App. July 30, 2025) (per curiam) (Schenck, P.J., concurring and dissenting). In particular, given the distinct and elevated life interest involved in such cases, I noted my belief that application of our “clear and convincing” proof standard from Elizondo (and the same applicable under Section 5) was inappropriate to the question unique to death cases of whether the punishment of a person shown post-judgment unlikely to be guilty is so obviously disproportionate as to be cruel and unusual given the acute risk of erroneous deprivation of the constitutionally protected “life” interest. Id. at 11–15; see Ex parte Elizondo, 947 S.W.2d 202, 209 (Tex. Crim. App. 1996).

I see no intention in the text of the Section 5 writ bar (or constitutional authority) to foreclose a person who has made such a showing access to this Court, as I do not believe either the legislature or suspension clause would allow it. I suspect the same problem obtains here with our resort to Blue in escalating the burden to show intellectual disability by clear and convincing evidence on account of an earlier submission that did not present the question or adequately present it. This is especially true when the submission was pro se and filed by a person found by the trial court and experts to be intellectually disabled. A habeas applicant is either mentally disabled or not. If he is, Atkins poses a substantive bar to his execution.

3. The Atkins Prohibition Appears Absolute and Might Be Able to Be Raised Any Time

I also find the Blue majority's resolution of the Marin, “systemic prohibition” argument unpersuasive. First, the majority refused to answer whether the Eighth Amendment prohibition on executing intellectually disabled people is a systemic prohibition of the type identified in Marin. Blue, 230 S.W.3d at 154. It probably is. If a category of people cannot be executed under any circumstances, that seems to be a system prohibition of the kind that “is not option.” See Marin v. State, 851 S.W.2d 275, 289 (Tex. Crim. App. 1993). Second, if Marin and systemic prohibitions can be raised for the first time on direct appeal by any party entitled to appeal, why would a postconviction writ be different? Applicants are entitled to file applications for postconviction writs of habeas corpus. I am not aware of a constitutional or statutory provision wholly preventing their filing.

So, to the extent that the Blue Court distinguished Marin and held that an intellectual disability claim made in a successive writ application is subject to Section 5 even if it were a systemic prohibition, I think this Court should reconsider its position. Instead, the Court should ask whether that construction can be supported under our Constitution and whether our past reading of Section 5 should be revisited to permit us to avoid that constitutional question.

D. Postconviction Intellectual Disability Claims Not Presented to a Jury Might Be Exempt from Section 5(a)(3).

It is not at all clear that Section 5 can be read to apply to the intellectual disability question if it is not put before a jury. While a jury answers the special issues, which touch upon the moral question of if someone is deserving of the death penalty, it is free to believe that a person is or is not morally deserving of a sentence of death even if they believe the defendant is intellectually disabled. See Tex. Code Crim. Proc. art. 37.071, § 1(e)(1).

In answering the mitigation question to which intellectual disability may be considered, jurors are not instructed that such a person may not be executed. Instead, they are instructed to “consider mitigating evidence to be evidence that a juror might regard as reducing the defendant's moral blameworthiness.” Id.; id. art. 37.071, § 1(f )(4). A defendant who is intellectually disabled is still “morally” blameworthy as a matter of both Texas and federal law to the point that may be deprived of their liberty.³ Jurors are thus free to answer the mitigation question however they see fit in the absence of any instruction that the intellectual disability question is unique among “mitigating” circumstances in wholly precluding a sentence of death. That legal question, while premised on its own rigorous factual inquiry, is governed by a standard that is never placed before jurors and, to my mind, might exclude the question from default under Section 5.

V. WE SHOULD DEFER TO THE “ORIGINAL” FACTFINDER'S FINDINGS UNLESS NO REASONABLE VIEW OF THE RECORD SUPPORTS THEM.

Turning to this case, I disagree with the Court's analysis. One reason is that the Court seems to be invoking its “ultimate” factfinder authority without any real guiding principles. We recently discussed this authority in Ex parte Reed, 670 S.W.3d 689, 744 (2023), and followed that discussion with twenty-four pages of reasons why the “original” factfinder was wrong and how we know better. See id. at 744–68. Here the reader is subjected to only nine pages. Our discussion of our “ultimate” factfinder authority in Reed started with lip service about deference to the “original” factfinder but ended with a comment that we would review each claim “with a view toward exercising our own judgment.” Id. at 744.

It seems obvious these are mutually exclusive. How can we defer to an “original” factfinder's findings of fact if any view of the record supports them “with a view toward exercising our own judgment” at the same time? Does it turn on how good we think the findings are? When are findings “good enough?” The former is how this Court has treated “original” factfinders and their findings for decades—any reasonable view of the record. The latter is no standard at all and seems to be akin saying “whenever we want.”

Instead, I would follow the course set in Ex parte Harleston, 431 S.W.3d 67, 70–71 (Tex. Crim. App. 2014). In Harleston, this Court observed that the proper approach recognizes that we have the authority to reject a finding of fact, but we are mindful we are conducting a review of the findings of fact of the “original” factfinder on a cold record. See id. at 70. That is why this Court cited Anderson v. Bessemer City, 470 U.S. 564 (1985), one of the United States Supreme Court's famous decisions about appellate courts deferring to lower court credibility findings, especially when the credibility of live testimony might be relevant.

Another problem is that we appear to be engaging in our own factfinding in ways that create confusion over our decision to remand the case, and the contrary outcome we now direct that creates the unfortunate (and I'm sure unintended) appearance of outcome-based willingness to defer on review. The practice smacks of the United States Tax Court “reviewing” factual findings of a special trial judge in Ballard v. C.I.R., 544 U.S. 40, 55–56 (2005). On certiorari, the United States Supreme Court concluded the Tax Court's system of “reviewing” only after it had received a result it approved of was aberrant and unsupported by any written rule. Instead, the Court read the existing rule to respect a separation of the courts’ functions and a review standard that transparently operates like summary judgment or directed verdict reversal where the decision to submit the facts to someone else has been made. Id. Harleston suggests the same answer here. This is why I believe we should defer to findings of fact if any view of the record supports them.

Here, the trial judge did not call or observe the experts, but he did not do so because he considered the intellectual disability answer to be so conclusive as to call for disposition in Gonzales’ favor. That the judge did not hear live testimony does not matter, and the fact that the trial court found the weight of the expert opinion evidence so overwhelming as to require no observation of these witnesses does not mean that we should reject his findings. This is ultimately akin to a disposition on summary judgment or directed verdict that would parallel the Anderson standard we embraced in Harleston. A trial judge may enter summary judgment under the same standards an appellate court would employ in reviewing a verdict. See Armstrong v. City of Dallas, 997 F.2d 62, 66 (5th Cir. 1992). When a trial court concludes, as our convicting court did here, that the evidence of intellectual disability is conclusively established by the weight of expert opinions, and thus renders what amounts to a “summary judgment” in that it pretermits a live hearing to observe the testimony, a rejection of that assessment does not compel a contrary disposition unless there was no plausible prospect to warrant the referral to the trial court in the first place. E.g., Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 256–57 (1986). That the majority spends around eighteen pages of its sixty-seven-page opinion dissecting the various experts’ written submissions hardly suggests the lack of a live question.

At this stage, it is difficult for me to imagine how this Court could make the necessary contrary findings without personally calling and seeing the witnesses the habeas court found so collectively persuasive as to require no live hearing at all, something this Court has repeatedly said it is ill equipped to do.⁴ E.g., Ex parte Charley, Nos. WR–77,427–05, –06, 2013 WL 2726306, at *1 (Tex. Crim. App. June 12, 2013) (per curiam) (not designated for publication) (“[T]he trial court is the appropriate forum for findings of fact.”). At least then, the Court would be in better position than the on-the-ground habeas court to decide the facts. Right now, we're not.⁵

If being the “ultimate” factfinder is a license to police habeas courts by canvassing their findings and exercising our “ultimate” factfinder authority whenever we feel like it, which I don't believe to be the case, what is the point of asking habeas courts to make findings of fact?

I worry that this “ultimate” factfinder authority has been wielded too often and without adequate explanation of its limitations, leaving Texans to wonder if this Court thinks it does not care what the lower courts—who do nearly all the work in our criminal justice system—think. I believe we should show those courts more respect.

VI. WHETHER A PERSON IS INTELLECTUALLY DISABLED UNDER THE EIGHTH AMENDMENT IS A LEGAL QUESTION BUT IS SO FACT-INTENSIVE THAT, OFTEN, THE ANSWER TO THE LEGAL QUESTION FOLLOWS FROM THE FINDINGS.

To reach its result, the majority seems to rely on the fact that whether a person is intellectually disabled under the Eighth Amendment is ultimately a legal question. And yet, the intellectual disability question, despite its legal nature, is as nuanced and intensively fact bound as any in criminal law, often requiring multiple hearings, briefing, live testimony, the opinion of several experts, appeals, and more. See, e.g., Hall v. Florida, 572 U.S. 701, 710 (2014); Petetan v. State, 622 S.W.3d 321, 328 (Tex. Crim. App. 2021).

If there is a credible question as to whether the death penalty can be applied on the account of a defendant's mental acuity, a hearing before someone equipped to observe the witnesses and keep a record of the testimony sufficient to permit subsequent consideration for fidelity to the substantive constitutional standard is required. Because we found that credibility threshold to be passed, we remanded for findings. Because the evidence before the habeas court was perceived as conclusively demonstrating Gonzales’ disability, it did not hold a hearing. Our

reversing of course under the banner of title of “ultimate” factfinder fits awkwardly with the notion of our precedent, our posture, and the way “facts” are “found.” See, e.g., Ballard, 544 U.S. at 63; Raddatz, 447 U.S. at 672–73; Louis, 630 F.2d at 1109–10.

VII. CONCLUSION

With these comments, I respectfully dissent.

FOOTNOTES

1.   I suspect I join many of my colleagues’ skepticism of expert opinion, but I acknowledge that judicial decision-making sometimes requires anchoring to some discernable reasoning. In Texas, plaintiffs cannot get beyond dismissal of a dental hygienist malpractice case without an expert opinion supporting a breach of the requisite duty of care. Wood v. Tice, 988 S.W.2d 829, 830 (Tex. App.—San Antonio 1999, pet. denied). Having been handed the task of separating the intellectually disabled from the ranks of our condemned population, it would seem strange that we are able to do so without anyone hearing from the experts at all. While there may be jurisdictions elsewhere that permit resolution of critical fact questions hinging on expert opinion without any observation of the witness(es), I am not aware of them and question whether a contrary practice is so aberrant as to raise serious federal constitutional concerns. Cf. Honda Motor Corp. v. Oberg, 512 U.S. 415 (1994).

2.   The due-process question would of course be most acute when the elevated standard is said to be triggered by conduct that is itself consistent with the type of intellectual disability the Eighth Amendment exempts from the death penalty in the first place. It may also be informed by the unique process we deploy to dispose of relevant expert opinion testimony no one has observed. Supra n.1.

3.   The United States Supreme Court has held that not all intellectually disabled people are exempted from being put to death by the Eighth Amendment. Atkins, 536 U.S. at 317 (“Not all people who claim to be mentally retarded will be so impaired as to fall within the range of mentally retarded offenders about whom there is a national consensus” that executing them would be cruel and unusual punishment).

4.   I do not question our authority as a terminal court to make original factual determinations. We can hold trials and call witnesses if we want. See, e.g., United States v. Shipp, 214 U.S. 386 (1909) (original trial proceedings before the assembled Supreme Court). What I question is how we can decide this matter in this posture without either declaring that there is no possible, genuine factual question in play—posing the question of why we called for findings in the first place—or remanding to the trial court with instructions to hold a hearing.

5.   The United States Supreme Court said in United States v. Raddatz, 447 U.S. 667, 681 n.7 (1980),[W]e assume it is unlikely that a district judge would reject a magistrate's proposed findings on credibility when those findings are dispositive and substitute the judge's own appraisal; to do so without seeing and hearing the witness or witnesses whose credibility is in question could well give rise to serious questions which we do not reach.Id. (emphasis in original); see Louis v. Blackburn, 630 F.2d 1105 (5th Cir. 1980) (finding the Due Process Clause prohibits district courts from rendering judgment at odds with a magistrate's disposition without conducting live hearing and observation of witnesses where credibility determinations are implicated).

Schenck, P.J., filed a dissenting opinion.