EX PARTE HARLEM HAROLD LEWIS III v. << (2026)

ORDER

This is an initial application for a writ of habeas corpus filed pursuant to the provisions of Texas Code of Criminal Procedure Article 11.071, § 5.¹

In July 2014, a jury convicted Applicant of capital murder. The evidence showed that, after engaging in a high-speed automobile chase to avoid a traffic stop, Applicant shot and killed Corporal Jimmie Norman, a Bellaire Police Department patrolman, and Terry Taylor, a bystander, during the same criminal transaction in December 2012. See Tex. Penal Code § 19.03(a)(7)(A). The jury answered the special issues submitted pursuant to Article 37.071, and the trial court, accordingly, set punishment at death. This Court affirmed the conviction on direct appeal. Lewis v. State, No. AP-77,045 (Tex. Crim. App. April 26, 2017) (not designated for publication).

Applicant's initial post-conviction application for a writ of habeas corpus, which he filed in the trial court in September 2016, presents ten claims challenging the validity of his conviction and resulting sentence. In Claim One, Applicant alleges that he is ineligible for execution because he is intellectually disabled. The issue of intellectual disability was not litigated at Applicant's trial. On habeas, without holding a hearing, Judge George Powell in 2019 entered findings of fact, conclusions of law, and a recommendation that we deny habeas relief on all ten of Applicant's claims.

In an order dated September 16, 2020, we stated that there had been “significant additions to the law regarding intellectual disability since the time of applicant's trial,” citing Moore v. Texas, 581 U.S. 1 (2017) (Moore I) and Moore v. Texas, 586 U.S. 133 (2019) (Moore II). We also determined that additional fact-finding was necessary with regard to Claim One. Therefore, we remanded this cause to the habeas court for consideration and resolution of Claim One, with instructions “to receive evidence from mental health experts and any witnesses whose evidence the court determines is germane to the question of intellectual disability.”

In 2023, Judge Natalia Cornelio held a hearing on Claim One. She thereafter entered findings of fact, conclusions of law, and a recommendation that we grant relief on this claim. Thus, we have separate recommendations from Judge Powell and Judge Cornelio that conflict on the issue of whether Applicant is intellectually disabled. Some of Judge Cornelio's findings and conclusions on Claim One are not supported by the record, and Judge Powell applied outdated standards in some of his. Therefore, we consider portions of both sets of findings on Claim One but decline to wholly adopt either of them. See Ex parte Reed, 271 S.W.3d 698, 727 (Tex. Crim. App. 2008) (“When our independent review of the record reveals that the trial judge's findings and conclusions are not supported by the record, we may exercise our authority to make contrary or alternative findings and conclusions.”).

In Atkins v. Virginia, 536 U.S. 304 (2002), the United States Supreme Court held that the Eighth Amendment bars the execution of the intellectually disabled. The Court explained that intellectually disabled persons “do not act with the level of moral culpability that characterizes the most serious adult criminal conduct.” Id. at 306. “But the Court recognized that ‘[n]ot all people who claim to be [intellectually disabled] will be so impaired as to fall within the range of [intellectually disabled] offenders about whom there is a national consensus’ regarding diminished moral culpability.” Petetan v. State, 622 S.W.3d 321, 327 (Tex. Crim. App. 2021) (citing Atkins, 536 U.S. at 317).

Applicant has the burden to prove by a preponderance of the evidence that he meets the current, medically accepted diagnostic criteria for intellectual disability. See Franklin v. State, 579 S.W.3d 382, 386 (Tex. Crim. App. 2019) (stating that “the issue of intellectual disability is like an affirmative defense and that the defendant has the burden to prove it by a preponderance of the evidence, whether the issue is raised at trial or on habeas”). Specifically, Applicant must prove three criteria: (A) deficits in intellectual functioning—primarily a test-related criterion; (B) adaptive deficits, “assessed using both clinical evaluation and individualized ․ measures”; and (C) the onset of these deficits while he was still a minor. Moore II, 586 U.S. at 135.

The evidence in this case shows that Applicant repeated first grade but otherwise made passing grades in school. He took Advanced Placement (AP) courses in high school, graduated in the top quarter of his class with a 3.24 grade point average, and attended college.² He committed the instant offense when he was twenty-one years old. He took two IQ tests at age twenty-two, while he was awaiting trial in this case. He obtained a full-scale IQ score of 70 on a Wechsler Adult Intelligence Scale, Fourth Edition (WAIS-IV) test administered by defense expert Dr. Shawanda Williams-Anderson in July 2013.³ He took another WAIS-IV test administered by State's expert Dr. Timothy Proctor in June 2014 which yielded a full-scale IQ score of 76. Proctor testified at the habeas hearing that “someone could or could not” dispute that Applicant meets Criterion A because his IQ is in a “gray area” where “clinical judgment comes into play.”⁴

Judges Powell and Cornelio both found that Applicant meets Criterion A. Judge Powell took both WAIS-IV scores into account and found that Applicant's “pre-trial, adult-tested IQ, once the standard error of measurement is applied ․ falls within the range of mild intellectual disability.” Judge Cornelio found that Proctor conducted “a more comprehensive and more valid administration” of the WAIS-IV than Williams-Anderson, yet concluded that Applicant “provided evidence of two valid IQ test scores that establish” he meets Criterion A.

Judges Powell and Cornelio differed on the issue of whether Applicant meets Criterion B. Judge Cornelio concluded that Applicant meets Criterion B; Judge Powell concluded that he does not. We also have a battle of the experts on this issue: defense expert Dr. James Patton opined that Applicant meets Criterion B; State's expert Proctor opined that he does not.⁵ Although Proctor agreed that Applicant “has deficits as you would expect someone [to have] with borderline intellectual functioning,” he opined that there was not “sufficient evidence to say any of those things are in the significant range.” See Atkins, 536 U.S. at 318 (stating that “clinical definitions of [intellectual disability] require ․ significant limitations in adaptive skills”) (emphasis added).

Judge Cornelio credited Patton over Proctor in part because Patton completed a standardized adaptive behavior assessment.⁶ However, Patton administered the Adaptive Behavior Diagnostic Scale (ABDS) to only one respondent—Applicant's aunt, Niesha Williams—in September 2022. Patton asked Williams to rate Applicant's adaptive behavior from twenty-two years prior to the administration of the ABDS, when Applicant was nine years old and Williams was sixteen years old. Patton reported that the ABDS scores “in the conceptual and social domains are more than two standard deviations below the mean.”

Patton acknowledged that the ABDS was not created to be used retrospectively. Proctor testified that he gave “[n]o weight” to the ABDS results in part because Williams “was thinking back to when [Applicant] was a child and when she was a child.” And Judge Cornelio “recognize[d] the [S]tate's well-founded concerns regarding Dr. Patton's retrospective use of the ABDS.” Further, we have acknowledged problems with performing retrospective assessments of adaptive behaviors using family members with a vested interest in saving their relative from the death penalty. See Petetan, 622 S.W.3d at 355–56 (Tex. Crim. App. 2021) (stating that the Adaptive Behavior Assessment System, Second Edition, is not normed for retrospective assessments and Petetan's family members had a vested interest in saving him from a possible death penalty); see also Ex parte Cathey, 451 S.W.3d 1, 20 (Tex. Crim. App. 2014) (recognizing that the Vineland Adaptive Behavior Scales test is not designed to be administered retrospectively and is susceptible to informants being “highly motivated to misremember”).

Judge Cornelio further credited Patton in part because he personally conducted interviews with people (including teachers, friends, and family members) who observed Applicant during his developmental period. Many of these people also signed affidavits in 2016 and/or 2022. Some of them testified at the habeas hearing and some had testified at the punishment phase of Applicant's trial. Proctor expressed concern about the lay witness affidavits in part because “there's places where they're very inconsistent ․ with other pieces of information.” Judge Powell found that multiple affiants lacked credibility and that Applicant's assertion of “inflated” grades was “unsupported by the evidence and not credible.” He also faulted Applicant's experts for: failing “to document the potential for bias and motive to exaggerate or lie inherent in the providers of the vague, anecdotal evidence they relied on in assessing [Applicant's] adaptive behavior”; failing “to seek independent corroboration of the stories about the [A]pplicant”; and ignoring “conflicts and contradictions between the anecdotes and factual evidence.” After reviewing the record, we agree that skepticism is warranted about some of the lay witnesses and the experts’ reliance upon them.

Based upon our own review, we determine that Applicant has failed to prove by a preponderance of the evidence that he meets the current, medically accepted diagnostic criteria for intellectual disability. He is not so impaired as to fall within the range of intellectually disabled offenders about whom there is a national consensus regarding diminished moral culpability. Therefore, we deny relief on Claim One.

In Applicant's remaining claims, he alleges that: he was subjected to prosecutorial misconduct (Claim Two); he was denied effective assistance of trial counsel during jury selection and at the guilt and punishments phases of trial (Claims Three and Four); juror misconduct occurred (Claim Five); he was denied effective assistance of counsel on appeal (Claim Six); his death sentence is unconstitutional due to “racial discrimination in the Harris County District Attorney's Office” (Claim Seven); the trial court was “prohibited from instructing the jury that a vote by one juror would result in a life sentence” (Claim Eight); the future dangerousness special issue is “unconstitutionally vague” (Claim Nine); and the jury instructions at punishment “restricted the evidence that the jury could determine was mitigating” (Claim Ten). We agree with Judge Powell's recommendation to deny Claims Two through Ten. Based upon his findings and our own review, we deny relief on those claims.

IT IS SO ORDERED THIS THE 8th DAY OF OCTOBER, 2026.

FOOTNOTES

1.   Unless otherwise indicated, all references to Articles are to the Texas Code of Criminal Procedure.

2.   We are mindful of the requirement that courts may not place “undue emphasis” upon adaptive strengths. Moore I, 581 U.S. at 18 n.9.

3.   Williams-Anderson initially reported that Applicant obtained a full-scale IQ score of 71, but she later re-examined “the original [test] protocol” and found “[s]everal over-scores and two errors” that lowered Applicant's score to 70.

4.   Proctor testified that Applicant “is right in that gray area, which the DSM says is 70 plus or minus five.” The American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, Text Revision (2022) (DSM-5-TR) provides that individuals with intellectual disability have IQ scores of approximately two standard deviations or more below the population mean, including a margin for measurement error (generally plus or minus five points). DSM-5-TR at 38. On tests with a standard deviation of 15 and a mean of 100, this involves a score of 65-75. Id.

5.   Several other defense experts who signed affidavits and/or testified at the hearing met with Applicant and/or reviewed his records, but did not formally assess his adaptive behavior. One of them performed a “readability grade level analysis” of Applicant's writing samples and another conducted “bedside screening tests” that do not diagnose intellectual disability.

6.   The DSM-5-TR provides that adaptive functioning “is assessed using both clinical evaluation and individualized, culturally appropriate, psychometrically sound measures.” DSM-5-TR at 42. It further provides that “[s]tandardized measures are used with knowledgeable informants (e.g., parent or other family member; teacher; counselor; care provider) and the individual to the extent possible.” Id.

Per curiam.