A panel of this court previously affirmed the district court’s denial of Billy Ray Nelson’s habeas corpus petition challenging his sentence on the ground that the Texas capital-sentencing procedure failed to give constitutionally sufficient effect to his mitigating evidence, in violation of Penry v. Lynaugh (Penry I), 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989). See Nelson v. Cockrell, 77 Fed.Appx. 209 (5th Cir. Aug. 12, 2003) (unpublished). Nelson petitioned the Supreme Court for writ of cer-tiorari. The Supreme Court granted the petition, vacated our judgment, and remanded the case to this court for reconsideration in light of the Supreme Court’s decision in Tennard v. Dretke, 542 U.S. 274, 124 S.Ct. 2562, 159 L.Ed.2d 384 (2004). Nelson v. Dretke, 542 U.S. 934, 124 S.Ct. 2905, 159 L.Ed.2d 808 (2004). On remand, a panel of this court once *290again affirmed the district court’s denial of Nelson’s habeas corpus petition. See Nelson v. Dretke, 442 F.3d 282 (5th Cir.2006). Having ordered rehearing en banc, Nelson v. Dretke, 442 F.3d 912 (5th Cir.2006), we again reconsider the application of Penry I and its progeny to Nelson’s case. We conclude that, on the facts presented here, there is a reasonable likelihood that the Texas capital-sentencing scheme precluded the jury from giving full effect to Nelson’s mitigating evidence as required by the Supreme Court; accordingly, we REVERSE the district court’s denial of habeas relief and REMAND with instructions to grant the writ of habeas corpus.
I. FACTUAL AND PROCEDURAL BACKGROUND
On December 11, 1991, a Texas jury found Nelson guilty of capital murder for the February 23, 1991, slaying and brutal sexual assault of his neighbor, Charla Wheat. Evidence presented during the guilt/innocence phase of trial revealed the following: Nelson gained entrance to Wheat’s apartment by asking if he could use her phone. Once inside, he cut the telephone cord to prevent her from calling for help and then proceeded to stab her. He then found Wheat’s roommate, Carol Maynard, who was five months pregnant at the time, and forced her to get out of bed and enter the living room, where Wheat was on her knees bleeding from her stab wounds. Nelson told the women to remove their clothing and threatened to kill them if they refused. He then forced the women to perform sexual acts on him and each other. Thereafter, he stabbed Maynard in the neck and proceeded to strike Wheat. Nelson left briefly but Wheat began screaming and he returned. While Maynard pretended to be dead, Nelson struck and stabbed Wheat until she died. He then left the women’s apartment.
At the sentencing phase of the trial, Nelson presented the following mitigating evidence, which we will discuss more fully infra: (1) he was rejected by his mother, who had completely abandoned him by age 14 (“abusive childhood” evidence); (2) he abused drugs and alcohol (“substance abuse” evidence); (3) he has troubled relationships with his brother and with women; (4) he had a child out of wedlock, with whom he was not permitted to have a relationship; and (5) a psychiatrist testified he was suffering from borderline personality disorder (“mental disorder” evidence). For a jury to impose the death penalty at the time of Nelson’s trial, Article 37.071(b) of the Texas Code of Criminal Procedure required the jury to answer two special issue questions concerning evidence presented in mitigation: “(1) whether the conduct of the defendant that caused the death of the deceased was committed deliberately and with the reasonable expectation that the death of the deceased or another would result” (“the deliberateness special issue”); and “(2) whether there is a probability that the defendant would commit criminal acts of violence that would constitute a continuing threat to society” (“the future-dangerousness special issue”).1 The jury answered both special-issue questions in the affirmative, sentencing Nelson to death. Nelson appealed his sentence and conviction to the Texas Court of Criminal Appeals; that court affirmed, Nelson v. Texas, 864 S.W.2d 496 (Tex.Crim.App. *2911993), and Nelson’s conviction became final when the Supreme Court denied certiorari review, Nelson v. Texas, 510 U.S. 1215, 114 S.Ct. 1338, 127 L.Ed.2d 686 (1994).
Nelson filed a state petition for writ of habeas corpus in September 1997, arguing that the Texas capital sentencing scheme, 1.e., the two special-issue questions, failed to ensure that the jury could give the constitutionally required consideration of and effect to his mitigating evidence of his mental disorder, abusive childhood, and substance abuse under Penry I, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256. He also filed a second petition in February 1998, alleging additional claims. The Texas Court of Criminal Appeals denied the writ based on the findings and recommendations of the trial court. Ex parte Nelson, No. 49,886-01 (Tex.Crim.App. Oct. 10, 2001). Specifically, with regard to Nelson’s Penry claims, the Texas Court of Criminal Appeals recognized that, to be constitutional, “a death penalty procedure must allow the jury to consider all relevant mitigating evidence.” Ex parte Nelson, No. 8,214 at 88 (118th Judicial District Howard County, Tex. July 10, 2001) (findings of fact and conclusions of law). The court also recognized that where the defendant’s mitigating evidence is beyond the scope of the special issues, and the jury is unable to give effect to its reasoned moral response to the mitigating evidence, the procedure is unconstitutional as applied to the defendant. Id. In applying the law to the facts of Nelson’s case, the court noted that Nelson’s evidence of drug and alcohol abuse had no mitigating relevance beyond the scope of the special issues. Id. at 89. Moreover, with regard to the other mitigating evidence presented,
[t]he Court instructed the jury on the charge on punishment, “You should consider and give effect in answering each issue to your evaluation of all of the evidence before you, including all aspects of the background and character of the defendant and the circumstances of the crime.” ... The jury charges and special issues allowed the jurors to give effect to all presented mitigating evidence in their answers to the special issues including the intoxication of [Nelson] at the time of the offense.
Id. at 90. Therefore, the court concluded that the procedure was constitutional as applied. The court dismissed Nelson’s subsequent habeas petition as an abuse of the writ. Ex parte Nelson, No. 49,886-02 (Tex.Crim.App. Oct. 10, 2001).
Nelson filed a petition for writ of habeas corpus in the federal district court in August 2002. The district court rejected Nelson’s Penry claim for failing to meet the requirements of our now-defunct “constitutional-relevance” test.2 A panel of this court granted Nelson a certificate of ap-pealability (“COA”) on this issue; however, the panel ultimately affirmed the district court’s denial of habeas relief. Nelson petitioned the Supreme Court for writ of certiorari, and the Supreme Court granted the petition, vacated the panel’s judgment, and remanded the case to this court for reconsideration in light of the Supreme *292Court’s decision in Tennard, 542 U.S. 274, 124 S.Ct. 2562, 159 L.Ed.2d 384. On remand, a panel of this court once again affirmed the district court’s denial of habe-as relief. All three panel members concurred in the judgment; however, there was no consensus on the correct methodology for analyzing Nelson’s claim.3 Accordingly, this court ordered rehearing en banc, and we once again reconsider the application of Penry in light of Tennard to the facts of Nelson’s case.4
II. DISCUSSION
A. Standard of Review
Because Nelson filed his § 2254 habeas petition after April 24, 1996, this habeas proceeding is governed by the Antiterrorism and Effective Death Penalty Act (“AEDPA”). See Fisher v. Johnson, 174 F.3d 710, 711 (5th Cir.1999). We have jurisdiction to resolve the merits of Nelson’s habeas petition because, as stated above, we granted him a COA on his Pen-ry claim. See Nelson v. Dretke, 442 F.3d at 284; see also 28 U.S.C. § 2253(c)(1).
Under AEDPA, a federal court may not grant a writ of habeas corpus “with respect to any claim that was adjudicated on the merits in State court proceedings” unless the petitioner shows that the state court’s adjudication “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 that the state court’s adjudication of a claim “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); Williams v. Taylor, 529 U.S. 362, 402-13, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000). A state court’s decision is “contrary to” clearly established federal law if (1) the state court “applies a rule that contradicts the governing law” announced in Supreme Court cases, or (2) the state court decides a case differently than the Supreme Court did on a set of materially indistinguishable facts. Mitchell v. Esparza, 540 U.S. 12, 15-16, 124 S.Ct. 7, 157 L.Ed.2d 263 (2003) (internal quotation marks omitted). A state court’s application of clearly established federal law is “unreasonable” within the meaning of AEDPA when the state court identifies the correct governing legal principle from Supreme Court precedent, but applies that principle to the case in an objectively unreasonable manner. Wiggins v. Smith, 539 U.S. 510, 520, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003).
A writ of habeas corpus may also issue if the state court’s adjudication of a claim “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)(2). Under AEDPA, a state court’s factual findings are “presumed to be correct” unless the habeas petitioner rebuts the presumption through “clear and convincing evidence.” Id. § 2254(e)(1); Miller v. Johnson, 200 F.3d 274, 281 (5th Cir.2000).
*293We review the district court’s conclusions of law regarding the state court’s application of federal law de novo, and we review the district court’s findings of fact, if any, for clear error. Collier v. Cockrell, 300 F.3d 577, 582 (5th Cir.2002).
B. Clearly Established Federal Ldtu
Under AEDPA, our duty is to determine whether the state court’s determination was contrary to or an unreasonable application of clearly established federal law as determined by the Supreme Court at the time that Nelson’s conviction became final in 1994. See Williams, 529 U.S. at 405, 120 S.Ct. 1495. In Tennard and Smith v. Texas, two recent cases involving Penry claims, the Supreme Court unequivocally stated that the relevant inquiry under its precedent was whether there was a reasonable likelihood that the jury would interpret the Texas special issues in a manner that precluded it from fully considering and giving full effect to all of the defendant’s mitigating evidence. See Tennard, 542 U.S. at 288-89, 124 S.Ct. 2562; see also Smith v. Texas, 543 U.S. 37, 38, 125 S.Ct. 400, 160 L.Ed.2d 303 (2004) (per curiam). This “full-effect” standard requires that a juror be able to express his reasoned moral response to evidence that has mitigating relevance beyond the scope of the special issues; i.e., a juror cannot be precluded from electing a sentence less than death if he believes that the mitigating evidence offered makes the defendant less morally culpable for the crime, even if he nonetheless feels compelled to answer the two special issues in the affirmative. See Penry v. Johnson (Penry II), 532 U.S. 782, 797, 121 S.Ct. 1910, 150 L.Ed.2d 9 (2001); Penry I, 492 U.S. at 320, 109 S.Ct. 2934. A review of the Supreme Court’s decisions in this area demonstrates that this “full-effect” standard was clearly established by the time that Nelson’s conviction became final.
1. Jurek v. Texas and the Immediate Post-Furman Cases
In Furman v. Georgia, 408 U.S. 238, 92 S.Ct. 2726, 33 L.Ed.2d 346 (1972), the Supreme Court held that state capital-sentencing schemes allowing the death penalty to be “wantonly and ... freakishly imposed” by permitting unbridled discretion in sentencing violated the Eighth and Fourteenth Amendments. Id. at 310, 92 S.Ct. 2726 (Stewart, J., concurring). After Furman, states began to rewrite their death penalty statutes, restricting the classes of death-penalty eligible offenders and channeling jurors’ discretion in sentencing in an attempt to comply with the Supreme Court’s directive. Specifically, Texas responded to Furman with the “special issues” capital-sentencing scheme at issue in this case, which was designed to guide jurors’ consideration of mitigating evidence offered in the sentencing phase of capital cases.
The immediate post-Furman Supreme Court cases addressing this and other sentencing schemes attempted to strike a balance between satisfying two competing constitutional requirements — the requirement of “individualized sentencing” that takes into account the unique facts of each case and each defendant, and the requirement of preventing the arbitrary imposition of the death penalty that can result from giving the sentencer unfettered discretion. These cases announced the principles that would underlie the Supreme Court’s later pronouncement that a capital sentencing scheme must allow the sentencer to give full effect to all of the defendant’s mitigating evidence.
In Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976), the Supreme Court upheld the facial constitutionality of the Texas special-issues sentencing scheme *294on the same day that it ruled on the validity of the post-Furman death penalty statutes of four other states. See Gregg v. Georgia, 428 U.S. 153, 96 S.Ct. 2909, 49 L.Ed.2d 859 (1976) (upholding the facial constitutionality of Georgia’s capital-sentencing scheme, which narrowed the class of death-eligible offenders and guided the sentencer’s consideration of mitigating and aggravating evidence); Proffitt v. Florida, 428 U.S. 242, 96 S.Ct. 2960, 49 L.Ed.2d 913 (1976) (upholding the facial constitutionality of Florida’s capital-sentencing scheme, which narrowed the class of death-eligible offenders and guided the sentencer’s consideration of mitigating and aggravating evidence); Woodson v. North Carolina, 428 U.S. 280, 96 S.Ct. 2978, 49 L.Ed.2d 944 (1976) (striking down North Carolina’s mandatory capital-sentencing scheme because it gave sentencers no discretion to impose the death penalty for certain crimes); Roberts v. Louisiana, 428 U.S. 325, 96 S.Ct. 3001, 49 L.Ed.2d 974 (1976) (striking down Louisiana’s capital-sentencing scheme requiring the imposition of the death penalty for certain crimes). In Ju-rek, a plurality of the Court explained that, while the Texas sentencing scheme was constitutional on its face, “[a] jury must be allowed to consider on the basis of all relevant evidence not only why a death sentence should be imposed, but also why it should not be imposed.” Jurek, 428 U.S. at 271, 96 S.Ct. 2950 (plurality opinion) (citing Woodson, 428 U.S. at 303-05, 96 S.Ct. 2978); see also Roberts, 428 U.S. at 334-36, 96 S.Ct. 3001 (plurality opinion). Therefore, “the constitutionality of the Texas procedures turns on whether the enumerated questions allow consideration of particularized mitigating factors.” Jurek, 428 U.S. at 272, 96 S.Ct. 2950. While observing that the future-dangerousness special issue allowed consideration of some types of mitigating evidence, the Jurek plurality also left room for as-applied challenges to the Texas sentencing scheme, noting that the Texas Court of Criminal Appeals had not yet interpreted the deliberateness and provocation special issues. Id. at 272 n. 7, 96 S.Ct. 2950 (“[I]t is as yet undetermined whether or not the jury’s consideration of those questions would properly include consideration of mitigating circumstances.”).
A Lockett v. Ohio and Eddings v. Oklahoma
Echoing these post-Furman concerns that the sentencer be able to consider and give effect to mitigating evidence in a constitutionally adequate way, the Supreme Court in Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978), struck down Ohio’s death penalty statute, which allowed the sentencer to impose a sentence less than death for certain crimes only if the mitigating evidence showed that (1) the victim induced or facilitated the offense, (2) the offense was a result of duress, coercion, or strong provocation, or (3) the offense was a product of psychosis or mental retardation. A plurality of the Court explained that this sentencing scheme, which allowed the sentencer to consider some aspects of the mitigating evidence presented but not others, was unconstitutional because
the Eighth and Fourteenth Amendments require that the sentencer, in all but the rarest kind of capital case, not be precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death.
Id. at 604, 98 S.Ct. 2954 (plurality opinion); see also Kansas v. Marsh, — U.S. -, 126 S.Ct. 2516, 2525, 165 L.Ed.2d 429 (2006) (noting that the Ohio sentencing scheme in Lockett was unconstitutional *295“because, by limiting a jury’s consideration of mitigation to three factors specified in the statute, it ‘prevented sentencers in capital cases from giving independent weight to mitigating evidence militating in favor of a sentence other than death ”) (emphasis added). Four years later, in Eddings v. Oklahoma, 455 U.S. 104, 102 S.Ct. 869, 71 L.Ed.2d 1 (1982), a majority of the Court adopted the Lockett plurality’s reasoning to vacate an Oklahoma death sentence where the sentencing judge refused to consider, as a matter of law, the defendant’s mitigating evidence of his abusive childhood and treatable emotional disturbance. The Court rejected the state appellate court’s application of a heightened-relevance standard to the mitigating evidence, noting that while the sentencer can “determine the weight to be given relevant mitigating evidence,” it “may not give it no weight by excluding such evidence from [its] consideration.” Id. at 115, 102 S.Ct. 869; see also Marsh, 126 S.Ct. at 2525 (observing that, in Eddings, “a majority of the Court held that a sentencer may not categorically refuse to consider any relevant mitigating evidence”).
3. Franklin v. Lynaugh
The Court considered an as-applied challenge to the Texas capital-sentencing scheme for the first time in Franklin v. Lynaugh, 487 U.S. 164, 108 S.Ct. 2320, 101 L.Ed.2d 155 (1988). There, the Court held that the special issues allowed the jury to give constitutionally adequate effect to the petitioner’s mitigating evidence of good behavior during a previous term of imprisonment. A plurality of the Court stated that, because the petitioner’s evidence of good prison behavior did not have mitigating significance independent of its relevance to the petitioner’s propensity to commit future crimes, “[i]n resolving the [future-dangerousness special issue] the jury was surely free to weigh and evaluate petitioner’s disciplinary record as it bore on his ‘character’ — that is, his ‘character’ as measured by his likely future behavior.” Id. at 177-78, 108 S.Ct. 2320. Justice O’Connor concurred separately, emphasizing that, although Jurek upheld the facial validity of the Texas capital sentencing scheme, and in this case the mitigating relevance of all of the petitioner’s evidence was within the scope of the special issues,
[i]f ... petitioner had introduced mitigating evidence about his background or character or the circumstances of the crime that was not relevant to the special verdict questions, or that had relevance to the defendant’s moral culpability beyond the scope of the special verdict questions, the jury instructions would have provided the jury with no vehicle for expressing its “reasoned moral response” to that evidence. If this were such a case, then we would have to decide whether the jury’s inability to give effect to that evidence amounted to an Eighth Amendment violation. In my view, however, this is not such a case.
Id. at 185, 108 S.Ct. 2320 (O’Connor, J., concurring).
A Penry I
The very next term, the Supreme Court considered just such a case in Penry I, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256. The Penry I Court held that habeas relief was appropriate because a juror presented with the Texas special issues could not have given effect to the full scope of the mitigating evidence that had been presented at the sentencing phase. Penry, a death-row habeas petitioner, had offered mitigating evidence at sentencing of (1) a low I.Q. indicating likely mental retardation; (2) an organic brain disorder that prevented him from appreciating the wrongfulness of his conduct or conforming *296his behavior to the law; (3) a troubled, abusive upbringing; and (4) an anti-social personality disorder. Penry argued that the Texas special issues, as applied in his case, were an inadequate vehicle to allow the jury to consider or give effect to this evidence, because the evidence had mitigating relevance beyond the scope of the special issues. The Court, with Justice O’Connor writing for the majority, first held that granting Penry the relief he requested would not announce a new rule on collateral review in violation of Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), because granting such relief was “dictated by Eddings and Lock-ett” Penry 1, 492 U.S. at 317, 109 S.Ct. 2934.
The Court then granted the habeas petition, emphasizing that “it is not enough simply to allow the defendant to present mitigating evidence to the sentencer. The sentencer must also be able to consider and give effect to that evidence in imposing sentence.” Id. at 319, 109 S.Ct. 2934. Only then can the sentence imposed “ ‘reflect a reasoned moral response to the defendant’s background, character, and crime.’ ” Id.(quoting California v. Brown, 479 U.S. 538, 545, 107 S.Ct. 837, 93 L.Ed.2d 934 (1987) (O’Connor, J., concurring)). Indeed, as “both the concurrence and dissent in Franklin understood,” Ju-rek, in which the Court upheld the facial validity of the Texas capital-sentencing scheme, “rest[ed] fundamentally on the express assurance that the special issues would permit the jury to fully consider all the mitigating evidence a defendant introduced that was relevant to the defendant’s background and character and to the circumstances of the offense.” Id. at 321, 96 S.Ct. 2950 (emphasis added).
In Penry’s case, however, the Court held that the evidence of mental retardation and abusive childhood had mitigating relevance beyond the scope of the deliberateness and future-dangerousness issues, because it also spoke to Penry’s moral culpability; therefore, the jury was unable to give effect to the mitigating evidence in a manner consistent with the Eighth Amendment. First, with regard to the deliberateness special issue, the Court reasoned that, although a jury could give partial effect to Penry’s mental retardation and abusive past by finding that his actions were not deliberate, a jury could also conclude that Penry acted deliberately but, because of his mental retardation and abusive childhood, “was less morally ‘culpable than defendants who have no such excuse,’ but who acted ‘deliberately’ as that term is commonly understood.” Id. at 322-23, 96 S.Ct. 2950 (quoting California v. Brown, 479 U.S. at 545, 107 S.Ct. 837 (O’Connor, J., concurring)). Without a special instruction enabling the jury to give effect to the impact of Penry’s mitigating evidence on his moral culpability, the jury lacked an adequate vehicle through which to express its “reasoned moral response” to this evidence. Second, the Court held that the future-dangerousness instruction was likewise constitutionally inadequate because, in this case, “Pen-ry’s mental retardation and history of abuse is ... a two-edged sword: it may diminish his blameworthiness for his crime even as it indicates that there is a probability that he will be dangerous in the future.” Id. at 324, 96 S.Ct. 2950. Because Penry’s mitigating evidence, viewed through the lens of future dangerousness, “is relevant only as an aggravating factory ... ‘[i]t did not allow the jury to consider a major thrust of Penry’s evidence as mitigating evidence.’ ” Id. at 323-24, 96 S.Ct. 2950 (quoting Penry v. Lynaugh, 832 F.2d 915, 925 (5th Cir.1987)).
Although the Court did not expressly use the words “full effect” in Penry I, its *297reasoning makes clear that “full effect” is what it meant. See, e.g., Penry I, 492 U.S. at 323, 109 S.Ct. 2934 (“In the absence of jury instructions defining ‘deliberately’ in a way that would clearly direct the jury to consider fully Penry’s mitigating evidence as it bears on his personal culpability, we cannot be sure that the jury was able to give effect to the mitigating evidence of Penry’s mental retardation and history of abuse in answering the first special issue.”) (emphasis added); id. at 318-19, 109 S.Ct. 2934 (“Penry argues that those assurances were not fulfilled in his particular case because, without appropriate instructions, the jury could not fully consider and give effect to the mitigating evidence of his mental retardation and abused childhood in rendering its sentencing decision.”). Further, even the dissent in Penry I recognized that the Court was applying a full-effect standard:
that the constitutionality turns on whether the questions allow mitigating factors not only to be considered (and, of course, given effect in answering the questions), but also to be given effect in all possible tvays, including ways that the questions do not permit. ... What the Court means by “fully consider” (what it must mean to distinguish Jurek) is to consider for all purposes, including purposes not specifically permitted by the questions.
Penry I, 492 U.S. at 355, 109 S.Ct. 2934 (Scalia, J., dissenting) (citation omitted). Thus there can be no doubt that the Penry I Court applied a standard requiring the jury to be able to give full consideration and full effect to a defendant’s mitigating evidence.
The State contends that the “full effect” language in Penry I and its progeny “is merely dicta, because it would otherwise overrule Jurek”-, however, this argument mischaracterizes the holding in Jurek, which upheld only the facial validity of the Texas special issues scheme. See Jurek, 428 U.S. at 272, 96 S.Ct. 2950 (stating that “the constitutionality of the Texas procedures turns on whether the enumerated questions allow consideration of particularized mitigating factors,” but also noting that “it is as yet undetermined whether or not the jury’s consideration of [the special issues] would properly include consideration of mitigating circumstances” in every situation). The Penry I Court’s holding was a case-specific application of Jurek, which expressly left room for as-applied challenges. See Penry I, 492 U.S. at 320, 109 S.Ct. 2934 (“[B]oth the concurrence and the dissent understood Jurek as resting fundamentally on the express assurance that the special issues would permit the jury to fully consider all the mitigating evidence a defendant introduced that was relevant to the defendant’s background and character and to the circumstances of the offense.”). That Jurek involved only a facial challenge to the Texas statute is apparent not only from the Court’s decision in Penry I, holding the Texas statute unconstitutional as applied, but also from the Court’s decisions in as-applied challenges to the constitutionality of the death penalty procedures in other states. As discussed above, the Supreme Court upheld facial challenges to the death penalty procedures in Georgia and Florida at the same time that it upheld the facial challenge to the Texas statute. See Gregg v. Georgia, 428 U.S. 153, 96 S.Ct. 2909, 49 L.Ed.2d 859; Proffitt v. Florida, 428 U.S. 242, 96 S.Ct. 2960, 49 L.Ed.2d 913. Nevertheless,
after Gregg and Proffitt and prior to Franklin, [the Court] held unconstitutional specific applications of the same Georgia and Florida statutes [it] earlier had approved. See Godfrey v. Georgia, 446 U.S. 420, 100 S.Ct. 1759, 64 L.Ed.2d 398 (1980) (vague and overly broad con*298struction of aggravating factor rendered death sentence unconstitutional); Hitchcock v. Dugger, [481 U.S. 393, 107 S.Ct. 1821, 95 L.Ed.2d 347 (1987),] (holding it unconstitutional to restrict jury’s consideration of mitigating factors to those enumerated in the statute).
Johnson v. Texas, 509 U.S. 350, 384, 113 S.Ct. 2658, 125 L.Ed.2d 290 (1993) (O’Con-nor, J., dissenting). Further, applying the full-consideration and full-effect standard does not require overruling Jurek, because “some types of mitigating evidence can be fully considered by the sentencer in the absence of special jury instructions.” Penny I, 492 U.S. at 315, 109 S.Ct. 2934 (citing Franklin v. Lynaugh, 487 U.S. at 175, 108 S.Ct. 2320 (plurality opinion); Franklin v. Lynaugh, 487 U.S. at 185-86, 108 S.Ct. 2320 (O’Connor, J., concurring in judgment)); see also Graham v. Collins, 506 U.S. 461, 521, 113 S.Ct. 892, 122 L.Ed.2d 260 (1993) (Souter, J., dissenting) (explaining that the petitioner’s evidence of “[v]ol-untary chores for and church attendance with a relative, and supplying some level of support for [his] children” could be considered through the future-dangerousness special issue). The Constitution requires a court to determine whether the special issues as applied enable the sentencer to give full consideration and full effect to the capital defendant’s mitigating evidence; the “full-effect” standard is not— and has never been — inconsistent with the holding in Jurek.
5. Graham v. Collins and Johnson v. Texas
After Penny I, the Court addressed in Graham, 506 U.S. 461, 113 S.Ct. 892, 122 L.Ed.2d 260, and Johnson, 509 U.S. 350, 113 S.Ct. 2658, 125 L.Ed.2d 290, two more as-applied challenges to the Texas special issues sentencing scheme, both of which denied relief to petitioners who claimed that the special issues failed to give effect to the mitigating evidence of their youth. In Graham, the Court held that Teague barred it from granting relief to a habeas petitioner who lodged a Penry challenge to his death sentence, which became final in 1984. The petitioner argued that the Texas special issues did not give constitutionally adequate effect to his mitigating evidence of good character and youth. Because the Court disposed of the case on Teague grounds, it did not address the substantive merits of the petitioner’s Pen-ry claim; instead, it considered whether granting the petitioner’s requested relief would have constituted a new rule at the time the petitioner’s sentence became final in 1984, holding that
even if Penny reasonably could be read to suggest that Graham’s mitigating evidence was not adequately considered under the former Texas procedures, that is not the relevant inquiry under Teague. Rather, the determinative question is whether reasonable jurists reading the case law that existed in 1984 could have concluded that Graham’s sentencing was not constitutionally infirm. We cannot say that all reasonable jurists would have deemed themselves compelled to accept Graham’s claim in 1984 .... The ruling Graham seeks, therefore, would be a “new rule” under Teague.
Id. at 477, 113 S.Ct. 892.
Later that term, in Johnson, 509 U.S. 350, 113 S.Ct. 2658, 125 L.Ed.2d 290, the Court considered a similar challenge on direct review. In Johnson, the only mitigating evidence that the petitioner offered was that of his youth at the time he committed the crime. The Court noted that, unlike other mitigating evidence that the Court had considered in previous cases, “[t]he relevance of youth as a mitigating factor derives from the fact that the signature qualities of youth are transient; as individuals mature, the impetuousness and *299recklessness that may dominate in younger years can subside.” Id. at 368, 113 S.Ct. 2658 (emphasis added). Given these unique characteristics of youth, the Court held that this evidence did not lie beyond the reach of the sentencer applying the Texas special issues because “there is ample room in the assessment of future dangerousness for a juror to take account of the difficulties of youth as a mitigating force in the sentencing determination.” Id. The Court applied the standard set forth in Boyde v. California, 494 U.S. 370, 110 S.Ct. 1190, 108 L.Ed.2d 316 (1990), to “determine ‘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.’ ” Johnson, 509 U.S. at 367, 113 S.Ct. 2658 (quoting Boyde, 494 U.S. at 380, 110 S.Ct. 1190). “Although the reasonable likelihood standard does not require that the defendant prove that it was more likely than not that the jury was prevented from giving effect to the evidence, the standard requires more than a mere possibility of such a bar.” Id. The Court, again emphasizing the unique qualities of youth as a mitigating factor, distinguished Penry I, noting that “[ujnlike Pen-ry’s mental retardation, which rendered him unable to learn from his mistakes, the 111 effects of youth that a defendant may experience are subject to change and, as a result, are readily comprehended as a mitigating factor in consideration of the second special issue.” Johnson, 509 U.S. at 369, 113 S.Ct. 2658. Further, unlike the evidence of mental retardation at issue in Penry I, a juror’s consideration of the impact of youth on the petitioner’s conduct “is not independent of an assessment of personal culpability .... If any jurors believed that the transient qualities of petitioner’s youth made him less culpable for the murder, there is no reasonable likelihood that those jurors would have deemed themselves foreclosed from considering that in evaluating petitioner’s future dangerousness.” Id. at 369-70, 113 S.Ct. 2658. Thus Graham and Johnson stand for the proposition that youth, which is different in kind and in mitigating effect from Penry’s evidence of mental retardation and abusive childhood, can be fully considered and given effect through the special-issues sentencing scheme.
6. Penry II
In Penry II, 532 U.S. 782, 121 S.Ct. 1910, 150 L.Ed.2d 9, Justice Kennedy, the author of Johnson, joined the majority, and the Court reaffirmed that the standard is full effect, once again invalidating the application of the Texas special issues to Penry’s mitigating evidence of mental retardation and abusive upbringing. After the Court vacated Penry’s death sentence in Penry I, the State of Texas retried Penry, who was again found guilty of capital murder and sentenced to death. During the sentencing phase of the second trial, the court submitted the same special issues to the jury that were the focus of Penry I, only this time the court also provided a supplemental “nullification” instruction. This instruction directed the jury to consider the effect of all of the mitigating evidence on Penry’s personal culpability, and,
[i]f you determine, when giving effect to the mitigating evidence, if any, that a life sentence, as reflected by a negative finding to the issue under consideration, rather than a death sentence, is an appropriate response to [Penry’s] personal culpability ..., a negative finding should be given to one of the special issues.
Id. at 790, 121 S.Ct. 1910.
The Court, fully aware of the analytical constraints imposed by the deferential AEDPA standard of review, held that the Texas Court of Criminal Appeals had un*300reasonably applied the holding of Penry I when it held that the special issues and the nullification instruction were constitutionally adequate vehicles to give effect to Penny’s mitigating evidence. Justice O’Connor, writing for the Court, stated:
the key under Penry I is that the jury be able to “consider and give effect to [a defendant’s mitigating] evidence in imposing sentence.” 492 U.S., at 319, 109 S.Ct. 2934, 106 L.Ed.2d 256 (emphasis added). See also Johnson v. Texas, 509 U.S. 350, 381, 113 S.Ct. 2658, 125 L.Ed.2d 290 (1993) (O’CONNOR, J., dissenting) (“[A] sentencer [must] be allowed to give full consideration and fall effect to mitigating circumstances” (emphasis in original)). For it is only when the jury is given a “vehicle for expressing its ‘reasoned moral response’ to that evidence in rendering its sentencing decision,” Penry I, 492 U.S. at 328, 109 S.Ct. 2934, that we can be sure that the jury “has treated the defendant as a [‘uniquely individual human being]’ and has made a reliable determination that death is the appropriate sentence,” id., at 319, 109 S.Ct. 2934 (quoting Woodson v. North Carolina, 428 U.S. 280, 304, 305, 96 S.Ct. 2978, 49 L.Ed.2d 944 (1976)).
Penry II, 532 U.S. at 797, 121 S.Ct. 1910. As the Court held in Penry I, the deliberateness and future-dangerousness issues were not broad enough to provide a vehicle that allowed the jury to express its reasoned moral response to the full mitigating impact of all of the evidence; neither was the State’s attempted fix — the nullification instruction — constitutionally sufficient, because “it made the jury charge as a whole internally contradictory and placed law-abiding jurors in an impossible situation.” Id. at 799, 121 S.Ct. 1910. Under this scheme, there was still “at the very least, ‘a reasonable likelihood that the jury ... applied the challenged instruction in a way that prevented] the consideration’ of Pen-ry’s mental retardation and childhood abuse.” Id. (quoting Boyde, 494 U.S. at 380, 110 S.Ct. 1190). Because the State failed to define either special issue “in a way that would clearly direct the jury to consider fully Penr/s mitigating evidence as it bears on his personal culpability,” the Texas special-issues scheme was still unconstitutional as applied to Penry’s mitigating evidence, and the Texas Court of Criminal Appeals’ conclusion otherwise was an unreasonable application of clearly established federal law. Penry II, 532 U.S. at 803, 121 S.Ct. 1910 (emphasis added).
7. Tennard v. Dretke and Smith v. Texas
The Supreme Court’s decision in Ten-nard, in light of which this court must assess Nelson’s Penry claim, reaffirms that a jury cannot be precluded from giving full effect to a defendant’s mitigating evidence and leaves no doubt that this standard was in effect at the time that Nelson’s conviction became final.5 The *301Supreme Court handed down Tennard on June 24, 2004, reversing a panel of this court that had applied the aforementioned “constitutional-relevance” test to deny a COA on a death-row inmate’s petition for habeas relief on Penry grounds. The Court explained that the petitioner, who argued that the Texas special issues sentencing scheme did not enable the sentencer to give full effect to his mitigating evidence of impaired intellectual functioning and low I.Q. score, was entitled to a COA, and that the lower courts had erred by applying the Fifth Circuit’s “constitutional-relevance” test.
Specifically, the Supreme Court excoriated the Fifth Circuit for invoking its own restrictive gloss on the Court’s Penry I decision, uniformly applying to Penry claims a heightened-relevance standard that “has no foundation in the decisions of this Court.” Tennard, 542 U.S. at 284, 124 S.Ct. 2562. The Court then reiterated that the appropriate-relevance standard in a capital case — as in any other case — is a low one:
When we addressed directly the relevance standard applicable to mitigating evidence in capital cases in McKoy v. North Carolina, 494 U.S. 433, 440-41, 110 S.Ct. 1227, 108 L.Ed.2d 369 (1990), we spoke in the most expansive terms. We established that the “meaning of relevance is no different in the context of mitigating evidence introduced in a capital sentencing proceeding” than in any other context, and thus the general evidentiary standard — “ ‘ “any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence” ’ ” — applies. Id. at 440, 110 S.Ct. 1227 (quoting New Jersey v. T.L.O., 469 U.S. 325, 345, 105 S.Ct. 733, 83 L.Ed.2d 720 (1985)).... Thus, a State cannot bar “the consideration of ... evidence if the sentencer could reasonably find that it warrants a sentence less than death.” 494 U.S. at 441, 110 S.Ct. 1227.
Tennard, 542 U.S. at 284-85, 124 S.Ct. 2562.
Then, “[o]nce this low threshold for relevance is met, the ‘Eighth Amendment requires that the jury be able to consider and give effect to’ a capital defendant’s mitigating evidence.” Id. at 285, 124 S.Ct. 2562 (quoting Boyde v. California, 494 U.S. 370, 377-78, 110 S.Ct. 1190, 108 L.Ed.2d 316 (1990) (citing Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973 (1978); Eddings v. Oklahoma, 455 U.S. 104, 102 S.Ct. 869, 71 L.Ed.2d 1 (1982); Penry I, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989))); see also Payne v. Tennessee, 501 U.S. 808, 822, 111 S.Ct. 2597, 115 L.Ed.2d 720 (1991) (“We have held that a State cannot preclude the sen-tencer from considering ‘any relevant mitigating evidence’ that the defendant proffers in support of a sentence less than death ... [VJirtually no limits are placed on the relevant mitigating evidence a capital defendant may introduce concerning his own circumstances.” (quoting Eddings, 455 U.S. at 114, 102 S.Ct. 869)).
The Court emphasized that, in assessing the relevance of mitigating evidence, a reviewing court should not weigh the severity or sufficiency of the evidence, except
insofar as evidence of a trivial feature of the defendant’s character or the circumstances of the crime is unlikely to have any tendency to mitigate the defendant’s culpability. See Skipper [v. South Carolina, 476 U.S. 1,] 7, n. 2, 106 S.Ct. 1669, 90 L.Ed.2d 1 (“We do not hold that all facets of the defendant’s ability to adjust to prison life must be treated as relevant *302and potentially mitigating. For example, we have no quarrel with the statement ... that ‘how often [the defendant] will take a shower’ is irrelevant to the sentencing determination^”).] However, to say that only those features and circumstances that a panel of federal appellate judges deems to be “severe” (let alone “uniquely severe”) could have such a tendency is incorrect. Rather, the question is simply whether the evidence is of such a character that it “might serve ‘as a basis for a sentence less than death,’ ” Skipper, [476 U.S.] at 5, 106 S.Ct. 1669.
Tennard, 542 U.S. at 286-87, 124 S.Ct. 2562.
The Court concluded:
the Fifth Circuit’s screening test has no basis in our precedents and, indeed, is inconsistent with the standard we have adopted for relevance in the capital sentencing context. We therefore hold that the Fifth Circuit assessed Tennard’s Penry claim under an improper legal standard. Cf. Miller-El v. Cockrell, 537 U.S. [322, 341, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003)] (holding, on certio-rari review of the denial of a COA, that the Fifth Circuit had applied an incorrect standard by improperly merging the requirements of two statutory sections).
Tennard, 542 U.S. at 287, 124 S.Ct. 2562.
Although the decision in Tennard principally focused on rejecting the “constitutional-relevance” standard, the Court also indicated that Tennard’s evidence may have had relevance beyond the scope of the special issues, and that a jury might have been precluded from giving effect to that aspect of Tennard’s mitigating evidence. The Court explained that a COA should have issued because
[t]he relationship between the special issues and Tennard’s low IQ evidence has the same essential features as the relationship between the special issues and Penny’s mental retardation evidence. Impaired intellectual functioning has mitigating dimension beyond the impact it has on the individual’s ability to act deliberately. See Penry I, 492 U.S. at 322, 109 S.Ct. 2934. A reasonable jurist could conclude that the jury might well have given Tennard’s low IQ evidence aggravating effect in considering his future dangerousness ....
Id. at 288-89, 124 S.Ct. 2562.
In its most recent pronouncement on the Penry issue, the Supreme Court in Smith v. Texas, 543 U.S. 37, 125 S.Ct. 400, 160 L.Ed.2d 303, once again reiterated that, to comply with the Eighth Amendment, a capital sentencing scheme must give full effect to all of a defendant’s mitigating evidence. In a per curiam opinion issued shortly after Tennard, the Court reversed the Texas Court of Criminal Appeals’ denial of state habeas relief, holding that the Texas special issues and a supplemental nullification instruction similar to the one at issue in Penry II did not give full effect to the petitioner’s mitigating evidence of the petitioner’s (1) learning disabilities; (2) low I.Q. scores; and (3) childhood abuse and troubled upbringing.
First, the Court held that, in light of Tennard, the Texas Court of Criminal Appeals erred when it relied on the Fifth Circuit’s “constitutional-relevance” test to dispose of the petitioner’s Penry claim. Second, the Court held that, under its precedent, the Texas Court of Criminal Appeals erred when it held that the special issues and nullification instruction gave sufficient mitigating effect to the petitioner’s mitigating evidence. The Court, reviewing its case law, stressed that “[i]n Penry v. Johnson, 532 U.S. 782, 121 S.Ct. 1910, 150 L.Ed.2d 9 (2001) (Penry II), we held a similar ‘nullification instruction’ con*303stitutionally inadequate because it did not allow the jury to give [full consideration and full effect to mitigating circumstances’ in choosing the defendant’s appropriate sentence. Id. at 797, 121 S.Ct. 1910 (quoting Johnson v. Texas, 509 U.S. 350, 381, 113 S.Ct. 2658, 125 L.Ed.2d 290 (1993) (O’CONNOR, J., dissenting)).” Smith, 543 U.S. at 38, 125 S.Ct. 400. The Smith Court therefore once again reaffirmed that the standard is full consideration and full effect.
The State’s contention that Smith and Penry II are inapposite to the instant case because they involved a nullification instruction is not well taken. As we explained above, the nullification instruction was not an adequate solution to the problem the Court identified in Penry I— namely, that the jurors could not give Penny’s mitigating evidence full effect through the special issues. Penry II, 532 U.S. at 797, 121 S.Ct. 1910 (“[T]he key under Pen-ry I is that the jury be able to ‘consider and give effect to [a defendant’s mitigating] evidence in imposing sentence.’ ” (emphasis and alteration in original) (quoting Penry I, 492 U.S. at 319, 109 S.Ct. 2934)). Accordingly, the focus of our inquiry is not whether there was a nullification instruction, but whether the procedure, whatever it was, allowed the jury to express its reasoned moral response to the defendant’s mitigating evidence. See id. And the standard for making that determination is whether there is a reasonable likelihood that the procedure precluded the jury from giving full consideration and full effect to the defendant’s mitigating evidence.
This review of the relevant Supreme Court case law therefore establishes that, at the time Nelson’s conviction became final in 1994, the clearly established law as announced by the Supreme Court was a full-effect standard. The Penry II Court left no doubt that full effect was the applicable standard, or that this was the standard that applied in Penry I. The debate has long since been over. Today, we make clear that we are following the Supreme Court’s directive and applying the standard it articulated; i.e., whether there is a reasonable likelihood that the special issues precluded the jury from giving full consideration and full effect to the defendant’s mitigating evidence, including evidence that has mitigating relevance outside the scope of the special issues because it speaks to a defendant’s moral culpability. This standard was “dictated by” the Supreme Court’s earlier decisions in Eddings and Lockett, see Penry I, 492 U.S. at 317, 109 S.Ct. 2934, and Graham and Johnson are not to the contrary. Moreover, the Court’s most recent decisions in Tennard and Smith reaffirm that this standard was clearly established federal law at the time Nelson’s conviction became final. Accordingly, we turn to the question presented in this case' — whether the state court’s determination that the Texas capital-sentencing scheme was constitutional as applied in Nelson’s case was contrary to or an unreasonable application of clearly established law as announced by the Supreme Court.
C. Application of Clearly Established Federal Laiv to Nelson’s Case
The Texas Court of Criminal Appeals concluded that the special issues were constitutional as applied to Nelson. Because there is a reasonable likelihood that the jury was precluded from giving full effect to Nelson’s mitigating evidence, we hold that the Texas Court of Criminal Appeals’ determination was an unreasonable application of clearly established law as announced by the Supreme Court.
1. Nelson’s Mitigating Evidence
The parties agree that, at the punishment phase of the trial, Nelson presented *304the following mitigating evidence of: (1) an abusive childhood; (2) substance abuse; (3) troubled relationships with his brother and with women; (4) having had a child out of wedlock with whom he was not permitted to have a relationship; and (5) a mental disorder. Specifically, Nelson offered the testimony of his father, who described in great detail the emotional abuse and rejection that Nelson suffered at the hands of his mother while he was growing up. Nelson’s father explained that Nelson was the second of two boys, and Nelson’s mother, who had always wanted a girl, rejected Nelson from birth, refusing to care for him, “change him or feed him [or] anything.” After Nelson’s parents separated when Nelson was fourteen years old, his mother completely abandoned him, leaving and refusing to take him with her.
Nelson also presented testimony from Dr. John Hickman, a psychiatrist who personally interviewed and assessed Nelson. Dr. Hickman testified extensively about the symptoms of borderline personality disorder, which can manifest themselves in “psychotic outburst[s]” and a “lack of impulse control.” According to Dr. Hickman, a person with borderline personality disorder has little insight into his own illness and may “periodically go through an outburst of feelings which can become very violent, very destructive,” even though he exhibits normal behavior “75 to 80 percent” of the time. Dr. Hickman noted that Nelson in particular experiences “a lot of impulse and a lot of raw energy and anger ... [that] he has no [insight] into whatsoever” as a result of his borderline personality disorder.
He further explained that borderline personality disorder can be especially “severe” in eases of maternal abandonment, and, in this case, Nelson’s abusive upbringing and rejection by his mother engendered a “rage toward women” that was evidenced by the nature of the crime that he committed. Dr. Hickman observed that Nelson’s borderline personality disorder was a consequence of growing up in a home where Nelson did not learn to control his anger and where he was subjected to psychologically abusive treatment by his mother, who told him that “he couldn’t do anything right” and that “she didn’t want him.” In Dr. Hickman’s judgment, at the time he committed the crime, Nelson “had a psychotic outburst” and was under the influence of “either a mental or physical form of duress” resulting from “his physical and psychological makeup.” Dr. Hickman also stated that, in addition to being “psychologically abused” by his mother, Nelson had “some family history which indicates disregard and abuse for women” and that “it is almost as if he is trained to be that way.” Additionally, Dr. Hickman noted that Nelson’s substance abuse likely exacerbated the effects of Nelson’s borderline personality disorder, describing “eruptive episodes, generally influenced by alcohol or cocaine, where all that primitive impulse comes out,” which were “guaranteed to be self-destructive.” In sum, Dr. Hickman observed that Nelson “has a morass of anger, hostility, given the combination of a borderline personality, given stress factors, given alcohol, given cocaine, all hell is going to break loose with him.”
Although Dr. Hickman testified that borderline personality disorder can be treated in some cases, he indicated that borderline personality disorder is difficult to treat because persons with borderline personality disorder do not want to “admit they are weak and vulnerable” and often refuse to undergo therapy. Dr. Hickman estimated that in Nelson’s case, it could take at least a year just to break down Nelson’s “defenses” and convince him to participate in treatment; after that, Nelson would require “long psychotherapy— and I’m talking about two to five years. *305That is standard for borderline. And ... medication.” Dr. Hickman emphasized that this intensive psychotherapy would require “two or three times a week with ... a therapist that can work with him” in addition to “the proper drug medication” and “a strict environment” where Nelson could “learn internal controls.” Dr. Hickman noted that, even with such treatment, he could not guarantee Nelson’s success, and “if he doesn’t get treatment, I think we can predict dangerousness.”
2. The Special Issues as Applied to Nelson’s Mitigating Evidence
As a threshold matter, the State contends that Penry and its progeny apply only to a very narrow set of cases in which the mitigating evidence is “double-edged,” i.e., has both aggravating and mitigating effect, and the future-dangerousness special issue gives the evidence only aggravating effect. Thus, according to the State, a Penry analysis in this case is not necessary. We disagree. The Supreme Court has never limited the applicability of Penry- — either explicitly or implicitly — to cases involving “double-edged” mitigating evidence. In Penry I, the Court’s observation that Penry’s evidence of mental illness was “two-edged” was just one of many reasons that the special issues were inadequate vehicles to give Penry’s evidence full mitigating effect; it was not the determining factor. See Penry I, 492 U.S. at 324, 109 S.Ct. 2934 (listing the “two-edged sword” nature of Penry’s evidence as one of a number of reasons that the future-dangerousness issue could not give Penry’s evidence full mitigating effect). Justice O’Connor’s dissent in Johnson explains that placing too much weight on the Court’s description of Penry’s evidence as “two-edged” mischaracterizes the Penry I Court’s reasoning:
The second special issue was not inadequate because evidence worked only against Penry; it was inadequate because it did not allow the jury to give full effect to Penry’s mitigating evidence. Penry, 492 U.S. at 323, 109 S.Ct. 2934. Our discussion of the third special issue- — whether the defendant’s conduct was unreasonable in response to the provocation — also focused on the inability of a juror to express the view that Penry lacked “the moral culpability to be sentenced to death” in answering the question. Id. at 324-25, 109 S.Ct. 2934. The point of Penry is clear: A death sentence resulting from application of the Texas special issues cannot be upheld unless the jurors are able to consider fully a defendant’s mitigating evidence. Accord, id. at 355, 109 S.Ct. 2934 (SCALIA, J., concurring in part and dissenting in part) (The Court today holds that “the constitutionality turns on whether the [special] questions allow mitigating factors not only to be considered ..., but also to be given effect in all possible ways, including ways that the questions do not permit”).
See Johnson, 509 U.S. at 386, 113 S.Ct. 2658 (O’Connor, J., dissenting). Indeed, Chief Justice Rehnquist dissented from the Court’s decision in Tennard arguing that Tennard’s evidence was not “two-edged.” Tennard, 542 U.S. at 292-93, 124 S.Ct. 2562 (“In either case — contrary to Penry I — the evidence could be given mitigating effect in the second special issue. In short, low intelligence is not the same as mental retardation and does not necessarily create the Penry I ‘two-edged sword.’ ”). A majority of the Court declined to accept that argument in Tennard and, therefore, we cannot accept it here.
Further, the Court has indicated that Penry applies — or at least potentially could apply — in cases involving evidence that is not double-edged. See, e.g., Smith, 543 U.S. 37, 125 S.Ct. 400, 160 L.Ed.2d 303 *306(reversing the state court’s denial of habe-as relief because the special issues could not give full effect to mitigating evidence of low I.Q. and troubled upbringing); Tennard, 542 U.S. 274, 124 S.Ct. 2562, 159 L.Ed.2d 384 (holding that habeas petitioner was entitled to a COA on his Penry claim based on mitigating evidence of low I.Q. and impaired intellectual functioning). In short, the State urges this court to wrench one component of the Court’s reasoning in Penry I out of context and use it as a dispositive screening test in our assessment of Penry claims. In effect, the State asks this court to develop another “restrictive gloss on Penry similar to the “constitutional-relevance” test that the Court struck down in Tennard. Tennard, 542 U.S. at 283, 124 S.Ct. 2562. The Court has never used the consideration of whether evidence is double-edged as a single-issue screening test as the State urges us to do; and, given the Court’s strong admonition in Tennard, we decline to do so. Consequently, we turn now to the State’s alternative argument that Nelson’s evidence could be adequately considered through the two special issues.
a. Deliberateness Special Issue
Nelson’s mitigating evidence of borderline personality disorder and abandonment by his mother had relevance beyond the scope of the deliberateness special issue. As the Supreme Court observed in Pemy I, a reasonable juror could have concluded that, while the murder was deliberate, Nelson was less morally culpable as a result of his borderline personality disorder and abusive childhood than a murderer without such a mental illness and similar upbringing might have been. See Penry I, 492 U.S. at 323-24, 109 S.Ct. 2934; see also Skipper v. South Carolina, 476 U.S. 1, 13-14, 106 S.Ct. 1669, 90 L.Ed.2d 1 (1986) (Powell, J., concurring in judgment) (stating that evidence concerning a defendant’s “emotional history ... bear[s] directly on the fundamental justice of imposing capital punishment”). Because a major mitigating thrust of evidence of a mental disorder and an abusive childhood is that such afflictions could reduce an offender’s moral culpability, it is “reasonably likely” that a juror would not have been able to give full effect to his “reasoned moral judgment” regarding the full mitigating impact of Nelson’s evidence through the narrowly worded deliberateness instruction. See, e.g., Penry II, 532 U.S. at 797, 121 S.Ct. 1910; Penry I, 492 U.S. at 322, 109 S.Ct. 2934. Significantly, the Supreme Court has never held that the deliberateness issue alone is broad enough to give full effect to mitigating evidence that also bears on a defendant’s moral culpability; indeed the Court’s most recent opinion in Smith v. Texas suggests the contrary. There, the Court characterized Smith’s evidence as follows:
(1) he had been diagnosed with potentially organic learning disabilities and speech handicaps at an early age; (2) he had a verbal IQ score of 75 and a full IQ of 78 and, as a result, had been in special education classes throughout most of his time in school; (3) despite his low IQ and learning disabilities, his behavior at school was often exemplary; (4) his father was a drug addict who was involved with gang violence and other criminal activities, and regularly stole money from family members to support a drug addiction; and (5) he was only 19 when he committed the crime.
Smith, 543 U.S. at 41, 125 S.Ct. 400. Considering the nature of this evidence, the Court noted that, “just as in Penry II, the burden of proof on the State was tied by law to findings of deliberateness and future dangerousness that had little, if any*307thing, to do with the mitigation evidence petitioner presented.” Id. at 48, 125 S.Ct. 400. Likewise, Nelson’s mitigating evidence had relevance beyond the deliberateness special issue insofar as it bore on his moral culpability for the crime. Consequently, although the jury may have been able to give partial effect to this evidence through the deliberateness special issue, there is a reasonable likelihood that it was unable to give full effect to this evidence, because it had relevance beyond whether Nelson acted deliberately.
b. Future-Dangerousness Special Issue
Likewise, the future-dangerousness special issue cannot give Nelson’s evidence full mitigating effect. The jury heard conflicting evidence about the treatability of Nelson’s borderline personality disorder and about the efficacy of any possible treatment. According to the expert testimony, even assuming that Nelson’s borderline personality disorder were treatable, success would depend on many factors. Based on this evidence, the jury could have easily concluded that it was unlikely that Nelson would successfully complete treatment. The State’s expert, Dr. Grigson, testified that there was insufficient information to make a diagnosis of antisocial personality disorder, but repeatedly emphasized that “in [his] opinion there is no question whatsoever that [Nelson] will commit future acts of danger.” In contrast, Nelson’s expert, Dr. Hickman, diagnosed Nelson with borderline personality disorder. He further testified that, with treatment consisting of incarceration, two to five years of intensive psychotherapy two to three times a week, medication, and refraining from drug and alcohol abuse, Nelson may not be continuing threat. He opined that if Nelson did not receive treatment, he would pose a danger to society. He also explained that “the last thing a borderline wants to do is admit they are weak and vulnerable,” and thus borderline patients often resist treatment. Indeed, in its own closing, the State emphasized the strong possibility that Nelson would not receive the treatment he needed to keep his borderline personality disorder in check, and even if he did receive such treatment, there were no guarantees that the therapy would be effective -to prevent future violence:
Dr. Hickman said, if, if, if, if he is imprisoned long enough, if he undergoes psychotherapy, if he chooses to take his medication, and if he leaves dope and alcohol alone, then maybe, maybe he won’t be a future danger. Look at Special Issue Number Two, ladies and gentlemen. There is not an asterisk next to that, there is not something referring you down here that says if, if, if, if. We look at the defendant right now, and right now even their witness [said], yes, he may be a danger.
Based on the expert testimony at trial, the jury might have concluded that Nelson could be treated, and therefore, it could have given some effect to this mitigating evidence within the context of the future-dangerousness special issue. But if the jury concluded that the condition was not treatable or that treatment was improbable, as the State argued, it would necessarily have to answer “yes” to the special issue. Just as in Penry I and Penry II, it is likely that a juror considering Nelson’s evidence of borderline personality disorder would have felt that he could give the evidence only one possible effect via the future-dangerousness issue: Such a juror would have seen the evidence as only aggravating, because Nelson’s borderline personality disorder and the difficulty of treating it increase the likelihood that Nelson will act out violently again. Consequently, there would be no vehicle to give *308mitigating effect to his evidence of borderline personality disorder, i.e., no way for the jury to express its conclusion that even though he is likely to be dangerous in the future, his mental illness makes him unworthy of the death penalty. Cf. Penry I, 492 U.S. at 302, 109 S.Ct. 2934 (“[A] reasonable juror could well have believed that there was no vehicle for expressing the view that Penry did not deserve to be sentenced to death based upon his mitigating evidence”). And, also similar to Pen-ry I, the jury was likely precluded from interpreting the future-dangerousness issue in a way that gave effect to the major mitigating thrust of the evidence, that it tends to lessen Nelson’s moral culpability for the crime. See Penry I, 492 U.S. at 322-24, 109 S.Ct. 2934.
The State and the dissenting opinions of Chief Judge Jones and Judge Owen argue that the evidence at issue here is more comparable to the evidence of youth at issue in Johnson and Graham. Specifically, they contend that, because borderline personality disorder can be a “transient” condition like youth, a jury could believe that Nelson would be less dangerous in the future, thereby giving full mitigating effect to the evidence. We disagree. This argument erroneously analogizes evidence of youth and evidence of mental illness. The Supreme Court in Johnson held that the future-dangerousness issue could give effect to both mitigating aspects of youth — likelihood of future violent behavior and moral culpability — due to the uniquely transient nature of youth. See Johnson, 509 U.S. at 368, 113 S.Ct. 2658 (“The relevance of youth as a mitigating factor derives from the fact that the signature qualities of youth are transient; as individuals mature, the impetuousness and recklessness that may dominate in younger years can subside .... [T]here is ample room in the assessment of future dangerousness for a juror to take account of the difficulties of youth as a mitigating force in the sentencing determination.”); see also Eddings, 455 U.S. at 115, 102 S.Ct. 869 (“[Yjouth is more than a chronological fact. It is a time and condition of life when a person may be most susceptible to influence and to psychological damage.”). In this sense, the Supreme Court, which has never endorsed the extension of Johnson and Graham to treatable mental illness, has treated youth as sui generis, because it is a condition that is certain to pass.6 In contrast, as acknowledged by Nelson’s own expert witness, there was no guarantee that Nelson’s borderline personality disorder would diminish over time. Dr. Hickman noted that, although borderline personality disorder is treatable, success is by no means certain and is expressly conditioned on intensive therapy that, a juror could conclude, the Texas prison system is unlikely to provide. In fact, Dr. Hickman’s trial testimony indicated that, because of the severity of borderline personality disorder and patients’ common resistance to therapy, successful treatment is often the exception rather than the rule. *309Unlike a jury considering evidence of youth, therefore, a reasonable likelihood existed that a jury considering Nelson’s mitigating evidence of borderline personality disorder would have felt foreclosed from giving full mitigating effect to Nelson’s evidence of his disorder via the future-dangerousness issue. Thus, based on the principles announced in Penry I and its progeny, the future-dangerousness special issue, like the deliberateness special issue, provided a constitutionally insufficient vehicle to allow a jury to express its reasoned moral response and give full effect to Nelson’s mitigating evidence. The Texas Court of Criminal Appeals’s holding to the contrary is an unreasonable application of clearly established federal law as announced by the Supreme Court.
This is not simply a matter of disagreement with the state court’s conclusion that the jury could consider and give effect to Nelson’s mitigating evidence through the special-issues sentencing scheme. We are mindful that under AEDPA a federal court may not grant habeas relief simply because it disagrees with the state court’s resolution of an issue; it may grant relief only if the state court’s decision was contrary to or an unreasonable application of clearly established Supreme Court precedent. See 28 U.S.C. § 2254(d)(1). Indeed, Chief Judge Jones’s dissent invokes this standard, asserting that our approach to the future-dangerousness issue improperly “relies upon a string of hypotheticals to create [a] Penry violation” and adopts an “attenuated theory of the jury deliberations [that] extends Penry I far beyond its intended boundaries.” Chief Judge Jones’s Dissent at 21 n. 19; see also Judge Owen’s Dissent. But rather than extending the reach of Penry I or any other case in violation of the AEDPA standard of review, our approach merely follows the Supreme Court’s longstanding directive to determine only “whether the evidence is of such a character that it ‘might serve as a basis for a sentence less than death,’ ” which was clearly established federal law at the time that Nelson’s conviction became final. Tennard, 542 U.S. at 287, 124 S.Ct. 2562 (quoting Skipper, 476 U.S. at 5, 106 S.Ct. 1669) (emphasis added).
In contrast, the alternative approach, upon which the dissenting opinions of Chief Judge Jones and Judge Owen base their conclusions that the Texas Court of Criminal Appeals did not unreasonably apply clearly established federal law, would require us, sitting as a federal appellate habeas court, to weigh the evidence presented at sentencing in a manner that the Supreme Court in Tennard held was an unreasonable application of clearly established federal law at least as far back as 1991. See id. at 286-87, 124 S.Ct. 2562. Like Nelson’s case, Tennard reached the Supreme Court on federal habeas review and was governed by the AEDPA standard. As noted above, we measure clearly established federal law for AEDPA purposes as of the date that the defendant’s conviction became final. While Tennard’s conviction became final in 1991, Nelson’s conviction did not become final until 1994. See supra note 4. Therefore, the principles that the Supreme Court in Tennard held had been clearly established in 1991 were certainly clearly established by the time that Nelson’s conviction became final in 1994. Athough Chief Judge Jones and Judge Owen correctly recite the AEDPA standard in their dissenting opinions, they simply fail to accept that Tennard — which stood for the propositions that (1) a reviewing court may not reweigh or reassess the mitigating evidence presented at sentencing, and (2) a jury must be able to give effect to the impact of that mitigating evidence on the defendant’s moral culpability via the special issues — also set forth the *310federal law that was clearly established for the purposes of Nelson’s case.
Specifically, the dissenting opinions of Chief Judge Jones and Judge Owen run afoul of Tennard by assuming that the jury in Nelson’s case found that Nelson’s borderline personality disorder was treatable, and that the Texas Court of Criminal Appeals would therefore not have acted unreasonably in treating it as akin to the mitigating evidence of youth at issue in Graham and Johnson. However, we know from the record only that the jury determined that Nelson was a future danger after hearing conflicting expert testimony about whether he suffered from borderline personality disorder and, if so, whether it could be treated. Despite the purportedly definitive reading of the record contained in Chief Judge Jones’s dissent, we cannot be certain of the precise reasons for the jury’s future-dangerousness determination. Instead, we know that the jury could have arrived at its conclusion for any of the following reasons: (1) the jury believed that Nelson suffered from borderline personality disorder but that the disorder was not treatable; (2) the jury believed that Nelson suffered from borderline personality disorder that was treatable but that some other factor rendered Nelson a future danger; or (3) the jury did not believe that Nelson actually suffered from borderline personality disorder. To conclude that the mental illness at issue was treatable in the face of these multiple possibilities, the dissenting opinions of Chief Judge Jones and Judge Owen reassess and reweigh the evidence presented at sentencing, even though we, sitting as a federal appellate habeas court, have no way of knowing why the jury determined that Nelson was a future danger.7
Weighing the evidence in this manner violates the Supreme Court’s express admonition in Tennard that we not substitute our own interpretation of the evidence for that of the jury or assess the strength of the mitigating evidence presented except “insofar as evidence of a trivial feature of the defendant’s character or the circumstances of the crime is unlikely to have any tendency to mitigate the defendant’s culpability.” Tennard, 542 U.S. at 286, 124 S.Ct. 2562. Just as the Supreme Court in Penry I made no determination as to whether the jury actually believed that Penry was mentally retarded based on the conflicting trial evidence, we may not conduct an independent review of the conflicting evidence in this ease to make a determination as to whether the jury actually believed that Nelson’s mental illness was treatable. In short, under Tennard, which *311clarified the clearly established law in this area as of 1991, we may not graft a treata-bility test based on our view of the strength of the evidence onto the low relevance threshold as the dissenting opinions of Chief Judge Jones and Judge Owen propose, and neither may the Texas Court of Criminal Appeals. Rather, the only question we may ask regarding the jury’s interpretation of the mitigating evidence presented at trial is “simply whether the evidence is of such a character that it ‘might serve as a basis for a sentence less than death.’ ” Id. at 287, 124 S.Ct. 2562 (quoting Skipper, 476 U.S. at 5, 106 S.Ct. 1669) (emphasis added).
Further, the Supreme Court has made it clear in Boyde and in Johnson (both issued before Nelson’s conviction became final) that once the low relevance threshold is satisfied, rather than inquiring into or second guessing the jury’s interpretation of the trial evidence, all a court must determine is whether a reasonable likelihood exists that the jury applied the instructions in a manner that precluded it from giving effect to the defendant’s mitigating evidence as it pertains to the defendant’s moral culpability. In the instant case, given the conflicting testimony regarding the treatability of Nelson’s mental illness, there is certainly a reasonable likelihood that the jury felt precluded from giving full effect to the impact of the evidence on Nelson’s moral culpability via the future-dangerousness issue because it found that Nelson’s illness could not be treated. See Johnson, 509 U.S. at 367, 113 S.Ct. 2658 (explaining that the Boyde “reasonable likelihood standard does not require that the defendant prove that it was more likely than not that the jury was prevented from giving effect to the evidence”).
Therefore, rather than “extending] Penny I far beyond its intended boundaries, without instructions from the Supreme Court,” Chief Judge Jones’s Dissent at 346 n. 19, our approach is firmly grounded in Supreme Court precedent and consistent with the AEDPA standard of review. The alternative upon which the dissenting opinions of Chief Judge Jones and Judge Owen rely to affirm the state court’s denial of habeas relief in this case — that we scour the trial record for evidence of treatability and substitute our interpretation of the evidence for that of the jury’s — is not merely incorrect, but is an unreasonable application of clearly established federal law as announced by the Supreme Court. See Tennard, 542 U.S. at 288-89, 124 S.Ct. 2562; see also Smith, 543 U.S. at 38, 125 S.Ct. 400; Penny II, 532 U.S. at 803, 121 S.Ct. 1910; Penry I, 492 U.S. at 323, 109 S.Ct. 2934; Skipper, 476 U.S. at 5, 106 S.Ct. 1669.
This case is therefore different from the Supreme Court’s recent decision in Brown v. Payton, 544 U.S. 133, 147, 125 S.Ct. 1432, 161 L.Ed.2d 334 (2005), which Judge Clement discusses in her dissenting opinion. In Payton, the Supreme Court reversed the Ninth Circuit’s grant of habeas relief to a death-row petitioner who challenged the constitutionality of California’s “factor (k)” jury instruction, concluding that the Ninth Circuit did not give proper deference to the state court’s decision. Specifically, the Court held that “[i]t was not unreasonable for the state court to determine that the jury most likely believed that the evidence in mitigation, while within the reach of the factor (k) instruction, was simply too insubstantial to overcome the arguments for imposing the death penalty.” Id. (emphasis added). In Payton, the state court held that the mitigating evidence of the defendant’s religious conversion fell within the reach of the catch-all instruction directing the jury to consider “ ‘[a]ny other circumstance which extenuates the gravity of the crime even though it is not a legal excuse for the *312crime,’ ” although the prosecutor argued to the jury that it could not consider this evidence. Payton, 544 U.S. at 137, 125 S.Ct. 1432 (alteration in original) (quoting CaLPenal Code Ann. § 190.3 (West 1988)). In reversing the Ninth Circuit’s determination that the state court erred in denying habeas relief, the Supreme Court emphasized that the state court’s holding was a reasonable interpretation of its prior decision in Boyde, 494 U.S. 370, 110 S.Ct. 1190, 108 L.Ed.2d 316, in which the Court upheld the validity of the factor (k) instruction in similar circumstances. See Payton, 544 U.S. at 144, 125 S.Ct. 1432 (“As the California Supreme Court recognized, like in Boyde, for the jury to have believed it could not consider Payton’s mitigation evidence, it would have had to believe that the penalty phase served virtually no purpose at all.”). Accordingly, the Ninth Circuit had deviated from the deferential AEDPA standard when it reversed the state court’s determination.
Nevertheless, Judge Clement’s dissenting opinion, which relies on Payton to conclude that this court should defer to the Texas Court of Criminal Appeals’s denial of habeas relief, fails to recognize that “[t]he [AEDPA] standard is demanding but not insatiable; ... ‘[d]eference does not by definition preclude relief.’ ” Miller-El v. Dretke, 545 U.S. 231, 125 S.Ct. 2317, 2325, 162 L.Ed.2d 196 (2005) (third alteration in original) (quoting Miller-El v. Cockrell, 537 U.S. 322, 340, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003)). In contrast to the circumstances at issue in Payton, Nelson’s mitigating evidence clearly has relevance beyond the issues of deliberateness and future dangerousness under Penry I and its progeny. If the jury concluded that Nelson was likely to be dangerous in the future based on his mental disorder and abusive childhood, but also concluded that this evidence rendered him less morally culpable, it had no way to give effect to the mitigating aspect of that evidence through the two special issues. See Smith, 543 U.S. at 38, 125 S.Ct. 400; Tennard, 542 U.S. at 288-89, 124 S.Ct. 2562; Penry II, 532 U.S. at 803, 121 S.Ct. 1910; Penry I, 492 U.S. at 323, 109 S.Ct. 2934. Moreover, Tennard precludes a reviewing court from reweighing the evidence presented at trial to determine whether the alleged mitigating circumstance is treatable and therefore transient. Tennard, 542 U.S. at 286-87, 124 S.Ct. 2562. Thus, unlike the state court’s determination in Payton, where the Supreme Court in Boyde had previously held that the challenged California instruction was broad enough to allow the jury to consider the impact of the mitigating evidence on the defendant’s moral culpability, it was unreasonable for the Texas Court of Criminal Appeals in this case to conclude that Nelson’s mitigating evidence was within the reach of the jury through the narrow special issues, given the law clearly established by the Supreme Court in Penry I and its progeny.
Finally, in support of its argument that evidence of a potentially treatable mental disorder should be analyzed similarly to the Court’s consideration of youth, the State relies on Fifth Circuit case law that has erroneously interpreted Penry as requiring that the mitigating evidence be given only “some effect.” Specifically, it relies on this court’s opinion in Lucas v. Johnson, 132 F.3d 1069 (5th Cir.1998), in which the panel held that the special issues gave constitutionally sufficient effect to Lucas’s evidence of schizophrenia coupled with a troubled upbringing. See id. at 1083 (“[The] prospect of medical treatment placed the evidence of his mental illness and abusive childhood within ‘the effective reach of the sentenced as a potential mitigating factor with respect to the second issue, that is, the jury could *313have considered whether, in an institutional setting, the probability that Lucas posed as a future danger to society was not so great as to merit imposition of the death sentence.”)• In reaching this conclusion, Lucas cited the Supreme Court’s decisions in Johnson and Graham for the proposition that “Penry’s application has since been limited to that narrow class of situations in which the petitioner’s mitigating evidence was placed beyond the jury’s effective reach,” and that the evidence in that case was within the jury’s effective reach, because the jury could have given it partial effect. Id- at 1082. As explained above, the Supreme Court has clearly held that the standard is full effect. Thus, continued reliance on the partial-effect methodology is erroneous, because that standard fails to take into account, as Penry I and its progeny require, a jury’s inability to give mitigating effect to a defendant’s moral culpability via the future-dangerousness issue. See Smith, 543 U.S. at 38, 125 S.Ct. 400; Tennard, 542 U.S. at 288-89, 124 S.Ct. 2562; Penry II, 532 U.S. at 803, 121 S.Ct. 1910; Penry I, 492 U.S. at 323, 109 S.Ct. 2934. Moreover, and most importantly, AEDPA requires us to determine whether the state court unreasonably applied “clearly established federal law as announced by the Supreme Court,” not by the Fifth Circuit. 28 U.S.C. § 2254(d)(1). Accordingly, to the extent that this court’s cases have applied a less-than-full-effeet standard to Penry claims in the past, i.e., to the extent that past cases failed to account for the jury’s ability to give effect to the impact of mitigating evidence on a defendant’s moral culpability via the special issues, those cases were based on an erroneous interpretation of Supreme Court precedent. See Smith, 543 U.S. at 38, 125 S.Ct. 400 (holding that a sentencing scheme that fails to “give full consideration and full effect to mitigating circumstances in choosing the defendant’s appropriate sentence” is “constitutionally inadequate” under Penry I and its progeny) (internal quotation marks and citations omitted).
8. Sufficiency of Nelson’s Mitigation Evidence
We also reject the argument that Nelson’s evidence of borderline personality disorder is insufficient to warrant relief based on Penry. The Supreme Court has recognized that
gravity has a place in the relevance analysis, insofar as evidence of a trivial feature of the defendant’s character or the circumstances of the crime is unlikely to have any tendency to mitigate the defendant’s culpability. See Skipper [v. South Carolina, 476 U.S. 1,] 7, n. 2, 106 S.Ct. 1669, 90 L.Ed.2d 1 (“We do not hold that all facets of the defendant’s ability to adjust to prison life must be treated as relevant and potentially mitigating. For example, we have no quarrel with the statement .. that ‘how often [the defendant] will take a shower’ is irrelevant to the sentencing determination[.”).].
Tennard, 542 U.S. at 286-87, 124 S.Ct. 2562. The Tennard Court was discussing evidence that had no probative worth in the jury’s consideration of a defendant’s moral culpability, not evidence that the jury may choose to believe or disbelieve. In contrast, the strength of Nelson’s evidence of borderline personality disorder and abusive childhood “goes to the credibility of [Nelson’s] mitigation evidence, which should be judged by the jury in answering effective supplemental instructions addressing the mitigation evidence.” Blue v. Cockrell, 298 F.3d 318, 322 (5th Cir.2002), abrogated on other grounds by Tennard, 542 U.S. 274, 124 S.Ct. 2562, 159 L.Ed.2d 384. Further, any argument that this court should dispose of Nelson’s Pen-ry claim on grounds that the evidence is *314insufficient endorses precisely the type of judicial evidence-weighing that the Court in Tennard expressly warned against:
[T]o say that only those features and circumstances that a panel of federal appellate judges deems to be “severe” (let alone “uniquely severe”) could have such a tendency is incorrect. Rather, the question is simply whether the evidence is of such a character that it “might serve ‘as a basis for a sentence less than death,’ ” Skipper, [476 U.S.] at 5, 106 S.Ct. 1669.
Tennard, 542 U.S. at 286-87, 124 S.Ct. 2562. Nowhere does Tennard prescribe (or even allow for) a balancing test that weighs the strength of the mitigating evidence against that of the aggravating evidence. Such reasoning runs afoul of the low relevance standard that the Court emphasized in Tennard, i.e., any tendency to mitigate the defendant’s culpability, and comes perilously close to applying a heightened-relevance test similar to the one that the Court struck down in Ten-nard. Accordingly, we also reject this argument.
A Harmless Error
Finally, we reject the State’s argument that any Penry error in this case is subject to harmless-error analysis under Brecht v. Abrahamson, 507 U.S. 619, 622-23, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993), which applies to error that is “amenable to harmless-error analysis because it ‘may ... be quantitatively assessed in the context of other evidence presented in order to determine [the effect it had on the trial].’ ” Id. at 629, 113 S.Ct. 1710 (omission and alteration in original) (quoting Arizona v. Fulminante, 499 U.S. 279, 309, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991)). The State advances this harmless-error theory for the very first time on en banc rehearing in a discussion that consumes less than a page of its brief; it did not argue the applicability of harmless error before this court during Nelson’s original habeas appeal, before the Supreme Court on certiorari review, or before this court when we initially reconsidered Nelson’s habeas appeal on remand in light of Ten-nard. It was not until a concurring panel member in the most recent Nelson panel opinion suggested that Brecht might be applicable that the State argued harmless error in its en banc brief. The State’s failure to argue this point prior to now is understandable because the Supreme Court has never applied a harmless-error analysis to a Penry claim or given any indication that harmless error might apply in its long line of post -Furman cases addressing the jury’s ability to give full effect to a capital defendant’s mitigating evidence. See generally Tennard, 542 U.S. 274, 124 S.Ct. 2562, 159 L.Ed.2d 384; Penry II, 532 U.S. 782, 121 S.Ct. 1910, 150 L.Ed.2d 9; Penry I, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256; Eddings, 455 U.S. 104, 102 S.Ct. 869, 71 L.Ed.2d 1; Lockett, 438 U.S. 586, 98 S.Ct. 2954, 57 L.Ed.2d 973. Indeed, the Penry II Court applied the Brecht harmless-error test to Penry’s claim that the prosecution’s use of a psychiatrist’s report violated his Fifth Amendment rights, see Penry II, 532 U.S. at 795, 121 S.Ct. 1910. Conspicuously absent from the discussion regarding Penry’s Eighth Amendment claim, however, is any mention of the harmless-error test in either the majority or the dissenting opinions.
Implicit in the Court’s failure to apply harmless error in cases where the jury has been precluded from giving effect to a defendant’s mitigating evidence is the recognition that a Penry error deprives the jury of a “vehicle for expressing its ‘reasoned moral response to the defendant’s background, character, and crime,’ ” which precludes it from making “ ‘a reli*315able determination that death is the appropriate sentence.’ ” Penry II, 532 U.S. at 797, 121 S.Ct. 1910 (quoting Penry I, 492 U.S. at 328, 319, 109 S.Ct. 2934) (internal quotation marks omitted) (emphasis added). This reasoned moral judgment that a jury must make in determining whether death is the appropriate sentence differs from those fact-bound judgments made in response to the special issues. It also differs from those at issue in cases involving defective jury instructions in which the Court has found harmless-error review to be appropriate. Cf. Neder v. United States, 527 U.S. 1, 8-15, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999) (applying harmless-error review where the jury instructions omitted an element of the offense, reasoning that, given the evidence presented, the verdict would have been the same had the jury been properly instructed); Johnson v. United States, 520 U.S. 461, 468-69, 117 S.Ct. 1544, 137 L.Ed.2d 718 (1997) (applying harmless-error review where the jury instructions omitted the materiality element of the perjury charge, noting that the error did not warrant correction in light of the “overwhelming” and “uncon-troverted” evidence supporting materiality). Given that the entire premise of the Penry line of cases rests on the possibility that the jury’s reasoned moral response might have been different from its answers to the special issues had it been able to fully consider and give effect to the defendant’s mitigating evidence, it would be wholly inappropriate for an appellate court, in effect, to substitute its own moral judgment for the jury’s in these cases. See Tennard, 542 U.S. at 286-87, 124 S.Ct. 2562 (“[T]o say that only those features and circumstances that a panel of federal appellate judges deems to be ‘severe’ (let alone ‘uniquely severe’) could have such a tendency [to serve as a basis less than death] is incorrect. Rather, the question is simply whether the evidence is of such a character that it ‘might serve “as a basis for a sentence less than death” ’ ”) (quoting Skipper, 476 U.S. at 5, 106 S.Ct. 1669); cf. Sullivan v. Louisiana, 508 U.S. 275, 281, 113 S.Ct. 2078, 124 L.Ed.2d 182 (1993) (refusing to apply harmless error where the jury was improperly instructed on the burden of proof at the guilt/innocence phase, noting that “the essential connection to a ‘beyond a reasonable doubt’ factual finding cannot be made where the instructional error consists of a misdescription of the burden of proof, which vitiates all the jury’s findings. A reviewing court can only engage in pure speculation — its view of what a reasonable jury would have done. And when it does that, ‘the wrong entity judge[s] the defendant guilty’ ”) (quoting Rose v. Clark, 478 U.S. 570, 578, 106 S.Ct. 3101, 92 L.Ed.2d 460 (1986)).
Therefore, given the Supreme Court’s refusal to allow an appellate court to substitute its own moral judgment for a moral judgment that the jury was unable to make in a Penry case, we decline to do so now.8
III. CONCLUSION
At the time that Nelson’s conviction became final, the Supreme Court had clearly established that the relevant inquiry is whether there was a reasonable likelihood *316that the jury would interpret the Texas special issues in a manner that precluded it from fully considering and giving full effect to all of the defendant’s mitigating evidence. For the foregoing reasons, we conclude that there is a reasonable likelihood that the jury was precluded from giving full consideration and full effect to Nelson’s mitigating evidence via the Texas special issues; therefore the state court’s determination that the special issues were constitutional as applied to Nelson’s case was unreasonable. Accordingly, we REVERSE the district court’s denial of habe-as relief and REMAND with instructions to grant the writ of habeas corpus.
. Although the Texas legislature amended the special issues sentencing scheme in 1991, Nelson was sentenced under the pre-amendment version of the special issues. In some cases, the jury was also given a third special issue addressing provocation. Nelson’s jury did not receive a provocation instruction, and therefore we do not address that aspect of the pre-amendment special issues sentencing scheme here.
. The "constitutional-relevance” test required that petitioner’s evidence show "(1) a 'uniquely severe permanent handicap[] with which the defendant was burdened through no fault of his own' and (2) that the criminal act was attributable to this severe permanent condition.” Davis v. Scott, 51 F.3d 457, 460-61 (5th Cir.1995) (citations omitted) (alteration in original). As discussed below, in Tennard v. Dretke, 542 U.S. at 275, 124 S.Ct. 2562, the Supreme Court rejected the constitutional relevance test and reaffirmed that the standard for relevance is "any tendency to make the existence of any fact that is of consequence to the determination of the action more probable or less probable than it would be without the evidence.” Id. (internal quotations omitted).
. Chief Judge Jones authored an opinion; Judge Stewart concurred in the judgment only; and Judge Dennis filed a concurring opinion. Nelson v. Dretke, 442 F.3d 282.
. We note that the Supreme Court granted certiorari in Brewer v. Dretke, 442 F.3d 273 (5th Cir.), cert. granted, - U.S. -, 127 S.Ct. 433, 166 L.Ed.2d 307 (2006), Cole v. Dretke, 418 F.3d 494 (5th Cir.2005), cert. granted sub nom. Abdul-Kabir v. Quarterman, - U.S. -, 127 S.Ct. 432, 166 L.Ed.2d 307 (2006), and Ex Parte Smith, 185 S.W.3d 455 (Tex.Crim.App.) cert. granted sub nom. Smith v. Texas,-U.S. -, 127 S.Ct. 377, 166 L.Ed.2d 265 (2006).
. Tennard’s conviction became final when the Supreme Court denied certiorari on his direct appeal on June 28, 1991. Tennard v. Texas, 501 U.S. 1259, 111 S.Ct. 2914, 115 L.Ed.2d 1077 (1991). Under AEDPA, therefore, the Supreme Court's duty in Tennard was to determine whether the Texas Court of Criminal Appeals unreasonably applied federal law that was clearly established as of June 28, 1991. In light of AEDPA’s mandate, the Tennard Court's insistence that a jury be able to consider and give effect to evidence with mitigating relevance to a defendant’s moral culpability in addition to the .special issues indicates that the “full-effect" standard was well in place by 1991; indeed, as explained above, this standard, which is the same standard that the Court applied in Penry I, was "dictated by Eddings and Lockett." Penry I, 492 U.S. at 317, 109 S.Ct. 2934. Nelson’s conviction became final in 1994, three years after Ten-*301nard’s. Nelson v. Texas, 510 U.S. 1215, 114 S.Ct. 1338, 127 L.Ed.2d 686 (1994).
. The sui generis nature of youth in the death penalty context is perhaps best evidenced by the Supreme Court's categorical holding in Roper v. Simmons, 543 U.S. 551, 125 S.Ct. 1183, 161 L.Ed.2d 1 (2005), that the Constitution prohibits the execution of persons who were under eighteen years of age at the time of their crime. See id. at 569, 125 S.Ct. 1183 (noting that three unique characteristics of youth mitigate juveniles’ moral culpability for certain behavior: ''[a] lack of maturity and an underdeveloped sense of responsibility!], which] ... often result in impetuous and ill-considered actions and decisions”; increased "vulnerabfility] or susceptibility] to negative influences and outside pressures, including peer pressure”; and "that the character of a juvenile is not as well formed as that of an adult. The personality traits of juveniles are more transitory, less fixed”) (citing Johnson, 509 U.S. at 367, 113 S.Ct. 2658; Eddings, 455 U.S. at 115, 102 S.Ct. 869).
. Compare Johnson, in which the Supreme Court singled out youth, as opposed to other conditions that could be transitory, because its ephemeral nature is bound up in its mitigating impact such that a juror could not reasonably assess youth as a mitigating factor without taking into account this aspect of transience. See Johnson, 509 U.S. at 368, 113 S.Ct. 2658 (emphasizing that the impact of youth on the defendant’s conduct "is not independent of an assessment of personal culpability”). Because this transient quality is so subsumed within the mitigating relevance of youth, the Court did not inquire whether the jury might have found that Johnson was likely to mature as he grew up before it held that the jury could give full effect to youth through the future-dangerousness issue; the undisputed chronological fact of the defendant’s age was enough. In contrast, under the approach favored by the dissenting opinions of Chief Judge Jones and Judge Owen in this case, when presented with mitigating evidence of a possibly treatable mental illness, an appellate habeas court must conduct such an inquiry into the jury’s findings and weigh the evidence to determine whether the illness is treatable. Perhaps for this very reason, the Supreme Court, which spoke about youth in very specific terms in Johnson, has never extended Johnsons reasoning to any other mitigating evidence — including possibly treatable mental illness — that might have transient characteristics.
. The State's reliance on Calderon v. Coleman, 525 U.S. 141, 119 S.Ct. 500, 142 L.Ed.2d 521 (1998), in support of its argument that the Brecht harmless-error test is applicable is misplaced. Coleman involved a jury instruction that gave the jury inaccurate information on the governor’s power to commute a sentence, which the lower court found might have misled the jury and distracted it from the mitigating evidence presented. Coleman is not at all comparable to cases involving Penry violations, where the jury is precluded from giving its reasoned moral response to the defendant's mitigating evidence.