join dissenting from the majority opinion:
I. BACKGROUND
This court voted to rehear Nelson’s case en banc because we are divided over how to interpret recent Supreme Court cases— Penry II, Tennard, and Smith — concerning Texas’s pre-1991 death penalty statute. Three years ago, we reheard the Robertson case en banc because we were divided over interpretation of the Supreme Court’s Texas death penalty case law leading up to and including Penry II.1 The Court’s continuing mixed signals on issues of critical importance to Texas’s criminal justice system are unfortunate. It is to be hoped that, for the sake of certainty, the *338Court will clarify its jurisprudence in the cases on which it just granted certiorari.2
The majority opinion grants habeas relief to Nelson based on an adjective. It concludes that Nelson’s mitigating evidence could not be given “full effect” by the jury at sentencing due to the inadequacy of the pre-1991 Texas death penalty special issues. It concludes, based on some language in the Court’s opinions, that “full effect,” not just “some effect,” is now the baseline for constitutionally adequate jury evaluation of a defendant’s mitigating evidence.
This conclusion marks a surprising result in a habeas petition governed by AEDPA, which mandates affirmance of state criminal convictions unless the state court’s decision was contrary to, or an unreasonable application of, federal law. First, Nelson proffered mitigating evidence of a sort that this court has frequently encountered: (1) his mother rejected him and he had no relationship with a child he had sired; (2) he was intoxicated by drugs and alcohol when he committed the crime; (3) he had troubled relationships with his brother and women; and (4) he suffered from a treatable borderline personality disorder. This court has upheld numerous capital sentences against claims that similar evidence could not be given sufficient effect by Texas juries under the pre-1991 statutes. The Supreme Court has frequently refused to review those decisions, and prisoners were executed.3 Today’s result suggests a “sea change”4 from those decisions and their understanding of the Court’s case law.
Second, the majority’s reasoning implies that the Penry line of cases, which was described by the Court as an “exception” to the “rule,” commencing with Jurek, of the overall constitutionality of the Texas sentencing issues,5 has become the “new *339rule” to which Jurek, Franklin, Graham, and Johnson are now exceptions. Yet Penry I is self-described as “not a new rule” (which means that it may be applied retroactively in habeas cases),6 and none of its progeny has altered that characterization. Even more potently, neither Penry II, Tennard, nor Smith overruled the other line of cases. If, however, “full effect” has become the test for mitigating evidence, rather than “some effect” or “within the effective reach of the jury,” then the majority’s decision is irreconcilable with the Jurek-Franklin-Johnson-Graham line of cases.
This court cannot “underrule” the Supreme Court. Our duty is to harmonize its decisions as well as possible. We are always bound by the force of stare decisis, which caused Justice Kennedy to comment in Johnson that
[t]he interests of the State of Texas, and of the victims whose rights it must vindicate, ought not to be turned aside when the State relies upon an interpretation of the Eighth Amendment approved by this Court, absent demonstration that our earlier cases were themselves a misinterpretation of some constitutional command.
Johnson, 509 U.S. at 366, 113 S.Ct. at 2668 (citations omitted).
II. THE “CLEARLY ESTABLISHED” LAW
With this preface, a closer analysis of the majority’s opinion can begin. Billy Ray Nelson’s habeas petition was rejected by the state courts for reasons that had nothing to do with this court’s now-abandoned “constitutional relevance” and “uniquely severe” evidentiary thresholds. See Tennard v. Dretke, 542 U.S. 274, 124 S.Ct. 2562, 159 L.Ed.2d 384 (2004). The state courts conducted thoughtful and thorough analyses of Nelson’s proffered mitigating evidence, and determined that all such evidence was sufficiently encompassed by the former Texas special issues and did not run afoul of Penry v. Lynaugh, 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989) (Penry I).
Nevertheless, and despite the demanding AEDPA “unreasonableness standard,”7 the majority now holds that Nelson is entitled to relief because there was a “reasonable likelihood” that Nelson’s jury was prevented from giving “full effect” to his mitigating evidence. Whether the standard is that of “full effect” or something else is the principal issue before this court. Only last year, the author of today’s majority opinion stated the test without a “full effect” gloss: “To grant relief on a Penry claim, we must determine (1) whether the mitigating evidence has met *340the ‘low threshold for relevance,’ and, if so, (2) that the evidence was beyond the effective scope of the jury.” Bigby v. Dretke, 402 F.3d 551, 564-65 (5th Cir.2005) (Stewart, J.) (citations omitted). The constitutional relevance of Nelson’s mitigating evidence is not at issue here. But to say that a death penalty must be upheld unless such evidence was “beyond the effective scope of the jury,” as Bigby does (and as this dissent advocates), is a much different test than whether such evidence could be given “full effect” by the jury.
The majority opinion cites every instance in which opinions of the Court — in dicta or dissents — have employed the term “full effect”. Unfortunately, the course of the Court’s jurisprudence, in our view, is far more complex than reliance on one adjective — -“full”—would suggest.
In the beginning, in Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976), the Supreme Court upheld the constitutionality of the Texas special issues, noting that Texas’s sentencing scheme permitted the jury to “consider whatever evidence of mitigating circumstances the defense can bring before it.” Id. at 273, 96 S.Ct. at 2957. The special issues were not seen to preclude the consideration of mitigating evidence, but rather, served to “guide[ ] and' focus[ ] the jury’s objective consideration of the particularized circumstances of the individual offense and the individual offender.” Id. at 274, 96 S.Ct. at 2957. Such focusing was seen as beneficial, as it promoted evenhandedness by the jury, allowed an individualized assessment of the defendant’s culpability, and guarded against arbitrary results. Cf. Lockett v. Ohio, 438 U.S. 586, 605-06, 98 S.Ct. 2954, 2965, 57 L.Ed.2d 973 (1978) (invalidating Ohio death penalty statute that altogether prevented the jury from considering relevant mitigating evidence; the Ohio statute was explicitly compared unfavorably to the Texas statute). There is thus no basis to conclude as a general matter that the Texas special issues will fail to allow a jury to weigh a petitioner’s mitigating evidence.
This assessment of the special issues was confirmed in Franklin v. Lynaugh, 487 U.S. 164, 108 S.Ct. 2320, 101 L.Ed.2d 155 (1988), as the Supreme Court again rejected a challenge to the constitutionality of the special issues. In that case the petitioner argued that mitigating evidence of his good behavior while in prison presented in his defense had relevance beyond the special issues, particularly the second special issue, which concerns “future dangerousness.” In denying habeas, the Court held that all “relevant aspects” of the petitioner’s character could be encompassed by the second special issue. Id. at 178, 108 S.Ct. at 2329. More important, in commenting on the adequacy of the special issues, the plurality qualified the broad statement in Eddings v. Oklahoma, 455 U.S. 104, 102 S.Ct. 869 (1982), that the sentencing jury may not be precluded from considering “any relevant, mitigating evidence.” In the plurality’s view, Ed-dings and Lockett did not prevent a state from “structuring or giving shape to the jury’s consideration of ... mitigating factors.” Franklin, 487 U.S. at 179, 108 S.Ct. at 2330. The Court thus rejected the contention that a catch-all instruction allowing the jury an independent basis for rendering a sentence other than death was necessary, as such an instruction would overrule Jurek. Id. at 180 & n. 10, 108 S.Ct. at 2330 & n. 10. Jurek had approved the Texas special issues, and the Court had repeatedly referred with approval to Texas’s sentencing scheme, see Franklin, 487 U.S. at 182 n. 11, 108 S.Ct. at 2331 n. 11 (citing cases), precisely because it reconciled the Court’s twin concerns for statutory structuring and for jury flexibility to consider mitigating evidence. Justice O’Connor’s concurrence in the judgment *341presaged her view in Penry I that Jwrek did not preclude a “claim that, in a particular case,” the special issues were constitutionally inadequate. Penry I, 492 U.S. at 321, 109 S.Ct. at 2948. However, from Franklin, including Justice O’Connor’s special concurrence, it is clear that the Texas special issues ought to be constitutional in the vast majority of cases.
Ultimately, the question of what exactly it means for a court to give “full consideration” to a habeas petitioner’s mitigating evidence was answered in the cases of Graham v. Collins, 506 U.S. 461, 113 S.Ct. 892, 122 L.Ed.2d 260 (1993), and Johnson v. Texas, 509 U.S. 350, 113 S.Ct. 2658, 125 L.Ed.2d 290 (1993). Although Graham came to the Supreme Court on collateral, as opposed to direct, review, and was thus subject to analysis under Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), the case was nevertheless instrumental in explaining the sufficiency of state death penalty statutes. In that case, the petitioner argued that evidence of his youth and transient upbringing had mitigating impact beyond the special issues. The Court rejected this contention, again turning to Jwrek. Death penalty statutes only had to supply the defendant with a “constitutionally adequate” consideration of his mitigating evidence, which Texas’s special issues did. Graham, 506 U.S. at 470, 113 S.Ct. at 899. The majority explained Lockett, Eddings, Skipper,-8 Hitchcock,9 and Penry I as being constitutionally defective because “relevant mitigating evidence was placed beyond the effective reach of the jury.” Id. at 475, 113 S.Ct. at 902. The fact that the defendant’s evidence might have “some arguable relevance” beyond the special issues did not invalidate the special issues. Id. at 475-76, 113 S.Ct. at 902. This is because “virtually any mitigating evidence is capable of being viewed as having some bearing on the defendant’s ‘moral culpability’ apart from its relevance to the particular concerns embodied in the Texas special issues.” Id. at 476, 113 S.Ct. at 902. Again, citing Franklin and Jurek, the Court determined that Texas’s death penalty statute allowed mitigating evidence to be adequately considered while permissibly focusing the considerations of the sentencing jury. Id. Graham, a majority opinion, thus stands for the proposition that a Texas jury may constitutionally render a sentence of death even where a defendant presents mitigating evidence that has some arguable relevance beyond the special issues.
Just months later, a majority of the Court in Johnson reaffirmed the reasoning of Graham, in a direct appeal in which the appellant’s youth as an offender was his major mitigating quality. Justice Kennedy’s opinion drew heavily from Graham, re-emphasizing that while
Lockett and Eddings prevent a state from placing relevant mitigating evidence beyond the effective reach of the sentencer, ... we have held that there is no ... constitutional requirement of unfettered sentencing discretion in the jury, and states are free to structure and shape mitigating evidence in an effort to achieve a more rational and equitable administration of the death penalty.
Johnson, 509 U.S. at 362, 113 S.Ct. at 2666 (citations and internal quotation marks omitted). Recapitulating the cases construing Texas’s special issues, the Court confirmed a narrow interpretation of Pen-ry I, “making it clear that [Jurek, Lockett and Eddings] can stand together with Pen-*34211/.” Id. at 365, 113 S.Ct. at 2667-68. The Court closely analyzed youthfulness as a mitigating factor and held that “there is ample room in the assessment of future dangerousness for a juror to take account of the difficulties of youth .... ” Id. Pen-ry’s condition, in contrast, rendered him unable to learn from his mistakes and could only be considered to aggravate, not lessen, his future dangerousness.10
The Court concluded Johnson with the observation, originating in Graham, that Jurek would have to be overruled if, whenever a defendant proffers mitigating evidence “that has some arguable relevance beyond the special issues,” a fourth jury issue in mitigation would be required. Id. at 372, 113 S.Ct. at 2671. Such an issue, as the Court reasoned, would effectively abrogate the state’s power, repeatedly affirmed by the Court, to structure the consideration of mitigating evidence.
Graham and Johnson are majority opinions of the Court.11 Penry I is also a majority opinion, but Pemy I represented a fact-specific exception to the Jurek line of cases. This was made abundantly clear in Graham, 506 U.S. at 475, 113 S.Ct. at 902. What distinguished Penry I from the aforementioned cases was that, according to Penry’s experts, he had extremely poor impulse control, and, owing to his limited mental abilities, he was unable to appreciate the consequences of his actions or learn from his mistakes. Unlike the instant case, there was no suggestion that Penry’s condition would improve; his brain damage was allegedly permanent. Such evidence might have diminished Penry’s culpability, but it also served to indicate, as all sides agreed, that he would always be a threat to society. As such, with regard to the “future dangerousness” special issue, Penry’s evidence served “only as an aggravating factor” for the jury. Id. at 323, 109 S.Ct. at 2949. The defense found itself in the unenviable position of arguing that a “juror should vote ‘no’ on one of the special issues even if she believed the State had proved the answer should be ‘yes.’ ” Id. at 325, 109 S.Ct. at 2950. The prosecution in turn stressed that “the jurors had taken an oath to follow the law, and that they must follow the instructions.” Id. This created a uniquely unfortunate situation in which a reasonable juror could credit the mitigating evidence and feel a sentence other than death was warranted for Penry, yet nevertheless be compelled to answer the special issues in the affirmative and render a sentence of death. Unlike Graham and Johnson, in which the juries had the ability to give at least “some effect” to the mitigating evidence presented by the defendants, it was “impossible to give meaningful mitigating *343effect” to Penry’s evidence through the special issues. Graham, 506 U.S. at 474, 113 S.Ct. at 901. The Penry I jury had “no vehicle for expressing the view that Penry did not deserve to be sentenced to death.” Penry I, 492 U.S. at 326, 109 S.Ct. at 2951.12
To quote Graham again: “In Penry, the defendant’s evidence was placed before the sentencer but the sentencer had no reliable means of giving mitigating effect to that evidence.” Graham, 506 U.S. at 475, 113 S. Ct at 902 (emphasis added). Penry I was thus “limited ... [in] its scope,” as otherwise, it could not be consistent with Jurek and Lockett, both of which were repeatedly reaffirmed by the Court. Johnson, 509 U.S. at 354, 113 S.Ct. at 2668. In short, the “clearly established law” as of 1994 is not, as the majority argue, the Penry I “full effect” test, but instead consists of Penry I together with Graham, Johnson, Franklin, and Jurek.
The Court’s subsequent decisions in Penry II, Tennard v. Dretke, and Smith v. Texas have muddied the waters, but they have not replaced, much less overruled, Jurek, Franklin, Graham, and Johnson. Each of the more recent cases resolves a narrow procedural issue. Penry II considered the sufficiency of a “nullification instruction” to the jury that Texas courts thought would alleviate the problem in Penry’s case. The Court explained why the nullification instruction would cause jurors to violate their oaths if they felt, notwithstanding that Penry’s condition required a positive answer to his deliberateness and future dangerousness, he was less culpable because of his mental retardation. The Court’s opinion mentions “full effect” once, but its overruling of the nullification instruction was not tied to whether the jury could give “full effect” to Penry’s mitigating evidence. The jurors’ catch-22 was independent of the amount of the mitigating effect.
In Tennard, the Court held that the Fifth Circuit’s “uniquely severe permanent handicap” and “nexus” tests for identifying Penry evidence were incorrect, and that for COA purposes, “reasonable jurists would find debatable or wrong the District Court’s disposition of Tennard’s low-IQ-based Penry claim.” Tennard, 542 U.S. at 289, 124 S.Ct. at 2573. Indeed, Tennard found that the petitioner’s low IQ evidence had “the same essential features” as Pen-ry’s mental retardation evidence: His low IQ could be considered irrelevant to mitigation while having only aggravating relevance to his future dangerousness. Id. at 288, 113 S.Ct. at 2572. Tennard did not cite Graham or Johnson. Because the decision expressly models its analysis on Penry I, it cannot be said to extend Penry I or to undercut Graham or Johnson.
Nowhere does Tennard require that the jury be able to give “full effect” to mitigating evidence in its sentencing deliberations. Instead, the Court quotes a potpourri of earlier decisions requiring states to enable the jury to “consider and give effect to” mitigating evidence;13 forbidding states to “preclude the sentencer from con*344sidering any ‘relevant mitigating evidence’ ”;14 and asserting virtually no limits on a defendant’s ability to proffer relevant mitigating evidence.15 Tennard compels Texas courts confronted with low IQ evidence to submit a proper special issue; Tennard also counsels fact-specific evaluation of petitioners’ mitigation evidence for its application within the pre-1991 Texas special issues.
A final word about Tennard: Justice Kennedy concurred. Does this mean that he had changed his mind since he wrote Johnson, or that he viewed Tennard as reconcilable with Johnson? A “reasonable jurist” would draw the latter conclusion, since one can hardly assume Justice Kennedy would have failed to explain his departure from the very explicit cabining of Penry I that he accomplished with the majority opinion in Johnson.
Smith v. Texas, 543 U.S. 37, 125 S.Ct. 400, 160 L.Ed.2d 303 (2004), is the most recent case in the Penry line, and it, too, represents a narrow procedural holding. The Court reversed a Texas Court of Criminal Appeals decision that utilized the “constitutional relevance” tests adopted by the Fifth Circuit but rejected in Tennard, and purported to distinguish a nullification instruction from the instruction overruled by the Supreme Court in Penry II. That the Court would enforce its prior decisions with a per curiam reversal is hardly surprising. That the Court would employ such a brief opinion to expand the reach of Penry I and undermine Graham and Johnson sub silentio is unlikely. The majority in this case points to language supporting the “unlikely” reading. Smith initially quotes Penry II as holding a similar nullification instruction inadequate to enable a jury to give “full consideration” and “full effect” to defendant’s mitigating circumstances. In the third paragraph of the decision, the Court states: “Approximately two years prior to the trial, we had held that presenting only these two special issues, without additional instructions regarding the jury’s duty to consider mitigation evidence, violated the Eighth Amendment.” Id. at 39, 125 S.Ct. at 402 (citation omitted). After explaining the plain errors in the state court’s decision, the Smith Court states that: “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 anything, to do with the mitigation evidence petitioner presented.” Id. at 48, 125 S.Ct. at 407 (footnote omitted). Smith’s mitigation evidence included potentially organic learning disabilities and speech handicaps; low IQ and special education in school; good behavior in school; a drug-addicted criminal father; and his age of nineteen at the date of the offense. Smith concludes, without analysis of the types of mitigating evidence, that because it was “relevant mitigation evidence for the jury under Tennard and Penry the nullification instruction was inadequate under Penry II. Id. at 48-49, 125 S.Ct. at 407.
This court may not overlook the potentially broad language in Smith. On the other hand, Smith failed to cite or distinguish Jurek, Franklin, Graham, or Johnson. Since Chief Justice Rehnquist and Justice Kennedy joined Smith, the question again arises whether they did so in deference not just to a limited view of Penry II and Tennard but also, and without explanation, to a de facto overruling of Graham and Johnson through Smith’s casual incorporation of the appellant’s youth, good school behavior, and disadvantaged *345(not abused) childhood as Penry mitigation evidence.
Finally, notwithstanding Smith’s two references to “full effect,” the opinion also quotes Penry II as recognizing that “the key under Penry I is that the jury be able to ‘consider and give effect to [a defendant’s mitigation evidence] in imposing sentence.’ ” Smith, 543 U.S. at 46, 125 S.Ct. at 406 (emphasis in original).
“Giving effect” to mitigating evidence is not the same as allowing a jury to give “full effect.” The latter formulation, in effect, rejects a state’s ability to focus the jury’s consideration of mitigating evidence. Here lies the crux of our difference with today’s majority opinion. Despite its efforts to turn narrow procedural decisions and imprecise language into a constitutional mandate of “full effect,” the Supreme Court’s case law will not support that conclusion. As an inferior court, we can overlook neither Jurek, Franklin, Graham, and Johnson, nor Penry I, Penry II, Ten-nard and Smith. Sadly, for the State of Texas, for certainty and stare decisis, and for defendants who deserve to know their fate before the last minute, we seem no further along in understanding the Court’s pronouncements today than we were fifteen years ago when we reheard Graham en banc. See Graham v. Collins, 950 F.2d 1009 (5th Cir.1992) (en banc), aff'd, 506 U.S. 461, 113 S.Ct. 892, 122 L.Ed.2d 260 (1993).
The interrelated rules we believe must be holistically drawn from the Court’s decisions — until we are told otherwise — are as follows: First, courts must consider all mitigating evidence for its comprehensibility within the Texas special issues. Second, if, as with Penry I and Tennard low IQ evidence, the proffered evidence has only aggravating force beyond the issues of deliberateness and future dangerousness, re-sentencing is required. In such cases, the proffered evidence was “beyond the effective reach of the jury” such that “the jury was precluded from considering the evidence.” Third, evidence of such qualities as a defendant’s youthfulness at the date of the crime and a “transient” upbringing16 can, however, be considered within the special issues.
III. THE MITIGATING EVIDENCE
Nelson offered in evidence that (i) his mother rejected him; (2) he had troubled relationships with his brother and women; (3) he was denied a relationship with his child; (4) he was intoxicated by drugs and alcohol when he committed the crime; and (5) he suffered from a treatable borderline personality disorder. The majority opinion dwells principally on the last element, subsidiarily on the rejection by his mother, and not at all on Nelson’s substance abuse or other troubled relationships. Consequently, we focus on the first two characteristics. It must be pointed out, though, that the majority’s “full effect” test apparently renders the pre-1991 Texas sentencing hearing constitutionally inadequate for any mitigating evidence except for youthfulness (and good behavior in prison).17 After all, nearly any mitigating evidence can be said to have “some arguable relevance” beyond the deliberateness and future dangerousness inquiries.
Nelson’s expert, Dr. Hickman, testified that Nelson had anger issues resulting from his childhood experiences, and that treatment for his borderline personality disorder would require long-term psychotherapy and medication. Hickman also suggested that individuals with borderline *346personality disorder tend to be difficult to treat, and success with Nelson was not guaranteed. However, Hickman further testified that if successfully treated, Nelson would no longer represent a danger to society.
Nelson’s evidence is fundamentally distinguishable from that of Penry, who was presented as being beyond treatment because of an insufficient mental acuity and inability to learn from his mistakes. In contrast, Nelson’s defense offered the jury evidence that Nelson could get better, and that if he spent the rest of his life in prison, he would no longer represent a future danger to society. Unlike Penry, but like the defendant in Graham, Nelson’s attorneys could honestly and “vigorously urge[ ] the jury to answer ‘no’ to the special issues based upon” the evidence presented. Graham, 506 U.S. at 475, 113 S.Ct. at 902.
With regard to the “deliberateness” of the crime, Nelson’s jury could have concluded, based on his maternally-deprived upbringing, his “anger issues” and his poor impulse control, that he did not sexually abuse his victims and murder Charla Wheat “deliberately.” He was, in other words, too warped to have acted responsibly. Alternatively, the jury could have balanced these mitigating factors against his self-induced drug abuse and intoxication, and the speculation embodied in Dr. Hickman’s connecting his behavioral problems to the crime, and found this crime to be “deliberate.”
Nelson’s jury was also presented with clear alternatives in regard to future dangerousness. It could believe Hickman’s testimony and conclude that Nelson was less morally culpable, given his mental illness, and that with proper treatment, Nelson would not present a future danger. Alternatively, the jury could follow the prosecution’s theory that Nelson was fully culpable for his actions and would continue to be dangerous even in prison.18 That the jury chose the latter assessment of Nelson does not mean that habeas relief must issue. Indeed, in order to even make a plausible argument that a Penry violation occurred in the instant case, the majority recasts the record to suggest that Nelson would be untreatable. This is simply not the case. Dr. Hickman’s purpose for testifying was not just to illustrate Nelson’s condition but to demonstrate his potential for change. That potential clearly found mitigating expression in both the “deliberateness” and “future dangerousness” issues.19
Because this case is reviewed under AEDPA, we must, as the majority acknowledges, find the state courts’ resolution of the Penry issue not simply wrong, but unreasonable. Further, the “unreasonableness” must here stem from a conclusion that there is a “reasonable likelihood” — not a “mere possibility” — that the *347jury applied the two issues “in a way that prevents the consideration of constitutionally relevant evidence.” Johnson, 509 U.S. at 367-68, 113 S.Ct. at 2669 (paraphrasing Boyde, 494 U.S. at 380, 110 S.Ct. at 1198). The “reasonable likelihood” standard is applied according to a “eommonsense understanding of the record in the light of all that has taken place at the trial.” Id. at 381, 110 S.Ct. at 1198. Finally, the fact that “a juror might view the evidence ... as aggravating, as opposed to mitigating, does not mean that the rule of Lockett is violated.” Johnson, 509 U.S. at 368, 113 S.Ct. at 2669. As illustrated above, there is no reasonable likelihood that the jury applied the special issues in an unconstitutional manner; in expressing its “reasoned moral response” to Nelson’s evidence, the jury could have relied upon Hickman’s testimony and concluded that Nelson would not remain a danger in prison. Based on our interpretation that Johnson and Graham remain good law, coexisting with Pen-ry and its progeny, we cannot subscribe to the unreasonableness of the state courts’ determination.
Clearly, the evidence of a treatable mental condition and a deprived family background could be afforded decisive, if perhaps not “full,” mitigating effect under the pre-1991 sentencing scheme. The Court stated in Graham:
We see no reason to regard the circumstances of Graham’s family background and positive character traits in a different light [from Franklin]. Graham’s evidence of transient upbringing [while his mother spent long periods hospitalized for a “nervous condition”] more closely resembles Jurek’s evidence of age, employment history, and familial ties than it does Penry’s evidence of mental retardation and harsh physical abuse.
Graham, 506 U.S. at 476, 113 S.Ct. at 902.
The Court, of course, held in Graham that to require an additional jury instruction would be a “new rule” of constitutional law. We do not pretend that Nelson’s evidence of personality disorder and maternal rejection is on all fours with Jtvrek, Franklin, Graham, or Johnson. But the majority cannot pretend that such evidence — of a treatable mental condition and not “harsh physical abuse” — compels habe-as relief based on Penry I, Penry II, Ten-nard or Smith.20
Reinforcing our conclusion is the inconsistency between the majority’s analysis of a treatable mental disorder today and our court’s analysis of an untreatable mental condition — schizophrenia—a year ago. See Bigby v. Dretke, 402 F.3d 551 (5th Cir.2005). Today’s majority overrules the decision in Lucas v. Johnson, 132 F.3d 1069 (5th Cir.1998), which held that the Texas special issues furnished sufficient scope for a jury to give effect to evidence of a treatable mental condition. Id. at 1082-83. Last year, in Bigby, the author of today’s opinion distinguished Lucas because of the different ramifications of a treatable mental disorder under the Texas special issues. Bigby, 402 F.3d at 571. If Bigby found no conflict between Lucas and the Court’s decisions in the Penry line, how can the majority assert today that a comparable decision by the Texas courts was “unreasonable?”21
*348IV. CONCLUSION
Nelson’s evidence had constitutionally adequate mitigating effect as to both of the special issues, and his jury was neither foreclosed from giving effect to the evidence by the Texas special issues, nor was it put in the position of rendering a false verdict, as in Penny I and Penny II. If the majority’s expansive reading of Penny compels the result reached today, it is to be hoped that the Supreme Court will so inform us definitively in the cases now pending before it. Because none of the Court’s precedents to date compels the “full effect” test or the result reached by the majority, it cannot be said that the state courts unreasonably applied federal law. I would deny habeas relief, and therefore, I respectfully dissent.
. In 1992, we reheard the Graham case en banc for the same reason. Graham v. Collins, 950 F.2d 1009 (5th Cir.1992) (en banc), aff'd, 506 U.S. 461, 113 S.Ct. 892, 122 L.Ed.2d 260 (1993).
. See Cole v. Dretke, 418 F.3d 494 (5th Cir.2005), cert. granted, - U.S. -, 127 S.Ct. 432, 166 L.Ed.2d 307 (2006); Brewer v. Dretke, 442 F.3d 273 (5th Cir.2006), cert. granted, - U.S. -, 127 S.Ct. 433, 166 L.Ed.2d 307 (2006).
. See, e.g., Hernandez v. Johnson, 248 F.3d 344 (5th Cir.), cert. denied sub nom. Hernandez v. Cockrell, 534 U.S. 1043, 122 S.Ct. 621, 151 L.Ed.2d 543 (2001); Emery v. Johnson, 139 F.3d 191 (5th Cir.1997), cert. denied, 525 U.S. 969, 119 S.Ct. 418, 142 L.Ed.2d 339 (1998); Davis v. Scott, 51 F.3d 457 (5th Cir.), cert. denied, 516 U.S. 992, 116 S.Ct. 525, 133 L.Ed.2d 432 (1995); Jacobs v. Scott, 31 F.3d 1319 (5th Cir.1994), cert. denied, 513 U.S. 1067, 115 S.Ct. 711, 130 L.Ed.2d 618 (1995); Lackey v. Scott, 28 F.3d 486 (5th Cir.1994), cert. denied, 513 U.S. 1086, 115 S.Ct. 743, 130 L.Ed.2d 644 (1995); Clark v. Collins, 19 F.3d 959 (5th Cir.1994), cert. denied, 513 U.S. 966, 115 S.Ct. 432, 130 L.Ed.2d 344 (1994); Motley v. Collins, 18 F.3d 1223 (5th Cir.), cert. denied, 513 U.S. 960, 115 S.Ct. 418, 130 L.Ed.2d 333 (1994); Madden v. Collins, 18 F.3d 304 (5th Cir.1994), cert. denied, 513 U.S. 1156, 115 S.Ct. 1114, 130 L.Ed.2d 1078 (1995); Russell v. Collins, 998 F.2d 1287 (5th Cir. 1993), cert. denied, 510 U.S. 1185, 114 S.Ct. 1236, 127 L.Ed.2d 580 (1994); Callins v. Collins, 998 F.2d 269 (5th Cir.1993), cert. denied, 510 U.S. 1141, 114 S.Ct. 1127, 127 L.Ed.2d 435 (1994); Drew v. Collins, 964 F.2d 411 (5th Cir.), cert. denied, 509 U.S. 925, 113 S.Ct. 3044, 125 L.Ed.2d 730 (1993); Lincecum v. Collins, 958 F.2d 1271 (5th Cir.), cert. denied, 506 U.S. 957, 113 S.Ct. 417, 121 L.Ed.2d 340 (1992); Barnard v. Collins, 958 F.2d 634 (5th Cir.1992), cert. denied, 506 U.S. 1057, 113 S.Ct. 990, 122 L.Ed.2d 142 (1993); Mayo v. Lynaugh, 893 F.2d 683 (5th Cir.1990), modified sub nom. Mayo v. Collins, 920 F.2d 251 (5th Cir.1990), cert. denied, 502 U.S. 898, 112 S.Ct. 272, 116 L.Ed.2d 225 (1991).
. But cf. Johnson v. Texas, 509 U.S. 350, 365, 113 S.Ct. 2658, 2668, 125 L.Ed.2d 290 (1993) (stating that Penry did not “effec[t] a sea change in this court’s view of the constitutionality of the former Texas death penalty statute”) (quoting Graham, 506 U.S. at 474, 113 S.Ct. at 901).
. Graham, 506 U.S. at 491, 113 S.Ct. at 910.
. Penry v. Lynaugh, 492 U.S. 302, 315, 109 S.Ct. 2934, 2945, 106 L.Ed.2d 256 (1989) (Penry I).
. The fact that a federal habeas court would have reached a different conclusion than did the state court is insufficient to merit habeas relief pursuant to AEDPA. See Brown v. Payton, 544 U.S. 133, 147, 125 S.Ct. 1432, 1442, 161 L.Ed.2d 334 (2005); Woodford v. Visciotti, 537 U.S. 19, 27, 123 S.Ct. 357, 361, 154 L.Ed.2d 279 (2002). The Court in Williams was careful to note that "an unreasonable application of federal law is different from an incorrect application of federal law,” and, as such, the state court’s application of federal law must be "objectively unreasonable,” as opposed to merely incorrect, for habeas relief to be granted. Williams, 529 U.S. at 409-10, 120 S.Ct. at 1521-22 (emphasis in original); see also Penry v. Johnson, 532 U.S. 782, 793, 121 S.Ct. 1910, 1919, 150 L.Ed.2d 9 (2001)(Penry II). Consequently, this court overlooks the erroneous reasoning of state courts, and reviews the reasonableness of their ultimate decision. Neal v. Puckett, 286 F.3d 230, 236 (5th Cir.2002)(en banc), cert. denied, 537 U.S. 1104, 123 S.Ct. 963, 154 L.Ed.2d 772 (2003).
. Skipper v. South Carolina, 476 U.S. 1, 106 S.Ct. 1669, 90 L.Ed.2d 1 (1986).
. Hitchcock v. Dugger, 481 U.S. 393, 107 S.Ct. 1821, 95 L.Ed.2d 347 (1987).
. The Court quoted Justice Brennan's dissent in Blystone, which acknowledged the ability of the Texas special issues to afford jury consideration of a defendant's moral culpability:
[The two special issues] require the jury to do more than find facts supporting a legislatively defined aggravating circumstance. Instead, by focusing on the deliberateness of the defendant's actions and his future dangerousness, the questions compel the jury to make a moral judgment about the severity of the crime and the defendant’s culpability. The Texas statute directs the imposition of the death penalty only after the jury has decided that the defendant’s actions were sufficiently egregious to warrant death.
Id. at 371, 113 S.Ct. at 2671 (quoting Blystone v. Pennsylvania, 494 U.S. 299, 322, 110 S.Ct. 1078, 1091, 108 L.Ed.2d 255 (1990) (Brennan, J., dissenting)).
. Notably, in both Graham and Johnson, spirited dissents capture the same debate over "full effect” and "some effect” that preoccupies us still; but the advocates of "full effect” lost. See, e.g., Graham, 506 U.S. at 504, 113 S.Ct. at 917 (Souter, J., dissenting); Johnson, 509 U.S. at 374, 113 S.Ct. at 2672. (O'Con-nor, J., dissenting).
. This reading of Penry I is entirely consistent with, and indeed anticipates, the Court's later decision in Penry v. Johnson, 532 U.S. 782, 121 S.Ct. 1910, 150 L.Ed.2d 9 (2001)(Penry II). As with Penry I, the Penry II Court rejected as “arbitrary” a death penalty system that would encourage a juror to provide a "false answer” to one of the special issues, thereby violating his oath as a juror. id. at 801, 121 S.Ct. at 1923. It is only in these rare circumstances that a jury finds itself without a vehicle to provide a “reasoned moral response” to the defendant's evidence.
. Boyde v. California, 494 U.S. 370, 377-78, 110 S.Ct. 1190, 1196, 108 L.Ed.2d 316 (1990). Boyde held that jury discretion could be guided by the States.
. Payne v. Tennessee, 501 U.S. 808, 822, 111 S.Ct. 2597, 2606, 115 L.Ed.2d 720 (1991).
. Eddings v. Oklahoma, 455 U.S. 104, 114, 102 S.Ct. 869, 877, 71 L.Ed.2d 1 (1982).
. See Graham, 506 U.S. at 476, 113 S.Ct. at 902.
. See Franklin, 487 U.S. at 179 n. 9, 108 S.Ct. at 2330 n. 9.
. The prosecution did not agree with Hickman's assessment of Nelson’s mental condition, as it did not have sufficient evidence to make a diagnosis. Its expert, Dr. Grigson, concluded only that Nelson would continue to pose a threat.
. The majority further relies upon a string of hypotheticals to create its Penry violation. If his jury believed that Nelson suffered from borderline personality disorder; if that jury believed that Nelson was untreatable or would not receive proper treatment in prison; and if that jury concluded that Nelson's mental illness had aggravating effect as to the special issues, only then is it possible that the jury might have felt compelled to answer "yes” as to the future dangerousness special issue, even if the jury wished a sentence other than death due to Nelson’s borderline condition. This attenuated theory of the jury deliberations extends Penry I far beyond its intended boundaries, without instructions from the Supreme Court.
. The holding of Graham, based on Teague, is that Penry I did not dictate constitutional relief based on the defendant's youthfulness. How, then, could the different evidence of a treatable mental disorder have become so indistinguishable from Penry as to render the state court's decision in this case unreasonable?
. Today's majority decision is also squarely contrary to the recent decision in Cole v. Dretke, 418 F.3d 494 (5th Cir.2005) cert. granted, that the “Texas special issues allowed *348the jury to give 'full consideration and full effect’ ” to Cole's mitigating evidence of a destructive family background. Id. at 511. Nevertheless, two members of today’s majority panel joined the Cole decision.