dissenting:
I respectfully dissent from the majority’s decision because it mischaracterizes the record and contravenes Supreme Court precedent. Fernando Garcia presented evidence at the sentencing phase of his capital trial that he suffered severe and frequent sexual abuse before the age of ten. Texas law, however, did not permit the jury to give proper effect to this mitigating evidence in deciding between life and death. The majority nonetheless concludes that no Eighth Amendment violation occurred. It reasons that Garcia is now estopped from using the abuse evidence in mitigation because his trial lawyer did not explicitly demand that the jurors disregard Texas law and violate their oaths.
To put the majority’s decision in perspective, I offer the following chronology: (1) Texas law did not allow the jury to consider abuse mitigation; (2) Garcia vigorously objected that this was unconstitutional under Penry and requested an additional special issue; (3) the trial judge rejected Garcia’s objection on the record; (4) Garcia, having preserved his Penry objection, attempted (unsuccessfully) to mold his summation to comply with the judge’s order and with Texas law; (5) Garcia maintained his Penry claim throughout direct and collateral review; (6) Garcia is now precluded from complaining under Penry. Before exploring the flaws in the majority’s holding, I will explain why Garcia is entitled to relief under a proper Penry analysis.
1. The “Proper Legal Analysis” Under Penry
When considering Penry claims, “the proper legal analysis ... is for the court to answer two questions: (1) whether the defendant’s evidence held relevance to the jury’s capital deliberations; and if so, (2) whether, the jury was able to give constitutionally-sufficient mitigating effect to that evidence.” Tennard v. Dretke, 442 F.3d 240, 257 (5th Cir.2006) (“Tennard III"). If the Court answers “yes” on the first prong and “no” on the second, then Supreme Court precedent requires habeas relief. See id.
a. Garcia Presented Relevant Mitigating Evidence
The State correctly concedes, “There is no question as to the relevancy of Garcia’s mitigating evidence.” (Dretke Suppl. Br. 2.)1 Dr. Powitzky testified that Garcia was repeatedly drugged by his mother at a young age so he would have sex with her and her boyfriend; forced to perform oral sex at the age of five on a friend of his brother’s and at six on his brother; forced to perform oral sex at the age of eight on a fourteen-year-old female cousin; and sexually abused in the third grade by a nun who was supposed to be providing him with spiritual guidance. Penry v. Lynaugh (“Penry I”) establishes that evidence of this kind properly mitigates. See 492 U.S. 302, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989); see also supra p. 470 (Maj.Op.) (“[E]vidence that a defendant was a victim of childhood abuse may constitute mitigating evidence .... ”). The Supreme Court’s decision in Tennard clarified that general relevance is all that is required. See Ten-*474nard v. Dretke, 542 U.S. 274, 283-84, 124 S.Ct. 2562, 159 L.Ed.2d 384 (2004) (“Tennard II”).
b. The Jury Could Not Give Sufficient Effect to Garcia’s Mitigating Evidence of Child Abuse
i. The Texas Special Issues Were Inadequate
I now turn to the second prong of the Tennard III test. Supreme Court precedent conclusively refutes the State’s position that Garcia’s evidence of childhood abuse could be given constitutionally-sufficient mitigating effect under the deliberateness and future dangerousness special issues.2 In Penry I, the Court held that the Texas special issues did not permit the jury to give proper mitigating effect to Penry’s “abused background.” 492 U.S. at 315, 109 S.Ct. 2934. “[Evidence of ... childhood abuse has relevance to [a defendant’s] moral culpability beyond the scope of the [Texas] special issues.” Id. at 322, 109 S.Ct. 2934. This is because “[p]ersonal culpability is not solely a function of a defendant’s capacity to act ‘deliberately.’ ” Id. A defendant who acts deliberately but who has suffered childhood abuse may be “less morally culpable than defendants who have no such excuse.” Id. at 323, 109 S.Ct. 2934 (internal quotation marks omitted). Additionally, a “history of abuse” has only aggravating effect as to the second special issue, since “it indicates that there is a probability that he will be dangerous in the future.” Id. at 324, 109 S.Ct. 2934.3
Although Penry presented evidence of diminished mental capacity along with evidence of childhood abuse, the Penry I Court treated Penry’s history of abuse as something with independent mitigating effect. See id. at 322, 109 S.Ct. 2934. At no point in the opinion does the Court remark that it is only because Penry is mentally retarded that his past history of abuse was not given sufficient mitigating effect under the special issues. See Hernandez v. Johnson, 248 F.3d 344, 377 (5th Cir.2001) (Dennis, J., dissenting) (“The [Supreme] Court did not hold or suggest that either the factor of mental retardation or childhood abuse by itself would fail to constitute relevant mitigating evidence that the jury must be able to consider and give effect to in deciding Penry’s fate.”).4 In Tennard II, the High Court discussed Penry’s mental retardation and history of abuse claims in wholly independent terms: “As to the evidence of childhood abuse, we held that the two special issues simply failed to provide a vehicle for the jury to give it mitigating effect.” 542 U.S. at 279, *475124 S.Ct. 2562 (citing Penry I) (internal modifications and quotation marks omitted). In this case, as in Penry I, the special issues did not provide a meaningful vehicle for the jury to give effect to Garcia’s mitigating evidence of child abuse.
ii. The “Nullification Instruction” Did Not Cure the Problem
The “nullification instruction” given in this case did not fix the problem. Texas courts adopted a nullification instruction in an attempt to correct the special issues’ inadequacies identified by the Supreme Court in Penry I. Such an instruction permits the jurors to answer “no” to one of the special issues (even though the proper answer is “yes”) if they still believe that the death penalty is not warranted. In Penry v. Johnson (“Penry II”), the Supreme Court found this innovation insufficient. See 532 U.S. 782, 121 S.Ct. 1910, 150 L.Ed.2d 9 (2001).
Although the nullification instruction given in Garcia’s case differed in some respects from the nullification instruction condemned by Penry II, Garcia’s instruction still had a primary flaw identified by the Supreme Court: it required jurors to answer a special issue “dishonestly in order to give effect to ... mitigating evidence.” Id. at 802-03, 121 S.Ct. 1910 (emphasis in original); (Dretke Br. 14 n.15) (conceding that the instruction was “substantially similar to the ‘nullification instruction’ at issue in Penry II”).5 Thus, under Penry II, the nullification instruction did not bring Garcia’s mitigating evidence within the effective reach of the jury.6
In sum, Garcia’s claim satisfies the second prong of the test that Tennard III derived from the Supreme Court’s opinion in Tennard II. The jury was unable to give constitutionally-sufficient mitigating effect to Garcia’s evidence. This should be the end of the matter. This Court, following Supreme Court precedent, has identified the two questions that a proper Penry analysis entails. Both favor Garcia. The majority, however, denies relief through a novel theory. This theory was not argued by the State,7 has no foundation in the decisions of this Court, and contravenes Penry II and Tennard II.
2. The Majority’s Estoppel/Default/Nul-lification Rationale
The majority here holds that Garcia is somehow estopped from arguing that his mitigating evidence went beyond the special issues because his trial lawyer struggled to shoehorn that evidence within the special issues at summation. See supra p. 471 (Maj.Op.). It is worth noting that the majority’s creative approach is the fourth *476separate rationale that has been offered to uphold Garcia’s death sentence. The Supreme Court has rejected the first three justifications.8 The latest approach is to deny Garcia relief because his trial lawyer failed to argue the proper “mitigation theory” before the jury, supra p. 468, a “mitigation theory” that Texas law did not even allow the jury to consider at the time.
The majority’s refusal to consider Garcia’s “new arguments in federal court” because they do not reflect the “case [he] presented in the [state] trial court,” supra 468, 471 (Maj.Op.), sounds a lot like procedural default. There is a reason why the majority merely gestures in that direction without ever fully going down the procedural-default path. It is because Garcia’s lawyer clearly made a Penry objection in the Texas trial court. After the court submitted a proposed jury charge, Garcia objected as follows:
In the Court’s Charge on mitigation, which is page 2 of the Court’s Charge, we would object for the following reasons:
[W]e would submit, Your Honor, that drug and alcohol abuse and child abuse have been brought into evidence. They are universally considered to be mitigating factors. And that the Court does not set those out in applying the law to the facts and relating them to the relevant questions in either Special Issue No. 1 or Special Issue No. 2.
We would submit, Your Honor, that a fourth special issue is required. What I mean by a fourth special issue is an additional special issue besides Special Issues 1 and 2. We would, of course, concede that Special Issue 3 is not raised by the evidence.9
But the fourth special issue, that has been talked about in Penry, is mitigating evidence that is relevant to the blameivorthiness and moral culpability of the Defendant in general, yet not directly relevant to either one of the first two special issues.
(R. at 21:2890-91 (emphasis added).) The trial court denied the objection “with some reluctance.” Id. at 2895.
The majority could not rest its analysis on any established version of procedural default because Garcia made a Penry objection and maintained it throughout direct and collateral review. See Breard v. Greene, 523 U.S. 371, 375, 118 S.Ct. 1352, 140 L.Ed.2d 529 (1998) (defining the procedural default rule: “assertions of error in *477criminal proceedings must first be raised in state court in order to form the basis for relief in habeas”).10 Whether or not the majority’s theory should be understood as a novel extension of procedural default doctrine, that theory is flawed for a number of reasons.
The majority reasons that Garcia should be estopped from making a Penry claim because his trial lawyer cabined his closing argument within the deliberateness special issue. The premise of this argument misstates the record. After the trial court denied his explicit Penry objection, Garcia’s lawyer nonetheless did present the jury with a “mitigation theory” that went outside the scope of the special issues. He argued that “the abuse,” among other things, “turn[ed] Fernando Garcia from a victim to a predator.” Id. at 2921. The logic of this argument goes outside the deliberateness special issue. It invited the jury to shift some of the moral responsibility for the crime away from Garcia himself to Garcia’s abusers, whose conduct was at least partly responsible for turning him into a predator. Under this “mitigation theory,” the jury would attribute some of the blame to Garcia’s abusers whether or not Garcia murdered his victim deliberately. Nor did counsel’s vague rhetorical flourish (“I ask for no sympathy for Fernando Garcia”) negate this argument. Contra supra p. 471-72 (Maj.Op.). The jury did not have to conjure feelings of sympathy for Garcia to acknowledge that others shared moral responsibility for the crime and to show Garcia mercy on that basis.
In its rebuttal, the State clearly did not think Garcia’s “mitigation theory” had been entirely bound up with whether he acted deliberately. The State argued that Garcia’s “abuse[] as a child” did not excuse his actions because he had the opportunity to participate in rehabilitative counseling but chose not to do so. Id. at 2927. The State essentially contended that Garcia’s failure to seek rehabilitation broke the causal chain between Garcia’s original abusers and his crime: “His decision, his moral decision, his moral culpability, his life. Nobody else’s.” Id. at 2928 (emphasis added). Under the State’s theory, despite his history of abuse, Garcia became “solely responsible” for being a “rabid dog” when he “rejected every single attempt to ever make him better.” Id. at 2934, 2930. In short, the State argued not only that Garcia acted deliberately but also that Garcia’s attempt to shift some of the moral responsibility to his abusers should be rejected.
Thus, the arguments of the defense and prosecution presented the jury with a significant moral controversy. The Texas special issues, however, “provided an inadequate vehicle for the jury to make a reasoned moral response” to the dispute. Penry II, 532 U.S. at 800, 121 S.Ct. 1910. They did not allow the jury to give effect either to the mitigating evidence Garcia introduced or to the “mitigation theory” he presented at summation.
Stepping back from the record in this case, dissecting a defense attorney’s closing argument to determine whether it presented a “mitigation theory” that went outside the scope of the special issues is a fundamentally misguided inquiry. To begin with, the search for “the case as pre*478sented in the trial court,” supra p. 471 (Maj.Op.), cannot be limited to ten pages of closing argument in a three-thousand page trial transcript. Good advocates offer their theories of a case not just through summation but also through questioning of witnesses and through the very substance of the evidence they present. See Taylor v. United States, 495 U.S. 575, 601, 110 S.Ct. 2143, 109 L.Ed.2d 607 (1990) (stating that courts might discern the Government’s “theory or theories of the case presented to the jury” in “the indictment or other charging paper” or in the “actual proof at trial”). It makes no sense to ignore relevant mitigating evidence that the defendant presented to the jury simply because his counsel, at least according to the majority, failed to draw attention to it under an appropriate theory in closing argument. The defendant is not even required to present a summation. Under the majority’s theory, however, a defendant’s decision to forego summation would invariably forfeit his right to present a Penry claim in this Court “because he did not do so before the jury.” Supra p. 472 (Maj.Op.).
More importantly, requiring the defendant to adopt a “mitigation theory” that goes outside the special issues is paradoxical. As applied, the special issues violate the Eighth Amendment precisely because they limited the defendant’s possible mitigation theories. Under Texas law, Garcia could only argue either that he did not commit his crime deliberately or that he was not a danger to society. So that is what he did. The majority has simply confused the cause with the effect. Garcia’s lawyer did not restrict the mitigation inquiry; rather, Texas’s limited mitigation inquiry under the special issues restricted Garcia’s lawyer.
This brings me to a problematic effect of the majority’s logic. Texas law did not recognize the mitigation theory the majority claims Garcia should have argued to the jury. Therefore, the majority’s theory amounts to a requirement that counsel argue jury nullification at closing to preserve Penry error. Compare United States v. Funches, 135 F.3d 1405, 1409 (11th Cir.1998) (“[Djefense counsel may not argue jury nullification during closing argument”) and United States v. Thomas, 116 F.3d 606, 614 (2d Cir.1997) (“We categorically reject the idea that, in a society committed to the rule of law, jury nullification is desirable or that courts may permit it to occur when it is within their authority to prevent.”).11
The majority’s mandate that defense counsel argue nullification is impossible to square with Penry II. The Supreme Court held in Penry II that courts cannot require jury nullification of the special issues. The majority now requires that defense counsel insert the same “element of capriciousness” into the sentencing process that the Penry II Court found problematic. 532 U.S. at 800, 121 S.Ct. 1910. The majority faults Garcia’s lawyer for not specifically arguing to the jury that Garcia’s life should be spared out of mercy even though doing so, as in Penry II, would have been tantamount to demanding that the jurors “answer the special issues dishonestly” and “violate their oath to render a true verdict.” Id. at 802, 800, 121 S.Ct. 1910 (internal quotation marks and emphasis omitted). By shifting the actor who introduces the nullification element from *479the judge to defense counsel, the majority basically makes an end run around Penry II.
In addition, the majority’s approach is at odds with at least the spirit of Tennard II. In Tennard II, the Supreme Court rejected our constitutional relevance inquiry as creating an impermissible “threshold ‘screening test’” with “no foundation in the decisions of this Court.” 542 U.S. at 283-84, 124 S.Ct. 2562. This rebuke overstated somewhat the novelty of our constitutional relevance test. See Tennard III, 442 F.3d at 253 n. 16. Yet the critique certainly applies here to the majority’s innovation.
As with constitutional relevance, the majority uses its estoppel theory as a threshold screening test, refusing to “consider[ ] whether the jury instructions comported with the Eighth Amendment,” Tennard II, 542 U.S. at 284, 124 S.Ct. 2562, because defense counsel did not explicitly argue outside the special issues at summation. Moreover, the majority’s approach has no foundation in Supreme Court precedent. The Court’s jurisprudence has invariably looked to “the mitigating evidence presented” rather than to the closing arguments of counsel. Tennard II, 542 U.S. at 284, 124 S.Ct. 2562 (discussing Penry I). Indeed, the majority cannot cite a single authority, Supreme Court or otherwise, in support of its estoppel theory during the eleven pages that it discusses that approach. Supra p. 468-72, 472-73 (Maj. Op.). Premises that are fundamental to the majority’s analysis, such as that we must focus on the “mitigation theory Garcia actually made to the jury,” Id. at 468 (emphasis in original), are left completely unsupported.
As previously stated, the Supreme Court has effectively told us that the only Penry inquiries we should undertake are (1) whether the petitioner presented relevant mitigating evidence and (2) whether the jury was able to consider and give sufficient effect to that evidence. Instead of following this mandate, the majority creates another threshold screen.
In conclusion, the trial court failed to present to the jury an adequate vehicle that could give meaningful mitigating effect to Garcia’s history of child abuse. Supreme Court precedent establishes that this failure violated his Eighth Amendment rights. I would hold that the state court’s opposite conclusion was contrary to the Supreme Court’s precedent or, alternatively, an unreasonable application of clearly established law. In accord with Penry I, Penry II, and Tennard II, I would reverse the district court’s decision and remand with instructions to grant ha-beas relief. I respectfully dissent.
. Although Garcia’s history of abuse mitigates, I agree with the majority that Garcia's evidence that he suffered from a pedophilic personality disorder does not meet even the low threshold of relevance set by Tennard II. In my view, instances of pedophilic abuse that Garcia himself suffered as a victim mitigate independently from his alleged pedophilic personality disorder.
. The majority incorrectly suggests that Garcia may not have fully asserted an argument in this Court based on his history of abuse. See Supra p. 470 (Maj.Op.) (stating that Garcia’s abuse argument was limited to a footnote). On the contrary, Garcia made an argument based on his abuse in the main text of at least four different documents submitted to this Court. (Garcia Br. 6-7 & 29 (describing in detail Garcia’s history of abuse and arguing that "evidence of ... his childhood neglect and abuse fell outside the scope of the submitted punishment issues, and therefore required an adequate mechanism — one other than a nullification instruction — in order to permit each juror to give the evidence particularized consideration and effect”)); (Garcia Reply Br. 14); (Garcia Suppl. Br. at 9-12); (Garcia Letter-Brief 3).
. Therefore, Garcia’s history-of-abuse evidence, just like Penry’s, forced him to face the "two-edged sword” that worried the Supreme Court. See Penry I, 492 U.S. at 324, 109 S.Ct. 2934.
. The majority in Hernandez held that, because the child abuse was "unlinked to the offense, it is not mitigating.” 248 F.3d at 349. Since then, the Supreme Court has explicitly rejected this Circuit’s nexus requirement. See Tennard II, 542 U.S. at 284, 124 S.Ct. 2562.
.In this case, as in Penry II, the prosecution's closing argument compounded problems with the nullification instruction by emphasizing that the jurors had to answer the special issue questions honestly: “you told us [at voire dire] ..., ‘if you people bring the kind of evidence that convinces me the answer is yes, I’ll answer the questions yes.' ... You told us very clearly and very emphatically that, 'if you people right here bring the kind of case that convinces me that those questions should be answered yes, I'll answer those questions [the two special . issues] yes.’" (R. at 23:2902-03.) Then later, "Did he do it deliberately? Yes he did. I'm going to ask you to answer that question yes because the evidence says that answer is yes.” Id. at 2906.
. Paradoxically, the majority simultaneously acknowledges that nullification instructions are “unconstitutional” yet denies relief in part because “Garcia had the benefit of a nullification instruction.” Supra p. 471 (Maj.Op.).
. As discussed above, the State's primary argument on appeal was that Garcia's history-of-abuse evidence “could be fully considered within the special issues.” (Dretke Suppl. Br. 2). The majority declines to consider that issue. Supra p. 470 (Maj.Op.).
. The State defended the sentence on the ground that evidence of child abuse is within the scope of the special issues. (Dretke Br. 36 (arguing that Garcia's jury could give effect to his evidence of sexual abuse through the "deliberately” special issue).) The Supreme Court rejected this rationale in Penry I. The Texas Court of Criminal Appeals affirmed Garcia's sentence on the ground that the nullification instruction cured any problem with the special issues. See Garcia v. State, 887 S.W.2d 846, 860 (Tex.Crim.App.1994). The Supreme Court rejected this rationale in Pen-ry II. The district court below applied then-controlling Fifth Circuit precedent and upheld Garcia's sentence on the ground that his mitigating evidence was not "constitutionally relevant.” Garcia v. Dretke, 2003 WL 22144058, *4-5 (N.D.Tex.2003) (unpublished) (holding that Garcia's evidence of "sexual abuse he received as a child” did not give rise to a "uniquely severe permanent handicap”). The Supreme Court rejected this rationale in Temará. II.
. Under Texas law, a third special issue was sometimes submitted. " ‘[I]f raised by the evidence,’ " the court would submit, " 'whether the conduct of the defendant in killing the deceased was unreasonable in response to provocation, if any, by the deceased.’ " Tennard III, 442 F.3d at 246 n. 3 (quoting Tex. Crim. Proc.Code Ann. § 37.071(b) (Vernon 1981)).
. Similarly, traditional principles of equitable estoppel do not bar Garcia from pressing his Penry claim. In a unanimous decision, the Supreme Court recently confirmed the limitations on that doctrine. Ordinarily, it only applies where a party has "succeeded in persuading a court to accept” a position that is "clearly inconsistent” with the one it now espouses. See Zedner v. United States, 547 U.S. -, 126 S.Ct. 1976, 164 L.Ed.2d 749 (2006). That has not happened here.
. It is not clear whether the majority acknowledges this problematic aspect of its reasoning: In one part of its opinion, the majority reasons explicitly that defense counsel should have made arguments "beyond the special issues” pursuant to the "nullification instruction” given in this case. Supra p. 471 (Maj.Op.). Later in its opinion, it denies that its reasoning requires that "defense counsel must argue jury nullification.” Id. at 472.