dissenting from the denial of rehearing en banc:
A panel of the court grants habeas relief to a recidivist armed robber based on its determination that the New York Court of Appeals does not understand, and therefore unreasonably applied, a precept of criminal law known to every law student: the independent-source rule pronounced in United States v. Wade, 388 U.S. 218, 240-41, 87 S.Ct. 1926, 18 L.Ed.2d 1149 (1967), which requires that an in-court identification following a tainted lineup have an independent basis. The panel charges New York’s highest court with operating under the “mistaken impression” that an independent-source inquiry is an issue of fact, rather than a mixed question of fact and law. Young v. Conway, 698 F.3d 69, 84-85 (2d Cir.2012). That conclusion is belied by the New York Court of Appeals’ own precedent. See, e.g., People v. Jackson, 98 N.Y.2d 555, 559, 750 N.Y.S.2d 561, 564, 780 N.E.2d 162 (2002) (stating that “Wade hearing ... involves mixed questions of law and fact.”).
More troubling still, in granting habeas relief based on its own Wade analysis of petitioner’s Fourth Amendment challenge, the panel puts this court at odds with controlling Supreme Court precedents in three respects. First, the panel grants habeas relief to a state prisoner who does not allege, and who cannot demonstrate, that he was denied a full and fair opportunity to litigate ' his Fourth Amendment claim in state court as required by Stone v. Powell, 428 U.S. 465, 96 S.Ct. 3037, 49 L.Ed.2d 1067 (1976). Moreover, by invoking forfeiture to justify this departure from Stone, the panel creates a “circuit split.” See Woolery v. Arave, 8 F.3d 1325, 1326 (9th Cir.1993) (holding that Stone v. Powell precludes federal court enforcement of exclusionary rule through writ of habeas corpus “even though the state has failed to raise the Stone issue”). Second, the panel’s independent-source analysis fails to accord New York courts the AED-PA deference demanded by Harrington v. Richter, — U.S.-, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011). - Third, the panel supports its independent-source determination by reference to social science materials outside .the state court record in contravention of Cullen v. Pinholster, — U.S. -, 131 S.Ct. 1388, 179 L.Ed.2d 557 (2011).
These are errors of exceptional importance that misdirect our habeas jurisprudence, and we should correct them en banc before they are corrected for us. See Fed. R.App. P. 35(a)(2).1 Insofar as the court declines to accord the case further review, I respectfully dissent.
*88I. Background
A. The Charged Robbery
On the night of March 29, 1991, a man armed with a sledgehammer and an axe forcibly entered the home of Lisa and William Sykes and, wielding the axe over the head of the wheelchair-bound Mr. Sykes, threatened to kill him if Mrs. Sykes did not surrender the couple’s money and valuables. Not surprisingly, Mrs. Sykes complied. Among items taken that would later link petitioner Rudolph Young to the robbery were binoculars bearing the name “Sykes,” three watches, and a pair of gloves.
B. Fourth Amendment Challenge to Mrs. Sykes’s Identification
At issue on habeas review was Mrs. Sykes’s in-court identification of Young as the robber. The problem was not that her identification was infected by suggestive police procedures, so as to raise Fourteenth Amendment due process concerns. Rather, it was that she first identified Young at a lineup found to be the fruit of an unlawful arrest. See People v. Young, 255 A.D.2d 905, 683 N.Y.S.2d 677 (4th Dep’t 1998) (holding that Young’s participation in lineup resulted from arrest lacking probable cause).2
The admission of this tainted line-up identification at Young’s first trial prompted reversal of his conviction on direct appeal. See id. at 906, 688 N.Y.S.2d at 678. To ensure that this Fourth Amendment error would not also taint any in-court identification on retrial, the trial court had to determine whether Mrs. Sykes retained a basis for making such an identification *89independent of the lineup. See United States v. Wade, 388 U.S. at 240-41, 87 S.Ct. 1926.3
After a Wade hearing, the trial court determined that Mrs. Sykes had the requisite independent basis for making a trial identification. Although the robber had worn a blanket over his body and a scarf over the lower part of his face, the court credited Mrs. Sykes that she “really stared” at him for five tó seven minutes at close range in well-lit conditions, to see if he was someone she recognized. Hr’g Tr. 34, People v. Young, Ind. No. 91-0402 (N.Y.Sup.Ct. Mar. 11, 1999). He was not. Nevertheless, as a consequence of her attention, Mrs. Sykes formed a strong mental image of the robber’s eyes. See id. at 35 (stating that, for nights after robbery, she would awaken “seeing those eyes in nightmares”). Mrs. Sykes’s deliberate focus on the robber’s face was evident even to him, because he told her at one point, “Don’t look at my face.” Id. at 32.
In initially describing the robber, Mrs. Sykes told police that he was a black male in his twenties, approximately five-feet-ten-inches tall, with a medium build. See id. at 34. Although Young is black, he was 34 years old at the time of the Sykes robbery, and is six-feet to six-feet-one-inch tall. See Trial Tr. 302-03, People v. Young, Ind. No. 91-0402 (N.Y.Sup.Ct. Jan. 10, 12-14, 2000). The state trial court, with the advantage of seeing and hearing Mrs. Sykes, did not think these discrepancies indicated that her in-court identification would be influenced by the lineup. Nor was it so persuaded by Mrs. Sykes’s inability to help create a composite sketch of the robber soon after the crime, crediting her explanation that she had mistakenly understood that she would have had to have seen the robber’s full face to help with a sketch. See Mar. 11, 1999 Hr’g Tr. 34. The court similarly credited Mrs. Sykes that she was unable to identify Young from a photo array two days before the lineup because a “photograph is not a real person” and “did not look real” to her. Id. at 64. She stated that she was able to identify Young from among six men in the lineup based on his eyes and voice. Indeed, she asserted that she could have made the lineup identification based on Young’s eyes alone. See id. at 77-78. She maintained that, even eight years after the crime, she had a recollection of the robber’s eyes that was independent of seeing Young in the lineup, testimony credited by the trial court. See id.
C. Retrial and Sentencing
At Young’s retrial, the defense vigorously cross-examined Mrs. Sykes about the *90reliability of her identification. See Trial Tr. 59-88. ■ It was not, however, permitted to offer expert testimony on the reliability of eyewitness identifications. Meanwhile, the prosecution linked Young to the Sykes robbery through two witnesses, one of whom testified that, in the month following the robbery, Young had given her binoculars bearing the name “Sykes” and three watches to sell,4 and- the other, of whom identified a pair of gloves belonging to the Sykeses as ones that she had found left in her home at a time when Young was a frequent guest. See id. at 136, 261.
After the jury found Young guilty, the sentencing judge heard evidence of Young’s extensive criminal history — including convictions for murder in North Carolina, robbery in Georgia, and robbery and burglary in New York—as well as his admission to committing 140 to 150 burglaries between 1988 and 1991. See Sent. Hr’g Tr. 76, 114-15, 128, People v. Young, Ind. No. 91-0402 (N.Y.Sup.Ct. July 14, 2000); App’x of State Court Records 181, Young v. Conway, 761 F.Supp.2d 59 (W.D.N.Y.2011) (No. 07-CV-6047 DGL). The court sentenced Young as a persistent felony offender, see N.Y. Penal Law § 70.10; N.Y.Crim. Proc. Law § 400.20, to a prison term of 15 years to life. See Sent. Tr. 8-10, 45, People v. Young, Ind. No. 91-0402 (N.Y.Sup.Ct. Sept. 7, 2000).
D.Appellate Review
On direct appeal, Young argued, among other things, that Wade precluded Mrs. Sykes’s in-court identification. The Appellate Division rejected the argument, holding by a vote of three to two that the prosecution had carried its burden to establish that the trial identification ■ rested on a basis independent of the Fourth Amendment-tainted lineup. See People v. Young, 20 A.D.3d 893, 798 N.Y.S.2d 625 (4th Dep’t 2005).
The New York Court of Appeals granted review and affirmed by a vote of six to one. See People v. Young, 7 N.Y.3d 40, 817 N.Y.S.2d 576, 850 N.E.2d 623 (2006). In a decision largely devoted to explaining why the failure to admit defense expert testimony did not warrant reversal, see id. at 44-46, 817 N.Y.S.2d at 578-80, 850 N.E.2d 623, the court recognized Young’s Wade argument to be that, “as a matter of law ... it was impossible to find by the requisite clear and convincing evidence that the lineup would not influence [Mrs. Sykes’s] in-court identification,” id. at 44, 817 N.Y.S.2d at 578, 850 N.E.2d 623. The Court concluded that the argument failed “on an issue of fact,” i.e., the lower courts’ finding that Mrs. Sykes possessed a recollection of the robbery independent from the tainted lineup. Id. It explained, “[t]here is support in the record for their finding, and we may not disturb it.” Id.
E.Section 225k Petition
On consent of the parties, Young’s habe-as petition was, decided by a magistrate judge, who ruled in Young’s favor and vacated his convictions for the Sykes robbery. See Young v. Conway, 761 F.Supp.2d at 77. A panel of this court affirmed but remanded to allow the state to retry Young without Mrs. Sykes’s identification. See Young v. Conway, 698 F.3d at 89.
II. Discussion
A. Stone v. Powell Precludes Habeas Review of Young’s Fourth Amend- ■ ment Challenge
For almost 40 years, it has been well established that “where the State has pro*91vided an opportunity for full and fair litigation of a Fourth Amendment claim, a state prisoner may not be granted federal habe-as corpus relief on the ground that evidence obtained in an unconstitutional search or seizure was introduced at his trial.” Stone v. Powell, 428 U.S. at 494, 96 S.Ct. 3087 (emphasis added; footnote omitted). In this case, Young did not plead — much less prove — that he was denied a full and fair opportunity to litigate his Fourth Amendment-based Wade claim in New York’s courts. To the contrary, the record shows that he was afforded such an opportunity at all three levels of state court review. In these circumstances, I respectfully submit that his Fourth Amendment Wade challenge is not a proper subject for habeas review.
In concluding otherwise, the panel observes that (1) Stone’s bar is not jurisdictional, and (2) the state forfeited its claim to the bar by raising it on appeal without having done so in the district court. See Young v. Conway, 698 F.3d at 85-87. While the jurisdictional point is beyond dispute, see Stone v. Powell, 428 U.S. at 494 n. 37, 96 S.Ct. 3037; accord Withrow v. Williams, 507 U.S. 680, 686, 113 S.Ct. 1745, 123 L.Ed.2d 407 (1993), the same cannot be said about the panel’s reliance on forfeiture.
In support, the panel quotes the following language from Stone:
In sum, we hold only that a federal court need not apply the exclusionary rule on habeas review of a Fourth Amendment claim absent a showing that the- state prisoner was denied an opportunity for a full and fair litigation of that claim at trial and on direct review. Our decision does not mean that the federal court lacks jurisdiction over such a claim....
428 U.S. at 494 n. 37, 96 S.Ct. 3037, quoted in Young v. Conway, 698 F.3d at 85. In fact, the last sentence continues with language that defeats the panel’s conclusion:. “Our decision does not mean .that the federal court lacks jurisdiction over such a claim, but only that the application of the [exclusionary] rule is limited to cases in which there has been both such a showing and a Fourth Amendment violation.” Id. (emphasis added). In short, Stone’s limitation may be prudential rather than jurisdictional, but it is categorical. Habeas review of Fourth Amendment claims is “limited” by a prerequisite: “a showing that the state prisoner was denied an opportunity for a full and fair litigation of that claim” in the state courts. Id. Unless and until that showing is made, a federal habeas court “may not” apply the exclusionary rule to upset a state conviction. Id. at 494, 96 S.Ct. 3037.5
*92Our precedent has heretofore adhered to this limitation, recognizing that “Fourth Amendment claims are not reviewable by the federal courts when raised in a petition brought under § 2254 unless the state prisoner shows that he or she has not had a full and fair opportunity to litigate that claim in the state court.” Graham v. Costello, 299 F.3d 129, 133-34 (2d Cir.2002) (emphasis added). Indeed, Graham signals that courts may not excuse petitioners from this burden: “[T]he bar to federal habeas review of Fourth Amendment claims is permanent and incurable absent a showing that the state failed to provide a full and fair opportunity to litigate the claim.” Id. at 134 (emphasis added); see also Gates v. Henderson, 568 F.2d 830, 840 (2d Cir.1977) (en banc) (“Stone v. Powell ... holds that we have no authority to review the state record and grant the writ simply because we disagree with the result reached by the state courts.”). Sister circuits agree that it is a petitioner’s burden “to plead and prove” that his case fits within the narrow exception to the Stone bar. Davis v. Blackburn, 803 F.2d 1371, 1372 (5th Cir.1986); see Sanna v. DiPaolo, 265 F.3d 1, 8 (1st Cir.2001); Doleman v. Muncy, 579 F.2d 1258, 1266 (4th Cir.1978).
The panel’s reliance on respondent’s failure to invoke Stone in the district court to absolve Young from this burden not only lacks support in Stone jurisprudence but also creates a “circuit split.” In Woolery v. Arave, the Ninth Circuit acknowledged that Stone’s bar was prudential, but nonetheless concluded that it was “founded in policy considerations that oblige the court to raise the issue sua sponte if the state neglects to assert it.” 8 F.3d at 1327. Thus, “absent a showing that the state denied a full and fair opportunity to litigate the Fourth Amendment claim, the rule of Stone v. Powell precludes the federal court from enforcing the exclusionary rule through the writ of habeas corpus even though the state has failed to raise the Stone issue.” Id. at 1326.6
Notably, Woolery reached this conclusion in the case of a petitioner who had at least pleaded the denial of an opportunity to litigate his Fourth Amendment claim in state court. In dissent, Judge Reinhardt thought that the state’s failure to respond to this pleading permitted the district *93court to find that the state had “conceded” the Stone exception and to reach the merits. Id. at 1331 (Reinhardt, J., dissenting). The majority, however, concluded that petitioner’s pleading meant simply that-he had not “waived” his claim to a Stone exception; thus, on remand, he was entitled to demonstrate the pleaded denial. Id. at 1328 n. 3. By contrast, Young has never alleged that he was denied an opportunity to litigate his Fourth Amendment claim in state court.7 Nor would the record permit him to do so.
As further justification for not holding Young to his Stone burden, the panel suggests that identification challenges fall outside Stone’s paradigm because identification evidence “lacks the typical indicia of reliability that ordinarily weigh against re-litigating a Fourth Amendment claim on collateral review.” Young v. Conway, 698 F.3d • at 87; but see id. at 80 (acknowledging that “much eyewitness identification testimony is reliable”); ante at 80 (reiterating that “eyewitness identification testimony is typically reliable”). The contention is not persuasive and contravenes the instructions of the Supreme Court. Stone does not admit exceptions based on the types of evidence subject to Fourth Amendment challenges. Having made the general determination that application of the exclusionary rule to collateral Fourth Amendment challenges would deprive fact-finders of “typically reliable” evidence that was “often the most probative information bearing on ... guilt or innocence,” Stone v. Powell, 428 U.S. at 490, 96 S.Ct. 3037 (emphasis added), while having only a “minimal” deterrent effect on Fourth Amendment violations, id. at 486, 96 S.Ct. 3037, the Supreme Court erected a categorical bar: “a state prisoner may not be granted federal habeas corpus' relief on the ground that evidence obtained in an unconstitutional search or seizure was introduced at his trial” except in those circumstances where the state has failed to provide “an opportunity for full and fair litigation” of the claim, id. at 494, 96 S.Ct. 3037. Stone specifically rejected the contention that “isolated cases,” in which the costs and benefits of applying the exclusionary rule might be different, could justify lifting this categorical bar. Id.
In sum, because Young has -not pleaded, and the record would not permit him to prove, that he was denied an opportunity fully and fairly to litigate his Fourth Amendment Wade claim in the New York courts, this court should clarify en banc that the claim is barred from federal habe-as review by Stone v. Powell.
B. The Panel’s Identification of Wade Error Misconstrues New York Precedent and Defies Harrington v. Richter
Even if Stone did not bar habeas review of Young’s Fourth Amendment challenge, the panel’s award in his favor fails to comport with 28 U.S.C. § 2254(d)(1). That section states that federal habeas relief “shall not be granted” to a state prisoner with respect to any claim adjudicated on the merits in state court unless the adjudication “resulted in a decision that was [1] contrary to, or [2] involved an unreasonable application of, clearly established Federal law,” as pronounced by the Supreme Court. Although the panel purports to find both conditions satisfied here, see Young v. Conway, 698 F.3d at 84-85, the first conclusion is defeated by New York precedent; the second by Harrington v. Richter, 131 S.Ct. 770.
*941. The New York Court of Appeals Does Not Misunderstand the Wade Rule
In order to find New York’s rejection of Young’s in-court identification challenge “contrary to” established Supreme Court precedent, ‘ the panel concludes that the New York Court of Appeals misunderstands the Wade rule, operating under the “mistaken impression that' the independent source inquiry [is] an ‘issue of fact,’ ” rather than a mixed question of law and fact. Young v. Conway, 698 F.3d at 84-85. It is no small matter for a panel of this court to charge New York’s highest court with failing to understand so well-established and frequently applied a-Supreme Court precedent as Wade. In fact, the conclusion is wholly unwarranted in light of precedent of the New York Court of Appeals — ignored by the panel — expressly recognizing that “a Wade hearing dealing with the propriety of a lineup identification involves mixed questions of law and fact.” People v. Jackson, 98 N.Y.2d at 559, 750 N.Y.S.2d at 564, 780 N.E.2d 162.
The New York Court of Appeals effectively acknowledged that Young’s Wade challenge raised a question of law as well as fact when it summarized his argument on appeal to contend that it was “impossible” as a matter of law “to find by the requisite clear and convincing evidence that the lineup .would not influence [Mrs. Sykes’s] in-court identification.” People v. Young, 7 N.Y.3d at 44, 817 N.Y.S.2d at 578, 850 N.E.2d 623; see generally People v. Léonti 18 N.Y.2d 384, 389, 275 N.Y.S.2d 825, 829, 222 N.E.2d 591 (1966) (holding that, despite limits on its review of pure questions of fact, see N.Y. Const, art. 6, § 3(a), when presented with mixed question of law and fact, Court of Appeals could decide “whether the evidence adduced meets the [legal], standard required”). Although our court’s panel concludes otherwise based on the New York Court of Appeals’ observation that Young’s Wade claim was effectively resolved by the lower courts’ resolution of “an issue of fact,” People v. Young, 7 N.Y.3d at 44, 817 N.Y.S.2d at 578, 850 N.E.2d 623, quoted in Young v. Conway, 698 F.3d at 85, that language of New York’s highest court echoes rather than departs from controlling Supreme Court precedent. In United States v. Crews, the Supreme Court explained its reversal of the District of Columbia Court of Appeals’ en banc preclusion of a Fourth Amendment-tainted identification by reference to trial court factfinding that was supported by the record: “[T]he trial court expressly found that the witness’ courtroom identification rested on an independent recollection of her initial encounter with the assailant, uninfluenced by the -pretrial identifications, and this determination finds ample support in the record.” 445 U.S. at 473, 100 S.Ct. 1244. In short, the Supreme Court itself recognizes that once a factual finding of an independent source is made, a reviewing court need only determine that the finding has support in the record to satisfy the Wade inquiry.
Our own precedent follows this approach. In Karavos Compania Naviera S.A. v. Atlantica Export Corp., 588 F.2d 1 (2d Cir.1978) (Friendly, J.), at the same time that we cautioned against conflating questions of law and fact, we added a “qualification”: a lower court’s resolution of a mixed question “will ordinarily stand unless the lower court manifests an incorrect conception of the applicable law.” Id. at 8 (internal quotation marks omitted). In other words, where there is no question as to the applicable legal standard, the resolution of a mixed question of law and fact is often determined by the latter, as long as there is record support for the finding. That is all I understand the New *95York Court of Appeals to have been saying in Young’s case.8
Accordingly, this court should withdraw the unwarranted charge that the New York Court of Appeals does not understand Wade and the conclusion the panel derives therefrom, i.e., that New York’s affirmance of Young’s conviction was “contrary to” federal law as clearly established by the Supreme Court. 28 U.S.C. § 2254(d)(1).
2. The Panel’s Failure To Adhere to Harrington v. Richter
Insofar as the panel also charges the New York Court of Appeals with “an unreasonable application of the correct standard” enunciated in Wade, Young v. Conway, 698 F.3d at 85, that conclusion defies Harrington v. Richter, 131 S.Ct. 770. In there construing the “unreasonable application” requirement of 28 U.S.C. § 2254(d)(1), the Supreme Court reiterated that not every erroneous application of clearly established federal law is an unreasonable application of that law. See id. at 786; Williams v. Taylor, 529 U.S. 362, 365, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (“[T]he most important point” in applying AEDPA deference “is that an unreasonable application of federal law is different from an incorrect application of federal law.” (emphasis in original)). A state court’s application of federal law cannot be deemed unreasonable under § 2254(d)(1) “so long as ‘fairminded jurists could disagree’ on the correctness of the state court’s decision.” Harrington v. Richter, 131 S.Ct. at 786 (quoting Yarborough v. Alvarado, 541 U.S. 652, 664, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004)).
Here, one Court of Appeals and two Appellate Division judges thought that the record did not satisfy the independent-source requirement. But six Court of Appeals judges, three Appellate Division judges, and the trial judge — the only one who actually saw the identifying witness— concluded that it did. A review of the various state court opinions precludes a conclusion that any of these jurists was other than “fairminded” in reaching the conflicting determinations. In such circumstances, the fact that the panel shares the minority view is not enough to denominate the majority view “unreasonable.” See id. (“It bears repeating that even a strong case for relief does not mean the state court’s contrary conclusion was unreasonable.”). Indeed, the three judges who join in this dissent think it entirely reasonable for the state courts to have resolved the Wade question in favor of allowing Mrs. Sykes to make an in-court identification. See infra at 96-97.9
*96Judge Parker plainly disagrees with Harrington’s strict standard.of unreasonableness, concerned that few, if any, habe-as cases will satisfy it. See ante at 85. His concern, however, does not render that standard any less controlling on this court; As Harrington bluntly states, “[i]f this standard is difficult to meet, that is because it was meant to be.” 131 S.Ct. at 786. “Section 2254(d) reflects the view that habeas corpus is a guard against extreme malfunctions in the state criminal justice systems, not a substitute for ordinary error correction through appeal.” Id. (internal quotation marks omitted). Thus, a state prisoner cannot secure § 2254 relief unless “the state court’s ruling on the claim being presented in federal court was so lacking in justification that there was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Id. at 786-87.
New York’s refusal to preclude Mrs. Sykes’s in-court identification of Young cannot be deemed “beyond any possibility for fairminded disagreement.” Id. That conclusion is reinforced by the'fact that Wade’s rule — that an in-court identification following a tainted (as opposed to suggestive) lineup must have an independent basis — is general rather than specific. “[T]he more general the rule” at issue in a § 2254 petition, “the more leeway [state] courts have in reaching outcomes in case-by-case determinations.” Yarborough v. Alvarado, 541 U.S. at 664, 124 S.Ct. 2140. To be sure, Wade identifies factors that properly inform the independent-source analysis, see 388 U.S. at 241, 87 S.Ct. 1926, but it does not tell courts how to weigh those factors' or the record evidence that informs them. Thus, it cannot be deemed an unreasonable application of “a specific legal rule ... squarely established by th[e Supreme] Court,” Knowles v. Mirzayance, 556 U.S. 111, 122, 129 S.Ct. 1411, 173 L.Ed.2d 251 (2009), that the state courts assigned more weight than the panel would have to Mrs. Sykes’s conscious attention to the robber, her identification of him from among six men in a non-suggestive lineup, and her professed strong recollection of his eyes,10 while assigning less weight than the panel would have to the fact that the robber was masked,11 that Mrs. Sykes underestimated both his age and height, and that she did not assist in producing a composite sketch, see generally Marshall v. Rodgers, 133 S.Ct. at 1450 (holding that state court ruling is neither *97contrary to nor an unreasonable application of general standards established by Supreme Court because state court’s approach to standards differed from that of federal court on direct appeal).12 As Harrington explains, “[u]nder § 2254(d), a ha-beas court must determine what arguments or theories supported or, ... could have supported, the state court’s decision; and then it must ask whether it is possible fairminded jurists could disagree that those arguments or theories are inconsistent with the holding in a prior decision of this Court.” 131 S.Ct. at 786.
Although the panel doubts that Mrs. Sykes — a woman it has never seen or heard testify — could have formed a sufficient independent basis for identifying Young during the robbery, there is considerable record evidence that “could have supported” the state courts’ contrary conclusion. Id. To begin, Mrs. .Sykes’s unhesitating identification of Young from among six men in a non-suggestive lineup necessarily derived from a source independent of the lineup. See United States v. Crews, 445 U.S. at 473 n. 18, 100 S.Ct. 1244 (recognizing that identification made during non-suggestive, Fourth Amendment-tainted lineup can support independent-basis finding). That the independent source was the robbery itself was, moreover, reinforced by evidence corroborating Young’s identity as the Sykes robber. In addition to trial evidence linking Young to items stolen from the Sykes’s home, this corroborating evidence included Young’s extensive record for robbery and burglary, which was known to the state courts at the time of the Wade ruling. See Brisco v. Ercole, 565 F.3d 80, 94-95 (2d Cir.2009) (recognizing state courts’ authority to consider corroborating evidence in deciding whether to exclude identification evidence); id. at 96 (Korman, J., concurring) (“Corroborative evidence of guilt goes to the heart of the Supreme Court cases that initially addressed the problem posed by eyewitness identifications.”). Indeed, when ruling on Young’s Wade challenge, the state courts knew that, within days of the Sykes robbery, Young had burglarized two other homes under similar circumstances, including masking his face during at least one of the robberies. See People v. Young, 94 N.Y.2d 171, 175, 701 N.Y.S.2d 309, 311, 723 N.E.2d 58 (1999) (noting that, at initial sentencing for these other crimes, trial judge characterized Young as “confrontational burglar,” whose willingness to engage in violence made him “any homeowner’s worst nightmare” and “clear threat to the community” (alteration and internal quotation marks omitted)).
On this record, the state courts’ Wade determination — at all three levels of review — that Mrs. Sykes’s in-court identification of Young would rest on her independent recollection of the crime and not the tainted lineup can hardly be said to reflect an error so “well understood and comprehended in existing law” as to be “beyond any possibility for fairminded disagreement.” Harrington v. Richter, 131 S.Ct. at 787. Much less does it manifest the sort of “extreme malfunetion[ ] in the state criminal justice system[]” that warrants habeas relief. Id. at 786. Thus, the court should correct the panel’s Harrington error en banc.
*98C. Reliance on Social Science Literature Outside the State Court Record Violates Cullen v. Pinholster
In holding that New York courts unreasonably applied Wade, the panel cites extensively to social science literature discussing risks inherent in eyewitness identifications. See Young v. Conway, 698 F.3d at 77-84. The vast majority of this literature, largely referenced in a brief filed by amicus curiae, The Innocence Project, was not part of the state court record.13 Indeed, much of the literature had not even been published when the New York Court of Appeals rendered its decision in this case. See id. (citing six studies published after May 9, 2006). The same holds true for cited state court analyses of literature on eyewitness identifications. See id. at 79 & n. 9, 82 (citing State v. Guilbert, 306 Conn. 218, 49 A.3d 705 (2012), State v. Henderson, 208 N.J. 208, 27 A.3d 872 (2011), and People v. Santiago, 17 N.Y.3d 661, 934 N.Y.S.2d 746, 958 N.E.2d 874 (2011)). The review of such extrinsic' evidence by a federal habeas court is foreclosed by Cullen v. Pinholster, which holds “that review under § 2254(d)(1) is limited to the record that was before the state court that adjudicated the claim on the merits.” 131 S.Ct. at 1398 (emphasis added); see id. at 1399 (“Our cases emphasize that review under § 2254(d)(1) focuses on what a state court knew and did.”).
In an effort to evade Pinholster, the panel suggests that its “conclusion that Mrs. Sykes’s in-court identification lacked an independent source is reinforced, but not compelled or controlled by, the literature we discuss.” Young v. Conway, 698 F.3d at 79 n. 8. The distinction is unconvincing; the panel opinion is largely devoted to a discussion of social science. See id. at 77-84. In any event, Pinholster1 s holding is not confined to the determinative bases for a habeas award; it applies generally to habeas “review.” 131 S.Ct. at 1398-99. Thus, however much the panel may disclaim its social science discussion as dictum, see Young v. Conway, 698 F.3d at 79 n. 8, the fact remains that it “review[ed]” materials not before the state courts and cited those materials to support its habeas decision, Cullen v. Pinholster, 131 S.Ct. at 1398-99. This is impermissible after Pinholster. See id.; see also Ryan v. Gonzales, — U.S.-, 133 S.Ct. 696, 704, 708 & n. 14, 184 L.Ed.2d 528 (2013) (citing Pinholster in unanimously rejecting argument that prisoner’s mental incompetency prevented counsel from pursuing habeas challenge (based, in part, on tainted identification) because § 2254(d) proceedings are wholly “backward-looking” and “record-based”).14
The panel’s Pinholster error is compounded, moreover, by the fact that the cited extrinsic materials do not speak to the independent-source question at issue. Their focus is the general reliability of eyewitness identifications. Reliability is a *99question that the law generally leaves for jury resolution, see Perry v. New Hampshire, — U.S.-, 132 S.Ct. 716, 727-28, 181 L.Ed.2d 694 (2012), except when the government subjects a witness to suggestive procedures, in which case due process conditions the witness’s in-court identification on a showing of independent reliability, see Neil v. Biggers, 409 U.S. 188, 198-99, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972); Stovall v. Denno, 388 U.S. 293, 302, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967). That is not this case. Young specifically disavowed any suggestivity argument, complaining only of Fourth Amendment taint. See supra at 88 n. 2. This triggered Wade’s independent-source inquiry, not due process’s independent-reliability inquiry. While the distinction between the two is not always easily discerned, see 6 Wayne R. LaFave, Search & Seizure: A Treatise on the Fourth Amendment § 11.4(g) (5th ed.2012), that distinction informs the Supreme Court’s recognition that an identification made from a non-suggestive lineup can support an independent-basis finding, see United States v. Crews, 445 U.S. at 473 n. 18, 100 S.Ct. 1244, whereas a suggestive lineup is the essence of the problem requiring a showing of independent reliability.15 Thus, the panel injects considerable confusion into our Wade jurisprudence by relying on social science literature addressing the general reliability of eyewitness identifications to make an independent-source assessment.16
Further, even with respect to reliability, the cited social science literature speaks only in terms of probabilities, not absolutes.17 At best, social science has determined that in some cases, certain identified variables might make some eyewitness identifications less reliable. Such probabilities hardly translate into legal maxims for appellate — much less collateral — review of in-court identification challenges.18
What such social science literature might inform are trial court decisions as to *100whether juries should hear expert testimony about the general reliability of eyewitness identifications or be given special instructions on the subject.19 Courts have reached different answers on these questions. Compare State v. Henderson, 208 N.J. at 298, 27 A.3d 872 (concluding that special jury instruction was preferable to expert testimony in assisting jury to assess identification evidence), with State v. Guilbert, 306 Conn, at 263, 49 A.3d 705 (holding that defendant should be permitted to offer expert testimony if state relies on eyewitness identification, deeming comprehensive jury instructions inadequate substitute); see also Daniel Patrick Moynihan, Social Science & the Courts, 54 Pub. Int. 12, 19-20 (Winter 1979) (observing that “social science is rarely dispassionate, and social scientists are frequently caught up in the politics which their work necessarily involves”).
In the ongoing debate about the proper trial use of social science studies on the reliability of eyewitness identifications, our own court has thus far come out in favor of district courts’ broad discretion to exclude such evidence, see United States v. Lumpkin, 192 F.3d 280, 289 (2d Cir.1999), and to deny eyewitness-specific jury instructions, see United States v. Luis, 835 F.2d 37, 41 (2d Cir.1987). But see United States v. Brownlee, 454 F.3d 131, 144 (3d Cir.2006) (identifying error in exclusion of expert testimony when primary issue before jury was reliability of four eyewitnesses); United States v. Hall, 165 F.3d 1095, 1119-20 (7th Cir.1999) (Easterbrook, J., concurring) (supporting eyewitness-specific jury charge). It seems curious, to say the least, that, at the same time we do not compel the presentation of such expert opinions to federal juries that actually see and hear in-court identifications, we would ourselves rely extensively on such opinions to conclude that an in-court identification by an eyewitness whom we have never seen or heard necessarily lacked an independent basis. That we make this determination on collateral review of state courts to which we owe AEDPA deference is more curious still. That we make it with respect to a Fourth Amendment claim barred from federal habeas review by Stone v. Powell is most curious of all.
III. Conclusion
In sum, this case warrants en banc review to correct panel errors with respect to three Supreme Court precedents:
*1011. By reviewing a Fourth Amendment claim in circumstances where the petitioner has neither pleaded nor proved the denial of a full and fair opportunity to litigate that claim in state courts — and where the record in fact shows that he had that opportunity — the panel fails to adhere to Stone v. Powell, 428 U.S. 465, 96 S.Ct. 3037, and creates a “circuit split.”
2. To justify conducting its own Wade analysis and granting habeas relief, the panel both (a) mistakenly charges the New York Court of Appeals with failing to understand Wade and (b) fails to accord the state courts’ Wade analysis of the independent basis for Mrs. Sykes’s in-court identification the AEDPA deference required by Harrington v. Richter, — U.S.-, 131 S.Ct. 770, 178 L.Ed.2d 624.
3. The panel supports its habeas grant by reference to social science studies not in the state court record, contrary to Cullen v. Pinholster, — U.S.-, 131 S.Ct. 1388, 179 L.Ed.2d 557.
Accordingly, I respectfully dissent from the denial of rehearing en banc.
. Although the panel hints that the lineup was suggestive because Mrs. Sykes had previously viewed a photo array including Young's picture, see Young v. Conway, 698 F.3d at 83-84, Young affirmatively waived that argument in both the state and district courts. See Hr’g Tr. 3-4, People v. Young, Ind. No. 91-0402 (N.Y.Sup.Ct. May 4, 1999) (“This is not an issue of a suggestive identification procedure which is under the ... right to due process.”); Young v. Conway, 761 F.Supp.2d 59, 73 (W.D.N.Y.2011) ("[Young] emphasizes that he is ‘arguing that his lineup was illegal, not suggestive, therefore he is under the [Wade] standards and -not the standard set forth in Neil v. Riggers, 409 U.S. 188 [93 S.Ct. 375, 34 L.Ed.2d 401 (1972)].'”). Moreover, courts have consistently held that a witness’s viewing of a suspect in successive pre-trial identification procedures is not inherently suggestive. See, e.g., Dunlap v. Burge, 583 F.3d 160, 165 (2d Cir.2009); United States v. Concepcion, 983 F.2d 369, 379 (2d Cir.1992); People v. Peterkin, 81 A.D.3d 1358, 1359, 921 N.Y.S.2d 744, 745 (4th Dep’t 2011); People v. Munoz, 223 A.D.2d 370, 370, 636 N.Y.S.2d 313, 314 (1st Dep’t 1996).
. Wade established this independent-source rule in the context of a lineup tainted by denial of the Sixth Amendment right to counsel. See 388 U.S. at 221, 240-41, 87 S.Ct. 1926. Among the factors Wade identified as relevant to making an independent-source determination are (1) the witness's opportunity to observe the alleged criminal act, (2) the existence of any discrepancy between any pre-lineup description by the witness and the defendant’s actual appearance, (3) any identification by the witness prior to the lineup of a person other than defendant, (4) the witness’s identification by picture of the defendant pri- or to the lineup, (5) the witness's failure to identify the defendant on a prior occasion, and (6) the lapse of time between the alleged crime and the lineup identification. See id. at 241, 87 S.Ct. 1926 (stating that factors are illustrative).
The Supreme Court subsequently applied Wade to an identification challenge grounded in a Fourth Amendment violation in United States v. Crews, 445 U.S. 463, 472-73, 477, 100 S.Ct. 1244, 63 L.Ed.2d 537 (1980) (reinstating conviction based on in-court identification supported by independent recollection of crime untainted by intervening identifications that were fruits of unlawful arrest). Thus, references to the Wade rule in this opinion should' be understood to incorporate Crews as well.
. The witness explained how, at police direction, she recovered the binoculars from the person to whom she had sold them, and the binoculars were put into evidence at trial. See Trial Tr. 136.
. Any number of non-jurisdictional rules are categorical. AEDPA categorically precludes habeas grants except where state rulings are "contrary to” or "an unreasonable application of” clearly established federal law. 28 U.S.C. § 2254(d)(1); see also Eze v. Senkowski, 321 F.3d 110, 120-21 (2d Cir.2003) (rejecting petitioner's argument that state waived AEDPA deference by not referencing standard before district court, and observing that § 2254(d)(1) “contains unequivocally mandatory language”). Harrington v. Richter, 131 S.Ct. at 786, categorically requires a ha-beas petitioner to show that a challenged ruling is unreasonable as well as erroneous. Cullen v. Pinholster, 131 S.Ct. at 1398, categorically limits habeas review to the state court record. These rules not only allocate burdens between parties to the litigation; they also ensure proper deference to state courts within our federal system to resolve matters — even constitutional matters — pertaining to their own criminal law. See Calderon v. Thompson, 523 U.S. 538, 555-56, 118 S.Ct. 1489, 140 L.Ed.2d 728 (1998) ("Federal habeas review of state convictions frustrates both the States' sovereign power to punish offenders and their good-faith attempts to honor constitutional rights.” (internal quotation marks omitted)); see also Harrington v. Richter, 131 S.Ct. at 786 (cautioning that federal habeas guards “against extreme malfunc*92tions in the state criminal justice systems” and is "not a substitute for ordinary error correction through appeal” (internal quotation marks omitted)). Such deference plainly informs Stone’s demand for a showing that petitioner was denied a full and fair opportunity to litigate a Fourth Amendment claim in state courts as a precondition to federal review. In the absence of such a showing, the state courts are no less entitled to deference simply because a respondent custodian was negligent in the timing of his invocation of Stone.
. The panel suggests that the Ninth Circuit is but one of four courts to have addressed— and the only one to have found — a sua sponte obligation by courts to apply the Stone bar. See Young v. Conway, 698 F.3d at 86 n. 10. In fact, in Davis v. Blackburn, the Fifth Circuit recognized an obligation "to apply Stone as a prudential limitation on the exercise of [its] jurisdiction ..., even if it must be raised sua sponte.” 803 F.2d at 1372-73. The other two cases cited by the panel reference the issue only in dicta and hardly support ignoring the Stone bar in this case. See Tart v. Massachusetts, 949 F.2d 490, 497 n. 6 (1st Cir.1991) (describing it as "perilous” for petitioner to rely on state's failure to raise Stone bar in light of caselaw suggesting court may do so sua sponte, but concluding no need to do so because petitioner's Fourth Amendment claim was meritless in any event); Wallace v. Duckworth, 778 F.2d 1215, 1219 n. 1 (7th Cir.1985) (observing that Stone's non-jurisdictional bar need not be raised sua sponte, but denying petition for failure to exhaust remedies); but see id. at 1226 (Posner, J.,.concurring) (identifying no need to consider exhaustion because petition "is in any event frivolous in light of Stone v. Powell ”).
. Young failed to assert that he was denied an opportunity to litigate his Fourth Amendment claim in the state courts not only in his initial pro se § 2254 petition, but also in his counseled brief to this court responding to the state's invocation of the Stone bar.
. The New York Court of Appeals’ decision precludes any suggestion that it failed to understand that the Wade standard requires "clear and convincing evidence that the lineup would not influence [Mrs. Sykes’s] in-court identification.” People v. Young, 7 N.Y.3d at 44, 817 N.Y.S.2d at 578, 850 N.E.2d 623. The panel may disagree with the Court of Appeals’ conclusion that the evidence was sufficient to satisfy this standard, but that goes to the reasonableness of the court's application of the Wade rule, see infra Part B.2; it does not demonstrate that the ruling was "contrary to” Wade, 28 U.S.C. § 2254(d)(1).
. In urging otherwise, Judge Parker suggests that even the district attorney effectively conceded on appeal that Mrs. Sykes’s identification was not supported by an independent source. See ante at 83. Respectfully, this overstates the record. In the district court, respondent argued that the state courts’ independent-source finding was not a clearly unreasonable application of federal law, but failed to argue that the point was barred from habeas review by Stone. On appeal, he argued only the Stone point, without repeating the merits argument. However flawed these strategic choices, I do not think they can fairly be construed to concede error in the state courts’ independent-source ruling — let *96alone the "unreasonable” error necessary to support habeas. Harrington v. Richter, 131 S.Ct. at 786.
. Raheem v. Kelly, 257 F.3d 122 (2d Cir. 2001), cited .by the panel to discount Mrs. Sykes's independent recollection of the robber, see Young v. Conway, 698 F.3d.at 80, is distinguishable in, that it involved a highly suggestive lineup, triggering due process concerns. with reliability that went beyond the deterrence goal in applying the exclusionary rule to Fourth Amendment violations, see Raheem v. Kelly, 257 F.3d at 137. It is precisely because the deterrence achieved by collateral review of Fourth Amendment challenges is minimal that such claims are generally barred from habeas review. See Stone v. Powell, 428 U.S. at 493-95, 96 S.Ct. 3037. That the panel fails to accord proper AEDPA deference to New York’s independent-source determination, in circumstances where it should not have conducted habeas review at all, only heightens the need for en banc review.
. Even when reliability rather than deterrence is the concern, a witness’s inability to see a person’s face does not necessarily preclude an in-court identification. See, e.g., Willis v. Garrison, 624 F.2d 491, 494 (4th Cir. 1980) (concluding that victim who was unable to see robber’s face nevertheless could make identification based on height, weight, clothing, and complexion); Hamele v. Manson, 577 F.Supp. 439, 442-43 (D.Conn.1983), aff’d 738 F.2d 418 (2d Cir.1984) (table decision); People v. Spinks, 37 A.D.2d 424, 425-27, 326 N.Y.S.2d 261, 263-64 (3d Dep’t 1971).
. Judge Parker contends that because “every single Wade factor turned against a finding of an independent source” in this case, it "makes no difference” how New York courts weighed the factors. Ante at 85. The conclusion that every Wade factor turned against an independent-source finding depends on a federal habeas court assigning different weight to relevant evidence than New York courts did in reaching a contrary Wade determination. It is that action by the district court, repeated by the panel, that cannot be reconciled with Harrington and Yarborough.
. Three studies cited by the panel are referenced in footnotes to an article that was submitted to the state trial court as part of the defense's proffer of expert identification testimony that it sought to introduce at trial. See App’x of State Court Records 132 n. 6, 134 n. 16 & 136 n. 36. Neither the studies themselves nor any of the other social science materials cited by the panel were ever made part of the state court record or even cited to state courts in support of Young's Wade challenge.
. Judge Parker suggests that such a construction of Pinholster would preclude judges from even reading briefs that might reference any extrinsic materials. See ante at 82. Not so. A habeas court presented with extrinsic materials in a brief or appendix need only make clear that the materials will play no part in its deliberations to satisfy Pinholster. That is hardly what occurred in this case.
. As noted supra at 96-97, Mrs. Sykes's unhesitating ability to identify Young from among six men in a non-suggestive lineup is powerful evidence that she possessed a basis for identifying him at trial independent of the lineup.
. Judge Parker suggests that, by pointing out that the extrinsic materials cited by the panel do not speak to the independent-source question at issue, I effectively acknowledge that the panel “could not have relied” on them in deciding this case. Ante at 82-83. I respectfully disagree. The panel’s extensive citation to these materials demonstrates its reliance. See Young v. Conway, 698 F.3d at 77-84. My point is that the panel's reliance was erroneous not only because the materials were extrinsic in violation of Pinholster, but also because they were not relevant to the particular identification challenge at issue.
. See, e.g., Charles A. Morgan III, et at, Accuracy of Eyewitness Memory for Persons Encountered During Exposure to Highly Intense Stress, 27 Int’l J.L. & Psychiatry 265, 274 (2004) (observing that, while high stress had negative impact in about half of cases studied, stress had no impact on memory of many individuals, and for minority of subjects, eyewitness memory was better for high-as compared with low-stress conditions); Nancy Mehrkens Steblay, A Meta-Analytic Review of the Weapon Focus Effect, 16 Law & Human Behav. 413, 417 (1992) (stating that presence of weapon during crime has only a "relatively small effect,” reducing correct identifications by approximately 10%); Robert K. Bothwell et al., Cross-Racial Identification, 15 Personality & Soc. Psychol. Bull. 19, 23 (1989) (concluding that own-race bias accounts for "11% of the variance in recognition ability of Black subjects and 10% of the variance in recognition ability of White subjects”).
.A Special Master Report prepared for the Supreme Court of New Jersey reviewed much of the same social science literature as the panel, found it reliable, but nevertheless cautioned that the impact variables such as a perpetrator’s disguise, use of a weapon, victim stress, race differentials, multiple pre-trial identifications, time lapses between crime and identification, etc., have on the reliability of *100eyewitness identification is necessarily "only probabilistic.” Special Master Report 73-74, State v. Henderson, 208 N.J. 208, 27 A.3d 872 (No. A-8-08) ("[NJeither science nor scientists can say, at least at present, whether a real-life identification is accurate or not.... The science has simply identified variables that have an unquantifiable capacity or tendency to impair or contaminate memory and thus bring into question the reliability of a real-life eyewitness identification.”).
. Insofar as Judge Parker suggests that the panel’s citation to extrinsic materials was intended only to alert “the bench and bar to the existence of the studies and to go no further," ante at 81 (emphasis added); see also id. at 80 (stating that citations to extrinsic materials rendered "important service to the bench and bar”), that narrow purpose is hardly evident in an opinion that integrates extrinsic materials throughout a Wade assessment that Pin-holster required be based only on the record before the state court. See e.g., Young v. Conway, 698 F.3d at 80-82 (following one paragraph discussion of facts regarding independent identification with six-paragraph discussion of extrinsic studies indicating that presence of such facts contributes to mistaken identifications). Further, trial courts are usually ahead of us in their awareness of developments pertaining to evidentiary matters; most of the studies cited by the panel are already referenced in published opinions, see, e.g., State v. Henderson, 208 N.J. at 230-279, 27 A.3d 872; and the Innocence Project is dedicated to raising judicial awareness of their existence.
. "AEDPA deference” is, of course, shorthand for the requirements of the Antiterrorism and Effective Death Penalty Act, 28 U.S.C. § 2254.