concurring in result only.
What I perceive to be the dispositive issue in this case has somehow disappeared from the discussion at the current stage of the litigation: the procedural default of the Miranda issue that resulted from the petitioner’s failure to present the issue in state court. When this case was heard on appeal by the original panel, it resulted in a split decision. The majority chose to “deem this [procedural default] argument forfeited” based on the state’s failure to raise it in the district court. Garner v. Mitchell, 502 F.3d 394, 401 (6th Cir.2007). Citing Trest v. Cain, 522 U.S. 87, 89, 118 S.Ct. 478, 139 L.Ed.2d 444 (1997) (“procedural default is normally a defense that the State is obligated to raise and preserv[e] if it is not to lose the right to assert the defense thereafter”), as support for the decision to impose a forfeiture, the majority nevertheless took account of the fact that a procedural default not only may be recognized for the first time on appeal, White v. Mitchell, 431 F.3d 517, 524 (6th Cir.2005), but also may be raised by the reviewing court sua sponte. See Elzy v. United States, 205 F.3d 882, 886 (6th Cir.2000). The panel majority decided not to recognize the defense on two grounds: first, because the district court had “expended considerable resources in deciding Garner’s Miranda claim,” Garner, 502 F.3d at 401, and, second, because “Garner faces the death penalty.” Id. But, the latter basis could be justified only if the resources expended in the district *272court had actually produced an adequate record on which to base a reasoned judgment about the alleged Miranda violation in this ease. In fact, the record is insufficient to make a reasoned judgment precisely because there was a procedural default of that issue in this case.
Because the Miranda issue was not raised and litigated in state court, the only record we have with regard to the petitioner’s waiver comes from the suppression hearing in state court. See id. at 410 nn. 7-8. The testimony of the interrogating officers was limited to the question of whether the proper Miranda warnings had been given and, to only a superficial extent, whether the petitioner had appeared to understand what was being communicated by the officers before he was asked to sign a waiver of his rights. We know from the record that Garner appeared “perfectly normal” and “very coherent” to the officers who interrogated him. We do know from the record that prior to trial in state court, Garner underwent assessment by a psychiatrist and a clinical neuropsychologist in order to determine his competency to stand trial. Based on their assessment, a mental-health expert appointed by the state trial court to assist with Garner’s defense submitted a report questioning whether he would have been capable of understanding the language used in the waiver or the consequences of signing it. See id. at 411. But, because the issue of the petitioner’s competency to make a valid waiver was never directly or indirectly addressed in the state courts, the prosecution was prevented from countering the implication that the waiver was flawed — for example, by securing additional testimony from the interrogating officers concerning the events preceding the petitioner’s confession, or by securing an additional examination that would have addressed his intellectual functioning near the time of his interrogation, rather than many years afterward.
In addition to the insufficiency of the record resulting from the procedural default in the state courts, the decision to ignore that default is unjustified as a matter of law. Indeed, the legal analysis was cogently laid out by Judge Rogers in his dissent from the majority opinion in Garner, in which he contended that “even if we have the discretion to disregard the procedural default because of the state’s failure to argue procedural default in the district court, it is inconsistent with the principles of AEDPA to exercise that discretion in the context of this case.” Id. at 423. In support of this contention, he articulated three legally rock-solid reasons. First, he noted, the procedural default was clear. See id. at 424. Second, the absence of a state court ruling on the issue was due to “lack of opportunity to pass on the merits, [and] not the result of, for instance, a state court’s erroneous application of some procedural hurdle or the ineffective assistance of counsel appointed by the state courts.” Id. Because there was no state court review of the issue, he argued, the majority’s exercise of discretion in the petitioner’s favor flew in the face of both “comity and federalism principles.” Id. (quoting Perruquet v. Briley, 390 F.3d 505, 518 (7th Cir.2004)). Third, and most significantly, Judge Rogers noted the paradox inherent in the application of a de novo standard of review in this case:
[I]f we were to reach the merits of [petitioner’s] constitutional claim, we necessarily would have to do so de novo, as there is no state-court decision we can look to for an evaluation of this claim. This would be inconsistent with the high level of deference to state-court decisions that Congress mandated when it passed the Antiterrorism and Effective Death Penalty Act of 1996. It would *273also amount to a windfall for [petitioner], who would win plenary review of a claim that he never presented to the [state] courts, whereas habeas petitioners who properly present their claims to state courts first are entitled only to the extremely narrow review mandated by [28 U.S.C.] section 2254(d).
Id. (quoting Perruquet, 390 F.3d at 518 (citations omitted)).
For these reasons, I conclude that the petitioner’s procedural default of the Miranda issue should prevent us from reviewing that question en banc, and I would therefore affirm the district court’s judgment, but for reasons other than those expressed by the en banc majority.