dissenting:
I would grant a stay of execution and an evidentiary hearing on petitioner’s claim of incompetence to stand trial, and oh his Caldwell and Brady claims. I would grant a stay of execution on petitioner’s Grigsby claim until the Supreme Court decides Lockhart v. McCree, cert. granted, — U.S. -, 106 S.Ct. 59, 88 L.Ed.2d 48 (1985).
I. Competence to stand trial.
Thomas did not raise his competency claim in his first federal habeas petition. He argues that his competency claim is unwaivable and may not be barred on abuse of the writ grounds. Based on precedent in this Circuit, I find this argument compelling.
In Adams v. Wainwright, 764 F.2d 1356, 1359 (11th Cir.1985), this Court held that the procedural default rule of Wainwright v. Sykes, 433 U.S. 72, 97 S.Ct. 2497, 53 L.Ed.2d 594 (1977) does not operate to preclude a defendant who failed to request a competency hearing at trial or pursue a claim of incompetency on direct appeal from contesting his competency to stand trial and be sentenced through post-conviction proceedings. See also Horace v. Wainwright, 781 F.2d 1558, 1565 (11th Cir. 1986); accord Silverstein v. Henderson, 706 F.2d 361, 367 (2d Cir.1983), The same reasoning which led to that conclusion supports a holding here that a defendant who failed to pursue a claim of incompetence in his first federal habeas petition may not be barred from presenting it in a second such petition on abuse of the writ grounds.
The abuse of the writ doctrine would place on the defendant’s shoulders the responsibility of raising the competency issue on his first habeas petition; if the defendant does not do so, he effectively waives that claim. This conflicts directly with the Supreme Court’s conclusion in Pate v. Robinson, 383 U.S. 375, 384, 86 S.Ct. 836, 841, 15 L.Ed.2d 815 (1966), that “it is contradictory to argue that a defendant may be incompetent, and yet knowingly or intelligently ‘waive’ his right to have the court determine his capacity to stand trial.” If the Court’s reasoning logically extends to direct appeal, as in Adams, then it must also obtain in other post-conviction proceedings. Thus, a defendant may not “waive” his constitutional right to a competency hearing by his failure to raise the issue in his first habeas petition.
Although the petitioner is not barred from raising his competency claim here by his failure to advance it earlier, he is not automatically entitled to a hearing on this claim. Adams, supra, at 1359. Under Pate, only where the court has a “bona fide” doubt as to the defendant’s competency must it sua sponte conduct a hearing on his competence to stand trial.” Pate, supra, 383 U.S. at 385, 387, 86 S.Ct. at 842, 843. See also Hance v. Zant, 696 F.2d 940, 948 (11th Cir.1983), cert. denied, 463 U.S. 1210, 103 S.Ct. 3544, 77 L.Ed.2d 1393 (1983). It is our task to satisfy ourselves that such a doubt existed. In my view, this standard was met when the court itself, after noting the defendant’s persistent and unaccountable silence — and in response to defense counsel’s request for a pretrial evaluation — ordered such an evaluation from a psychiatrist, Dr. Burt Kaplan. Once this doubt was acknowledged, the court was obliged under Pate to conduct a hearing. Hance, supra at 948. There is no authority to permit the judge on his own initiative, and on the unchallenged opinion of one expert, thereafter to resolve this doubt against the defendant.
But even if we were to authorize the trial judge to reconsider his initial doubt without a hearing, we would surely ask him to do so on evidence more compelling than a con-clusory report based on an entirely one-sided hour-long “interview” between a psychiatrist and a defendant. It is true that the trial judge “is only required to act reasonably based on the facts before him.” Id. at 949. But when he or she takes the initiative in securing those facts, the judge has some obligation — even under a reasonableness standard — to insure that the information that will persuade him to forego a hearing is reliable and complete. This is no *691more than what we customarily expect a reviewing court to consider in assessing the factfinder’s decision to disregard or to credit an expert opinion on a defendant’s mental condition. Strickland v. Francis, 738 F.2d 1542, 1552 (11th Cir.1984).
Two factors relevant to that analysis are particularly important here:
(1) the correctness or adequacy of the factual assumptions on which the expert opinion is based;
(2) possible bias in the expert’s appraisal of the defendant’s condition.
Id. As to the first factor, it was patently unreasonable in this case for the trial judge to conclude that an expert could find a defendant competent to stand trial based on one interview — and an exceedingly brief interview, at that — in which the defendant never spoke. That fact alone insured that this expert opinion (termed “useless” by another psychiatrist) was based on no facts at all; — and in particular, on no historical or clinical evidence. Such evidence did not require extensive investigation, but was available in welfare documents from Thomas’ own hometown.
Those documents show not only an extraordinary history of abandonment, physical and emotional abuse, malnutrition, retardation and psychological trouble, but they also specifically flag Thomas’ habit of silent withdrawal as a potential symptom of serious disturbance. For example, these papers indicate that as a boy, Thomas was “shy and [did] not talk freely,” that he was “very quiet and withdrawn,” that he spoke mostly with his brother in a made-up language that only they could understand, and that he would “clam up” when he got in trouble. In affidavits, his brothers and sisters have since indicated that that as a very young child Thomas was “slow to learn, and especially slow about talking,” that he would often “go off in a trance and get inside himself,” and that after his father and mother were injured in an automobile accident and a tornado hit their house, he “kind of went into himself and stopped talking to anyone.” On subsequent review of these records and Thomas’ behavior at trial, one distinguished psychiatrist concluded that Thomas’ silence was by no means mere intransigence, but rather an easily recognizable symptom of mental illness — a “schizotypal personality disorder with psychotic-like features and paranoia.”
Even if Dr. Kaplan’s report were not on this ground patently inadequate, it was undoubtedly biased. A simple inquiry from the trial judge would have uncovered the fact that Kaplan had worked on the Thomas case before — as a kind of psychological forecaster speculating about the sort of people responsible, in his own words, for the “spree of terrorist crimes in Central Florida.” (Kaplan thought that these criminals would be “hostile individuals ... with a need to exploit and control”; “marginal” persons who were probably high school drop-outs and whose behavior compared with that of motorcycle gang members.) Not only was Kaplan thus originally employed to work for the state against Thomas, but he also had an obvious stake in vindicating his own psychological forecast. His evaluation could hardly have been less objective.
Thomas is legally entitled to a competency hearing to determine whether, at the time of trial, he had “sufficient present ability to consult with his lawyer with a reasonable degree of rational understanding.” Hance, supra at 948. This is a matter of crucial importance, since “[t]rial of a criminal defendant while he is mentally incompetent violates due process.” Id. At a minimum, Thomas is entitled to an evidentiary hearing on his competency claim.
II. Grigsby claim.
Thomas argues that the process of “death-qualification” for jurors in his case, and specifically the exclusion of certain death-scrupled jurors for cause or on peremptory challenges in his voir dire, produced a conviction-prone jury. The majority observes that the Supreme Court considered precisely this issue in Thomas’ case only days ago and declined to stay his execution.
*692The majority's discussion of this claim makes its conclusion seem inevitable and certain. This is misleading. If the Supreme Court’s action in Grigsby cases is to be our guide, then in all candor we are at this point groping in the dark.
In some Grigsby cases with abuse of the writ problems the Court has not stayed execution. See Jones v. Smith, 786 F.2d 1011 (11th Cir.1986), stay denied, — U.S. -, 106 S.Ct. 1393, 89 L.Ed.2d 709 (1986). In others, stays were granted even to petitioners in procedural default or those who proffered successive petitions. See Moore v. Blackburn, 774 F.2d 97 (5th Cir.1985), stay granted, No. 85-5555 (October 4, 1985); Celestine v. Blackburn, 750 F.2d 353 (5th Cir.1984), stay granted, — U.S. -, 106 S.Ct. 31, 87 L.Ed.2d 707 (1985); Bowden v. Kemp, 774 F.2d 1494 (11th Cir. 1985), stay granted, — U.S. -, 106 S.Ct. 213, 88 L.Ed.2d 182 (1985).
These conflicting stay decisions do not appear to turn on the variety of Grigsby claim presented in each case — that is, whether what is challenged is the actual removal of death-scrupled jurors (for cause or on peremptory strikes), or simply the process of death-qualification itself. For example, on Grigsby cases from this Circuit, the high court has stayed the executions of petitioners asserting only the general death-qualification claim. See Adams v. Wainwright, stay denied, — U.S.-, 106 S.Ct. 1254, 89 L.Ed.2d 362 (1986), stay granted, — U.S.-, 106 S.Ct. 1371, 89 L.Ed.2d 598 (1986), stay lifted, — U.S. -, 106 S.Ct. 1505, 89 L.Ed.2d 906 (1986); Bowden v. Kemp, supra. But see Harich v. Wainwright, stay denied, — U.S.-, 106 S.Ct. 1392, 89 L.Ed.2d 707 (1986). Yet the Court has denied a stay in a case (other than that at bar) presenting the strongest kind of Grigsby claim, where death-scrupled jurors were excused for cause. See Jones, supra. And on one petition for stay narrowly granted by the Court, James v. Wainwright, stay granted, — U.S.-, 106 S.Ct. 1393, 89 L.Ed.2d 707 (1986), stay continued, — U.S. -, 106 S.Ct. 1512, 89 L.Ed.2d 912 (1986), one justice voted for the stay because “persons on the venire who expressed reservations as to capital punishment were removed by peremptory challenges,” Harich, supra, (Powell, J. concurring) — while another justice noted that the petitioner “relied solely on the juror’s exposure to death qualification; he nowhere claimed that the exclusion of jurors via peremptory strikes brought his case within the scope of McCree,” id. (Marshall, J. concurring).
Ours is not the only circuit left in confusion by the high court’s decisions. The Fifth Circuit has, in sequence: (1) denied a stay and rejected a Grigsby claim in a case in which the Supreme Court later granted a stay, Celestine, supra; (2) rejected a stay on a second Grigsby claim, Watson v. Blackburn, 756 F.2d 1055 (5th Cir.1985); (3) stayed an execution to permit petitioner to apply to the Supreme Court for certiora-ri and a further stay on the possibility that certiorari would be granted in Grigsby, Berry v. King, 765 F.2d 451 (5th Cir.1985); (4) rejected another Grigsby claim, vacating a stay by the lower court and dismissing the appeal, Rault v. State of Louisiana, 772 F.2d 117, 133-34 (5th Cir.1985); (5) rejected yet another Grigsby claim and denied a stay, which the Supreme Court in turn granted one day later, Moore, supra; (6) revisited Rault, denying a rehearing but granting a stay based on the Supreme Court’s stays in Celestine and Moore, Rault v. State of Louisiana, 774 F.2d 675 (5th Cir.1985); (7) denied a stay in a Grigbsy case, observing that “until the Supreme Court affirms [Grigsby ], we are bound by the law of this circuit,” Wicker v. McCotter, 783 F.2d 487 (5th Cir.1986); and (8) and stayed an execution “[bjecause the Supreme Court now has under consideration the Grigsby issue and has consistently stayed the execution of sentence in all pending petitions for habeas corpus raising that issue ...,” Wingo v. Blackburn, 783 F.2d 1046 (5th Cir.1986).
Widely varying results of this sort are clearly intolerable to a system committed not to impose or execute sentences of death “in an arbitrary and capricious manner,” Gregg v. Georgia, 428 U.S. 153, 188, 96 *693S.Ct. 2909, 2932, 49 L.Ed.2d 859 (1975) (opinion of Stewart, Powell and Stevens, JJ.) — a system that seeks to “produce evenhanded justice,” Furman v. Georgia, 408 U.S. 238, 399, 92 S.Ct. 2726, 2808, 33 L.Ed.2d 346 (Burger, C.J., dissenting). In the face of this confusion, I urge this Court to pause in its rush to dispose of these petitions and to think carefully about our responsibilities.
The practice of this Court in Grigsby cases has been simply to ignore the Supreme Court’s grant of certiorari in McCree and to punt — there is no better word — to that Court for any contrary stay decisions. See Bowden, supra (observing that the law of this Circuit would oblige us to reject a Grigsby claim on its merits — and that “[t]he grant of the writ of certiorari in Grigsby is no authority to the contrary; any implications to be drawn therefrom may be discerned by application to the Supreme Court”). I submit that this is a cynical, and indeed improper, abdication of responsibility. Nothing in our Congressional mandate or in the jurisprudence of the “great writ” permits us to sidestep issues properly before us on habeas in favor of the Supreme Court. Indeed, the authority is to the contrary: “There is no higher duty of a court, under our constitutional system, than the careful processing and adjudication of petitions for writs of habeas corpus____” Harris v. Nelson, 394 U.S. 286, 292, 89 S.Ct. 1082, 1086, 22 L.Ed.2d 281 (1968). If we had no obligation to exercise our judgment in such cases, then federal habeas and stay petitions could simply be filed with the Supreme Court, bypassing the lower federal courts entirely.
This obligation to decide certainly may not permit us to conclude that the Supreme Court’s grant of certiorari in McCree means that it will disagree with our Circuit’s substantive law on the Grigsby claim. But neither does it support the contrary view. What we can say with confidence in such a situation — and what we not only have an obligation to say, but have traditionally said — is “that the state of the law with respect to these issues is unsettled.” Adams v. Wainwright, 734 F.2d 511 (11th Cir.1984). And where such issues of constitutional magnitude are unsettled, and human life is at stake, we have routinely thought it prudent to stay executions on our own authority pending resolution of these issues. See, e.g., id. To do otherwise is to hold, sub rosa, that as far as this Circuit is concerned, only McCree himself will benefit from a favorable decision in his case.
The dissent in Rault advocated our own ill-chosen practice of lying low and hoping that the Supreme Court will act: “Before staying cases of this kind we should await an express, affirmative statement by the Supreme Court that executions should be stayed in cases presenting the Grigsby issue. Our precedents clearly establish that a prisoner is not entitled to relief on a Grigsby claim. The grant of certiorari in Lockhart v. McCree does not suggest that a Grigsby claim has validity.” Rault, supra (Garwood, J., dissenting). But that view was squarely rejected by the Fifth Circuit, which determined to decide rather than to abdicate in such cases. Wingo demonstrates that the Fifth Circuit will now stay executions in which the petitioner presents a valid Grigsby claim until the Supreme Court resolves the issue.
Our practice should be no less responsible.
III. Other claims.
Petitioner raises two other claims that I am persuaded are sufficiently meritorious to warrant a stay of execution and an evi-dentiary hearing.
1. Caldwell claim.
Caldwell v. Mississippi, — U.S. -, 105 S.Ct. 2633, 86 L.Ed.2d 231 (1985), which controls petitioner’s claim that certain statements by the prosecutor and the judge in his case diminished the jury’s sense of responsibility for its verdict in violation of the Eighth Amendment, was issued after Thomas filed his first habeas petition. Thus, petitioner’s Caldwell claim is properly before this Court.
On voir dire the judge and the prosecutor repeatedly let the jurors at Thomas’ trial *694know that their sentencing recommendation was only advisory, not binding on the court. The trial judge also made this point in his preliminary sentencing instructions to the jury. This representation was misleading. As this Court said in Adams v. Wainwright, 764 F.2d 1356, 1364 (11th Cir. 1985):
The jury’s role in an advisory sentencing proceeding is critical. A verdict recommending life imprisonment establishes an important set of parameters beyond which a trial judge may exercise his discretion in reaching a sentence of death only if “the facts suggesting a sentence of death [are] so clear and convincing that virtually no reasonable person could differ. Tedder v. State, 322 So.2d 908 (Fla.1975).
In my view, this iteration of an incorrect standard by the judge and the prosecutor— suggesting that the judge alone was responsible for Thomas’ death sentence— may have improperly diminished the jury’s sense of its actual responsibility for sentencing in Thomas’ case. Although Thomas does not present a pure Caldwell claim — that is, one in which jurors are specifically told by the prosecutor, “your job is reviewable” — his constitutional argument deserves searching consideration on the record. I would grant an evidentiary hearing.
2. Brady claim.
I am persuaded that Thomas does not abuse thé writ in claiming for the first time here that the state suppressed exculpatory FBI evidence showing that the fingerprints of Lee 0. Martin, not those of Thomas, were found on the murder weapon. The state also failed to produce certain prison and medical records showing that Martin, as well as Thomas, had scars or welts on his legs — a feature emphasized in the victim’s description of her assailant. I am not convinced that Thomas’ failure to raise these claims earlier may be attributed to his own “intentional abandonment” or “inexcusable neglect.” Witt v. Wainwright, 755 F.2d 1396 (11th Cir.1985). Petitioner relied on the trial court’s assurance, after an in camera review, that no exculpatory evidence was withheld. Only this year did petitioner actually obtain the fingerprint report under the Freedom of Information Act.
To justify a new trial on a Brady violation Thomas must show that “there' is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.” United States v. Bagley, — U.S.-, 105 S.Ct. 3375, 3384, 87 L.Ed.2d 481 (1985). The defense’s theory was that Martin committed the crime. In my view the fingerprint evidence is more probative than the district court believed, since the gun found in Martin’s house not only disclosed Martin’s fingerprints but also yielded none of Thomas’. The scar or welt evidence may also be material. Whether this evidence, taken together, would have changed the trial’s outcome simply cannot be assessed thoughtfully in these summary proceedings. I would grant an evidentiary hearing on this claim.
Accordingly, I dissent.