dissenting.
The underlying facts in every death penalty case present a sordid story, and this case is no exception. Zettlemoyer’s murder of DeVesco, uncontested at trial and before us, was base. Although Zettlemoyer’s ha-beas counsel makes a valiant effort to challenge the conviction on the ground of ineffective assistance of counsel, I agree with the majority that in the end this argument is not persuasive.
Zettlemoyer’s trial counsel, a public defender appointed to defend him, apparently conceived the strategy of relying upon one expert witness, Dr. Schneider, to substantiate a defense of diminished capacity. Dr. Schneider was not competent to testify to support such a defense which, under Pennsylvania law, applies to a defendant who does “not possess sufficient mental capacity to form the specific intent required for a conviction of murder of the first degree.” Commonwealth v. Walzack, 468 Pa. 210, 360 A.2d 914, 916 (1976). Instead of presenting a witness who could testify to any possible “mental disorders affecting the cognitive functions necessary to formulate a specific intent,” Commonwealth v. Weinstein, 499 Pa. 106, 451 A.2d 1344, 1347 (1982), the defense relied on Dr. Schneider who could only testify to Zettle-moyer’s personality disorder. Zettlemoyer does not challenge the correctness of the trial court’s ruling that such evidence, going to the “irresistible impulse” defense which Pennsylvania does not accept, was inadmissible. I agree with the majority that in the absence of any showing of the availability of evidence to support the diminished capacity defense and in the face of overwhelming evidence that Zettlemoyer had the specific intent to kill, there is simply no basis for the claim of ineffective assistance of counsel at the guilt phase.
I part company with the majority, however, over the treatment of the claim of *311ineffective assistance of counsel at the penalty phase. Zettlemoyer argues that the paltry case presented on his behalf was the product of ineffective assistance of counsel. The majority, without the benefit of any evidence presented at a hearing, either state or federal, has decided that counsel acted reasonably in the sentencing phase. Although, as I suggest hereafter, I believe it is probable that I would reach a contrary conclusion, I am reluctant to draw any conclusion without the usual hearing. Therefore, I believe that the principal question before us is whether the district court had a duty to hold an evidentiary hearing so that this issue could be explored.
Counsel for the state cites no capital case in which a defendant who claims ineffective assistance of counsel at the sentencing stage has not been afforded at least one evidentiary hearing on that claim, either as part of the state post-conviction proceeding or in federal court on a habeas petition. Surprisingly, the majority overlooks the singularity of its decision to bypass the usual process by which a trial judge decides whether counsel’s performance fell “within the wide range of reasonable professional assistance,” Strickland v. Washington, 466 U.S. 668, 689, 104 S.Ct. 2052, 2065, 80 L.Ed.2d 674 (1984), after an evi-dentiary hearing at which trial counsel explains the strategy of his or her trial conduct. Instead, without the benefit of such a post hoc explanation the majority undertakes to make the requisite analysis itself, and reaches what for me are two startling conclusions: first, that Zettlemoyer’s attack on trial counsel for making a “perfunctory” case for his life “is completely unfounded,” Maj. op. at 300, and second, that “it would have been improper” to have had an evidentiary hearing on the claim of counsel’s ineffectiveness at the sentencing stage. Maj. op. at 300. I proceed to explore both conclusions, which are central to the disposition of this appeal.
The Supreme Court established the general standards to be used by a federal habeas court in deciding whether to grant an evidentiary hearing in Townsend v. Sain, 372 U.S. 293, 83 S.Ct. 745, 9 L.Ed.2d 770 (1963). Of particular relevance to this case, a hearing is required if “for any reason it appears that the state trier of fact did not afford the habeas applicant a full and fair fact hearing.” Id. at 313, 83 S.Ct. at 757.
In United States v. Dawson, 857 F.2d 923 (3d Cir.1988), we outlined the more particularized standard to be used when considering whether a hearing was required on a claim of ineffective assistance of counsel as follows:
First, we must determine whether the district court considered as true all appellant’s nonfrivolous factual claims. This step requires that we review whether the district court properly found certain allegations frivolous. Second, we must determine whether, on the existing record, those claims that are nonfrivolous conclusively fail to show ineffective assistance of counsel.... If a nonfrivolous claim does not conclusively fail either prong of the Strickland test, then a hearing must be held.
Id. at 927-28 (emphasis in original).
The district court dealt with this particular claim of Zettlemoyer’s in a single paragraph. Although the court rejected it as without merit, it did not state that it regarded the claim as frivolous. Because the majority does not remand for an evidentia-ry hearing, it apparently views Zettlemoyer’s allegations as frivolous or deems that they failed to show ineffective assistance of counsel at the penalty phase. I believe that either conclusion is unsupported by the record before us.
The externalities give some indication of the extent of the case made for Zettlemoyer’s life. On April 24, 1981, at 11:36 a.m., the jury found Zettlemoyer guilty of first-degree murder. App. at 215. The case proceeded immediately to the penalty phase. There was a sidebar conference at which counsel and the court discussed stipulations as to aggravating and mitigating circumstances, defense counsel presented one witness, Zettlemoyer’s father, whose direct testimony takes eight transcript pages, there was cross-examination occupying three pages, the case recessed so that *312the jury could lunch, counsel presented final arguments to the jury, the court gave the jury its charge, the jury deliberated, and it returned with a death sentence at 5:31 p.m. on the same day.
Of course, the fact that Zettlemoyer’s fate was decided in the span of six hours is not alone evidence of “perfunctoriness.” We must instead examine the substance of the case presented by counsel to the jury in an attempt to convince them that there were mitigating circumstances to outweigh the one aggravating circumstance present, that DeVesco was to have been a witness against Zettlemoyer on a robbery charge.
Zettlemoyer argues that Dr. Schneider should have been recalled to testify at the penalty hearing. The majority holds that the failure to do so was not ineffective assistance because Dr. Schneider was the last witness at the trial phase, and the court instructed the jury to consider in the penalty phase all of the evidence from both sides that it heard during the trial in chief. However, the majority overlooks that the focus of Dr. Schneider’s testimony at trial was to establish a diminished capacity defense, not to show mitigating circumstances.
Dr. Schneider testified that he interviewed Zettlemoyer on three separate occasions for a total of eight hours and conducted seven different tests on him, and that based on these tests, he concluded that Zettlemoyer showed normal intelligence overall but “deficits” in common sense and practical judgment, particularly as they relate to interpersonal relationships. Dr. Schneider testified that Zettlemoyer showed no signs of organic damage to his brain or central nervous system and that he is not psychotic, meaning he does not grossly distort reality, but that “under stress this man will lose control and lose judgment.” App. at 133. Dr. Schneider continued, saying that Zettlemoyer “has a limited ability to tolerate adversity, that too much emotional stimulation may precipitate a loss in judgment, perhaps even to a psychotic proportion.” App. at 134-35. He described Zettlemoyer as “self centered” and “autistic” and “an individual who has been protected and indulged to the point where he has not developed adequate skills to cope or meet with the normal pressures, demands and responsibilities of everyday living.” App. at 135. The overall “diagnostic impression” was of “a schizoid personality with paranoid features.” App. at 137. Dr. Schneider explained that schizoid meant a personality that was unusually detached, unemotional, and reclusive, with little social involvement.
Dr. Schneider described Zettlemoyer’s inability to complete anything successfully as a “pressure cooker syndrome,” and that “[i]f you don’t have a release valve it will blow and that’s where I believe that the emotional disturbance, in addition to the personality disorder, may have resulted in the behavior.” App. at 142. Again, “[tjhere is sufficient evidence to suggest that he has not met with success in any field of endeavor. If you take that kind of foundation and build upon that a number of stressful situations, a number of powerful factors over which-he had no control, then you are going to get a result which can be catastrophic.” App. at 144. Dr. Schneider stated that the stressful situations pertinent to Zettlemoyer immediately before October 13, 1980 were the publicity over “the robbery,” the embarrassment he brought upon his family, particularly his mother, his debts, and rejection by his girlfriend/fiance. App. at 145.
Finally, defense counsel asked Dr. Schneider whether Zettlemoyer was capable of fully forming the specific intent which is required for a willful, deliberate and premeditated act, the state objected, and, after a sidebar conference, the court sustained the state’s objection. Thus, Dr. Schneider’s testimony on Zettlemoyer’s behalf in the guilt phase ended with a discredited whimper, rather than a bang, particularly since it was followed by grueling and effective cross-examination designed to show that Zettlemoyer knew the nature and quality of his act.
At sentencing, the state had stipulated to two statutory mitigating factors, the absence of a prior criminal record and Zettle-moyer’s age (25). Thus it appears that the *313entire sentencing phase strategy of defense counsel was to try to establish the additional mitigating factor that “[t]he defendant was under the influence of extreme mental or emotional disturbance.” 42 Pa.Cons. Stat.Ann. § 9711(e)(2) (Purdon 1982).
In denying an evidentiary hearing on Zettlemoyer’s claim of ineffective assistance of counsel at sentencing, the district court held that it was unnecessary to recall Dr. Schneider because “the petitioner’s father testified during the sentencing phase about petitioner’s mental and emotional health, thus reinforcing in the minds of the jury petitioner’s asserted diminished capacity defense.” Opinion of May 31, 1988, slip op. at 30 (emphasis added). It is thus evident that the district court failed to recognize that the diminished capacity defense which Dr. Schneider had sought unsuccessfully to establish during the guilt phase, and which itself can be the basis of a mitigating circumstance, see 42 Pa.Cons. Stat.Ann. § 9711(e)(3), is distinct from the mitigating circumstance of “extreme mental or emotional disturbance.” Id. § 9711(e)(2). Zettlemoyer makes this distinction in paragraphs 139 through 142 of his petition for habeas corpus, quoted in Maj. op. at 302 n. 14.
Although Dr. Schneider’s testimony may have contained the evidentiary predicate for an “emotional disturbance” mitigating circumstance, the jury did not hear Dr. Schneider’s testimony during the guilt phase from the perspective of whether Zettlemoyer was troubled enough so that his life should be spared, and thus I find it difficult based on what is before us to characterize the defense counsel’s decision not to recall Dr. Schneider as a reasonable one.
The majority relies on Clozza v. Murray, 913 F.2d 1092 (4th Cir.1990), to support its conclusion that counsel was not ineffective in failing to recall Dr. Schneider. Clozza is inapposite. There the court rejected the claim of ineffective assistance of counsel for failing to recall a psychiatrist who had testified at the guilt phase because the state court had made findings and reached the conclusion that counsel’s tactic was reasonable trial strategy. That decision was made only after the state court held an evidentiary hearing on the ineffective assistance of counsel claim in a post-conviction habeas corpus proceeding. During that hearing, the doctor testified that his testimony at sentencing would not have added anything new, and it was revealed that the doctor’s testimony had been based on the understanding that the defendant had been drunk at the time of the offense. Because Clozza had testified during the guilt phase that he was sober, the Court of Appeals discussed the strategic dilemma Clozza’s counsel would have faced had the psychiatrist been recalled to testify. Id. at 1102.
In contrast, no such state court hearing was held in this case, and thus trial counsel has never been called upon to explain his enigmatic decision to rest on Dr. Schneider’s testimony given during the guilt phase. I find it significant that in the majority’s reiteration of its conclusion that it was reasonable for defense counsel not to recall Dr. Schneider, the majority never once suggests that, as in Clozza, Dr. Schneider’s testimony at the sentencing phase could have been harmful to defendant. Recalling Dr. Schneider would have given the defense the opportunity to rehabilitate him in the eyes of the jury. The majority has not and cannot reasonably disagree that such a move would have enabled the defense to emphasize Zettlemoyer’s extreme mental and emotional disturbance at the time of the crime. Thereafter, the jury would have proceeded to its deliberations focused on Dr. Schneider’s testimony about this statutory mitigating circumstance rather than focused on his earlier testimony that the defendant was unable to form a specific intent to kill, which the jury had already found unpersuasive. I cannot discount the possibility that Dr. Schneider’s recall might have tipped the scales in favor of a decision by the jury to spare Zettlemoyer’s life.
Zettlemoyer also claims that trial counsel should have presented additional evidence at the penalty phase. As I noted previously, the only live testimony during the sentencing phase was that of Zettlemoyer’s father, who, in his brief testimony, reit*314erated his wife’s earlier testimony that Zettlemoyer had personality problems when he was a boy and had poor relations with the rest of the family members. He testified that he had one or two discussions with his son during the two weeks before the murder, that Keith Zettlemoyer discussed “all the problems that he was having and how he felt overwhelmed,” App. at 184, and that he should have taken responsibility to get psychiatric or psychological help for his son. Zettlemoyer’s father also mentioned that his son “has shown considerable remorse” over the crime, that “he’s taken to religion to some extent which is something he never did before,” id., and that he believed his son was suffering under extreme mental or emotional disturbance.
It is not apparent from the record why counsel would have relied solely on the testimony of Zettlemoyer’s father. Zettle-moyer’s father’s testimony shows that he was only peripherally involved with his son and had “only mild” knowledge that his son’s fiance had broken off with him, App. at 187, a patently traumatic event. He was obliged to admit that he was only aware of the problems in his son’s personal life and his financial problems “[t]o some degree, not to the degree that my wife would have been aware of them but I know he had them.” App. at 186. Under these circumstances, it is not difficult to see why the jury would not be overly impressed with the only testimony put before them on the critical issue of life or death sentencing.
I cannot understand why counsel would not have recalled Zettlemoyer’s mother, since her testimony, like Dr. Schneider’s, during the guilt phase was more inclined to the diminished capacity defense, whereas her penalty phase testimony could have emphasized the mitigating circumstance of “extreme mental or emotional disturbance.” It is also not evident why counsel would not have called Zettlemoyer’s friends, co-workers, or his former fiance to support the claim of overwhelming emotional disturbance. Certainly the absence of such supporting testimony could not have escaped the jury's notice.
Even more inexplicable is the complete failure of the defense to attempt to make any kind of case at sentencing to support the catch-all mitigating circumstance, which permits the defense to present and the jury to consider “[a]ny other evidence of mitigation concerning the character and record of the defendant and the circumstances of his offense.” 42 Pa.Cons.Stat. Ann. § 9711(e)(8). If this was the result of a defense trial strategy, its logic is opaque and defense counsel should be obliged to provide elucidation.
The majority rules that no evidentiary hearing is called for on this issue first, because it believes the petition for habeas corpus does not allege failure to produce other witnesses as a basis for the claim of counsel’s ineffectiveness and second, because petitioner has failed to point to specific significant mitigating evidence. The habeas petition does in fact complain of ineffectiveness at the sentencing phase on this basis.1 Although I agree that the failure to point to specific evidence that would have been available is troubling, I do not find it dispositive in this case. The habeas court is free to “direct that the record be expanded by the parties by the inclusion of additional materials relevant to the determination of the merits of the petition.” Rule 7(a), Rules Governing § 2254 Cases. Both the state court and the district court considered only the failure to recall Dr. Schneider. Had the district court considered paragraph 143 of the habeas petition and found it insufficiently specific, the rules provide ample opportunity for further elucidation.
I do not agree that our decision in Mayberry v. Petsock, 821 F.2d 179 (3d Cir.), cert. denied, 484 U.S. 946, 108 S.Ct. 336, 98 *315L.Ed.2d 362 (1987), compels the denial of an evidentiary hearing at which counsel would be obliged to explain his underlying trial strategy.2 In Mayberry, where we affirmed the district court’s denial of a hearing on Mayberry’s habeas corpus petition at which he sought to compel discovery from the state on his contention that state officials had obstructed him from pursuing his appeal, we held, and reiterate here, that “bald assertions and conclusory allegations do not afford a sufficient ground for an evidentiary hearing.” Id. at 185.
In this case, we have substantially more. We have before us a plainly unenthusiastic case made on behalf of Zettlemoyer’s life. I find it incomprehensible that there were no character witnesses who could testify to some redeeming virtue on behalf of a young man who had no criminal record. Is it possible that there was no clergyman, former employer, teacher, neighbor, or friend who was willing to come forward to show that despite his crime, Zettlemoyer’s life is still worth preserving? If so, then counsel should explain that he searched for such testimony, and that it was unavailable, or that there was a reasonable basis not to present any such testimony.
Surely even the majority must acknowledge that as appellate judges we have the experience to spot a nonchalant, languid and lackluster defense at sentencing. The question before us is not whether this was the case, but why. Unlike the majority, I will not put an explanation into defense counsel’s mouth which he has never given in either testimony or by affidavit filed at any time in any place.
Based on what is before us, this does not appear to be a case like Burger v. Kemp, 483 U.S. 776, 107 S.Ct. 3114, 97 L.Ed.2d 638 (1987), where the failure of the attorney to present any mitigating evidence at all at the penalty phase was the result of trial counsel’s strategic decision that the witnesses, including the petitioner himself, necessary to present such evidence would ultimately be more damaging than helpful. Nor is this a case like Coleman v. Brown, 802 F.2d 1227 (10th Cir.1986), cert. denied, 482 U.S. 909,107 S.Ct. 2491, 96 L.Ed.2d 383 (1987), where the court ruled that an evi-dentiary hearing by the habeas court was not warranted because it was clear from trial counsel’s testimony at the state post-conviction proceedings that he had made a strategic choice not to put on certain witnesses at the penalty phase. If this is a case like Burger or Coleman where counsel made such strategic decisions, then the reasonableness of that conduct can be explored at a hearing, such as .the one the district court held in Burger or the state court held in Coleman where the respective counsel testified.
On the other hand, it may be that an evidentiary hearing will show, as Zettle-moyer charges, that counsel was dispirited following the jury verdict of guilty to first-degree murder and, as a result, presented what appears to me to be a plainly anemic case. This may be a case such as Blake v. Kemp, 758 F.2d 523 (11th Cir.), cert. denied, 474 U.S. 998, 106 S.Ct. 374, 88 L.Ed.2d 367 (1985), where the court ruled, based on evidence presented at the habeas hearing, that trial counsel’s strategy to concentrate on winning the guilt phase rather than prepare for the sentencing phase and his failure to seek out and prepare any witnesses who could present character evidence at the most critical stage of the proceeding warranted the grant of a writ of habeas corpus.
As the court stated in Kubat v. Thieret, 867 F.2d 351, 369 (7th Cir.), cert. denied, — U.S. -, 110 S.Ct. 206, 107 L.Ed.2d 159 (1989),
[Djefense counsel must make a significant effort, based on reasonable investigation and logical argument, to ably present the defendant’s fate to the jury and to focus the attention of the jury on any mitigating factors. Mitigating *316factors brought out at trial might be emphasized, a coherent plea for mercy might be given, or new evidence in mitigation might be presented. But counsel may not treat the sentencing phase as nothing more than a mere postscript to the trial. While the Strickland threshold of professional competence is admittedly low, the defendant’s life hangs in the balance at a capital sentencing hearing. Indeed, in some cases, this may be the stage of the proceedings where counsel can do his or her client the most good.
Perhaps the most analogous case to that before us is Wilson v. Butler, 813 F.2d 664, 671-72 (5th Cir.), modified on other grounds, 825 F.2d 879 (5th Cir.1987), cert. denied, 484 U.S. 1079, 108 S.Ct. 1059, 98 L.Ed.2d 1021 (1988). There, as here, the issue was the effectiveness of trial counsel at the sentencing stage of a state trial culminating in a death sentence. The court noted that while trial counsel may choose not to pursue certain avenues of investigation or strategies, those decisions must be informed ones. Because there was no evi-dentiary hearing held in the state court, the federal court held that the record was inadequate to establish whether trial counsel “made a considered strategic decision or whether this decision was reasonable.” Id. at 672. Therefore, the court held that an evidentiary hearing by the habeas court was necessary.
Under Pennsylvania law, sustained by the United States Supreme Court, “the verdict must be a sentence of death ... if the jury unanimously finds one or more aggravating circumstances which outweigh any mitigating circumstances.” 42 Pa.Cons. Stat.Ann. § 9711(c). Given the single ag-
gravating circumstance that the victim was a potential witness, and the two stipulated mitigating circumstances of age and the absence of any prior significant criminal record, there was a clear chance that a forceful, well conceived, and adequately supported case at the penalty phase directed to one or two additional mitigating circumstances would have led the jury to return with a life imprisonment sentence for Zettlemoyer. There has never been any explanation why such a defense was not provided, and thus my confidence in the outcome has been undermined.
I believe that our opinion in Dawson required the district court to hold a hearing because Zettlemoyer’s claim of ineffectiveness “does not conclusively fail either prong of the Strickland test.” 857 F.2d at 928 (emphasis in original). The majority expresses some displeasure because a decade has passed since this murder. The majority has already satisfactorily explained the reason for the passage of time. See Maj. op. at 303. It is, however, ironic that in all this time no court has sought an explanation for the pallid case made for Zettlemoyer’s life. When the decision between life and death is as fine as a razor’s edge, I cannot agree with the majority that neither society nor Zettlemoyer is entitled to an objective evaluation of that explanation.3
. The petition alleges:
143. Mr. Zettlemoyer's defense counsel was ineffective in not presenting non-statutory evidence and argument, in support of mitigation, that the imposition of the death penalty in this case (a) would be cruel and inconsistent with the developed standards of civilized life; (b) would serve no demonstrable deterrent purpose; or (c) would be an inappropriate penalty for any other reasons.
App. at 39.
. In any event, Mayberry, a non-capital case, cannot control our decision with respect to when an evidentiary hearing is appropriate for an attack on the sentencing phase of a capital case because "the qualitative difference of death from all other punishment requires a correspondingly greater degree of scrutiny of the capital sentence determination.” California v. Ramos, 463 U.S. 992, 998-99, 103 S.Ct. 3446, 3452, 77 L.Ed.2d 1171 (1983).
. Zettlemoyer’s claims based on Caldwell v. Mississippi, 472 U.S. 320, 105 S.Ct. 2633, 86 L.Ed.2d 231 (1985), and Mills v. Maryland, 486 U.S. 367, 108 S.Ct. 1860, 100 L.Ed.2d 384 (1988), also raise disturbing issues. I do not agree with the majority that the trial court's charge does not present a Caldwell violation. The court charged, inter alia,
Ladies and gentlemen, you must now decide what sentence is to be imposed upon the defendant, whether it be death or life imprisonment. In a very proper sense, you are not really making that decision. You are not deciding whether he should be sentenced to death or life imprisonment. That was the law years ago and the Supreme Court of the United States declared such death penalties to be unconstitutional. I won’t go into the reason. One of the theories was that it placed discretion on the jury. They could decide whether a particular individual could suffer death or life imprisonment. They have removed that burden from you. That is not what you are to decide. You are to decide whether there are certain aggravating circumstances or mitigating circumstances and depending upon how you find those circumstances, as I will explain to you, your decision follows. It must follow. *317If you find a certain way, a certain penalty must follow. That is the law. If, for example, as I will explain in a little more detail, you find unanimously beyond a reasonable doubt, that there is an aggravating circumstance and no mitigating circumstances or that the aggravating circumstance outweighs the mitigating circumstances, you must return a verdict of death. So the burden is not really yours.
App. at 200-01 (emphasis added).
Caldwell held that "the uncorrected suggestion that the responsibility for any ultimate determination of death will rest with others bears an intolerable danger that the jury will in fact choose to minimize the importance of its role." 472 U.S. at 333, 105 S.Ct. at 2642. I believe the charge given in this case may have incorrectly led the jury to believe that their discretion, which is equivalent to their “burden,” had been removed and therefore it violated the underlying principle of Caldwell. However, in light of the decisions in Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), and Sawyer v. Smith, -U.S. -, 110 S.Ct. 2822, 111 L.Ed.2d 193 (1990), limiting the retroactive application of new rules on habeas review, I agree with the majority that this Caldwell violation cannot be the basis for the grant of a writ of habeas corpus.
With respect to Zettlemoyer's claim based on Mills, I, unlike the majority, believe that the jury charge ("if ... you find unanimously beyond a reasonable doubt, that there is an aggravating circumstance and no mitigating circumstances ...", App. at 204) and the verdict slip ("We the jury have found unanimously: at least one aggravating circumstance and no mitigating circumstances ... the aggravating circumstance outweighs any mitigating circumstance,” App. at 216) were ambiguous as to whether juror unanimity was required for each mitigating circumstance. Therefore, I believe that the instruction and verdict slip violated the holding in Mills and McKoy v. North Carolina, — U.S.-, 110 S.Ct. 1227, 108 L.Ed.2d 369 (1990). As with the violation of the Caldwell principle, I believe it is likely that the Supreme Court would view Mills as announcing a new rule and thus this claim also is subject to the almost insurmountable barrier on retroactive application announced in Teague.