Joseph John Cannon seeks permission for further collateral review of his conviction for capital murder and the resulting death sentence. Because the district court effectively granted Cannon a certificate of probable cause, he does not need our permission to appeal. We proceed to the merits, and with benefit of full briefing in the case, we affirm the denial of the writ and vacate the stay of execution.

In 1977, when he was only seventeen years old, Cannon emptied a .22 caliber revolver into Anne C. Walsh at close range, attempted to have sex with her dead body, and then drove off in her truck. As Cannon explained in his confession, he had no reason to kill Walsh. She was an attorney, and her brother, Dan Carabin, had been appointed Cannon's counsel in a burglary prosecution. Walsh had opened her home to Cannon because he had no place to stay and was unable to take care of himself, in part because of his illiteracy and poor cognitive skills.

At his first trial, in 1980, Cannon pled insanity. The jury rejected this defense. During the punishment phase, Cannon's defense counsel presented psychological experts who testified to Cannon's low intelligence and mental instability. The defense also had Cannon's mother testify about his troubled, violent childhood. The jury apparently found this mitigating evidence unpersuasive, and it sentenced him to death.

The trial court, however, granted him a new trial. At the second trial, in 1982, Cannon received new appointed attorneys who decided not to rely on an insanity theory. Instead, they tried to suppress Cannon's blood-chilling confession and, after the court admitted it into evidence, tried to convince the jury that it should not credit the confession because of inconsistencies with the indictment and with other evidence before them. This strategy also failed, and the second jury convicted Cannon. At the punishment stage, the defense decided not to use the parade of psychiatric experts that resulted in a death sentence in the first trial. Instead, Cannon's lawyers presented no mitigating evidence in the hope that the jury would view him as a confused, disadvantaged teenager who had a momentary loss of self-control and who no longer posed a threat to society. They managed to exclude testimony from the state's psychiatric expert. The prosecution's punishment evidence was limited to reports from a bailiff at the first trial and from Vincent Walsh, the victim's son, who was 13 at the time of the murder, that Cannon had threatened them. The state also told the jury that Cannon was on probation for burglary when he killed Walsh. But the defense's strategy resulted in the state's failure to inform the jury about the pattern of juvenile violence that surfaced in the first trial. Once again, the jury imposed the death sentence. The jury's decision has been upheld on direct appeal, see Cannon v. State, 691 S.W.2d 664 (Tex.Crim.App.1985), cert. denied 474 U.S. 1110, 106 S. Ct. 897, 88 L. Ed. 2d 931 (1986), and has survived five state petitions for habeas corpus.

The district court held a hearing on October 17, 1996, on Cannon's claim that his counsel was ineffective during the punishment phase of the second trial. On November 19, 1996, the court denied Cannon's application for a writ of habeas corpus. Its opinion addressed a variety of theories and applied the habeas law that was in place before enactment of the Antiterrorism and Effective Death Penalty Act of 1995 ("AEDPA"), Pub.L. No. 104-132, 110 Stat. 1214. With respect to the only theory that remains before us, the district court noted that "[a]t the time of Cannon's trial, there was a genuine legal question as to whether unadjudicated acts of juvenile misconduct were admissible" and thus that the defense's strategy had at least caused the state not to rebut Cannon's evidence with "unadjudicated acts of misconduct which might have had the tendency to infuriate the jury." The court did, however, grant Cannon's request for a certificate of appealability without specifying which issue or issues were worthy of appellate attention.

In keeping with the AEDPA, Cannon has asked this court to issue a certificate of appealability for the sole purpose of challenging the district court's ruling that his appointed attorneys at his second trial did not violate his right to effective assistance of counsel. Specifically, he asserts "that trial counsel's decision not to present available mental health evidence in mitigation at the punishment phase of Appellant's trial amounted to constitutionally ineffective assistance ... [and that] the deficiency prejudiced Appellant to the extent that a reasonable person would lose faith in the confidence of the outcome of the trial."

Because he filed his habeas petition in the district court on March 5, 1995, before the effective date of the AEDPA, Cannon's appeal is governed by the scheme of habeas corpus law that prevailed before the AEDPA's enactment. In Lindh v. Murphy, --- U.S. ----, 117 S. Ct. 2059, 138 L. Ed. 2d 481 (1997), the Supreme Court held that the AEDPA's standard for reviewing petitions by state prisoners, codified at 28 U.S.C. § 2254(d), does not apply retroactively to petitions filed before April 24, 1996.1 The AEDPA has amended § 2253 to require a certificate of appealability instead of a certificate of probable cause. Both types of certificates require Cannon to make a substantial showing of the denial of a constitutional right. Barefoot v. Estelle, 463 U.S. 880, 893, 103 S. Ct. 3383, 3394, 77 L. Ed. 2d 1090 (1983); Drinkard v. Johnson, 97 F.3d 751, 756 (5th Cir.1996), cert. denied, --- U.S. ----, 117 S. Ct. 1114, 137 L. Ed. 2d 315 (1997). But, in contrast to pre-AEDPA law, if a district court grants a certificate of appealability, it must "indicate which specific issue or issues satisfy the showing required." 28 U.S.C. § 2253(c)(3). See also Muniz v. Johnson, 114 F.3d 43, 45 (5th Cir.1997).

In light of Lindh, we have held that habeas petitioners who want to appeal need only a certificate of probable cause if they filed their petition in the district court before enactment of the AEDPA. United States v. Roberts, 118 F.3d 1071, 1072 (5th Cir.1997) (per curiam). We construe the district court's certificate of appealability as a certificate of probable cause. Thus, Cannon does not need further certification from a circuit judge before we can hear the merits of his appeal.

Cannon's 14-page motion for a certificate of appealability comes to us along with a 92-page brief in support of the motion. These documents lay out Cannon's ineffective-assistance theory in detail. We also have before us the record and accompanying exhibits as well as full briefing on the merits. See Garrison v. Patterson, 391 U.S. 464, 466, 88 S. Ct. 1687, 1688, 20 L. Ed. 2d 744 (1968) (per curiam) ("[N]othing we say here prevents the courts of appeals from considering the questions of probable cause and the merits together, and nothing said ... here necessarily requires full briefing and oral argument in every instance in which a certificate is granted."); Carafas v. LaVallee, 391 U.S. 234, 242, 88 S. Ct. 1556, 1562, 20 L. Ed. 2d 554 (1968) (indicating that a circuit court does not necessarily have to "give the parties full opportunity to submit briefs and argument in an appeal which, despite the issuance of a certificate of probable cause, is frivolous") (both discussing Nowakowski v. Maroney, 386 U.S. 542, 87 S. Ct. 1197, 18 L. Ed. 2d 282 (1967) (per curiam)).

Although we review findings of fact for clear error, the district court's ultimate conclusion that counsel was not constitutionally ineffective is a mixed question of law and fact that we review de novo. Boyle v. Johnson, 93 F.3d 180, 187 (5th Cir.1996), cert. denied, --- U.S. ----, 117 S. Ct. 968, 136 L. Ed. 2d 853 (1997).

Cannon's counsel at his 1982 trial were Fred G. Rodriguez and Gus Wilcox. Rodriguez had tried seven capital cases as a state prosecutor, although this was his first capital case on the defense side. Wilcox was an assistant district attorney in Bexar County from 1970 to 1977, when he entered private practice. Both participated actively at trial. They won important victories before a sympathetic judge, including the exclusion of testimony by the prosecution's psychiatric expert, Dr. James P. Grigson, during the punishment phase.

The district court found that Rodriguez conscientiously studied the first trial to learn from any mistakes that might have contributed to the initial death sentence. Affidavits attached to the state's reply to Cannon's first application for a state writ explain why Cannon's counsel chose not to present evidence of his mental health during the punishment phase of his second trial. Rodriguez provided the following account of his reasoning:

Before conferring with co-counsel on our trial strategy, I had the benefit of meeting with and consulting Mr. Cannon's prior counsel, William Brown, examining his file and reading the complete transcripts [of the 1980 trial].

It was my belief (and later co-counsel joined me in this belief) that the insanity defense was extremely weak, and had been quickly dismissed by a previous jury and there was no reason to believe that a subsequent jury would react any differently. Additionally, our client refused to admit to the offense. Furthermore, the introduction of psychological/psychiatric testimony would allow the prosecution to bring before the jury, every conceivable wrong, offense and referral to the probation office committed by young Cannon. In the previous trial, the prosecution, through cross-examination, brought out every offense which the defense had omitted addressing, of which there were many, including a manslaughter charge. All of these acts/offenses contributed to a quick verdict in the second phase of the trial. By staying away from this type of testimony, we sought to keep out of the record the defendant's prior criminal history. This we accomplished completely.

...

The same rationale for staying away from psychological/psychiatric testimony was applicable to the punishment phase of the trial. All of the [potential psychological] witnesses had examined young Cannon after he had committed some criminal act and been referred to them. We didn't want to place before the jury a pattern of anti-social behavior, aggressiveness and a long criminal history which would be considered by the jury on the question of future dangerousness. That information could have been elicited quite easily from any or all of these medical witnesses or from any reputation witnesses including Cannon's mother by way of "have you heard" questions. Through our strategy we were able to keep out of the record all the damning testimony elicited by both sides which portrayed Cannon as an individual who, because of his criminal past, would periodically continue a life of crime and pose a continuing threat to society, and also keep out the testimony of Dr. Grigso[n].

Wilcox filed an affidavit that expressed the same views. Their analysis proved accurate. The state sought to introduce Dr. Grigson's testimony on future dangerousness. The judge barred it, but at the same time he made it clear that the testimony would have been proper if the defense had made an issue of Cannon's psychological strengths and weaknesses.

Counsel points to Cannon's personal history suggesting that he is a victim of circumstance. A car hit him when he was four or five, and he spent three months in the hospital. Doctors now think he sustained a brain injury. He contends that he could not speak comprehensibly until he was about eight. He has learned how to read and write in prison, but he alleges that at the time of the murder he could barely write his name. When he was a child, doctors repeatedly suggested institutionalization, but Cannon's mother did not follow through.

Any defense team, however, would have trouble confining the personal history to these mitigating circumstances. At the first trial, his mother testified that schools would not keep him because he was so disruptive. He broke one girl's arm, and a boy drowned after Cannon threw him into a bayou. When he was fourteen and fifteen, he was arrested six times on burglary and theft charges. Even defense experts at the first trial portrayed Cannon as someone who needs constant supervision in order to control his violent and destructive impulses.

Cannon does not dispute that Rodriguez and Wilcox considered introducing psychological evidence, studied the transcript of the first trial, and concluded that Cannon's past was too checkered for that strategy to be effective. Nor does he claim that Rodriguez and Wilcox should have pursued an insanity theory in the second trial. Instead, he argues that Rodriguez and Wilcox should have mounted their own independent investigation into the strength and nature of the mitigating evidence. According to Cannon, they should not have assumed that a trial built on the same strategy as the first trial would fail, especially because, in Cannon's view, his defense attorneys at the first trial did not adequately convey his human qualities during the punishment phase. In the federal habeas proceeding below, Cannon presented legal experts who explained that Texas criminal defense clinics have been teaching the art of humanization of capital defendants since the late 1970s and that reasonable defense lawyers would never decline to present mitigating evidence when the state has offered evidence of a prior burglary conviction and the facts of the murder are so horrific.

But even if Rodriguez and Wilcox chose a bad strategy, and we make no such suggestion, their defense was not ineffective under the standard announced in Strickland v. Washington, 466 U.S. 668, 687, 104 S. Ct. 2052, 2064, 80 L. Ed. 2d 674 (1984). They had reason to avoid an insanity theory, not only because Cannon refused to admit his guilt, but because the facts showed advanced planning, deliberate cocking and perhaps reloading of the gun, and an awareness immediately afterward that the murder was wrong. At the punishment phase, therefore, they were relegated to arguing that the shooting was not deliberate and that Cannon would not pose a danger to the public in the future. The defense had every reason to think that once the jury learned about Cannon's personal history, they would find that he would pose a danger in the future. As Wilcox explained before the district court, the decision whether to use the insanity defense in the guilt phase practically determined whether to use Cannon's psychological history as a mitigating factor in the punishment phase. Once the jury was left with nothing but the question of future dangerousness, Cannon was arguably better off as a confused, disadvantaged juvenile than as a repeat offender whose aggressive behavior no one had been able to control. See Mann v. Scott, 41 F.3d 968, 983-84 (5th Cir.1994) (holding that the decision not to present evidence of low I.Q. and an abusive childhood during the punishment phase of a capital trial was an objectively reasonable strategic decision), cert. denied, 514 U.S. 1117, 115 S. Ct. 1977, 131 L. Ed. 2d 865 (1995); King v. Puckett, 1 F.3d 280, 284 (5th Cir.1993) (finding that a defense attorney acted reasonably by not offering mitigating evidence because he had a legitimate fear that it would open the door to rebuttal testimony about the capital defendant's poor reputation in the community); McInerney v. Puckett, 919 F.2d 350, 353 (5th Cir.1990) (explaining that the decision whether to raise an insanity defense is a matter of trial strategy and does not warrant a presumption of prejudice).

Even if Rodriguez and Wilcox performed below the Strickland standard, we cannot grant relief unless counsel's failings prejudiced Cannon. Strickland, 466 U.S. at 694, 104 S.Ct. at 2068. In this unusual case, we do not need to have a robust imagination to compare the second trial strategy of keeping the state from presenting harmful evidence at the punishment phase with the strategy at the first trial, which involved a detailed plea for the jury's understanding and sympathy. Counsel in the second trial had the benefit of studying the first trial and learning how the scenario that Cannon now seems to favor would likely have played out. Their choice to take a different tack did not make the second trial "fundamentally unfair or unreliable." Lockhart v. Fretwell, 506 U.S. 364, 369, 113 S. Ct. 838, 842, 122 L. Ed. 2d 180 (1993).

Cannon provides only a thin explanation of how Rodriguez and Wilcox could have prevented a replay of the first trial if they had chosen to emphasize his psychological and developmental problems. He has not specified what useful material an independent investigation might have revealed. He cites a medical reference manual for the claim that "[c]ertain sociological psychological characteristics that reveal themselves at an early age either disappear after the age of 15 or become significant for other purposes after the age of 18." But this statement is too general to be of much use, and it's unlikely that a jury would disregard Cannon's behavior based on such an amorphous psychological observation. Cannon points out that the defense made only a short conclusory statement at the close of the punishment phase in the first trial and insists that Rodriguez and Wilcox could have carried out the mitigation strategy more effectively in the second trial. But it's too speculative to conclude that this minor difference would have changed the outcome. Consequently, Cannon cannot show that Rodriguez's and Wilcox's strategic choices prejudiced him.

The denial of the writ is AFFIRMED, and the stay of execution is VACATED.