Howard Neal was sentenced to death by the state courts of Mississippi for the brutal rape and murder of his thirteen-year-old niece, Amanda Joy Neal. He also shot and killed his brother, Bobby Neal, and he raped and murdered his niece’s fourteen-year-old friend, all during the same episode. He now seeks federal habeas corpus relief on the grounds of ineffective assistance of counsel. Neal argues that his counsel failed to thoroughly investigate Neal’s background — including his horrid childhood of rejection, abandonment, and mental institutions, plus his torturous prison experience — to uncover evidence of mitigating circumstances that he could have presented to the jury during the sentencing phase of his trial. Neal raised this ineffective counsel claim before the Mississippi Supreme Court. That court denied relief, concluding that the additional evidence would have been cumulative of what actually was presented. Because we con-*685elude that the Mississippi Supreme Court’s conclusion, although incorrect, was not an unreasonable application of Strickland v. Washington, we deny Neal’s request for a writ of habeas corpus.
I
The facts of this case are discussed in detail in the published opinion by the Mississippi Supreme Court on Neal’s direct appeal. Neal v. State, 451 So.2d 743, 747-51 (Miss.1984). We restate the facts briefly here.
Neal is a moderately retarded man, with an IQ of between 54 and 60. The record indicates that he had a nightmarish childhood and young adulthood. We will discuss these facts in more detail in the body of this opinion. In short, as a youth he was discarded by his family, spent eight years in Mississippi state mental institutions, and then some time in an Oklahoma prison for assault and battery with a dangerous weapon, where, as a helpless individual, he apparently suffered sexual abuse by fellow prisoners.
In January 1981, Neal drove to the home of his half-brother, Bobby Neal, against whom he may have had a longstanding resentment. Bobby, Bobby’s thirteen-year-old daughter, Amanda Joy, and her friend, Melanie Sue Polk, were together in the house. The three left with Neal in Neal’s car, perhaps by force (but this is uncertain). During the drive, while they were on a logging road, Neal, according to his confession, began fondling Amanda Joy. Bobby told Neal to stop, and an argument ensued. Neal stopped the car, and he and Bobby got out and walked some distance away. At that point, Neal shot Bobby, killing him. Neal then returned to the car and drove to another deserted area with the two girls. He pulled a blanket from his car and proceeded to rape Amanda Joy. He then raped Melanie Sue and shot both girls.
After the bodies were found, the pathologist’s examination of Amanda Joy revealed bruises and lacerations about her face, head, and left wrist, and evidence of manual strangulation, in addition to the bullet hole in her abdomen. The pathologist concluded that Amanda Joy could have survived between five and thirty minutes given her wound.
The police began by canvassing the nearby communities. As part of their investigation, they showed some people a photograph of Neal and asked whether he looked familiar. The owner of a nearby motel said that he remembered Neal renting a room about the time of the murder. By this time, however, Neal was in California, where he was later arrested for shoplifting. During a standard background check, the California police discovered that Neal was wanted for questioning in Mississippi. After several days of interrogation in California, Neal admitted to the California authorities that he had committed the murders.
Neal was tried and convicted for Amanda Joy’s murder based on the confession he gave police,1 and the jury sentenced him to death. Under Mississippi law, the jury is required to balance statutorily-defined aggravating factors against any mitigating factors in determining whether the death penalty is warranted. Billiot v. Puckett, 135 F.3d 311, 315 (5th Cir.1998). In Neal’s case, the jury found that two aggravating circumstances — that the murder was committed in the course of a kidnaping and was “especially heinous, atrocious, or cruel” — were sufficient to impose the death penalty and were not outweighed by mitigating circumstances. See Miss.Code Ann. § 99 — 19—101(5)(d) and (h).
Neal appealed this conviction and sentence, both of which the Mississippi Supreme Court ultimately affirmed. Neal, 451 So.2d 743. Neal then sought habeas corpus relief. The Mississippi Supreme Court granted Neal an evidentiary hearing on whether he had been denied the opportunity to testify on his own behalf, Neal v. *686State, 525 So.2d 1279, 1283 (Miss.1987), but after this hearing, that court denied relief. Neal v. State, 687 So.2d 1180 (Miss.1996). Neal then filed a petition for a writ of habeas corpus in the United States District Court for the Southern District of Mississippi on July 7, 1997. In an unpublished order, the district court denied Neal’s petition on January 7, 1999, and then denied his request for a Certifícate of Appealability (“COA”) on October 7, 1999. Neal then filed a motion seeking a COA in this court. We denied his motion on all claims but one. We did grant a COA to determine whether Neal’s trial counsel was ineffective at the sentencing phase of the trial for failing to investigate evidence of mitigating circumstances and to present that evidence during the sentencing hearing. We now address that issue on the merits.
II
A
Neal contends that his trial counsel was ineffective for failing to investigate and gather, and consequently failing to present, mitigating evidence during the sentencing phase of the trial. The Sixth Amendment requires defense counsel to conduct a reasonably thorough pretrial inquiry into the defenses that might be offered in mitigation of punishment. Baldwin v. Maggio, 704 F.2d 1325, 1332-33 (5th Cir.1983). Neal argues that his lawyers failed to do so and, as a result, called only two witnesses during sentencing— Neal’s mother and a psychologist.
Neal argues that his lawyer should have interviewed members of the staff at the two institutions where Neal spent time as a youth, Ellisville and Whitfield.2 He also contends that his lawyer should have obtained the records from the prison in Oklahoma where Neal was incarcerated as a young man or at least consulted officials or medical personnel from that prison regarding his mental capabilities and character. Finally, he contends that he should have been evaluated by a neurologist to explain further his mental state. None of this occurred, so any possible mitigating evidence from these sources was unavailable to the jurors deciding his sentence. Neal further argues that he was prejudiced by his counsel’s performance because there is a reasonable probability that if this evidence had been before the jury, he would have received a life sentence instead of death.
B
Our first responsibility is to determine the standard of review. Because Neal filed his petition for habeas corpus relief on July 7, 1997, the Antiterrorism and Effective Death Penalty Act (“AEDPA”) governs this appeal. See Lindh v. Murphy, 521 U.S. 320, 324-26, 117 S.Ct. 2059, 2062-63, 138 L.Ed.2d 481 (1997)(confirm-ing that the AEDPA applies to federal habeas corpus petitions filed on or after April 24, 1996). The AEDPA standard for granting habeas corpus relief with respect to an adjudication on the merits in state court is stated in 28 U.S.C. § 2254(d):
“(d) An application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim — ”
“(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.... ”
In the context of federal habeas proceedings, adjudication “on the merits” is a term of art that refers to whether a court’s disposition of the case, was substantive as opposed to procedural. Green v. Johnson, 116 F.3d 1115, 1121 (5th Cir.1997). In the present case, the state supreme court denied habeas corpus relief on the grounds that any additional evidence that Neal could have uncovered and pre*687sented would have been “substantially redundant or cumulative when compared with the evidence Neal offered at trial.” Neal v. State, 525 So.2d 1279, 1281 (Miss.1987). Thus, that court’s disposition of this issue was substantive and therefore qualifies as a decision “on the merits.”
In this case, the “clearly established Federal law” is the Supreme Court’s decision in Strickland v. Washington, 466 U.S. 668, 687, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), and its progeny, which govern ineffective assistance of counsel claims. As the Supreme Court noted recently,
It is past question that the rule set forth in Strickland qualifies as “clearly established Federal law, as determined by the Supreme Court of the United States.” That the Strickland test “of necessity requires a case-by-case examination of the evidence” obviates neither the clarity of the rule nor the extent to which the rule must be seen as “established” by this Court.
Williams v. Taylor, 529 U.S. 862, 120, S.Ct. 1495, 1512, 146 L.Ed.2d 389 (2000).
The meaning of “unreasonable application,” however, is less clear. The Supreme Court has recently explained that a state court decision involves an unreasonable application of this Court’s precedent
if the state court identifies the correct governing legal principle from this Court’s decisions but unreasonably applies that principle to the facts of the prisoner’s case.
Id. at 1528. Although “unreasonable” is difficult to define, the Court offered some guidance. First, while acknowledging that earlier Supreme Court decisions may have caused confusion, the Court specifically rejected the subjective standard that we set out in Drinkard v. Johnson, 97 F.3d 751, 769 (5th Cir.1996). Instead, the Court held that the standard is objective: “Stated simply, a federal habeas court making the ‘unreasonable application’ inquiry should ask whether the state court’s application of clearly established federal law was objectively unreasonable.” Williams, 120 S.Ct. at 1521-22. Second, the Court emphasized that the “most important point” of the Williams decision is the critical distinction between an unreasonable application of federal law and a merely “incorrect” or “erroneous” application of federal law. Id. at 1522-23. Because section 2254(d) “places a new constraint” on a federal habeas court and demands greater deference to state courts, we have no authority to grant habeas corpus relief simply because we conclude, in our independent judgment, that a state supreme court’s application of Strickland is erroneous or incorrect. Id. at 1523.
Given this deferential standard of review under Section 2254(d), the question before us is whether the Mississippi Supreme Court’s decision to reject Neal’s ineffective assistance claim “involved an unreasonable application” (and not merely an incorrect application) of Strickland.
C
To establish an ineffective assistance of counsel claim, Neal must show both that his counsel’s performance was deficient and that the deficient performance prejudiced his defense. Strickland, 466 U.S. 668, 687, 104 S.Ct. 2052, 80 L.Ed.2d 674.
(1)
(a)
Counsel’s performance is considered deficient if it “falls below an objective standard of reasonableness” as measured by professional norms. Strickland, 466 U.S. at 688, 104 S.Ct. 2052. We must determine whether there is a gap between what counsel actually did and what a reasonable attorney would have done under the circumstances. In scrutinizing counsel’s performance, we make every effort to “eliminate the distorting effects of hindsight,” id. at 689, 104 S.Ct. 2052, and do not assume that counsel’s performance is deficient “merely because we disagree with trial counsel’s strategy.” Crane v. Johnson, 178 F.3d 309, 312 (5th Cir.1999). But with that said, we consider it indisputable *688that, in the context of a capital sentencing proceeding, defense counsel has the obligation to conduct a “reasonably substantial, independent investigation” into potential mitigating circumstances. Baldwin, 704 F.2d at 1332-33. In assessing counsel’s performance, we look to such factors as what counsel did to prepare for sentencing, what mitigating evidence he had accumulated, what additional “leads” he had, and what results he might reasonably have expected from these leads. Applying this standard to the facts of Neal’s case, we conclude that a reasonable attorney would have investigated further and put on a more compelling defense during sentencing.3
(b)
Neal’s evidence of mitigating factors presented during sentencing consisted of the testimony of only two witnesses: Neal’s mother, who gave an overview of Neal’s troubled background; and a psychologist, Dr. Dana Alexander, who testified about Neal’s mental and emotional difficulties. Reviewing this testimony does not take long.
The testimony by Neal’s mother covers nine pages in the transcript. She began with a brief description of his difficulties in school and his subsequent transfer to El-lisville and Whitfield:
He went to school till he was ten years old. He couldn’t learn in school. The teachers tried, I tried. He had a [hard] time remembering things. He couldn’t remember, and at the time his daddy and I separated, and because of his ability to learn, the welfare told me to send him to a State School at Ellisville, and he would have a trade, and he stayed there till he was sixteen, and then they took him from Ellisville to Whitfield, and he stayed there till he was eighteen or nineteen, and then he came home and stayed with me — lived with me for about a year or maybe two, and then my husband got disabled to work and he started drawing his Social Security, and he wasn’t — he couldn’t stay there with us because they told us that he couldn’t, and he got mad and left home, and I didn’t know where he even went from there, and we had five other kids, and I wasn’t able to take care of them when we separated.
This testimony also included several responses to counsel’s questions about Neal’s education level (second grade, but unable to read or write), his marital status (married twice, with one daughter), and whether he owned a home (he did not). Neal’s mother also testified that she tried to see him often when he was at Ellisville, but often could not get there. Finally, Neal’s mother explained that she had been unable to find anyone to adopt Neal, as she had done with her other ten children, and consequently placed Neal in the Ellisville institution.
Dr. Alexander’s testimony, excluding voir dire, includes twenty-four pages in the transcript, about a third of which is cross-examination. Dr. Alexander testified that she had tested Neal’s IQ, and that his score was 54, at the low end of mild retardation. According to her, Neal suffered from “organic brain syndrome,” a disability characterized by “slowness, perseveration, concreteness of perception, problems with memory, problems with orientation and poor impulse control ... impaired intellectual functioning, impaired social and vocational functioning ... and inability to use controls like you or I.” She concluded that Neal had the mental ability of an eight year old. Dr. Alexander also testified about Neal’s behavioral problems. She described him as “more irritable” and “more easily provoked” than normal people. She also explained that Neal suffered from “psycho-sexual confusion,” “a lack of a definite identity with either the male or the female role.” This is the essence of *689her testimony on direct and redirect examination.
The jury was therefore given the following picture of Neal: a man with an IQ of 54, with the mental ability of an eight year old, with conceptual deficiencies, with sexual identity problems, who, because of his mental deficiencies was less able to control himself and his impulses, including provocation, who had been denied any semblance of a homelife and virtually rejected by his mother who had placed him in state institutions for the retarded and mentally ill, where he grew up and spent eight years of his youth. That is the essence of the mitigating evidence that defense counsel presented to the jury. Although it seems to touch many relevant points, it was presented to the jury in an abbreviated form with no elaboration.
(e)
Neal’s habeas counsel now presents us with forty-two pages of affidavits and reports concerning Neal’s background as evidence of mitigating factors. The affidavits come primarily from doctors and employees at Ellisville, Whitfield, and the prison in Oklahoma. There is also one from Neal’s sister and another from a social worker. Neal contends that trial counsel was ineffective for failing to gather and present these materials that would have weighed against the death sentence. It is important to understand fully the additional evidence that could have been presented, and we turn now to the relevant material contained in the habeas record.
The additional evidence does, indeed, make disturbing reading. Neal’s sister, Maryann McNeese, describes his childhood household- — eleven children living with them parents in a two-bedroom house. Neal’s father was an abusive alcoholic who was particularly brutal towards Neal. An affidavit by Neal’s mother confirms this, as does an affidavit by a social worker who knew the family at the time, Marguerite McAulay. Neal was ten when he was sent to Ellisville, a school for retarded children. McNeese’s affidavit states that when that happened, “he was like a throwed away child. It was like he didn’t have parents.”
Lamar Collom, an Ellisville bus driver and cafeteria worker, befriended Neal and has provided another affidavit. It describes Neal as a “good worker” and “likeable kid.” Collom goes on to say that he “thought a lot” of Neal, but felt sorry for him because of his terrible family background. Collom’s affidavit also explains how Collom “used to give [Neal] a little money some weekends,” which Neal “would spend at the little canteen.” An affidavit by Lucille McIntosh, an Ellisville cafeteria employee, describes Neal as “a good worker,” “cooperative,” and a boy who “got along with the other kids.”
Two other items of evidence related to Neal’s time at Ellisville concern Neal’s level of intelligence. James Woolington, who founded Ellisville’s psychology program, tested Neal twice. In his affidavit, Mr. Woolington concludes that Neal had an IQ between 54 and 60 at that time. James Johnson, Ellisville’s Coordinator of Psychological Services at the time, also tested Neal’s intelligence. A report attached to his affidavit sets his IQ at 55. The report also discusses Neal’s behavioral problems, such as running away, picking on smaller children, and lack of remorse for misdeeds. It describes his behavior as “spasmodic” and questions Neal’s ability to control himself. Finally, the report mentions that Neal was suspected of having engaged in homosexual relationships while at Ellis-ville.
At age sixteen, Neal was transferred to Whitfield, a mental institution. Dr. A.G. Anderson, a psychologist who knew Neal at that facility, describes the conditions there in another affidavit:
The unit was not an enjoyable place for a young retardate to live and was not a good therapeutic environment. It was not a place that was beneficial to Howard’s mental prognosis. Neither Howard nor any of the other residents got the care they should have gotten. In fact it was not a good environment for *690anyone including the staff. Not many professionals wanted to work there.
The facility had little in the way of education, training, or recreation. The affidavit goes on to explain that between ages sixteen and eighteen, Neal lived in the maximum security unit with approximately 150 “chronics,” people with mental disorders that rendered them unable to stand trial, or who had shown aggressive tendencies.
After Neal was released from Whitfield, he returned home. He lived there several months until his mother forced him to leave because she feared losing her welfare benefits if he stayed. Neal then headed to Oklahoma, where he was soon arrested for assault and battery with a dangerous weapon.
The habeas record also contains several affidavits from people who knew Neal in the Oklahoma prison. The prison psychologist, Thomas Norwood, testifies that he knew Neal “about as well as [he had] known anybody [he has] worked with in [his] professional capacity.” Mr. Nor-wood’s affidavit goes on to explain that “the treatment of [Neal] by other inmates was so horrible [that he has] difficulty discussing it,” and that Neal was the “most tragic case [he] had” because Neal was so defenseless. As an example, Mr. Norwood describes an incident where thirty-to-forty inmates forced Neal under a table and to commit sodomy on each of them in succession. Finally, Mr. Norwood explains that he helped Neal get a place to live and social security benefits after Neal got out of prison.
Jack Cowley, currently a warden at another Oklahoma prison, was Neal’s case manager at the time. His affidavit asserts that he “more or less adopted [Neal]” in prison. Mr. Cowley’s affidavit also relates that he was concerned about what would happen to Neal if Neal returned to Mississippi, where there was no one there to care of him.
In addition, Neal has presented us with his prison records from Oklahoma. There is no report concerning the event Mr. Nor-wood described. The records do contain a report, however, about an incident in which two inmates raped Neal. The report presents Neal as defenseless, gullible, and at the mercy of the inmates generally: “Inmate Neal also made reference to several instances when his cell partner ... has tried to get Neal to sell himself so they could have some money.”
Several of the affidavits mention that Neal married Glenda Snow, who was also mildly retarded, after his release from prison. Though the two moved to Mississippi, her parents broke up the marriage and took their daughter home. McNeese’s affidavit describes her brother’s reaction: “Howard told me it hurt him when Glenda had to go back to Oklahoma.” A few of the affidavits mention that Neal subsequently remarried and had a child with his second wife.
(d)
In a later section of this opinion, we will address the question whether presentation of this additional testimony would have changed the outcome of the sentencing hearing. For now, we emphasize only the volume and easy availability of this additional mitigating evidence.
■Perhaps the most troubling aspect of these affidavits is that they indicate that counsel never contacted any of the other people (with the exception of Neal’s mother) who have provided the additional testimony we now have before us, and which would have added to and developed the skeletal evidence before the jury. For example, Neal’s sister, Maryann McNeese, states that she contacted defense counsel to ask about the case, and would have been willing to testify on her brother’s behalf, but that they never asked her to do so. And having examined the mother’s testimony at sentencing, it is not even clear how much information defense counsel collected from her before putting her on the stand. Dr. Alexander’s testimony, too, was surely limited by the fact that she had met with Neal just one time, three days *691before testifying, and that trial counsel failed to tell her about what specific crime Neal had been charged with or any facts about his personal history.
The only materials that defense counsel appears to have had, other than the assistance of Neal’s mother, are the records from Neal’s time at Ellisville and Whitfield.4 And while they had some indication of his difficult life in the institutions and in prison, Neal’s attorneys chose not to pursue these sources of evidence. In his affidavit, one of Neal’s attorneys at trial tried to explain that they did not contact these potential witnesses due to lack of funds and experience.
We did not have the time or money to properly investigate [Neal’s] case. We had no money to interview witnesses or travel.... I did not get Howard’s records or interview people who had dealt with him in Oklahoma. I did not interview any of his relatives other than his mother.... We did not have a complete psychological examination of Howard which would have included a thorough investigation of his past medical and psychological history and a neurological examination. Had we had the time and money we would have done the above investigations for use at both trial and sentencing.
This explanation does not fully address, however, the fact that most of the mitigating evidence was readily available and would have cost no more than several long distance telephone calls or postage stamps.
Because of the extent to which these available materials could reasonably have been expected to augment Neal’s case, we conclude that his trial counsel was deficient in failing to investigate, gather, and consider it for purposes of presentation at Neal’s sentencing hearing.5
(2)
(a)
Having concluded that counsel’s performance during sentencing was deficient, we now turn to the second prong of Strickland and determine whether the deficient performance prejudiced Neal’s defense during sentencing. To establish prejudice, Neal must show that there is at least “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland, 466 U.S. at 694, 104 S.Ct. 2052; see also United States v. Green, 882 F.2d 999, 1003 (5th Cir.1989) (“A defendant who alleges a failure to investigate on the part of his counsel must allege with specificity what the investigation would have revealed and how it would have altered the outcome of the trial.”) By reasonable probability, the Court means a probability sufficient to undermine confidence in the outcome. Id.
In determining prejudice, we are thus required to compare the evidence actually presented at sentencing with all the mitigating evidence contained in the postcon-viction record. Stated to the point: Is this additional mitigating evidence so compelling that there is a reasonable probability at least one juror6 could reasonably have *692determined that, because of Neal’s reduced moral culpability, death was not an appropriate sentence?
(b)
The additional mitigating evidence has been described in detail above. It seems indisputable that this new evidence, standing alone, presents a hugely sympathetic case for mitigating a death sentence. But the State argues that we must evaluate the evidence in context of the actual proceedings at sentencing; specifically, the State argues that no prejudice occurred because much of the additional testimony would not have been admitted, either for tactical or procedural reasons.
First, the State raises questions as to the admissibility of the testimony by Mr. Collom, Mr. Norwood, and Mr. Cowley because their statements about their personal affection for Neal, indicating a degree of personal worth, do not relate to the statutory mitigating factors. This argument creates a potential concern that such evidence would not be proper for the jury to consider. It seems clear, however, that this evidence would have been permissible for the jury to consider as mitigation. While he asserted two statutory mitigating factors — the offense was committed under the influence of extreme mental or emotional disturbance, and his capacity to conform his conduct to the requirements of law was substantially impaired, see Miss.Code Ann. § 99-19-101(6)(b) and (f) — Neal was not limited to presenting evidence that related to these statutory mitigating factors. In seeming contradiction to Furman v. Georgia, 408 U.S. 238, 92 S.Ct. 2726, 33 L.Ed.2d 346 (1972),7 the Supreme Court has repeatedly affirmed the portion of the plurality opinion in Lockett v. Ohio, 438 U.S. 586, 604, 98 S.Ct. 2954, 2964-65, 57 L.Ed.2d 973 (1978), holding that the sentencer may not be “precluded from considering, as a mitigating factor, any aspect of a defendant’s character or record and any of the circumstances of the offense that the defendant proffers as a basis for a sentence less than death.” See Buchanan v. Angelone, 522 U.S. 269, 276, 118 S.Ct. 757, 761, 139 L.Ed.2d 702 (1998)(reaffirming Lockett); Penry v. Lynaugh, 492 U.S. 302, 318, 109 S.Ct. 2934, 2946, 106 L.Ed.2d 256 (1989)(same); Eddings v. Oklahoma, 455 U.S. 104, 113-14, 102 S.Ct. 869, 876-77, 71 L.Ed.2d 1 (1982)(same). Such evidence appears clearly to be admissible under Mississippi law. See Evans v. State, 725 So.2d 613, 694 (Miss.1998)(explaining that § 99-19-101(1) of the Mississippi Code provides for the admission of nonstatutory mitigating evidence). The testimony in question here is evidence concerning Neal’s character. These potential witnesses found something worthy about Neal as a human being. Their testimony, therefore, would have been before the jury as non-statutory mitigating evidence and would have been weighed along with the statutory mitigating circumstances. See Billiot, 135 F.3d at 315.
Second, the state argues that Mr. Nor-wood’s testimony about the incident with the thirty-to-forty inmates would have been inadmissible hearsay, as Mr. Nor-wood was presumably not present at the time. Given this circuit’s narrow reading of Green v. Georgia, 442 U.S. 95, 99 S.Ct. 2150, 60 L.Ed.2d 738 (1979), we cannot assume that Norwood’s testimony would have been admitted. See Edwards v. Scroggy, 849 F.2d 204, 212 (5th Cir.1988)(exclusion of priest’s testimony about direct statements made by the defendant did not render the trial “fundamentally unfair” and thus did not violate Due Process); Barefoot v. Estelle, 697 F.2d 593, 597 (5th Cir.1983) (“We think that Green is limited to its facts, and certainly did not federalize the law of evidence .... [although] certain egregious evidentiary errors may be redressed by the due process clause.”). However, even if Norwood’s testimony would have been *693deemed inadmissible, the written report about the rape is similar in nature. That report discusses Neal’s defenselessness and that his own cellmate was trying to manipulate Neal to sell himself for the cellmate’s profit. Thus, we believe the flavor of the Norwood testimony, if not its details, would have been available to the jury in the form of that report.
Third, the State argues that defense counsel made a strategic decision to withhold some of the testimony. As we pointed out above, this argument could apply only to Mr. Johnson’s report, which was included with the Ellisville records that defense counsel had. As to the other evidence, Neal’s attorneys simply could not have made a decision strategically to withhold information that they had not obtained.
Fourth, the State maintains that even if defense counsel had obtained the new mitigating evidence, counsel would not have presented that evidence for strategic reasons. The State makes this argument primarily with respect to evidence about Neal’s imprisonment in Oklahoma. As the State points out, defense counsel successfully suppressed evidence of this conviction and imprisonment during the guilt phase of the trial. But the State also acknowledges that the prosecutors themselves could have presented evidence of Neal’s imprisonment and past offenses during sentencing under Miss. Code Ann. § 99-19-101(5)(b). In one sense, Neal would not have “opened the door” by putting this evidence on during sentencing because the door was already open. But if Neal had sought to introduce evidence from his Oklahoma imprisonment — if Neal had, in effect, invited the State to walk through this already-open door — then the State would have tried to put Neal’s Oklahoma past criminal experience in the worst possible light for Neal. In sum, we cannot tell whether withholding the evidence may have been strategically advantageous, and we find this argument inconclusive.
(c)
The State’s most persuasive argument concerning prejudice is that the additional mitigating evidence would have been, in the Mississippi Supreme Court’s words, “substantially redundant and cumulative.” Although the mitigating evidence the jury actually heard was skeletal, they were presented the basic evidence that Neal was moderately retarded, had been severely neglected by his family, spent several years in state institutions, and suffered from serious behavioral problems, including lack of self-control and sexual identity problems.8 In the State’s view, the additional evidence presented in the affidavits does little more than reinforce the testimony of Neal’s mother and Dr. Alexander and provide details to an otherwise sketchy portrait of Neal’s life. It must be conceded that the jury was presented a clear, if not fully portrayed, picture of Neal’s pathetic life. It was on this basis that the Mississippi Supreme Court concluded that Neal was not prejudiced by the omission of the evidence.
The State further argues that when we are considering the mitigating effect of this supplementary evidence, we must weigh it against the facts of the crime. The jury had heard how Neal had brutally raped and murdered Amanda Joy, his thirteen-year-old niece. The manner of death was unspeakably horrible. There were bruises and lacerations about her face, head, and left wrist, and there was also evidence of manual strangulation in addition to the bullet hole in her abdomen. The pathologist’s report states that Amanda Joy could have survived between five and thirty minutes after Neal had left her to die. To overcome the aggravating factors and avoid the death penalty, Neal’s mitigating evidence would had to have *694been overwhelming and specifically relevant to reducing his moral culpability for his heinous crimes. The state thus concludes that if the facts the jury had already heard were not sufficient to outweigh the aggravating circumstances, then the additional evidence would have been unlikely to sway a jury.
Although this question is extremely close, we are unable to agree with the State's argument because, as the Supreme Court has recently explained, courts must give due consideration to the quality and volume of the additional mitigating evidence. The Court’s evaluation of ultimate prejudice in that case is instructive:
[T]he graphic description of [the defendant’s] childhood, filled with abuse and privation, or the reality that he was “borderline mentally retarded,” might well have influenced the jury’s appraisal of his moral culpability.... [T]he entire postconviction record, viewed as a whole and cumulative of mitigation evidence presented originally, raised “a reasonable probability that the result of the sentencing proceeding would have been different” if competent counsel had presented and explained the significance of all the available evidence.
Williams, 120 S.Ct. at 1515-16. To be sure, there are several significant factual differences between Williams and Neal.9 But that does not allay our concern that the underlying principle of the Williams prejudice determination requires that we must assign significant weight to the quality of additional mitigating evidence. Specifically, Williams suggests to us that the correct analysis of our case is that with a more detailed and graphic description and a fuller understanding of Neal’s pathetic life, a reasonable juror may have become convinced of Neal’s reduced moral culpability.
Viewed together, these affidavits and other evidence augment Neal’s mitigating circumstances argument in at least five ways. First, they present additional details about Neal’s childhood, including the terrible living conditions with the alcoholic and abusive father. Second, they provide a description of the bleak, depressing, and hopeless life at the mental institutions. This is especially true with respect to Whitfield. Third, the affidavits describe Neal’s abuse and mistreatment in prison and his general helplessness there. Fourth, Mr. Woolington’s testimony supports Dr. Alexander’s limited testimony as to the level of Neal’s retardation and his inability to control much of his behavior. Fifth, the affidavits humanize Neal by demonstrating that there were people along the way who saw some worth in him and befriended him.
Our inquiry is obviously very difficult, but given the amount and character of the mitigating evidence in this case, we believe that there is a reasonable probability that a jury would not have been able to agree unanimously to impose the death penalty if this additional evidence had been effectively presented and explained to the sentencing jury. In our judgment, then, the Mississippi Supreme Court’s conclusion that the additional mitigating evidence was merely redundant and not prejudicial is erroneous.
D
Our conclusion that the state court’s prejudice determination is incorrect, however, is not enough to afford federal habe-as relief to Neal because, under AEDPA, we owe considerable deference to the Mississippi Supreme Court. The statute plainly states that we may not grant habe-as relief unless the state court’s decision unreasonably applies federal law. See 28 *695U.S.C. § 2254(d)(1); H.R. Conf. Rep. 104-518, 104th Cong., 2d Sess. Ill (1996), U.S. Code Cong. & Admin. News at 944 (Section 2254(d) generally “requires deference to the determinations of state courts”). Thus, Neal must go further yet and demonstrate that the prejudice determination of the Mississippi Supreme Court “involved an unreasonable application” of Strickland. We now turn to try to give meaning to the term “unreasonable application,” especially as it applies to this case.
(1)
The Supreme Court in Williams explained that “a federal court making the ‘unreasonable application’ inquiry should ask whether the state court’s application of law was objectively unreasonable.” Williams, 120 S.Ct. at 1521. But even after Williams, it is not immediately clear to us whether a federal habeas court looks exclusively to the objective reasonableness of the state court’s ultimate conclusion or must also consider the method by which the state court arrives at its conclusion. This question takes on some significance in a case such as Neal’s, where the state court’s holding (that Neal suffered no prejudice under Strickland) may be objectively reasonable, but in reaching that holding, the court did not adequately evaluate and weigh the substantial evidence-the implicit suggestion being that the state court may have reached a different, but still “reasonable,” conclusion if a more thorough method of reasoning had been applied.10
The Seventh Circuit, sitting en banc, appears to have concluded that federal courts must scrutinize a state court’s method of reasoning. “By posing the question whether the state court’s treatment was ‘unreasonable,’ § 2254(d)(1) requires federal courts to take into account the care with which the state court considered the subject.” Lindh v. Murphy, 96 F.3d 856, 871 (7th Cir.1996) (en banc), rev’d on other grounds, 521 U.S. 820, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997). The reasonableness of a court’s application of federal law must be measured, at least in part, by determining whether a state court provided “a responsible, thoughtful answer reached after a full opportunity to litigate.” Id.11
Certain passages in the Williams decision could be read to support this view. Writing for the Court, Justice Stevens explained that the Virginia Supreme Court’s *696“prejudice determination was unreasonable insofar as it failed to evaluate the totality of the available mitigation evidence — both that adduced at trial, and the evidence adduced in the habeas proceeding — -in reweighing it against the evidence in aggravation.” Williams, 120 S.Ct. at 1515. There is, therefore, at least some basis for the view that Section 2254(d)’s “unreasonable application” standard refers to the quality of the state court’s analysis.
On the other hand, this process-oriented view has been rejected by other circuits12 and challenged by Chief Judge Posner of the Seventh Circuit. In his view, scrutinizing state courts’ methods of reasoning “would place the federal court in just the kind of tutelary relation to the state courts that the [AEDPA was] designed to end.” Hennon, 109 F.3d at 334-35. Similarly, we do not interpret AEDPA in such a way that would require a federal habeas court to order a new sentencing hearing solely because it finds the state court’s written opinion unsatisfactory.13 It seems clear to us that a federal habeas court is authorized by Section 2254(d) to review only a state court’s “decision,” and not the written opinion explaining that decision.
In the absence of clear guidance from the Supreme Court, we conclude that our focus on the “unreasonable application” test under Section 2254(d) should be on the ultimate legal conclusion that the state court reached and not on whether the state court considered and discussed every angle of the evidence. The latter approach appears unduly formalistic considering that the federal habeas court has the full record before it and is competent to determine whether Strickland has been unreasonably applied to the case before it. Even though a thorough and well-reasoned state court opinion may be more likely to be correct and to withstand judicial review, it simply does not follow that “the criterion of a reasonable determination is whether it is well reasoned.” Id. at 334-35. Instead, the only question for a federal habeas court is whether the state court’s determination is “at least minimally consistent with the facts and circumstances of the case.” Id. at 335.
(2)
Thus, in making our unreasonable application determination, we look only to the substance of the Mississippi Supreme Court’s decision. The state court concluded that presentation of the additional mitigating evidence would probably not have changed the outcome of the case. The precise question, then, is whether the court’s ultimate conclusion' — that there was no prejudice and, consequently, no ineffective assistance of counsel under the Strickland test — is minimally consistent with the facts of Neal’s case.
As we have iterated throughout this opinion, the facts of this case are as horrible as one can imagine. The jury was reminded repeatedly that Neal killed his own brother in an argument that began when Neal fondled his young niece; that Neal, after killing his brother, kidnaped his niece and her friend and then brutally raped and killed both girls; that he shot his niece and left her to suffer for perhaps thirty minutes before she died. If any murder qualifies as “heinous, atrocious, or cruel,” the murder of Amanda Joy Neal does. We acknowledge that it was surely not unreasonable for the Mississippi Supreme Court to suggest (implicitly) that the evidence mitigating his moral culpability would have to be overwhelming.
The jury in mitigation heard Neal’s pitiful life story from his mother and a psychiatrist. The jury thus had before it evidence that Neal had an IQ of 54 and the mental ability of an eight year old; that he had been neglected by his family and spent *697much of his youth in state institutions for the retarded and mentally ill; that he had sexual identity problems; and that because of his mental deficiencies, he was less able to control his impulses. The evidence actually presented at the sentencing hearing did not persuade the jury to mitigate Neal’s sentence.
Neal’s supplemental mitigation evidence also relies almost exclusively on testimony that he was mentally retarded and had been badly mistreated by numerous people throughout his life. Much of this evidence had already been presented, even if in an outline form. The only new evidence was that some of the people whom he had encountered felt sorry for Neal and that their sympathy for his plight caused them to reach out to him. Moreover, the testimony about Neal’s life at Whitfield and the Oklahoma prison would have proved to have been double-edged in the sense that it would have reminded the jurors that Neal was a bully and had a criminal conviction. Although the additional mitigating evidence was of a significantly better quality than that actually presented, much of it was similar in nature to the original evidence. With those considerations in mind, the Mississippi Supreme Court concluded that the additional evidence was not substantial enough to outweigh the overwhelming aggravating circumstances.
Under the deferential standard of Section 2254(d), and given the circumstances of this case, we cannot conclude that the Mississippi Supreme Court unreasonably applied Strickland to the facts of Neal’s case. It was not unreasonable, in other words, to conclude that the outcome would have been the same because the additional evidence did not serve to reduce further his moral culpability for such a heinous and unforgivable crime beyond the evidence already presented. We thus hold that the state court’s prejudice determination was objectively reasonable in the sense that it is not inconsistent with the facts of this case — specifically, when those facts are viewed in the context of the extreme cruelty of the murder as an aggravating circumstance and that much of the mitigating evidence had already been presented to the jury, albeit in an abbreviated form.
Ill
Because we conclude that the Mississippi Supreme Court’s decision did not involve an unreasonable application of Strickland, Neal’s petition for a writ of habeas corpus must be DENIED and, accordingly, the judgment of the district court is
AFFIRMED.
. In a separate trial, Neal was tried and convicted for Bobby's murder but received only a life sentence. He was never tried for Melanie Sue's murder.
. Ellisville was a school for retarded children and Whitfield was the state mental hospital.
. The Mississippi Supreme Court appears to have assumed that counsel's performance was deficient, as the state court focused solely on Strickland's second prong, prejudice. But because the state does not concede the point, we address its arguments regarding deficiency.
.Of course, we do not assume that all of this evidence should have gone before the jury. A psychologist’s report, for example, was included in the Ellisville records that defense counsel obtained before sentencing. Some portions of that report are directly relevant to one the statutory mitigating factors: "[Neal's] ability to control his own behavior is questionable." But other portions of the report present an unflattering picture of Neal, describing him as a bully who felt no remorse for misdeeds. Given these facts, counsel may have made a strategic decision to withhold that report, which, in the context of the sentencing hearing that occurred, we will not second-guess.
. Several factually similar Fifth Circuit decisions have also found deficient performance where counsel have failed to investigate a defendant’s background in mental institutions and prisons. See Bouchillon v. Collins, 907 F.2d 589 (5th Cir.1990); Profitt v. Waldron, 831 F.2d 1245 (5th Cir.1987); Beavers v. Balkcom, 636 F.2d 114 (5th Cir.1981).
. In Mississippi, the jury must vote unanimously to impose the death penalty. Miss. Code Ann. § 99-19-103.
. See Walton v. Arizona, 497 U.S. 639, 656-69, 110 S.Ct. 3047, 3059-66, 111 L.Ed.2d 511 (1990)(Scalia, J., dissenlingXexplaining the conflict between Furman and Lockett).
. The evidence presented to the jury is summarized in more detail in section II.C.(l).(b) above.
. First, some of the additional mitigating evidence in Williams was not just of a better quality of evidence than that presented at sentencing but was, in fact, evidence of new mitigating factors. Second, the fact that Williams had turned himself in to the police was a significant piece of additional mitigating evidence that is not present in the case before us. Third, the circumstances of the murder in that case, which arose from a simple robbery, were less atrocious and unforgivable than those in the case before us.
. The Mississippi Supreme Court’s prejudice determination is fairly brief. After summarizing Neal's argument, the court stated that the additional evidence was
substantially redundant or cumulative when compared with the evidence Neal offered at trial. Specifically, Neal now wants to present evidence of his lack of mental capacity, a fact said to go to the voluntariness of his confession and to be in mitigation of sentence. But he went into these same matters at trial. He called Dr. Dana Alexander, a clinical psychologist. He showed that he had been in Ellisville State School for retarded youths and that he was later in the retardation unit at Mississippi State Hospital at Whitfield. He further proved that his IQ was 54. Because it is cummulative [sic], what Neal alleges and purports to show now that counsel should have developed and proved simply does not amount to a substantial showing of denial of a state or federal right. Miss.Code Ann. § 99-39-27(5).
The same is true of Neal's allegations that competent counsel should have done a better job at sentencing phase of proving the details of Neal's troubled life. In addition to the testimony described above regarding his prior institutionalization, Neal called his mother as a witness who told his life story. Perhaps the details could have been fleshed out more fully through additional witnesses. This may often be said after an unsuccessful trial experience.
Neal, 525 So.2d at 1282-83. Cf. Williams, 120 S.Ct. at 1502.
. In spite of the straightforward language of Lindh, a Seventh Circuit panel rejected the argument that the "unreasonable application” test requires a federal habeas court to consider the state court's process of reasoning. See Hennon v. Cooper, 109 F.3d 330 (7th Cir.1997). Chief Judge Posner contended that Lindh stands only for the proposition that "the better the job the state court does in explaining the grounds for its rulings, the more likely those rulings are to withstand further judicial review.” Id. at 335, 117 S.Ct. 2059.
. See, e.g., Long v. Humphrey, 184 F.3d 758, 760-61 (8th Cir.1999) (focusing on the reasonableness of the ''outcome”); O'Brien v. Dubois, 145 F.3d 16, 25 (1st Cir.1998) (same).
. In that situation, a habeas petitioner may not be the victim of constitutional error but only of a mere lapse in "judicial articulateness.” Hennon, 109 F.3d at 335.