concurring in part and dissenting in part.
I concur in all of the majority’s opinion except for its determination in part II.A.l. of Bertolotti’s ineffective assistance of counsel claim. I dissent because the dis*1529trict court erred by not requiring that the writ issue unless the state provides Berto-lotti a resentencing hearing so that evidence with respect to his mental condition can be considered as mitigating evidence. Because defense counsel failed to have Bertolotti undergo a court-ordered mental evaluation, the jury was prevented from considering two Florida statutory mitigating circumstances as well as non-statutory mitigating circumstances. Furthermore, defense counsel did not have sufficient information with which to properly prepare for the sentencing proceeding, interviewing family members, investigating Bertolotti’s mental health history, and preparing for argument. Cf. Magill v. Dugger, 824 F.2d 879 (11th Cir.1987).
There is no question but that counsels’ inadvertent failure to schedule the mental evaluation of the petitioner in this case constitutes deficient performance. Furthermore, the record reveals that this failure precluded the defendant from receiving the individualized sentencing proceeding that has been mandated by Lockett/Eddings/Skipper/Hitchcock and Penry. While the majority finds no prejudice in this case, it “voluntarily” goes out of its way to discuss counsels’ performance and to conclude that this performance was reasonable. Besides being contrary to the dictates of Strickland v. Washington, 466 U.S. 668, 697, 104 S.Ct. 2052, 2069, 80 L.Ed.2d 674 (1984), which states that a court should dispose of a claim of ineffective assistance of counsel on the prejudice inquiry if that is easier, this overreaching by the majority also creates law that conflicts with other cases of this circuit. See, e.g., 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); Magill, 824 F.2d 879; Stephens v. Kemp, 846 F.2d, 642 (11th Cir.), cert. denied, — U.S. —, 109 S.Ct. 189, 102 L.Ed.2d 158 (1988).
Bertolotti appeals the denial of his petition for writ of habeas corpus pursuant to 28 U.S.C. § 2254. Bertolotti alleges that his trial counsel rendered him ineffective assistance by not having Bertolotti examined by a psychiatrist in order to both develop a possible insanity defense and to develop mitigating evidence for use in the sentencing hearing in the event a guilty verdict was returned. In this case, after interviewing the petitioner and conducting some preliminary investigation, the attorneys for Bertolotti requested a psychiatric evaluation. The court granted this request and, about six months before trial, Dr. Pollack was appointed to evaluate Bertolotti. Counsel for Bertolotti, however, delayed making the arrangements necessary for the psychiatric exam until the morning of the sentencing hearing at which time Ber-tolotti refused to see the doctor. At the State rule 3.850 hearing, Bertolotti’s lawyers testified that in hindsight they should have had Bertolotti examined prior to trial and certainly would do so if they were currently handling the defense. See Berto-lotti v. State, 534 So.2d 386, 388 (Fla.1988).
INEFFECTIVE ASSISTANCE
As the majority recognizes, we evaluate claims of ineffective assistance of counsel under the test enunciated in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). To prevail in this case, Bertolotti must show that his lawyers’ performance fell outside the scope of reasonably professional assistance and that there is a reasonable probability that the results of either the guilt or penalty phase proceedings would have been different but for the alleged substandard performance. Id. at 104 S.Ct. 2052.
A. Deficient Performance.
Under the facts of this case, the failure to investigate Bertolotti’s mental condition through a psychiatric exam was clearly unreasonable. The district court’s conclusion (also reached by trial court) that “nothing in Petitioner’s history or in his behavior surrounding the murder and his subsequent confessions was so unusual as to call his mental condition into question,” (D.Ct.Op. at 31) is, as the Florida Supreme Court found,1 wholly refuted by the record: *1530First, the fact that a psychiatric evaluation was both requested and granted and the fact that a psychiatrist was obtained to conduct an evaluation right before the sentencing hearing belies a conclusion that counsel had no reason to suspect mental illness. Second, each of the three attorneys who were involved with the case at the preparation stage admitted at the 3.850 hearing that the petitioner should have been examined.
Mr. DuRocher, who was the elected Public Defender of the Ninth Judicial Circuit, initially interviewed the petitioner and assigned the case to Assistant Public Defenders, Mr. Wolf and Mr. Kenny. Mr. DuRo-cher’s qualifications are impeccable. He received his law degree from the University of Florida and was admitted to practice law in the State of Florida in 1967. After approximately two years in private practice, Mr. DuRocher became director of the Legal Aid Society of Orange County and served in that position until 1971 when he was elected to the bench to preside in juvenile court. Some five years later, Mr. Du-Rocher resigned to reenter private practice, a practice with an emphasis on criminal matters. Finally, in 1980, Mr. DuRocher ran for and was elected to the position of Public Defender, and he had been in that position for three years when he first interviewed Mr. Bertolotti.
When asked whether, as a result of his interview, he had formed an opinion as to whether Mr. Bertolotti should be mentally evaluated, Mr. DuRocher stated, “[Y]es, all the signals were there. You had to — he should have been evaluated.” Mr. DuRo-cher questioned Bertolotti’s mental condition based on several factors, one being the nature of Bertolotti’s story of what happened. “... I told him that if [his story were] true, then we must seriously investigate the mental condition, possible mental illness of the woman, Mrs. Ward. Because if what he was telling me was true, then she was crazy or vice versa.” He later elaborated “ — and [his story] was a clue to me because either she [had mental problems] or he did.” R. 3.850 H. at 274-306.
Mr. Kenny, one of the assistant attorneys on the case, agreed that, in this situation, a psychiatric evaluation was necessary. He testified as to numerous things that occurred during the preparation of the case that caused him to be concerned about Bertolotti’s mental state and summarized:
Along with everything else it was one of those things that just seemed to indicate that [Bertolotti] wasn’t all there, to use a colloquial term; he may have had a mental problem. I’m not saying that all these things are definitive, that [sic] show absolutely that he had a mental problem but they are the kinds of things I think ought to be explored by somebody who knows more about it than I do or [Mr. Wolf] did.
R. 3.850 H., Vol. 21 at 145. Additionally, as the Florida Supreme Court noted:
[N]otes taken by Mr. Wolf reflect that Sharon Griest, Bertolotti’s girlfriend at the time of the murder, told him that she “believes” that Bertolotti “needs psychiatric help,” that “he did not know what he was doing at the time of the offense” and that he might have a “split personality.” Griest also told Mr. Wolf that Ber-tolotti was discussing “suicide a great deal.” Even Mr. Wolf acknowledged that these statements in conjunction with other factors should have caused him to question Bertolotti’s mental condition.
534 So.2d at 389.
The above testimony is only a smattering of the evidence that reveals that the attorneys assigned to Bertolotti’s case had every reason to question their client’s sanity. Furthermore, this evidence clearly reveals that the failure of counsel to secure a mental health exam was not the result of any trial tactic or strategy. Rather, attorney *1531Kenny testified by deposition that it was the result of inadvertence and scheduling problems:
Q: So you have no idea why Mr. Wolfe didn’t have a mental health expert in regard to defending the felony murder charge? ...
A: No, I can’t [sic]. He may have told me and I can’t remember. At this date, I can’t remember. I think, and this is just speculation, it was one of those things that he didn’t think was real correct or just for some reason, never got around to doing until it was too late.
* * * * * *
Q. Apparently, effort [to obtain a mental health evaluation] was made before? ...
A: What it came down to in the end was a scheduling problem with [the psychiatrist] and, in hindsight, I think that we, A, should have gotten somebody else, although, quite honestly, we didn’t want to have somebody else....
R. 3.850 H. at 897-98; 941.
Upon reaching the conclusion that counsel failed to obtain available psychiatric evidence for no strategic reason, our inquiry into counsels’ performance should end with a finding that counsel performed deficiently. The majority in this case, however, credits counsels’ inadvertence as a “decision not to secure a psychiatric examination until the morning of Bertolotti’s sentencing hearing.” Maj. op. at pages 1511-12. It then proceeds to discuss the reasonableness of this “decision.” Characterizing counsels’ accidental failure to secure a mental exam until the morning of the sentencing hearing as a “decision” that must be evaluated for “reasonableness” oversteps the settled law of this Circuit regarding this issue.
It is true that any decision not to investigate must be reasonable. Armstrong v. Dugger, 833 F.2d 1430, 1433 (11th Cir. 1987). A failure to investigate that is not the result of any trial strategy, however, is no decision at all. Harris v. Dugger, 874 F.2d 756, 762 (11th Cir.1989). Thus, in Middleton v. Dugger, 849 F.2d 491, 494 (11th Cir.1988), we held that counsel’s failure to make any effort to investigate the petitioner’s background was unreasonable because it was not based on any discernible trial strategy. Likewise, in the present ease, counsel performed deficiently when for no reason, they failed to schedule an available mental health examination as part of an investigation into Bertolotti’s mental condition.
Although the majority’s depiction of counsels’ fortuitous actions as an actual “decision” sets up a dangerous precedent for future cases, the actual analysis the majority employs to find that this “decision” was reasonable is even more disheartening. In an attempt to dismiss the evidence as insufficient to alert the attorneys to the possibility of Bertolotti’s insanity, the majority dissects the evidence and discusses the insufficiency of each “alleged signal” of mental instability. This seriatim analysis wholly fails to address the bigger picture. The cumulative effect of all of the evidence undeniably points to the necessity of ordering a mental health evaluation. DuRocher and Bertolotti’s trial attorneys all testified that, on the whole, Bertolotti’s case required the assistance of a psychiatrist. As the majority opinion itself demonstrates, ample signs of Bertolotti’s mental impairment existed.
The majority further rejects as too specific the Florida Supreme Court’s holding that “where there is evidence calling into question a defendant’s sanity, defense counsel is bound to seek the assistance of a mental health expert.” Rather than embrace what it terms a state rule,2 the ma*1532jority prefers to evaluate counsels’ behavior under a “reasonableness” standard. Maj. op. at p. 1510. Although I disagree with the majority that Ake v. Oklahoma, 470 U.S. 68, 105 S.Ct. 1087, 84 L.Ed.2d 53 (1985) and its progeny does not require that counsel seek the assistance of a mental health expert whenever the defendant’s sanity is likely to be a significant factor at trial, I find that under any measure of reasonableness, trial counsel performed inadequately in this case.
The majority recognizes that Ake requires the state to provide access to a psychiatrist if an indigent defendant “makes a preliminary showing that his sanity at the time of the offense is likely to be a significant factor at trial.” After stating this observation, however, the majority goes on to say that Ake requires the state to provide a mental health expert only when the defendant “exhibits compelling evidence of incompetency or insanity.” Thus, reasons the majority, while Ake does imply that counsel would be deficient if he or she did not conduct a reasonable investigation into the possibility of raising an insanity defense where evidence of the defendant’s potential insanity is compelling, Ake does not imply that counsel must actually seek the assistance of a mental health expert where such evidence is less than “compelling.” Maj. op. at p. 1511.
Such a distinction is not in the Ake decision. Ake does not require “compelling evidence” of incompetency or insanity before a State must provide a mental health examination. Rather, the defendant need only make “a preliminary showing” that his sanity will be a significant issue at trial in order to invoke his constitutional right to a psychiatric exam. Ake, 105 S.Ct. at 1092. The right to psychiatric assistance also exists at the sentencing phase where psychiatric aid could serve as rebuttal to the State’s presentation of aggravating circumstances. Ake, 105 S.Ct. at 1097. If the defendant has a constitutional right to this exam, it follows that counsel does not act reasonably in foregoing this constitutional right without any investigation or strategy in mind. Cf. Elledge v. Dugger, 823 F.2d 1439, 1444-45 (11th Cir.), opinion withdrawn in part, 833 F.2d 250 (11th Cir. 1987), cert. denied, — U.S. —, 108 S.Ct. 1487, 99 L.Ed.2d 715 (1988) (counsel performed inadequately when, although he believed petitioner was “crazy,” he failed to seek expert psychiatric witness and to interrogate petitioner’s relatives).
Moreover, in Blake, 758 F.2d at 531, we noted that a critical link exists between minimally effective assistance of counsel and the need for psychiatric aid. See also Magill, 824 F.2d at 889-90 (counsel performed inadequately when he failed to secure testimony of available psychiatrist who would have testified favorably and instead called medical witness who testified unfavorably as to statutory mitigating circumstances). Thus, although we do evaluate counsel’s actions to determine whether they are “reasonable,” no counsel could justify a decision to forego a mental evaluation as part of an overall investigation into the defendant’s mental health where, as here, sufficient indications of mental incompetency exist. Finally, the trial court in this case had already granted the request for a psychiatric evaluation. Thus, a “preliminary showing” had already been made. Under these circumstances, the proper inquiry is whether the failure to capitalize on the existing opportunity to have a mental evaluation was reasonable, not whether one should have been ordered in the first instance. With no strategy to justify the failure to obtain this potentially *1533crucial evidence, the answer must be a resounding “no.”
While the majority holds that counsels’ failure to secure a mental health expert was a reasonable decision in regard to the guilt phase of Bertolotti’s trial, it concedes that “evidence of mental impairment could still have been used during the penalty phase of the trial.” Maj. op. at p. 1515. As the majority correctly points out, even if counsel believed Bertolotti was sane at the time he committed the crime, enough evidence of mental impairment existed so as to alert counsel to the possibility of presenting evidence of this impairment in mitigation of punishment. See Stephens, 846 F.2d at 653 (for purposes of guilt phase, trial counsel entitled to rely on psychiatric report stating that petitioner was sane when he committed crime, but counsel had enough indications of petitioner’s mental instability to require that he investigate the possibility of presenting evidence of this instability in mitigation of punishment at the penalty phase of trial).
Nonetheless, the majority concludes that counsel was not deficient for failing to present any evidence of Bertolotti’s mental health in mitigation of punishment. The majority bases its conclusion on nothing more than the fact that counsel secured a mental health expert to evaluate Bertolotti on the morning of his sentencing hearing. Thus, the majority concludes, if anyone was unreasonable it was Bertolotti because he refused to see this expert at that time. This conclusion totally begs the question of whether counsels’ actions were reasonable; that is, whether it was reasonable for them to procrastinate until immediately before the sentencing proceeding before scheduling Bertolotti’s psychiatric evaluation. Under the circumstances of this case, Bertolotti’s refusal was quite understandable; the decision as to his fate was only hours away. In addition, attorney Kenny testified that it is quite common for an indigent defendant to initially refuse to undergo such an exam. Bertolotti’s initial refusal to be seen by a psychiatrist on the morning of his hearing does not exonerate counsel from their totally inadequate representation in neglecting to secure an exam until the last possible moment.
In sum, the majority wholly fails to acknowledge that counsel unintentionally neglected to have Bertolotti evaluated by a psychiatrist. Furthermore, it ignores the cases of this Circuit which emphasize the importance of psychiatric evidence in capital punishment cases. See Blake, 758 F.2d at 531; Magill, 824 F.2d at 889-90; Stephens, 846 F.2d at 653-55. The results of a mental evaluation in this case would have allowed counsel to fully consider the possibility of an insanity defense as well as provide counsel with possible mitigating evidence to be used during the penalty phase. Clearly the failure to obtain such an evaluation constituted deficient performance.
B. Prejudice.
Agreeing as I do with the Florida Supreme Court that counsel for Bertolotti was ineffective in failing to have their client examined by a psychiatrist, I must also analyze whether that failure prejudiced Bertolotti’s defense in the sentencing phase of his trial. The majority holds that it does not; I must respectfully disagree. At the Rule 3.850 hearing, Bertolotti presented the testimony of Dr. James R. Merikangas, who testified that he had been a medical doctor for 18 years and is Board certified in neurology and psychiatry. He is an Assistant Clinical Professor of Psychiatry at Yale University School of Medicine. He has evaluated approximately 185 violent criminals and was qualified as an expert witness by the trial court. In addition to concluding that Bertolotti was insane at the time of committing the crime, Merikan-gas also found that the murder was committed while Bertolotti was under the influence of extreme mental or emotional disturbance, that he acted under extreme duress, and that his capacity to appreciate the criminality of his conduct and to conform his conduct to the requirements of law were substantially impaired.
The absence of this evidence deprived Bertolotti of an individualized sentencing hearing. The failure to obtain a mental health expert deprived the jury of an opportunity to consider two statutory mitigating circumstances — whether at the time of the crime Bertolotti was under the influence of *1534extreme mental or emotional disturbance and whether Bertolotti’s capacity to appreciate the criminality of his conduct or to conform his conduct to the requirements of law was substantially impaired. Just as importantly, this testimony could have served as nonstatutory mitigating evidence. Cf. Middleton, 849 F.2d at 495 (psychiatric evidence has potential to change whole evidentiary picture).
This court has recognized that “psychiatric evidence has the potential to totally change the evidentiary picture by altering the causal relationship that can exist between mental illness and homicidal behavior. ‘Thus, psychiatric mitigating evidence not only can act in mitigation, it also could significantly weaken the aggravating factors.’ ” Middleton, 849 F.2d at 495 (citation omitted). In the present case, the jury recommended death by a vote of nine to three. In addition, the trial court found three aggravating circumstances: (1) that Bertolotti had previously been convicted of three violent felonies; (2) that the murder occurred during the commission of a robbery and; (3) that the murder was especially heinous, atrocious and cruel. The trial court found no mitigating circumstances. Merikangas’ testimony, however, had “the potential to alter the entire evi-dentiary picture.” For example, the jury could have found that Bertolotti’s mental problems accounted for the heinous nature of the crime. The evidence could also have given the jury a context in which to understand Bertolotti’s prior violent behavior. Finally, Merikangas’ testimony could have given the jury two statutory mitigating circumstances to consider against the aggravating circumstances. That this could have altered the result in this case is beyond argument. The Florida Supreme Court in Bertolotti’s case stated:
As recognized by the United States Supreme Court, where a defendant’s mental condition is in question, “without the assistance of a mental health expert ... the risk of an inaccurate resolution of sanity issues is extremely high.” 105 S.Ct. at 1096. The trial court’s conclusion that defense counsel “had no reason to doubt Bertolotti’s sanity in any respect” is not supported by the testimony and other evidence adduced at the 3.850 hearing. Considering only these factors which the public defender’s office was aware of prior to trial, it is apparent that defense counsel had reason to question Bertolotti’s sanity at the time of the offense.
534 So.2d 386, 388. Given the fact that the jury voted 9 to 3 for the death penalty in the absence of two statutory mitigating circumstances and the accompanying non-statutory mitigating circumstances, no one is smart enough to know how the jury would have voted with respect to the penalty in this case. The nine jurors were adversely influenced by the prosecutor’s improper argument in the penalty phase of the case. The Florida Supreme Court found in its first review of this case that the “prosecutor clearly overstepped the bounds of proper argument on at least three occasions.... These considerations are outside the scope of the jury’s deliberation and their injection violates the prosecutor’s duty to seek justice, not merely ‘win’ a death recommendation.” 476 So.2d at 132-33.
Despite the strong precedent of this court to the contrary, the majority holds that there was no prejudice in the sentencing phase of this case in part because it finds Dr. Merikangas’ report internally inconsistent and contradicted by the State’s own experts. The majority concludes that a jury would likely have found the State’s experts more credible. Although I strongly contest the majority’s characterization of Merikangas’ testimony, I am more disturbed by the fact that, in reaching its conclusion, the majority has impermissibly invaded the province of the jury.
First, the majority’s characterization of Merikangas’ testimony as inconsistent or weak is without foundation. After presenting testimony regarding documents concerning Bertolotti’s history, Dr. Meri-kangas testified concerning his interview with the petitioner. He testified that Ber-tolotti suffers from schizophrenia of the chronic undifferentiated type. Testifying that Bertolotti was a typical patient suffering from schizophrenia, he found that as a young man he never dated women because *1535he was too socially crippled, that he was on the fringe of society and that his functioning had deteriorated.
Two pieces of documentary evidence that were relied upon by Merikangas in reaching his conclusions are of note. First, Ber-tolotti was incarcerated in the State of Georgia in 1973 and was reviewed for parole on November 19, 1973. Parole Supervisor David A. Kasriel noted that “subject’s psychological report shows a likelihood of ‘crazy,’ irrational behavior. I believe a more thorough psychological evaluation is needed before I could make a parole recommendation.” R. 3.850 H. at D. Ex. J. The referenced psychological report, if ever prepared, is not in the record. Second, Bertolotti was in prison in the State of Florida in 1982 and Walter H. Cary, Jr., Clinical Psychologist at the Baker Correctional Institution, summarized his conclusions in a document dated March 5, 1982:
Testing and interview indicate that this subject has a sociopathic personality. He has made a great improvement over previous testing. Previously, there were indications of the possibility of disorganization under stress, cyclic bizarre and/or aggressive behavior and sexual dysfunction. All of these indications have now disappeared, and it is likely that this subject will do well in a work release setting. However, it should be noted that persons with profiles similar to the subject’s present one, have extremely high recidivism rates, usually for crimes of a property offense nature.
R. 3.850 H. at D. Ex. K. In addition to this information, Merikangas testified that Ber-tolotti’s mother was a schizophrenic and that the disease has a genetic component making it relevant in reaching a diagnosis.
Concluding that Bertolotti was insane at the time of the offense and did not know what he was doing and did not know it was wrong at the time of the offense, he stated:
I believe my opinion is that he is a schizophrenic who had a catastrophic reaction to stress, that people with this disorder are predisposed to break down under conditions of stress and to go berserk, as this man apparently did; and that this is borne out not only by his recounting of the crime and the several different versions which he used, but by the facts that are documented in the autopsy and the police report of a berserk rage, stabbing multiple times with two different knives, for instance; his actions after the crime of leaving bloodstains all around and leaving the weapon there and going home and hiding these clothes; his girl friend, who is not a trained psychologist, observing that there was something weird and strange about him; his blubbering and whining and decompensating while giving a voluntary confession to the police the first time, and then coming back with another different confession that tried to implicate his girl friend after he had time to consider it and calm down; and his past history all point to the same conclusion.
R. 3.850 H. at 431. Dr. Merikangas further testified that schizophrenics do not always act strange or appear unusual, but have periods of remission and exacerbation. Finally, Dr. Merikangas opined that in committing the crime, Bertolotti was under the influence of extreme emotional distress, that he acted under emotional duress and that his ability to conform his conduct to the requirements of law was substantially impaired.
This evidence demonstrates that Meri-kangas relied on Bertolotti’s medical and prison records, other documents in Berto-lotti’s file, and his personal interviews of both Bertolotti and Bertolotti’s family members. Contrary to the findings of the majority, I find that Merikangas’ observation of Bertolotti’s “inappropriate emotional behavior” was corroborated. The confession tapes depict Bertolotti in an extremely emotional state. At the rule 3.850 hearing, trial counsel stated that Bertolotti should have been examined by a psychiatrist “[gjiven the nature of the crime, given some of the stuff that Bertolotti did both before and after the crime, some of his behavior, I would think that [an exam] would be more than just one of those things to protect your behind.” (Emphasis added). Finally, Bertolotti’s girl friend observed that Bertolotti was acting strangely after the crime and believed that he needed “psychiatric help.” There is no basis for *1536concluding that Merikangas' testimony was so unreliable or inconsistent as to dismiss it as irrelevant information for the jury to consider.
Second, the majority improperly evaluates the credibility of Merikangas vis-a-vis that of the State’s experts. The State presented the testimony of Dr. James D. Upson who did not personally examine Ber-tolotti but reached a conclusion based on a review of documents in the file. A summary of his testimony is reflected by the following:
Q. Doctor, did you review the report and deposition of a Doctor James Merikan-gas?
A. I did.
Q. And as a result of reviewing his deposition, his report and findings, and based upon your review of the records of Mr. Bertolotti, the witness statements, and did you listen to the taped confession of Mr. Bertolotti?
A. I did.
Q. Were you able to render some professional opinion as to whether or not you agreed with his findings and the basis for his findings in this case of, one, schizophrenia and, two, temporary insanity?
A. I don’t disagree with the basis of his findings because essentially he used the same basis I did with the main exception, he interviewed Mr. Bertolot-ti. But as near as I could tell, he used a fairly standard psychiatric approach which relies quite heavily on interview data and history.
I do disagree with the conclusions of his report, his conclusion was that Mr. Bertolotti is schizophrenic; exhibiting periods of delusions.
Again, I make these statements having never dealt with the gentleman. But from the record, I would come up with a little different diagnosis.
My feeling is that his characteristics, as documented in the records, more clearly reflect anti-social behavior and depression. I don’t see or have the same interpretation of events that the psychiatrist did as I don’t see the signs of delusions in his history.
Again, they may well be there through an interview or through testing.
R. 3.850 H. at 523-24. At another point in his testimony, Dr. Upson stated that he would feel more comfortable with his conclusion if he had had an opportunity to examine Bertolotti.
The State’s chief witness at the rule 3.850 hearing was Dr. Robert Kirkland, a Board certified psychiatrist who had a number of years of experience and had testified in court an estimated 100 times.
Dr. Kirkland interviewed Bertolotti at the jail, took a family history, asked him about the facts surrounding the crime, and observed his conduct. After stating various findings, he then concluded: “So in summary, what I saw was a young man in a very difficult situation who shows no evidence of having any major psychotic mental disorder or mental disorder of brain damage origin. Either yesterday, nor at any significant time in the past.” R. 3.850 H. at 566. When asked whether he had reviewed an opinion by Dr. James Merikan-gas, he described it as “hogwash.” When asked whether he had an opinion concerning Bertolotti’s sanity and insanity, at the time of the crime, Dr. Kirkland stated: “I believe that at that time, Mr. Bertolotti was legally sane and responsible for his actions.” Id. at 570.
The other witness for the State with respect to Bertolotti’s mental condition was John L. Cassady, a Staff Psychologist at the Orange County Jail, who had a Masters Degree in Psychology. Bertolotti had been placed on a suicide watch and Cassady observed him from time to time but did not notice anything unusual or bizarre about him at the time Cassady was present. The trial court did not allow Cassady to testify on the subject on whether Bertolotti was sane or insane, but did allow him to testify that from his observations he saw nothing which would lead him to believe that Berto-lotti had ever suffered from schizophrenia.
From the evidence presented at the hearing, one cannot conclude that the State’s experts are objectively any more believable than Dr. Merikangas. All of the State’s *1537experts relied on the same sources of information as Dr. Merikangas although, unlike Merikangas, each only looked at a part of the evidence available. Although the State’s expert testimony is subject to the same “well-considered attack on several fronts” that the majority undertakes with respect to Dr. Merikangas’ testimony, such attacks are inappropriate on appellate review. Our function is not to weigh the respective witnesses’ testimony.
I also disagree with the majority’s conclusion that there was no prejudice because, even if the mitigating evidence had been presented, a jury would have found that the aggravating circumstances outweighed any mitigating evidence. This evaluation is the kind of second-guessing that can deprive a defendant of an individualized sentencing determination, see Knight v. Dugger, 863 F.2d 705 (11th Cir.1988), and should only be undertaken when there is absolutely no doubt as to what the outcome would be. Because the effect of psychiatric evidence in the sentencing phase is so uncertain, such an evaluation cannot be made by an appellate court in this case.
The question of prejudice is a highly factual one. Some cases from our Circuit are instructive. For example, in Armstrong v. Dugger, we held that:
The major requirement of the penalty phase of a trial is that the sentence be individualized by focusing on the particularized characteristics of the individual. See Eddings v. Oklahoma, 455 U.S. 104, 112, 102 S.Ct. 869, 875, 71 L.Ed.2d 1 (1982); Gregg v. Georgia, 428 U.S. 153, 199, 96 S.Ct. 2909, 2937, 49 L.Ed.2d 859 (1976). The evidence before the district court plainly established that Armstrong’s trial counsel failed to provide the jury with the information needed to properly focus on the particularized characteristics of this petitioner.
833 F.2d at 1433. In addition, we held in Blake, 758 F.2d at 534-35, that although damaging evidence “very well could have persuaded a jury to impose the death sentence in any event, Blake was nonetheless prejudiced in the absence of character evi-dence_ ‘Certainly [it] would have provided some counterweight to the evidence of bad character which was in fact received.’ ” (Citation omitted). More recently, we held that the failure to present character evidence at the sentencing phase of trial, after counsel undertook no investigation, prejudiced the defense because “the jury did not assess ‘the information needed to properly focus on the particularized characteristics of this petitioner.’ ” Harris, 874 F.2d at 763 (citation omitted).
In Stephens, 846 F.2d at 646, this Circuit found prejudice where trial counsel conducted no investigation into the possibility of introducing evidence of the defendant’s mental history and mental capacity in the sentencing phase of defendant’s trial. The only testimony that the jury heard as to the defendant’s occasional bizarre behavior was that which was presented as an afterthought by the defendant’s mother in answer to a question by the trial judge. In addition, defense counsel made no comment on this testimony in closing argument. This court held that “the resulting prejudice is clear.... In light of the way that [the facts of mental impairment] came in, ... arguing their importance as mitigating circumstances may well have been futile. As a result, however, although the jury heard some testimony concerning the [defendant’s mental condition, the jurors were left with no guidance concerning how they might take such facts into consideration in mitigation of punishment.” Id. at 655.
Next, in Magill, 824 F.2d 879, we found that errors made during the sentencing phase of the petitioner’s trial, combined with errors made during the guilt phase, resulted in ineffective assistance of counsel in violation of the sixth amendment. In Magill, counsel failed to present available mitigating evidence in the form of psychiatric testimony. The psychiatrist in question had treated Magill and would have testified that the petitioner exhibited signs of serious emotional problems at the age of thirteen and could have been expected to commit a crime of serious magnitude. Because this psychiatrist would have been much more helpful than the one who actually testified, and because ample evidence existed in mitigation of a death sentence, we *1538found that the errors committed at trial sufficiently prejudiced the petitioner so as to entitle him to a new sentencing proceeding.
In yet another case, Middleton, 849 F.2d 491, we found that trial counsel’s almost total lack of background investigation despite conversations with the petitioner that such evidence existed, constituted deficient performance. Turning to the question of prejudice, we found that prejudice occurred because of the considerable amount of factual and record evidence which could have been used in mitigation of punishment. Further, we recognized that psychiatric testimony was available and this evidence could well have met the standard for statutory mitigating circumstances.
The present case is analogous to these cases. Like the Middleton case, Dr. Merikangas could have testified as to the existence of three statutory mitigating circumstances: extreme emotional disturbance, diminished capacity and duress. In Ake v. Oklahoma, the Court emphasized the importance of live psychiatric testimony as a meaningful way of presentation of mitigating evidence.
Psychiatrists can translate a medical diagnosis into language that will assist the trier of fact, and therefore offer evidence in a form that has meaning for the task at hand. Through this process of investigation, interpretation, and testimony, psychiatrists ideally assist lay jurors, who generally have no training in psychiatric matters, to make a sensible and educated determination about the mental condition of the defendant at the time of the offense.
470 U.S. at 80-81, 105 S.Ct. at 1095. Additionally, as in Magill, Dr. Merikangas could have aided the defense in the preparation and presentation of nonstatutory mitigating circumstances. See id. at 83, 105 S.Ct. at 1096 (“assist in evaluation, preparation, and presentation of the defense”). Based on the facts of this case and the cases discussed above, I conclude that Bertolotti was indeed prejudiced at the sentencing phase of his trial.
Thompson v. Wainwright, 787 F.2d 1447 (11th Cir.1986), a case cited by the majority as support for its conclusion, is easily distinguished. In Thompson, the petitioner did not offer any psychiatric evidence that could have been presented at the sentencing phase. Thus, the court held that the omission of evidence merely consisting of the petitioner’s poor school records and medical reports suggesting a personality disorder was not prejudicial. In this case, psychiatric evidence was available. Because of the importance of psychiatric evidence, the jury, and not this court, should be the one to weigh this evidence in determining the appropriate sentence for Berto-lotti.
Daugherty v. Dugger, 839 F.2d 1426, 1431 (11th Cir.1988), is similarly distinguishable. In Daugherty, the petitioner argued that counsel’s failure to offer psychiatric evidence of his domination by his girl friend prejudiced him in the sentencing phase of his trial. Id. at 1431. Unlike Bertolotti’s attorneys, Daugherty’s attorney did present other evidence in mitigation similar to the evidence that would have been produced by the expert testimony. Id. at 1432. Second, the nature of the evidence Daugherty sought to introduce is markedly different than Bertolotti’s. Daugherty’s attorney consulted a psychiatrist and a psychologist and concluded that their testimony would not be adequate to support the domination theory. Id. at 1431. In addition, the attorney did not present the evidence because he feared that the weak evidence he did have available would be worse than no expert evidence at all. Id. He also feared that presentation of the expert evidence would prompt the state to present Daugherty’s previous testimony that directly contradicted his domination claim. Id. In Daugherty’s case the evidence that was alleged as prejudicial was at best cumulative and was probably counterproductive. Thus, rather than weighing competing evidence to determine if Daugherty was prejudiced, the court was faced with a case where the defendant could not show that any additional evidence existed at the time of trial that would provide any credible support for his domination claim. Bertolotti, on the other hand, has proven that he would have had credible favorable testimony that sup*1539ported a mitigating factor not presented to the jury because of counsel’s failure.
The majority also relies on Bundy v. Dugger, 850 F.2d 1402 (11th Cir.1988). Bundy, unlike Bertolotti, refused to allow any evidence of a mental disorder to be introduced. Id. at 1412. Bundy’s decision made any evidence of mental disorder irrelevant to the case. If the Defendant would not have allowed the evidence to be presented, counsel’s failure to present it could not have prejudiced Bundy. Thus, Bundy can be seen as a case where no usable evidence existed rather than one where the court weighed conflicting testimony. As the majority points out, Berto-lotti did refuse to allow an examination immediately before his sentencing hearing. Bundy’s attorney, however, investigated the defense, and Bundy’s refusal to allow the evidence to be presented was based upon that investigation. Id. Bundy’s informed decision distinguishes his refusal from Bertolotti’s.
In addition, the majority relies on Elledge v. Dugger, 823 F.2d at 1447-48. Although a superficial reading of Elledge lends some support to the majority’s position, I find the present case sufficiently distinguishable. First, unlike, the defendant in Elledge, Bertolotti has shown that the contested mitigating evidence was reasonably available to counsel. The record reveals that Bertolotti’s father believed that Bertolotti was strange and peculiar almost from birth. The defendant’s girl friend, Sharon Griest, had told Mr. Wolf that Bertolotti might have a “split personality.” The availability of this information along with Merikangas’ testimony of what a psychiatric evaluation could have yielded, demonstrates that a failure to pursue these avenues resulted in prejudice to the petitioner.
Next, unlike Elledge, the overlooked evidence was nothing but favorable to the petitioner. In Elledge, some of the uncovered witness testimony would have been favorable but other testimony from those same witnesses would have been unfavorable.
Most importantly, unlike the Elledge ease, the defense psychiatrist’s testimony was credible. Merikangas based his findings on his own interview and neurological evaluation of the petitioner. In addition, his testimony is corroborated by several earlier evaluations that are part of the record. See R. 3.850 H. at D. Exs. J & K. Furthermore, in Elledge the several state experts that served to rebut the defense’s expert all personally examined 'Elledge over the years. In this case, only one of the State’s experts actually examined Ber-tolotti personally. Thus, this is not a case where the expert testimony was clearly weighted to one side. Rather it really came down to one expert versus the other. Moreover, although the State put on its own experts to refute Merikangas’ evaluation, the only qualified expert, Kirkland, testified only as to Bertolotti’s sanity, not to his mental condition as it might relate to the penalty phase of the trial. Merikangas testified favorably to the defense in regard to both of these issues.
When one critically looks at our cases in this area, it is evident that we have distinguished between the cases where credible and useful psychiatric evidence was available and ones where it was not. When it has been available, we have refrained from invading the province of the jury by weighing evidence and determine issues of credibility. See, e.g., Stephens, 846 F.2d at 655 (jurors left with no guidance as to how they might consider facts of mental instability in mitigation); Armstrong, 833 F.2d at 1434 (availability of expert who would have testified that petitioner was mentally retarded and suffered from organic brain damage enough to meet prejudice requirement). As the majority admits, “because psychiatry and psychology are ‘arts, not sciences,’ reasonable professionals could differ in their diagnosis.” Maj. op. at p. 1518. The majority’s reasoning discards this distinction.
CONCLUSION
The underlying issue in this case is whether Bertolotti’s sentencing hearing met constitutional standards. Penry v. Lynaugh, — U.S. —, 109 S.Ct. 2934, 106 L.Ed.2d 256 (1989) is the latest Supreme *1540Court decision to discuss the evolving Fur-man principle which teaches that sentencing in death penalty cases is governed by the particularized circumstances of the crime and the defendant. The Court in Penry stated:
In order to ensure “reliability in the determination that death is the appropriate punishment in a specific case,” Woodson [v. North Carolina], 428 U.S. [280], at 305, 96 S.Ct. [2978], at 2991 [49 L.Ed.2d 944 (1976)], the jury must be able to consider and give effect to any mitigating evidence relevant to a defendant’s background, character, or the circumstances of the crime.... Our reasoning in Lockett and Eddings thus compels a remand for resentencing so that we do not “risk that the death penalty will be imposed in spite of factors which may call for a less severe penalty.” Lockett [v. State of Ohio ], 438 U.S. [586], at 605, 98 S.Ct. [2954], at 2965 [57 L.Ed.2d 973 (1978) ]; Eddings [v. Oklahoma], 455 U.S. [104], at 119, 102 S.Ct. [869], at 879 [71 L.Ed.2d 1 (1982)] (concurring opinion). “When the choice is between life and death, that risk is unacceptable and incompatible with the commands of the Eighth and Fourteenth Amendments.” Lockett, 438 U.S., at 605, 98 S.Ct., at 2965.
109 S.Ct. at 2951-52.
I recognize that Lockett and its progeny, including Penry, govern a line of cases where state action prevented jury consideration of mitigating evidence. In Bertolot-ti’s case it was the ineffectiveness of counsel that prevented such consideration. Nevertheless, in appraising whether prejudice occurred it is essential that we consider the consequence of an attorney’s ineffectiveness which denies a defendant an individualized sentencing. It is literally the difference between life and death. No person can determine with hindsight whether this nine to three jury decision would have been the same if the jury had heard the psychiatric evidence. One can with some reason hypothecate that the three jurors thought Bertolotti was mentally deranged from the very facts of the killing itself. The proper judicial decision in this case is to remand the case for a new sentencing hearing as was done in Lockett, Eddings, Skipper, Hitchcock, and Penry.
. Petitioner contends that this finding of the Florida Supreme Court deserves a "presumption of correctness” under 28 U.S.C. § 2254(d). The presumption does apply to both appellate and trial court findings. See Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981). *1530Claims of ineffective assistance of counsel, however, are generally thought to embody “mixed questions” of law and fact not subject to the § 2254(d) presumption. Strickland, 466 U.S. at 698, 104 S.Ct. at 2070. The Florida Supreme Court concluded that "defense counsel had reason to question defendant's sanity." Bertolotti, 534 So.2d at 388 (emphasis added). I think that it is a close question whether this determination is a question of law or of fact. The Florida Supreme Court, however, seemed to make this determination in the context of applying the legal standard it set for deciding whether counsels’ conduct was in violation of Strickland's performance prong. I, therefore, do not accord the finding the § 2254(d) presumption.
. The majority characterizes the Florida Supreme Court’s finding that counsel for Bertolotti was deficient as an opinion based upon a "state-law standard.” Maj. op. at p. 1510. The majority bases this conclusion, not upon the opinion of the Florida Supreme Court, but upon a provision of the Florida Constitution giving the court the power to discipline attorneys. Id. at p. 1510. The Florida Supreme Court’s opinion, however, is totally devoid of reference to state law. Specifically, the majority takes issue with the Florida Supreme Court’s requirement that counsel seek the assistance of a mental health expert whenever there is evidence that calls the defendant’s sanity into question. Maj. op. at p. 1510; see Bertolotti, 534 So.2d at 388. Although the Florida Supreme Court cites Bush v. Wainwright, 505 So.2d 409 (Fla.), cert. denied, 484 U.S. 873, 108 S.Ct. 209, 98 L.Ed.2d 160 (1987) for this proposition, an examination of Bush shows that the court in that opinion was apply*1532ing the Supreme Court’s opinion in Ake v. Oklahoma. See Bush, 505 So.2d at 410.
I point this error out not because it effects the outcome of this case. As the majority correctly notes, we are not bound by state determinations that particular conduct is or is not effective assistance of counsel. Maj. op. at p. 1510. I point out the majority’s characterizations of the Florida opinions because the characterizations ignore important federalism concerns. The Supreme Court rejected the majority's approach in Michigan v. Long, 463 U.S. 1032, 1040-41, 103 S.Ct. 3469, 3476-77, 77 L.Ed.2d 1201 (1983), when it held that unless the state court clearly expresses that its decision rests on independent and adequate state grounds, federal courts are to presume that "the state court decided the case the way it did because it believed federal law required it to do so.” The majority’s characterization ignores this teaching.