Full opinion text
OPINION REINHARDT, Circuit Judge: I. Introduction In 1984, a house in California was burglarized and a number of items were stolen. K.H. and M.H. resided there with M.H.’s young children, a live-in babysitter, L.R., and her daughter. Petitioner John Doe, who was living at the time in a vacant house adjacent to the property, was arrested in connection with the burglary, but then released. Soon after, while K.H. and M.H. were not at home, their house was burglarized again. L.R. was murdered, having been beaten, stabbed, and strangled. Her body was found supine on the bed in the master bedroom, with her hands bound behind her back. She was naked from the waist down, with her legs open, and a vibrator near her body. A number of items were stolen. After an investigation, Doe was arrested. He was charged with one count of murder and two counts of burglarizing the home. Special circumstances of felony-murder-burglary and felony-murder-rape were alleged; also alleged was a prior felony conviction for an armed robbery committed in the Southern state where Doe grew up. J.B., who had never before worked on a case in which the death penalty was at issue, was appointed to represent Doe. He hired an investigator, D.S., who interviewed potential witnesses in California and in Doe’s home state. Doe pleaded not guilty to the charges and denied the allegations. The jury returned verdicts finding Doe guilty of murder and both counts of burglary. The jury also rendered a finding of true on the felony-murder-burglary special-circumstance allegation, and a finding of not true on the felony-murder-rape special-circumstance allegation. At the penalty phase, the jury returned a sentence of death. The California Supreme Court denied Doe’s direct appeal, and the Supreme Court denied his petition for certiorari. The California Supreme Court also denied Doe’s habeas petition, twice. Doe filed a federal habeas petition, which was also denied. The district court affirmed the conviction, rejecting a number of guilt-phase challenges. As for Doe’s claim that he had received ineffective assistance of counsel at the penalty phase of his trial, the court found that counsel for Doe had performed deficiently in failing to investigate and present various categories of mitigating evidence. However, the district court concluded that Doe could not establish that he had been prejudiced as a result, as required under Strickland v. Washington, 466 U.S. 668, 695, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). We agree with the district court that Doe is not entitled to reversal of his conviction on the basis of the claims presented in the petition before us. With respect to the penalty-phase claim, we agree that defense counsel was ineffective but disagree with the conclusion that Doe was not prejudiced. Accordingly, we affirm Doe’s conviction but reverse as to his sentence, and instruct the district court to grant the writ. II. Standard of Review This case is unusual in that Doe filed his federal habeas petition in 1995, prior to the passage of the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”). Accordingly, the parties agree that his petition is governed by pre-AEDPA standards of review. See Comer v. Schriro, 480 F.3d 960, 980 (9th Cir.2007). “Under these standards state court judgments of conviction and sentence carry a presumption of finality and legality and may be set aside only when a state prisoner carries his burden of proving that his detention violates the fundamental liberties of the person, safeguarded against state action by the Federal Constitution.” Id. (citations and internal quotation marks omitted). “A state court’s conclusion that a constitutional error was harmless is reviewed de novo.” Daniels v. Woodford, 428 F.3d 1181, 1196 (9th Cir.2005). In this analysis, the additional deference required by AEDPA does not apply. This court reviews de novo the district court’s denial of habeas relief. See Alcala v. Woodford, 334 F.3d 862, 868 (9th Cir. 2003). Underlying factual determinations made by the district court are reviewed for clear error. See Hovey v. Ayers, 458 F.3d 892, 900 (9th Cir.2006). Determinations by the district court of legal questions or mixed questions of law and fact are reviewed de novo. Frierson v. Woodford, 463 F.3d 982, 988 (9th Cir.2006). III. Guilt-Phase Claims In the petition before us, Doe raises a number of challenges to his conviction, all of which were rejected by the district court. We discuss these claims only briefly, as we agree with the result reached by the district court. A. Rule 60(b) First, Doe asserts that the district court abused its discretion in denying his motion to vacate the judgment in which it denied his habeas petition under Fed.R.Civ.P. 60(b). Doe requested relief under Rule 60(b) based on newly discovered physical evidence that was in the possession of the state. He alleges that the state withheld from his prior habeas counsel DNA and fingerprint evidence from the crime scene and from a related murder that it tested post-trial and was not a match to Doe. He also alleges that his prior habeas counsel was negligent in failing to pursue claims based on this evidence once she learned of it. Doe’s Rule 60(b) claims have a complicated procedural history: In March 2005, while the present petition was still pending before the district court, Doe sent a letter to the court stating that he no longer wanted his appointed attorneys to continue to represent him, in part because they refused to investigate his claims of actual innocence. Two weeks after he reiterated that request in June, the district court denied Doe’s request, and simultaneously denied his habeas petition. Doe appealed the denial of his motion for substitution of counsel, and we appointed new (present) habeas counsel, who filed his Rule 60(b) motion; the district court denied it. After consolidating Doe’s appeals, we held that the district court had abused its discretion in denying Doe’s request for substitution of counsel. We vacated the district court’s denial of this request, together with its denial of Doe’s petition for writ of habeas corpus, and remanded for further proceedings in which Doe’s newly-appointed counsel would have the opportunity to make additional submissions to the district court. The Supreme Court granted certiorari and reversed. It concluded that we had erred in holding that the district court abused its discretion in rejecting Doe’s request for new counsel. In so doing, it noted that the evidence at issue might have established a Brady claim, a claim of ineffective assistance of counsel for failure to adequately investigate the murder, or a claim of innocence, especially given that no physical evidence tied Doe to the crime and' that he was convicted based in part on recanted testimony. The Court went on to say, however, that all of those claims would have been new, and that as the district court subsequently found in ruling on the Rule 60(b) motion, the physical evidence was not related to the claims previously presented in Doe’s habeas petition. Because these claims were new claims for relief on the merits, and did not attack a defect in the integrity of the proceedings, Doe was required to raise them not in a Rule 60(b) motion, but in a successive habeas petition. Gonzalez v. Crosby, 545 U.S. 524, 529-82, 125 S.Ct. 2641, 162 L.Ed.2d 480 (2005). We therefore do not consider them here, but may do so in the future if Doe is subsequently granted the right to file a second or successive petition. B.Ineffective Assistance of Counsel Second, Doe alleges that his trial counsel, J.B., provided ineffective assistance of counsel during the guilt phase of his trial. J.B.’s performance at the guilt phase of Doe’s trial was certainly subpar. He failed to interview two young children who were the only eyewitnesses to the murder and who, in initial police reports, identified the killer as white (Doe is black). Although the failure to even interview the only eyewitnesses to the crime was unquestionably deficient performance, J.B. did offer a couple of reasonable justifications for his decision not to put them on as witnesses: the children appeared unreliable, and the prosecutor agreed not to tell the jury that children were present at the time of the murder if J.B. did not call them to testify. J.B. also failed to follow up on a potential alibi witness, C.L., with whom Doe claimed that he had gone drinking the night of the murder. During an interview with D.S., C.L. said that it was more likely than not that Doe was with him at a local bar on the night of the murder, given that the murder occurred on a Thursday and C.L. and Doe always went out drinking on Thursday nights. Because C.L. had no specific recollection of that particular evening, he told D.S. that he would attempt to find more definite support for the alibi. D.S. provided J.B. with a copy of a report summarizing his interview with C.L.; however, J.B. never attempted to contact C.L. again until, just before trial, when he tried to subpoena him as a trial witness. Then, when service was initially unsuccessful, J.B. made no further attempt to track him down. J.B. also acted in an objectively unreasonable way when he failed to call a blood spatter • expert who stated in his report that had Doe committed the murder, he would have been spattered with blood. J.B. never asserted a strategic reason for not calling the blood spatter expert, and the arguments raised by the state to undermine the probative value of this evidence (suggesting that Doe would have had time to wash the blood off, and that the witness who spent time with him later that evening did not see him in bright light) provide no reason not to present this testimony. Additionally, Doe argues that J.B. failed to investigate and challenge the reliability of one of the state’s witnesses, P.F., a girlfriend-turned-informant who testified that she bumped into Doe the night of the murder, that he left her alone during the time the crime was committed, and that he returned with a bit of blood on his hand and carrying distinctive items stolen from the home in which L.R. was killed. P.F. testified that he told her that he had “just finished beating up a woman.” Later, she taped a conversation with Doe, during which he made inculpatory statements. There was good reason to doubt the reliability of P.F.’s testimony. Two women who knew her told D.S. that she had a reputation for lying. However, neither D.S. nor J.B. interviewed B.P., one of the two people P.F. said she had been walking with when she encountered Doe that evening. When contacted later by habeas counsel, B.P. contradicted P.F.’s story, stating convincingly that she knew she had not been out with P.F. that night. P.F. had been in a bicycling accident shortly prior to the date of the crime, and a number of people stated in declarations that she had suffered from significant memory loss for months. P.F. essentially admitted in a declaration that, because she was still recovering from the accident, she could not have been with Doe on the night of the murder. It also appears that she was suffering cognitive deficits resulting from the combination of a medication and alcohol. Doe argues that in addition to impeaching P.F. based on her reputation for dishonesty and cognitive deficits, J.B. should have asked her about the extent to which the police appear to have helped her fill holes in her memory. However, J.B. did impeach P.F. to a significant degree. He elicited testimony about the seriousness of her head injury and the fact that she was taking medication and drinking alcohol on the night of the murder. He also elicited testimony that she had previously made false statements. He demonstrated that the moon was not full, as she had stated, the night of the murder, and that items she claimed to have seen that night in the vacant house had been removed previously. Finally, he prompted her to admit that she had not initially remembered the date of her interaction with Doe, and that the police had supplied her with it. We agree with the district court that while J.B. could have done a much better job of impeaching P.F., his efforts in this respect were not constitutionally inadequate. The additional impeachment evidence would have been largely cumulative, albeit stronger, but the failures regarding impeachment of P.F. are of comparatively little consequence, as the most important portion of her testimony was the introduction of her recorded conversation with Doe that served to corroborate the circumstantial evidence of his guilt. Lastly, Doe asserts that J.B. should have introduced evidence that K.H. was dealing drugs out of his home, that he had argued with L.R. shortly before her death and had previously assaulted someone, that neighbors reported domestic problems, and that L.R. had expressed to M.H. her fear that her wild life would end before her next birthday. The state is correct that evidence suggesting KH.’s culpability would have been excluded under People v. Hall, 41 Cal.3d 826, 833, 226 Cal.Rptr. 112, 718 P.2d 99 (1986), because for third party evidence to come in, it must demonstrate more than “mere motive or opportunity.” As for the evidence going more generally to the dangerous circumstances in which the victim lived, we do not believe it would have created significant doubt in the minds of the jurors. J.B. certainly did not provide high-quality representation to Doe at the guilt phase of his trial. However, he had a strategic justification for not calling the child witnesses. While he offered no such justification for his failure to follow up with the alibi witness, call the blood spatter expert, or demonstrate the dangerous environment in which the victim lived, it appears that none of this evidence would have been particularly persuasive. Failing to impeach P.F., the prosecution’s most important witness, would have been a very serious error, but J.B. did offer substantial impeachment evidence. Hence, we conclude that Doe has not shown prejudice. Decisively, the prosecution’s strongest evidence — which is not addressed by any of the claims Doe raises here — was the taped conversation between him and his girlfriend, P.F., during which, as the state argues, Doe made inculpatory statements. While Doe at one point denied involvement and never explicitly confessed, he made a number of very damning statements in regard to the murder. He warned: “They can’t prove a motherfuckin’ thing, not unless you open your motherfuckin’ mouth.” He added: “Baby what you fail to realize, how the motherfuckers they gonna prove I was there? ... There ain’t no motherfuckin’ fingerprints, ain’t no fuckin’ where in there, and ain’t no fuckin’ body seen me go in there and leave out of there.” In response to a request to tell her “what the fuck happened over there,” he said: “Why should I, so you can go back and tell [the police?]” When she stated that she had seen blood on him that night, he replied, “Ain’t on me no more.” Because of the strength of this evidence, we conclude that even if J.B. had performed adequately, there is not a reasonable probability that the jury would have acquitted Doe of murder. C. Batson Third, Doe claims that the prosecutor at his trial used peremptory strikes in a racially discriminatory manner, and that J.B. was ineffective for his failure to raise an objection. Four black veniremembers remained after excuses for hardship and death qualification; two were struck by the prosecutor, a third was removed for cause, and the remaining one was empaneled. Doe contends that J.B. was ineffective for failing to challenge these strikes under Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. 1712, 90 L.Ed.2d 69 (1986). In fact, J.B. did raise such a challenge, demanding reasons before the black jurors were struck, but the trial court ruled— correctly — that it was premature. For reasons passing understanding, J.B. never renewed his request after the black jurors were removed from the venire. This failure made it necessary for Doe to raise the issue of discriminatory jury selection through an ineffective assistance of counsel claim. Doe is correct that this failure constituted deficient performance. Additionally, though, Doe has the burden to demonstrate prejudice by showing that there is a reasonable probability that the claim J.B. failed to raise at trial would have prevailed, either at trial or on appeal. Strickland, 466 U.S. at 694, 104 S.Ct. 2052. He cannot do so. In order to prevail on a Batson claim, Doe would have needed to make a prima facie showing that the prosecutor exercised his peremptory strikes on the basis of race. To show that he could have done so, he relies on the statistically disparate use of strikes, and on the fact that the prosecutor asked black — but not white— veniremembers whether their race might influence their judgment. While the prosecutor’s disparate use of strikes and selective questioning is troubling, in a recent and similar case, Carrera v. Ayers, 699 F.3d 1104, 1110-11 (9th Cir.2012) (en banc), cert. denied, — U.S. -, 133 S.Ct. 2039, 185 L.Ed.2d 899 (2013), we concluded that under the standard that would have applied at Doe’s trial and on direct appeal, such a statistical disparity combined with questions about racial bias posed only to veniremembers of a particular racial or ethnic group was insufficient to show a “strong likelihood” that the strikes were made “because of [the veniremembers’] group association,” and therefore insufficient to demonstrate prejudice. We are bound by that precedent, so we deny relief with respect to the claim that J.B. was ineffective for failing to properly make a Batson challenge. D. Brady Fourth, Doe argues that the prosecutor improperly withheld impeachment evidence — namely, that the police working on his case had interceded on behalf of M.H. in a welfare fraud and perjury case. The extent of the intercession, if any, remains unclear; there is no evidence in the record of any deal, except for a notation in M.H.’s file by an unidentified person that she was “very important to [a] case.” Whether or not this constituted a violation of Brady v. Maryland, 373 U.S. 83, 83 S.Ct. 1194, 10 L.Ed.2d 215 (1963), however, Doe cannot establish prejudice. See Strickler v. Greene, 527 U.S. 263, 281-82, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999). M.H.’s testimony simply described items taken from the house during the murder-burglary, which the prosecution compared to the (nearly identical) list of items P.F. reported seeing in Doe’s possession. Especially given that the jury already knew that M.H. had been charged with perjury, it is unlikely to have further discounted her testimony upon learning that she received some indefinite benefit for her cooperation. What’s more, Doe’s recorded statements, discussed previously, were strong enough evidence to support a conviction even if the jury had some doubt about M.H.’s credibility. E. Evidence of Prior Crimes Fifth, Doe asserts that the jury received extraneous evidence of prior crimes he committed, and that this prejudiced him. Specifically, he complains that during an hour-long mid-trial examination of exhibits, but not during its eventual deliberations, the jury had access to unredacted transcripts of Doe’s recorded conversation with P.F. containing references to a prior burglary and a prior assault. However, these references were extremely brief and buried in a transcript containing much more compelling evidence of Doe’s guilt. Furthermore, the jurors were questioned about their review of the transcripts after the conclusion of the penalty phase, and none remembered reading anything about assaults or burglaries in the transcript. We agree with the district court that this evidence, even if the jury did see it, was harmless. Doe also challenges the trial judge’s decision to allow the prosecutor to impeach a character witness, D.P., Doe’s girlfriend in California, by asking her whether she had heard that he had been accused of rape in his home state and whether this affected her opinion of him. (It did not.) Impeachment of character witnesses with questions about prior bad acts of the defendant, even if unproven, is common practice. See Fed.R.Evid. 405(a). Thus, this claim fails. F. Cumulative Prejudice Sixth, and finally, Doe claims that these guilt-phase errors were cumulatively prejudicial.. Because Doe’s guilt-phase claims do not call into question the veracity or admissibility of the most damning evidence of his guilt — his own recorded, inculpatory statements — we hold that, on the record before us, he cannot demonstrate prejudice with respect to his conviction. IV. Penalty-Phase Ineffective Assistance of Counsel More important, for purposes of this opinion, Doe contends that his counsel was constitutionally ineffective for failing to investigate, and present at the penalty phase of his trial, certain mitigating evidence. That evidence relates to sexual abuse he suffered while previously incarcerated in a notorious prison in the South, as well as to mental illness, neglect and abuse he suffered during his childhood, and substance abuse. To prevail on this claim, Doe must show both that counsel was deficient and that he was prejudiced as a result. Strickland, 466 U.S. at 687-88, 104 S.Ct. 2052. Deficient performance requires a showing that “counsel’s representation fell below an objective standard of reasonableness.” Id. at 688, 104 S.Ct. 2052. Defense counsel is “strongly presumed to have rendered adequate assistance and made all significant decisions in the exercise of reasonable professional judgment.” Id. at 690, 104 S.Ct. 2052. To rebut this presumption, Doe must show that J.B. did not act “reasonably] considering all the circumstances.” Id. at 688, 104 S.Ct. 2052. “No particular set of detailed rules for counsel’s conduct can satisfactorily take account of the variety of circumstances faced by defense counsel or the range of legitimate decisions regarding how best to represent a criminal defendant.” Id. at 688-89, 104 S.Ct. 2052. However, “Restatements of professional standards ... can be useful as ‘guides’ to what reasonableness entails ... to the extent they describe the professional norms prevailing when the representation took place.” Bobby v. Van Hook, 558 U.S. 4, 7, 130 S.Ct. 13, 175 L.Ed.2d 255 (2009). At the time of Doe’s trial in 1987, the prevailing professional norms, as outlined by the ABA Standards, required that a lawyer “conduct a prompt investigation of the circumstances of the case and [ ] explore all avenues leading to facts relevant to the merits of the case and the penalty in the event of conviction,” and “called for [trial] counsel to cover several broad categories of mitigating evidence.” Id. at 7, 11, 130 S.Ct. 13 (citation omitted). The commentary to the standards made clear that “information concerning the defendant’s background, education, employment record, mental and emotional stability, family relationships, and the like, will be relevant....” Id. at 7-8,130 S.Ct. 13. “[D]eath is different!)]” Ring v. Arizona, 536 U.S. 584, 587, 122 S.Ct. 2428, 153 L.Ed.2d 556 (2002). So too are the lengths to which defense counsel must go in investigating a capital case. “The imperative to cast a wide net for all relevant mitigating evidence is heightened at a capital sentencing hearing because the Constitution prohibits imposition of the death penalty without adequate consideration of factors which might evoke mercy.” Frierson, 463 F.3d at 989 (citation and internal quotation marks omitted). “Although counsel’s duty to seek out evidence of mitigation is not limitless, the Supreme Court has recognized that the failure to pursue avenues of readily available information— such as school records, juvenile court and probation reports, and hospital records— may constitute deficient performance.” Id. (citing Rompilla v. Beard, 545 U.S. 374, 381-83, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005)). Although defense counsel may choose to ignore manifestly unfruitful lines of inquiry, a lawyer has not fulfilled his duties to his client if he ceases investigating because his client has not been forthcoming about his background or because counsel has acquired some relevant information. Rather, if what counsel knows or should know suggests that further investigation might yield more mitigating evidence, counsel must conduct that investigation. See Douglas v. Woodford, 316 F.3d 1079, 1088-89 (9th Cir.2003). “The presence of certain elements in a capital defendant’s background, such as a family history of alcoholism, abuse, and emotional problems, triggers a duty to conduct further inquiry before choosing to cease investigating.” Earp v. Omoski, 431 F.3d 1158, 1175-76 (9th Cir.2005). A. Deficient Performance We agree with the district court that counsel was ineffective in failing to investigate and present mitigating evidence. 1. Review of Investigation As a preliminary matter, J.B. was deficient in his acknowledged failure to consider even the evidence D.S.’s limited investigation had turned up: he did not “listen[ ] to any tapes of [D.S.]’s interviews, nor did [he] read transcripts of any taped interviews.” J.B. did read a few summary investigation reports produced by D.S., but these included virtually no material from the interviews conducted in Doe’s home state. Even if D.S. had conducted an adequate investigation — which he did not— J.B. would not have learned what D.S. had discovered. Because J.B. failed to review and follow up on the information presented to him by his investigator, he missed clear indications, for example, that his client was repeatedly raped in prison. An ex-girlfriend, I.R., told D.S. that Doe was “fresh meat” upon his arrival in prison at the age of 17, and that people in the community were well aware that fellow prisoners “got some” from him. J.B. also knew that Doe’s mother, C.G., had sought psychiatric help for her son after his release from prison. J.B.’s file contained a copy of a lengthy article from the inmate newspaper at the notorious Souther^ prison where Doe was incarcerated, which described in graphic detail the frequency and effects of prolonged sexual abuse. 2. Prison Records J.B. had a bit of Doe’s prison file — nine pages of criminal record, discharge papers, a rap sheet, and fingerprints. D.S. told J.B. that to get the complete file, which would shed additional light on Doe’s time in prison, J.B., as counsel, would need to request it. J.B., however, did not follow up on this most straightforward of leads, handed to him by his investigator: he did not request the records, and “do[es] not remember making any efforts to learn about [Doe]’s experience [in prison].” In addition, neither J.B. nor D.S. asked Doe whether he suffered abuse while incarcerated. Although Doe told them about being in prison, J.B. was “unaware of any allegation that [Doe] had been [redacted in original] abused [in prison],” and “did not specifically inquire” about any abuse. “I believe I should have known it,” J.B. said. J.B.’s failure to send off for Doe’s prison records — easy to obtain and very valuable — constituted deficient performance. See Correll v. Ryan, 539 F.3d 938, 945 (9th Cir.2008) (finding deficient performance based in part on defense counsel’s failure to obtain correctional records he knew existed). 3. Interviews with Doe J.B. had only one interview with Doe himself, during which he barely asked any questions about his upbringing. J.B. admitted that his interview with Doe was perfunctory: “I do not recall whether I discussed with [Doe] the abuse from his Uncle [J.C.], but if I did, it was on a superficial level.” He also acknowledged that this failure affected his penalty-phase presentation, because “the testimony elicited at trial reflected the extent of the abuse of which I was aware.” D.S. also spent a few sessions interviewing Doe. While Doe did speak positively about some aspects of his childhood, and said that “he would call everyone [in his family] basically sane,” Doe did mention— without disclosing its full extent — the physical abuse he suffered at the hands of his uncle. D.S. himself expressed the suspicion that Doe “avoid[ed]” or “ignor[ed]” his problems. He expressly informed J.B. that his interview of Doe was merely a preliminary inquiry into Doe’s personal and family history, noting that this “information was provided by [Doe] himself and is [in] no way meant to be a complete list of information available.... ” As a psychologist later retained by habeas counsel explained, based on professional experience, there are often reasons why a person who has been chronically abused and neglected might well decline to disclose the details of difficult and embarrassing personal history. That Doe did not volunteer more about the trauma he experienced during his childhood and in prison did not absolve J.B. of the need to conduct an adequate mitigation investigation, especially since Doe did identify (and D.S. recognized), if not in elaborate detail, a number of avenues for further investigation that would have proved fruitful. The district court was correct in concluding that the deficiencies it found in J.B.’s penalty-phase performance were not excused by the alleged failure of Doe (and his family) to be completely forthcoming. “[Doe] never put any limits on [the] penalty phase investigation,” J.B. acknowledged. Nor did he make false statements to J.B. or D.S. or obstruct their investigation. Doe simply did what most capital defendants — and most people — do, and did not volunteer deeply painful, shameful information when not pressed for details. Trial counsel has an affirmative duty not to simply accept the facts as they might be presented at first blush, but rather to “unearth[ ] for consideration” at the sentencing phase “all relevant mitigation information.” Wallace v. Stewart, 184 F.3d 1112, 1117 (9th Cir.1999) (quoting Caro v. Calderon, 165 F.3d 1223, 1227 (9th Cir.1999)); see also Daniels, 428 F.3d at 1209 (holding that counsel’s failure to investigate was not excused even by his client’s refusal to communicate). Anderson v. Calderon, 232 F.3d 1053, 1094 (9th Cir.2000), cited by the state, is not to the contrary. There, the defendant “did not disclose information relating to” the issues that later became central to habeas counsel’s mitigation presentation: evidence of the physical and emotional abuse the defendant suffered during childhood. Id. Thus, counsel had no clue as to the existence of these occurrences. In contrast, Doe and his family members and friends did disclose some evidence relating to childhood abuse and neglect, mental health problems, substance abuse, and even victimization in prison — all of the issues later drawn out by habeas counsel. Although no one delivered to J.B. a fully developed mitigation presentation, that does not excuse his failure to pursue the leads he did receive. Nor does Babbitt v. Calderon, 151 F.3d 1170, 1174 (9th Cir. 1998), help the state. There, we noted that “[o]ther courts have held that ‘counsel is not deficient for failing to find mitigating evidence if, after a reasonable investigation, nothing has put the counsel on notice of the existence of that evidence.’ ” Id. (quoting Matthews v. Evatt, 105 F.3d 907, 920 (4th Cir.1997)). By contrast, J.B. failed to conduct a reasonable investigation despite being on notice. [A]lthough counsel is not required “to investigate every conceivable line of mitigating evidence no matter how unlikely the effort would be to assist the defendant at senteneingt,]” they are in no position to decide, as a tactical matter, not to present mitigating evidence or not to investigate further just because they have some information about their client’s background. Moreover, ... the presence of certain elements in a capital defendant’s background, such as a family history of alcoholism, abuse, and emotional problems, triggers a duty to conduct further inquiry before choosing to cease investigating. Earp, 431 F.3d at 1175 (citing Wiggins v. Smith, 539 U.S. 510, 525, 527, 533, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003)). 4. Interviews with Others The interviews conducted by D.S. (and the few conversations J.B. himself had) with relatives and friends of Doe’s were unquestionably deficient. J.B. spoke at any length with only two people: Doe’s mother and his aunt, L.P. Had J.B. asked them more than admittedly cursory questions, he would have learned what the experts retained by habeas counsel did. Had he gone further than speaking “with them generally about the penalty phase,” he would have discovered powerful mitigating evidence. Instead, J.B. abandoned further investigation after “having acquired only rudimentary knowledge of [Doe’s] history from a narrow set of sourees[,]” and despite what he “actually discovered” in the course of his limited investigation. Wiggins v. Smith, 539 U.S. 510, 524-25, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003); see also Boyde v. Brown, 404 F.3d 1159, 1177 (9th Cir.2005); Douglas, 316 F.3d at 1082, 1088. J.B. did travel to Doe’s home state, but when asked if he visited Doe’s house, J.B. said: “I remember going to the neighborhood and going to the bar but not — I drove by the [family-owned] bar.... But I wasn’t comfortable being me just getting out and walking in there by myself so I decided not to go in.” He spoke on the phone with Doe’s mother (and possibly others) while he was there, but did not conduct any in-person interviews. At no point during his representation of Doe, J.B. reported, did he speak with anyone other than those he called to testify at the penalty phase. When D.S. went to Doe’s home state, he, at least, got out of his car. However, he repeatedly failed to ask obvious questions, including follow-up questions when interviewees revealed potentially significant information. He did not ask relatives and friends who had known Doe during his childhood about his upbringing or behavioral signs of mental illness. When the mother of one of Doe’s girlfriends (both mother and daughter were named V.M.) said she felt he was “like a second son,” and that he called to speak with her after he was released from prison about putting his sentence behind him, D.S. did not follow up to ask whether Doe had ever mentioned that he suffered abuse while incarcerated. When another interviewee, M.W., who had known Doe as he was growing up, described him as a “loner,” D.S. did not ask any follow-up questions about his mental health. Although Doe’s stepfather, B.G., had also served time in the same prison, D.S. did not ask him about abuse Doe might have suffered in prison. When a friend, J.A., reported that Doe told him over the phone that the prison was a “bad place,” D.S. did not ask whether Doe had said anything further. Other interviewees also mentioned that Doe was different after his release, or that they had spoken to him while he was incarcerated, but D.S. never asked them if they knew about any trauma Doe might have suffered in prison. As J.B. acknowledged, the important questions about Doe’s life were simply never asked. D.S. appears to have met only once with each interviewee, and many of the meetings were arranged by Doe’s mother at her home. Obviously, interviewees are less likely to be forthcoming about sensitive topics in the presence of family members and friends. See Correll v. Ryan, 539 F.3d 938, 945 (9th Cir.2008) (noting that “counsel testified that he met only once with Correll’s father, sister, and brother, ‘around the kitchen table at the same time,’ ” in concluding that the interviews he conducted were “worthless”). In the taped interviews, D.S. did not ask whether interviewees knew of any others who might have more or better information about Doe’s family history. Although D.S. said that he would normally conduct interviews in a way that would lead to identification of potential witnesses, he testified only that it was “possible” this was done in this case. Other witnesses, such as those whom habeas counsel was able to find, were “easily within [counsel’s] reach,” and would have been discovered by trial counsel, “[h]ad [he] only looked.” Wallace, 184 F.3d at 1116. Some of these witnesses were able to speak to Doe’s experiences in prison and their psychological effects on him. 5. Psychological Experts J.B. did retain a psychologist, Dr. M.R., to interview Doe. She had the professional expertise necessary to discover and present the compelling mitigating testimony regarding Doe’s mental health that went unheard at trial, but she was limited by the terms of her engagement. J.B. hired her, she reported, only to determine whether any mental state defenses based on “obvious signs of mental impairment” could be mounted at the guilt phase of the trial. Dr. M.R. reported that she “was not asked to do more,” and that she “was not asked to provide expert assistance in preparing or presenting a mitigation case at the penalty trial.” In addition to having a limited scope, the investigation that Dr. M.R. conducted was abbreviated. She was paid only for twelve and a half hours of her time and met with Doe only once, for an hour-long session largely devoted to filling out a questionnaire and discussing his criminal record. In her report to J.B., Dr. M.R. described her work as a “relatively brief evaluation,” and described her conclusions as “initial clinical impressions.” It appears that J.B. never spoke with Dr. M.R. after he initially hired her. The only background materials he provided her were police reports. When she left a message saying that she had “no documents on background,” he did not bother to return her call. This left J.B. effectively without the assistance of any expert at all at the penalty phase. J.B.’s failure to retain a psychological expert for the penalty phase was objectively unreasonable, given that he had sufficient notice of Doe’s mental health problems. He knew — or would have known, if he had reviewed the interviews that D.S. conducted — that Doe’s mother brought him to a hospital'for psychiatric help. D.S. stated in his assessment of Doe, which J.B. does appear to have read, that “he will not acknowledge that he has a psychiatric problem.” Doe’s mental illness was recognized by J.B.’s investigator, who was not a mental health expert. However, even this recognition was not enough to motivate J.B. to order a mitigation-related psychiatric examination of his client. The state’s assertion that Dr. M.R. addressed the issue of mitigating evidence in her report is incorrect. To the extent she commented on evidence relevant to the penalty-phase presentation, it was entirely in passing. “If in the course of performing [her limited-scope guilt-phase] evaluation [she had seen] issues that [she] thought would be useful for a penalty phase presentation, [she] would have flagged the issues for [J.B.].” Dr. M.R. was definitely not, however, “providing expert assistance in analyzing and developing a full-blown mitigation case.” Indeed, how could she have been? She “received no life history information regarding [Doe] from the defense.” Instead, she simply explained that “[she] had not seen any evidence that [she] believed to be mitigating during the, course of evaluating whether [Doe] had a guilt phase mental defense.” Hiring an expert to evaluate possible guilt-phase mental-state defenses does not discharge defense counsel’s duty to prepare for the penalty phase. Hendricks v. Calderon, 70 F.3d 1032 (9th Cir.1995), is directly on point. In Frierson, we explained: Because the evidence presented at each phase of a trial serves a markedly different purpose, we analyze- the reasonableness of counsel’s efforts to prepare for trial and sentencing differently. As we explained in Wallace v. Stewart: “Hendricks alludes to why the lawyer’s burden might differ at the guilt phase from that at the penalty phase: Mental state is relevant at the guilt phase for issues such as competence to stand trial and legal insanity — technical questions where a defendant must show a specific and very substantial level of mental impairment. Most defendants don’t have problems this severe, and counsel can’t be expected to know that further investigation is necessary to develop these issues. By contrast, all potentially mitigating evidence is relevant at the sentencing phase of a death case, so a troubled childhood and mental problems may help even if they don’t rise to a specific, technically-defined level.” Thus in Hendricks, we held that it was reasonable for counsel to rely on his experts’ findings that .no diminished capacity defense was available at the guilt phase, and to terminate his perfunctory investigation of his client’s known mental impairments. Our determination in Hendricks that counsel’s investigatory work was reasonable, however, did not extend into the penalty phase. Because a sentencing jury is given “broad latitude to consider amorphous human factors, in effect, to weigh the worth of one’s life against his culpability,” we have recognized that the presentation of relevant mitigation evidence is of vital importance to the jury’s penalty determination. Accordingly, we concluded that “counsel’s failure to investigate his client’s mental condition as a mitigating factor in a penalty phase hearing, without a supporting strategic reason, constitute^] deficient performance.” We therefore held that because evidence of Hendrick’s “nightmarish upbringing” and “mental problems” could have altered the jury’s decision to impose a death verdict, counsel was constitutionally ineffective. 463 F.3d at 993 (citations omitted). Compare Summerlin v. Schriro, 427 F.3d 623, 631 (9th Cir.2005) (counsel’s performance was deficient for relying exclusively on information developed at the defendant’s pretrial competency examination), with Stokley v. Ryan, 659 F.3d 802, 812-15 (9th Cir.2011) (counsel’s performance was not deficient, in choosing a neurological exam over a neuropsychological exam, when either was recommended, because counsel did pursue mental health evaluations pertinent to sentencing, as recommended by mental health experts, provided the documents the experts suggested, and presented their testimony). Based on this evidence, the district court concluded that defense counsel “did not fulfill his responsibility to [Doe] on the issue of investigating and presenting mental health testimony simply by retaining Dr. [M.R.], given the brief time she spent with him.” Especially given that J.B. failed both to instruct Dr. M.R. specifically to investigate penalty-phase mitigation and to provide her with any of the documents necessary to complete that task (which she requested), we firmly agree. 6. Substance Abuse . D.S.’s investigation did not reveal the extent of Doe’s substance abuse, but it did reveal enough to warrant expert follow-up. Doe acknowledged that he drank alcohol excessively, that he became aggressive when he was drunk, and that he had used both marijuana and cocaine. Other interviewees noted that he had problems with alcohol and that the robbery he committed was alcohol-related. Most important, D.S. concluded that although “[Doe] feels that he has no drinking problem, ... [i]t appears to this investigator that alcohol is a direct cause for [Doe]’s violence as he described his past life to me.” He went on to say that “[Doe] does not necessarily like to admit that he may have certain problems in particular areas ... [such as] his inability to control his drinking and drug abuse.... ” Despite his investigator’s clearly expressed disbelief in Doe’s representations about substance abuse — in a report he does appear to have read — J.B. did not follow up, and thus did not discover the fact that Doe consumed substances in a faded attempt to lessen the pain of the trauma he had suffered. 7. Penalty-Phase Witnesses In addition, Doe claims, J.B. did nothing to prepare his penalty-phase witnesses. J.B. does not dispute this: “I did not prepare any of the penalty phase witnesses for their testimony. I did not tell them what specific questions I was going to ask them nor did we discuss the responses I expected from them.” Although this failure, on its own, might have been insufficient to establish deficient performance, it serves here to reinforce other evidence of J.B.’s woefully incompetent mitigation investigation and presentation. The witnesses who did end up testifying were family and friends attending the trial. Some of them were not told that they would be testifying until they arrived. J.B. did little better with the two penalty-phase witnesses he “spent a lot of time with[,]” Doe’s mother and aunt. He acknowledged in his declaration that his failure to prepare them to testify was related to his failure to discover the powerful mitigating evidence about which they could have spoken. He said: “[M]y relationship with them was quite superficial, and I got very little useful or accurate information from them about [Doej’s life before his arrest for this offense.” J.B. also acknowledged that he “spoke with them only generally about the penalty phase [and] did not go over with them the particular questions they would be asked.” He explained that Doe’s “mother and aunt seemed like such nice, sweet ladies that I never got beyond consoling them about [Doe]’s plight so as to get to really understand his life.” The state’s argument that J.B.’s failure to prepare penalty-phase witnesses was an acceptable trial strategy is erroneous. Even if it were an intentional decision on his part — which is both doubtful and disturbing — spur-of-the-moment mitigation presentations form no part of constitutionally adequate representation. Witness preparation is a critical function of counsel. See Hamilton v. Ayers, 583 F.3d 1100, 1121 (9th Cir.2009) (“[T]he failure to prepare a witness adequately can render a penalty phase presentation deficient. This is especially true when the insufficiently prepared witnesses] [are] the only penalty phase witnesses] called to testifiy].” (emphasis and citations omitted)); Douglas, 316 F.3d at 1088-89 (failure to prepare defense mitigation witnesses led both to inadequate development of evidence and also to less-than-compelling testimony). A lawyer needs to know the nature of the testimony he will elicit, and a witness needs to understand the proceeding in which he is participating. Our case law, and an elementary understanding of the function of a trial lawyer in our adversary system, make plain that although there is no requirement of rehearsal, not preparing penalty-phase witnesses at all is not a legitimate defense method in a capital trial. 8. Conclusion The investigation here was facially inadequate. J.B. fell far short of his “sacrosanct duty to conduct a full and complete mitigation investigation.”. Earp, 431 F.3d at 1175. This conclusion is only strengthened by the fact that the limited investigation J.B. and D.S. did conduct put them on notice that further investigation was warranted. Unfortunately for Doe, they failed to perform it. We have repeatedly held, as has the Supreme Court, that this constitutes deficient performance. See Lambright v. Schriro, 490 F.3d 1103, 1117 (9th Cir.2007) (reaffirming the principle that “when ‘tantalizing indications in the record’ suggest that certain mitigating evidence may be available, those leads must be pursued” (emphasis added) (quoting Stankewitz v. Woodford, 365 F.3d 706, 719-20 (9th Cir.2004))); see also Wiggins, 539 U.S. at 525, 123 S.Ct. 2527 (“The scope of their investigation was also unreasonable in light of what counsel actually discovered in the [files he did obtain]____ [A]ny reasonably competent attorney would have realized that pursuing these leads was necessary to making an informed choice among possible defenses.... Had counsel investigated further, they might well have discovered the sexual abuse later revealed during state postconviction proceedings.”); Stankewitz v. Wong, 698 F.3d 1163, 1171 (9th Cir.2012) (“The state’s argument that [defense counsel’s] mere possession of [files containing leads to mitigating evidence] demonstrates that [he] conducted a reasonable investigation defies logic — if anything, that [he] had this evidence at his fingertips but did not investigate or present it is further proof of his deficiency.”). J.B. said: “I did not adequately prepare for a penalty phase in this case because of a combination of inexperience and overconfidence .... I did not ask the right questions of [Doe], his family, myself, or my investigator to obtain an adequate understanding of my client and his case.” We cannot help but agree. B. Strategic Judgment There is a “wide range of reasonable professional assistance!,]” Strickland, 466 U.S. at 689, 104 S.Ct. 2052, but J.B.’s performance was not within its outer bounds. Unlike most trial lawyers called to testify before a habeas court, J.B. never attempted to justify his actions as based in strategy; he admitted that he would have presented the extensive mitigating evidence habeas counsel discovered, had he found it himself. J.B. acknowledged that he did not make a strategic decision not to put on the sort of mitigating evidence later adduced; he simply didn’t know about it. “Looking back on the penalty phase of [Doej’s trial, it’s hard for me to say what my strategy was,” he said. Strickland tells us that “strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitations on investigation. In other words, counsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary.” 466 U.S. at 690-91, 104 S.Ct. 2052. In conducting this analysis, we “must indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance.” Id. at 689, 104 S.Ct. 2052. Here, J.B.’s “failure to investigate thoroughly resulted from inattention, not reasoned strategic judgment.” Wiggins, 539 U.S. at 526, 123 S.Ct. 2527. “[Djefense counsel failed to make a reasonable investigation into potential mitigating evidence. Therefore, his decision not to put on a mitigation case cannot be considered to be the product of a strategic choice. An uninformed strategy is not a reasoned strategy. It is, in fact, no strategy at all.” Correll, 539 F.3d at 949. The presumption that defense counsel’s conduct falls within the wide range of reasonable professional assistance is inapposite, or at least firmly rebutted, when, as here, we know for sure that defense counsel had no strategy, because he has unequivocally said as much. Representing a capital defendant without a strategy is per se unreasonable, and necessarily constitutes deficient performance. Generally, we credit the statements of defense counsel as to whether their decisions at trial were — or were not — based on strategic judgments. In Heishman v. Ayers, 621 F.3d 1030, 1040 (9th Cir.2010), we made clear that the trial strategy presumption does not apply when it “would contradict [defense counsel’s] testimony rather than filling a gap in memory, contravening the Supreme Court’s admonition against adopting ‘a post hoc rationalization of counsel’s conduct’ instead of relying on an ‘accurate description of their deliberations’ [when one exists].” Id. (quoting Wiggins, 539 U.S. at 526-27, 123 S.Ct. 2527). See also Williams v. Taylor, 529 U.S. 362, 373, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (crediting “trial counsel’s testimony before the state habeas court [that] counsel did not fail to seek [the defendant’s] juvenile and social services records because he thought they would be counterproductive ... [and his] acknowledgement] in the course of the hearings that information about [the defendant’s] childhood would have been important in mitigation”). In any event, it is self-evident that J.B.’s failure to conduct further mitigation investigation was objectively unreasonable. “This is [ ] a ease in which the defendant’s attorneys failed to act while potentially powerful mitigating ■ evidence stared them in the face, or would have been apparent from documents any reasonable attorney would have obtained.” Bobby, 558 U.S. at 11, 130 S.Ct. 13 (citing Wiggins, 539 U.S. at 525, 123 S.Ct. 2527; Rompilla, 545 U.S. at 389-93, 125 S.Ct. 2456). D.S. suggested that J.B. obtain Doe’s prison file, which contained readily apparent and powerful mitigating evidence. He failed to do so. D.S. noted that Doe was beaten as a child, and reported to J.B. his belief that Doe suffered more from mental health problems and substance abuse than he was willing to admit. Yet, no follow-up investigation to explore these issues was ever done. J.B. did not retain an expert to conduct a penalty-phase investigation, and when the psychologist he hired to consider mental state defenses at the guilt phase, Dr. M.R., called him to ask for records that would have allowed her to uncover the available mitigating evidence, he didn’t even bother to return her call. This slipshod work cannot be “immunized from Sixth Amendment challenges simply by attaching to it the label of ‘trial strategy.’” Silva v. Woodford, 279 F.3d 825, 846 (9th Cir.2002). In short, the evidence that J.B.’s performance at the penalty phase fell well below the constitutional minimum is overwhelming. V. Prejudice Although the district court found that J.B. had performed incompetently, it concluded that his deficient performance had not prejudiced Doe. That conclusion is erroneous. To establish prejudice, Doe must show “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. “A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id. “That requires a ‘substantial,’ not just a ‘conceivable,’ likelihood of a different result.” Pinholster, 131 S.Ct. at 1403 (quoting Harrington v. Richter, 562 U.S. 86, 131 S.Ct. 770, 792, 178 L.Ed.2d 624 (2011)). “[T]he question is whether there is a reasonable probability that, absent the errors, the sentencer ... would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.” Strickland, 466 U.S. at 695, 104 S.Ct. 2052. We therefore “reweigh the evidence in aggravation against the totality of available mitigating evidence.” Wiggins, 539 U.S. at 534, 123 S.Ct. 2527. Because death sentences in California must be imposed by a unanimous jury, we must find prejudice if there is a “reasonable probability that at least one juror would have” voted for life. Id. at 537, 123 S.Ct. 2527; Cal.Penal Code § 190.4(b). We hold that there is a substantial probability that there would have been a different result at the penalty phase had counsel’s performance during that phase of the trial not been ineffective. A. Aggravating Evidence at Trial The aggravating evidence the jury considered was, for a capital case, fairly minimal. Instead of any additional evidence about the murder for which Doe had just been convicted, beyond, what was presented during the guilt phase of the trial, a stipulation was read to the jury that said in its entirety that a doctor, if called to testify, would state that “the victim in this case [L.R.] at the time of her death had a mild hemiparesis of the brain as a result of congenital cerebral palsy which entailed a varying but never totally disabling of her impairment in motor coordination and muscular development on her left side only.” The remainder of the aggravating evidence consisted solely of testimony about two incidents: First, the prosecution called one of two women who were walking together in a park in Doe’s home state when Doe robbed them at gunpoint. The jury learned that Doe, then a juvenile, had grabbed the woman’s arm, pointed a gun at her, and demanded both of their purses. He fled immediately after acquiring them, leaving the women uninjured. A certified copy of Doe’s conviction, showing that he had served five years in prison for this crime, was also introduced. Second, the jury heard testimony about another incident, in which Doe was arrested on suspicion of residential burglary. (No conviction resulted.) Doe — who was homeless at the time — broke into an apartment after its residents left for work and got into bed. When he was found there, by a police officer, he was under the covers, wearing a sweatshirt and sweatpants but no socks or shoes. He initially gave a false last name and said that he had spent the previous night with his girlfriend, who, he said, lived in the apartment. After he was arrested, Doe was allowed to retrieve some óf his clothing, which he had hung in the closet. At the police station, Doe admitted that he had entered through the window, but said that he had touched nothing and only wanted to sleep. In fact, it appeared that he had riffled through the kitchen, and moved the TV and VCR away from the wall (a screwdriver was found nearby). A long knife was found on the floor of the bathroom; he had neglected to flush the toilet. This penalty-phase aggravating evidence is a far cry from that which the Supreme Court deemed “extensive” in Pinholster, 131 S.Ct. at 1408. There, the state presented evidence that the defendant had “threatened to kill the State’s lead witness, assaulted a man with a straight razor, and kidnapped another person with a knife.” Id. There was also evidence that he “had a history of violent outbursts, including striking and threatening a bailiff after a court proceeding at age 17, breaking his wife’s jaw, resisting arrest by faking seizures, and assaulting and spitting on police officers.” Id. Moreover, the jury in Pin-holster heard about the defendant’s involvement in juvenile gangs and his substantial disciplinary record in both county and state jails, where he had threatened, assaulted, and thrown urine at guards, [ ] fought with other inmates ... [and] had been segregated for a time due to his propensity for violence and placed on a ‘special disciplinary diet’ reserved only for the most disruptive inmates. Id. Doe’s criminal record — the only aggravating evidence presented by the state at the penalty phase — was light compared to those of many capital defendants; his only previous conviction was for an armed robbery, in which no one was injured, committed when he was a juvenile. When compared with the offenses of other death-eligible defendants, all of which necessarily involve egregious crimes of violence, the facts of the crime Doe committed were also not particularly aggravating; notably, although the jury rendered a finding of true on the felony-murder-burglary special-circumstance allegation, it rendered a finding of not true on the felony-murder-rape special-circumstance allegation. We have found prejudice from failure to present mitigating evidence in cases involving crimes substantially more heinous than Doe’s. Cornil, 539 F.3d at 951-55 (holding that the defendant was prejudiced by counsel’s failure to present mitigating evidence despite the fact that he kidnaped three people, bound their hands and feet with duct tape, drove them into the desert, shot one of them in the head execution-style, and watched as a friend of his killed the other two); Ainsworth v. Woodford, 268 F.3d 868, 870-71, 878 (9th Cir.2001) (holding that the defendant was prejudiced by counsel’s failure to present mitigating evidence despite the fact that he shot a woman in the hip, raped her as she bled from the gunshot wound, and confined her in her car, at times in the trunk, for twenty-four hours until she bled to death); Hendricks, 70 F.3d at 1044-45 (holding that the defendant was prejudiced by counsel’s failure to present mitigating evidence despite the fact that he was convicted of murdering two men who had paid him for sex by s