DANNY JONES V. CHARLES RYAN

FILED FOR PUBLICATION NOV 7 2022 UNITED STATES COURT OF APPEALS MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS FOR THE NINTH CIRCUIT DANNY LEE JONES, No. 18-99005 Petitioner-Appellant, D.C. No. 2:01-cv-00384-SRB v. ORDER AND AMENDED CHARLES L. RYAN, OPINION Respondent-Appellee. Appeal from the United States District Court for the District of Arizona Hon. Susan R. Bolton, District Judge, Presiding Argued and Submitted February 16, 2021 San Francisco, California BEFORE: S.R. THOMAS, and HAWKINS, and CHRISTEN, Circuit Judges Order; Amended Opinion by Judge S.R. Thomas; Dissent from Order by Judge Bennett; Dissent from Order by Judge Ikuta SUMMARY * Habeas Corpus / Death Penalty The panel filed an amended opinion, denied a petition for panel rehearing, and denied on behalf of the court a petition for rehearing en banc, in a case in which the panel, applying the appropriate standards pursuant to the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), reversed the district court’s judgment denying Danny Lee Jones’s habeas corpus petition challenging his Arizona death sentence, and remanded to the district court with instructions to issue the writ. In Claim 1, Jones asserted that his trial counsel was constitutionally ineffective by failing to request a mental health expert in advance of the sentencing hearing. The panel held that the state court record demonstrates that trial counsel was constitutionally ineffective by failing to secure a defense mental health expert, and that, pursuant to 28 U.S.C. § 2254(d)(1), the Arizona Supreme Court’s contrary conclusion was an unreasonable application of Strickland v. Washington, 466 U.S. 668 (1984), and its progeny. Holding that the state post-conviction review (PCR) court’s decision was also based on an unreasonable determination of the facts under 28 U.S.C. § 2254(d)(2), the panel agreed with Jones that (1) the PCR court employed a defective fact-finding process when it denied PCR counsel’s funding request for a defense neuropsychological expert, effectively preventing the development of Claim 1; and (2) the state court’s failure to hold a hearing on Claim 1 resulted in an unreasonable determination of the facts. The panel wrote that if the state court had reached the question of Strickland prejudice, the panel would be required to afford the decision deference under AEDPA, but because the PCR court did not reach the issue of prejudice, the panel reviewed the issue de novo. Noting that Jones was diligent in attempting to develop the factual basis for the claim in state court, the panel wrote that the district court did not err in its expansion of the record, and the panel considered the evidence developed in the district court in conducting its de novo review. The panel wrote that on de novo review, it must weigh the aggravating factors against the mitigation evidence, as developed in the state court record that was available, but not presented. The panel also considered the mitigation evidence that was presented. Reweighing the evidence in aggravation against the totality of * This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. the available mitigating evidence, the panel concluded that there is at least a reasonable probability that development and presentation of mental health expert testimony would have overcome the aggravating factors and changed the result of the sentencing proceeding. The panel therefore concluded on de novo review that Jones demonstrated Strickland prejudice, and, accordingly, reversed the district court’s denial of relief on Claim 1. In Claim 2, Jones asserted that his trial counsel was constitutionally ineffective by failing to seek neurological or neuropsychological testing prior to sentencing. The panel wrote that counsel’s failure to promptly seek neuropsychological testing ran contrary to his obligation to pursue reasonable investigations under Strickland, and in particular, his obligation to investigate and present evidence of a defendant’s mental defect. The panel therefore concluded that the PCR court’s decision that defense counsel’s performance did not fall below an objectively reasonable standard was an unreasonable application of Strickland, and that Jones satisfied § 2254(d)(1). The panel also held that the state PCR court’s decision was based on an unreasonable determination of the facts, satisfying § 2254(d)(2), where the PCR judge made factual findings regarding the necessity of neuropsychological testing, not on the basis of evidence presented by Jones, but on the basis of his own personal conduct, untested memory, and understanding of events—and by plainly misapprehending the record, which included a forensic psychiatrist’s testimony, six years earlier, strongly suggesting that neuropsychological testing was essential. Because the PCR court did not reach the issue of prejudice, the panel reviewed the issue de novo. Noting that Jones was diligent in attempting to develop the factual basis for the claim in state court, the panel wrote that the district court did not err in its expansion of the record, and the panel considered the evidence developed in the district court in conducting its de novo review. The panel concluded that Jones demonstrated Strickland prejudice because there is a reasonable probability that had such testing been conducted, and had the results been presented at sentencing, Jones would not have received a death sentence. The panel wrote that, in combination, the testing results and the presentation of contributing factors would have dramatically affected any sentencing judge’s perception of Jones’s culpability for his crimes, even despite the existence of aggravating factors. Because the panel determined that Jones is entitled to relief and resentencing on the basis of Claims 1 and 2, the panel did not reach whether new evidence presented at the federal evidentiary hearing fundamentally altered these claims such that they were unexhausted, procedurally defaulted, and excused in light of Dickens v. Ryan, 740 F.3d 1302 (9th Cir. 2014) (en banc), and Martinez v. Ryan, 566 U.S. 1 (2012). The panel likewise did not reach the merits of any of Jones’s other claims. Judge Bennett, joined by Judges Callahan, R. Nelson, Bade, Collins, Lee, Bress, Bumatay, and VanDyke, dissented from the denial of rehearing en banc. He wrote that the panel improperly and materially lowered Strickland’s highly demanding standard and failed to afford the required deference to the district court’s findings— essentially finding that no such deference was due. He wrote that the court should have taken this case en banc (1) to secure and maintain uniformity in our case law; (2) because this case involves issues of exceptional importance; and (3) so that the Supreme Court, which has already vacated this court’s judgment once in this case, does not grant certiorari a second time and reverse. Judge Ikuta, joined by Judges Callahan and VanDyke, dissented from the denial of rehearing en banc. She agreed with Judge Bennett that even if the panel had been correct in conducting a de novo review of the state court’s decision, it erred in failing to defer to the district court’s factual findings. In her view, however, the panel had no business conducting such a de novo review in the first place. She wrote that in reaching the issue of prejudice de novo, the panel mischaracterized the state court opinion and disregarded the admonitions of the Supreme Court to give such opinions proper deference. COUNSEL Amanda Bass (argued) and Leticia Marquez, Assistant Federal Public Defenders; Jon M. Sands, Federal Public Defender, District of Arizona; Federal Public Defenders’ Office, Tucson, Arizona; Jean-Claude André, Bryan Cave Leighton Paisner LLP, Santa Monica, California; Barbara A. Smith and J. Bennett Clark, Bryan Cave Leighton Paisner LLP, St. Louis, Missouri; Kristin Howard Corradini, Bryan Cave Leighton Paiser LLP, Chicago, Illinois; for Petitioner-Appellant. Jeffrey L. Sparks (argued), Assistant Attorney General, Capital Litigation Section; Lacey Stover Gard, Chief Counsel; Mark Brnovich, Attorney General of Arizona; Office of the Attorney General, Phoenix, Arizona; for Respondent-Appellee. — 1 — ORDER The opinion filed June 28, 2021, Jones v. Ryan, 1 F.4th 1179 (9th Circ. 2021) is amended and superseded by the opinion filed concurrently with this order. The full court has been advised of the petition for rehearing en banc. A judge of this Court requested a vote on the petition for rehearing en banc. A majority of the non-recused active judges did not vote to rehear the case en banc. Fed. R. App. 35. The petition for panel rehearing and for rehearing en banc is DENIED. No further petitions for panel rehearing or rehearing en banc will be entertained. Amended Opinion by Judge Sidney R. Thomas S.R. THOMAS, Circuit Judge: Danny Lee Jones, an Arizona inmate on death row, appeals the district court’s denial of his petition for writ of habeas corpus on remand from this court and the Supreme Court of the United States. Applying the appropriate standards pursuant to the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), we conclude that Jones was denied the effective assistance of counsel at sentencing. We reverse the judgment of the district court and remand for further proceedings consistent with this opinion. — 2 — I1 A On March 26, 1992, in Bullhead City, Arizona, Jones and his friend Robert Weaver spent the day drinking and using crystal methamphetamine in Weaver’s garage. At some point, a fight broke out, and evidence at trial indicated that Jones hit Weaver over the head multiple times with a wooden baseball bat, killing him. Jones then went inside the house where he encountered Weaver’s grandmother, Katherine Gumina. Jones struck Gumina in the head with the bat and knocked her to the ground. Jones then made his way to a bedroom where he found Tisha Weaver, Weaver’s seven-year-old daughter, hiding under the bed. Evidence showed that Jones hit Tisha in the head with the bat, and either strangled her or suffocated her with a pillow. Jones fled to Las Vegas, Nevada, where police eventually arrested him. He was indicted in Arizona on two counts of murder in the first degree, and one count of attempted murder.2 1 In accordance with our obligation under Cullen v. Pinholster, 563 U.S. 170 (2011), to consider only the state court record in conducting our 28 U.S.C. § 2254(d) analysis, this recitation of the facts looks only to that record. Evidence developed at the federal evidentiary hearing is included later in the limited contexts where Pinholster does not circumscribe our consideration of such evidence. 2 Gumina initially survived the attack and was in a coma for seventeen months before eventually dying from her injuries. The prosecution never amended (continued...) — 3 — B A public defender was assigned to Jones’s case. At the time, the public defender had been an attorney for a little more than three years, and he had never been a lead attorney on a capital case. He requested $5,000 from the trial court for expert witnesses. The court authorized $2,000, which the public defender split between a crime scene investigator and an addictionologist. The jury convicted Jones on all counts. Judge James Chavez scheduled the sentencing hearing for three months later. About six weeks before the hearing, counsel took his first trip to Reno, Nevada, in order to speak with Jones’s mother, Peggy Jones3, and Jones’s second step-father, Randy Jones, in order to investigate potential mitigation evidence. At sentencing, the public defender presented testimony from two witnesses: investigator Austin Cooper and Randy Jones. Cooper testified about evidence regarding an alleged accomplice. Randy explained that he married Peggy when 2 (...continued) the indictment after Gumina died. 3 To avoid confusion, we refer to the members of Jones’s family by their first names. — 4 — Jones was seven years old.4 He explained that Peggy gave birth to Jones when she was only fifteen years old and had numerous complications during the pregnancy and delivery. Randy testified that Jones suffered multiple head injures when he was growing up, and that when Jones was thirteen or fourteen his personality began to change drastically. Jones started lying, cutting classes at school, drinking, and doing drugs. Jones’s first step-grandfather introduced him to marijuana when he was about ten years old, and Jones was an alcoholic by the time he was seventeen. The trial court appointed the Chief of Forensic Psychiatry for the Correctional Health Services in Maricopa County, Dr. Jack Potts, to examine Jones and provide a report to the court pursuant to Rule 26.5 of the Arizona Rules of Criminal Procedure.5 Defense counsel called Dr. Potts to testify at sentencing. Dr. Potts stated that in conducting his review, he spent four hours interviewing Jones in prison, one and a half of which were spent administering a personality test. He also spoke to Jones for a couple of hours the day before testifying at the sentencing 4 The record is inconsistent whether Randy and Peggy married when Jones was seven or eight years old. 5 The Rule provides: “At any time before the court pronounces a sentence, it may order the defendant to undergo a mental health examination or diagnostic evaluation. Unless the court orders otherwise, any report concerning such an examination or evaluation is due at the same time as the presentence report.” — 5 — hearing. He interviewed Peggy by phone for thirty minutes, and he spoke to Randy for one hour the day before testifying. During Dr. Potts’s testimony, the following colloquy took place: Q. Do you feel you have been provided with adequate data, coupled with your in-person examination of the defendant, to make a conclusion for mitigating findings that you did? A. . . . . I believe everything I reviewed and what I have heard about the case and reviewed with the defendant, his comments to me. I would have liked, and I think I have – I think it would be valuable to have had some neurologic evaluations, not – by a neurologist, clinical exam, such as a CAT scan, possibly an MRI, possibly EEG, possibly some sophisticated neurological testing, because I think there’s very strong evidence that we have . . . , I believe, of traumatic brain injury, and there’s some other evidence that I believe we may have organic neurologic dysfunctions here that has gone on since he’s been about 13. So, there’s some other testing that I think would be valuable to have to pin down the diagnosis. . . . Q. And you think that further testing might shed some additional light on, perhaps, some of these factors you listed and maybe why Mr. Jones behaves in the way he did on March 26, 1992? A. Yes. I think it could help in clarifying and giving us etiology as the behavioral components, the explosive outbursts, the aggression, the mood changes, and the changes that occurred in his personality as noted by his mother when he was about 13, 14 years old. — 6 — Q. In your opinion, could that information possibly provide . . . a significant mitigating factor as to what would be relevant to the issues at this hearing? A. Clearly I think it would be corroborative of my clinical impressions and my diagnostic impressions in my report. Dr. Potts discussed the fact that Jones’s first step-father physically and verbally abused Jones, and stated that it was “unequivocal” that Jones carried that abuse with him into his adult life. Dr. Potts also stated that, given the long history of substance abuse and other psychological problems in Jones’s family, Jones was predisposed for substance abuse or a possible affective disorder. Dr. Potts did not, however, give a specific diagnosis, but stated: “I think . . . to a reasonable degree of medical certainty that the defendant suffers from a [cyclothymic] disorder, which is a mood disorder, possibly organic syndrome, secondary to the multiple cerebral trauma that he’s had as well as the prolonged substance abuse.” Dr. Potts testified that the drugs and alcohol Jones had on the day of the murders would have had a significant effect on Jones because “it’s real clear that the brain is much more susceptible when it’s been injured by drugs. Furthermore, when you’re on drugs, you are more susceptible to the acts of aggression under amphetamines.” He further stated that “I believe in my experience in cases like *7 this, is that had it not been for the intoxication, the alleged offense would not have occurred.” Dr. Potts also submitted a six-page report to the court. The report included approximately two pages describing Jones’s social development and history, including his medical history, one page of analysis, and one page of recommendations. Dr. Pott’s report was due to the court on November 29, 1993, but he did not complete it until December 3. He was late because he did not receive the Presentence Information Report (“PSR”) from the Mohave County probation department until December 1. Dr. Potts also testified at sentencing that he was under “significant time pressure” in preparing the report. Dr. Potts concluded that “Mr. Jones’ capacity to conform his conduct to that of the law was clearly impaired at the time of the offenses. . . .” He therefore recommended that an aggravated sentence should not be imposed. After Dr. Potts testified, counsel moved for a continuance so an expert could conduct psychological testing. Counsel stated: “It’s not a delay tactic . . . [I]t’s not something I planned on doing until . . . very recently after the report was done, after talking with Dr. Potts, after exploring all these issues.” Notably, however, counsel did not speak to Dr. Potts about the report until December 7, the night before sentencing. The sentencing judge considered and rejected the motion: — 8 — THE COURT: . . . . I also know that there were funds made available to the defense at some point and you used them to hire [an addictionologist]. . . . [I]f there were any follow-up questions of a psychological or neurological nature, I would think that the defense would have followed them up. COUNSEL: But, Your Honor, respectfully, . . . I didn’t realize this issue was that important until Dr. Potts brought it up or I would have certainly asked for the funds earlier. The judge found the following aggravating factors for Weaver’s murder: (1) Jones “committed the offense as consideration for the receipt, or in expectation of the receipt of anything of pecuniary value”; (2) Jones “committed the offenses in an especially heinous or depraved manner”; (3) Jones was “convicted of one or more other homicides . . . which were committed during the commission of the offense.” Under Arizona’s then-existing death penalty statute, the trial judge held an aggravation/mitigation hearing to determine whether a death sentence was warranted. Ariz. Rev. Stat. § 13-703(B) (1993). The judge had to impose a death sentence if he found “one or more of the [enumerated statutory] aggravating circumstances . . . and that there [were] no mitigating circumstances sufficiently substantial to call for leniency.” Id. § 13-703(E). — 9 — The judge found four non-statutory mitigating factors: (1) Jones suffered from long-term substance abuse; (2) he was under the influence of drugs and alcohol at the time of the offense; (3) he had a chaotic and abusive childhood; and (4) his longstanding substance abuse problem may have been caused by genetic factors and aggravated by head trauma. The judge found the same aggravating and mitigating circumstances for Tisha’s murder, but he also found that Tisha’s having been less than fifteen years old was an additional aggravating factor. The judge sentenced Jones to two death sentences for the murders, and twenty-five years without the possibility of parole for the attempted murder. The Arizona Supreme Court affirmed Jones’s conviction and sentence on direct review. State v. Jones, 917 P.2d 200 (1996). C Prior to filing Jones’s state post-conviction review (“PCR”) petition, PCR counsel sought authorization from the court for the funding of several experts. As relevant here, the PCR court rejected counsel’s request to appoint a neuropsychologist. The court stated that while Dr. Potts might not have been a defense expert, he did a good job, gave “defense opinions,” and there was no reason to believe that an expert appointed for the defense “would have been any different.” The court concluded by stating that based on Dr. Potts’s testimony, “I *10 don’t really see any grounds for any additional psychiatric or psychological testing.” On July 1, 1999, counsel filed the PCR petition, raising twenty-five claims. Among the petition’s exhibits were a declaration from defense trial counsel and an affidavit from Peggy. At an informal conference on February 23, 2000, the court ruled on several of Jones’s claims, and set others for an evidentiary hearing. In particular, the court denied Claim 1 (as numbered in this appeal) on the merits. The court set Claim 2 (as numbered in this appeal), as well as other claims, for evidentiary hearing. At the evidentiary hearing, Randy, Peggy, and defense trial counsel testified. Randy testified that he first spoke to counsel in July 1992, a few months after Jones’s arrest. During this conversation, Randy told counsel about Jones’s head injuries, as well as his struggles with substance abuse and stints in rehabilitation programs. Randy next spoke to counsel when he came to visit Peggy and Randy at their home in Reno in October 1993, about six weeks before sentencing. Peggy testified that she had provided counsel with a chronology of Jones’s life during counsel’s visit. Peggy remembered sharing about Jones’s difficult birth and the physical abuse she and Jones suffered at the hands of Jones’s biological father and first step-father. Peggy shared that Jones had a good home life and a *11 normal childhood once she married Randy, when Jones was about seven or eight years old. Trial counsel testified that at the time he was appointed to represent Jones, he had been an attorney for three and a half years and his experience with capital cases consisted of having been second chair at the penalty phase in one prior case. He stated that his strategy for defending the killing of Robert Weaver was self-defense, so he hired Dr. Sparks as an addictionologist to testify about Jones’s state of mind. Dr. Sparks opined at trial that because of the drugs Jones ingested, he was unable to premeditate the killings. Dr. Sparks was not called to testify at sentencing. When PCR counsel asked trial counsel if he visited Jones’s family early enough in the case to adequately develop mitigation evidence, trial counsel responded that Dr. Potts was able to make effective use of the information obtained from the family. He said that Dr. Potts was a “very favorable mitigation witness for the defense.” He stated that it felt to him like Dr. Potts was part of the defense team, even though he was appointed as a court expert. Finally, counsel stated that he did not consider the need for testing by a neuropsychologist until Dr. Potts suggested it to him on December 7, 1993, the evening before Jones’s sentencing hearing. — 12 — In the affidavit he provided as an exhibit to the PCR petition, trial counsel stated that he asked the court for $5,000 for expert witnesses at trial. When the trial court authorized only $2,000 of the $5,000 he requested, he “was of the opinion that it would be fruitless to ask the court for additional funding for any other needed experts such as an independent psychiatrist or psychologist.” Counsel added that he did not ask his supervisor for any money because he believed that the public defender’s office did not have sufficient funds for retaining expert witnesses. After the hearing, the PCR court denied Claim 2 as well as the remaining pending claims. As to Claim 2, the court stated that “[t]he report and testimony of Dr. Potts[,] who was appointed by the Court, adequately addressed defendant’s mental health issues at sentencing.” Jones filed a petition for review in the Arizona Supreme Court, which it denied on February 13, 2001. D Jones subsequently filed his federal petition for habeas relief. The district court granted an evidentiary hearing with regard to Claims 1 & 2 based on trial counsel’s failure to secure the appointment of a mental health expert and failure to move for neurological and neuropsychological testing. — 13 — The district court subsequently dismissed both ineffective assistance of counsel (“IAC”) claims. The court denied Claim 1 because counsel’s “failure to seek the appointment of a mental health expert in a more timely manner did not prejudice Petitioner.” The district court explained that “the Court has not been presented with evidence confirming that Petitioner suffers from neurological damage caused by head trauma or other factors. Therefore, Dr. Potts’s finding at sentencing remains the most persuasive statement in the record that neurological damage constituted a mitigating factor.” The district court dismissed Claim 2 after finding that Jones could only prove that he suffered from AD/HD residual type and possibly a low level mood disorder. The district court “conclud[ed] that the trial court would have assigned minimal significance to testimony indicating that Petitioner suffered from ADHD [sic] and a low-level mood disorder, and that this weight would not have outbalanced the factors found in aggravation.” E Jones timely appealed the district court’s denial of his petition for a writ of habeas corpus. We reversed the district court and concluded that Jones received IAC warranting relief on his claims regarding his counsel’s failure to secure the appointment of a mental health expert, failure to timely move for neurological and *14 neuropsychological testing, and failure to present additional mitigation witnesses and evidence. See Jones, 583 F.3d at 636. The State petitioned for certiorari. The Supreme Court granted the petition, vacated our judgment, and remanded the case for further consideration in light of Pinholster. See Ryan v. Jones, 563 U.S. 932 (2011). On remand from the Supreme Court, we remanded the case to the district court to consider, under Martinez v. Ryan, 566 U.S. 1 (2012), and Dickens v. Ryan, 740 F.3d 1302 (9th Cir. 2014) (en banc), “Jones’s argument that his ineffective assistance of counsel claims are unexhausted, and therefore procedurally defaulted, and that the deficient performance by his counsel during his post-conviction relief case in state court excuses the default.” Jones v. Ryan, 572 F. App’x 478 (9th Cir. 2014) (Mem.). We expressed “no opinion on any other issue raised on appeal,” and noted that “[t]hose issues are preserved for later consideration by the Court, if necessary.” Id. On remand, the district court rejected Jones’s arguments. The district court determined that Jones’s claims had not been fundamentally altered, and therefore, they had previously been exhausted and were not subject to de novo review. Additionally, the court concluded that PCR counsel was not ineffective as required by Martinez, so any default would not be excused anyway. 566 U.S. 1. Jones filed *15 a timely notice of appeal and stated that he was also appealing “all prior orders disposing of other claims, either on the merits or procedurally.” II We review de novo a district court’s dismissal of a habeas petition. Sexton v. Cozner, 679 F.3d 1150, 1153 (9th Cir. 2012). We review a district court’s findings of fact for clear error. Stanley v. Schriro, 598 F.3d 612, 617 (9th Cir. 2010). Because Jones filed his petition after April 24, 1996, AEDPA applies to our review of this petition. See Summers v. Schriro, 481 F.3d 710, 712 (9th Cir. 2007). Under AEDPA, habeas relief may not be granted unless the state court’s decision was: (1) “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States;” or (2) “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(1)–(2). “A state court decision is ‘contrary to’ clearly established Supreme Court precedent if the state court applies a rule that contradicts the governing law set forth in Supreme Court cases or if the state court confronts a set of facts materially indistinguishable from those at issue in a decision of the Supreme Court and, nevertheless, arrives at a result different from its precedent.” Lambert v. Blodgett, *16 393 F.3d 943, 974 (9th Cir. 2004) (citing Lockyer v. Andrade, 538 U.S. 63, 73 (2003)) (emphasis in original). A state court’s decision is an “unreasonable application” of federal law if it “identifies the correct governing principle from [the Supreme] Court’s decisions but unreasonably applies that principle to the facts of the prisoner’s case.” Id. (internal quotations and citation omitted). The Supreme Court has explained that the exceptions based on “clearly established” law refer only to “the holdings, as opposed to the dicta, of th[e] Court’s decisions as of the time of the relevant state-court decision.” (Terry) Williams v. Taylor, 529 U.S. 362, 412 (2000) (“Terry Williams”). Circuit precedent may not clearly establish federal law for purposes of § 2254(d), but we may “look to circuit precedent to ascertain whether it has already held that the particular point in issue is clearly established by Supreme Court precedent.” Marshall v. Rodgers, 569 U.S. 58, 64 (2013). With respect to § 2254(d)(2) claims, “a state-court factual determination is not unreasonable merely because the federal habeas court would have reached a different conclusion in the first instance.” Wood v. Allen, 558 U.S. 290, 301 (2010). If “‘[r]easonable minds reviewing the record might disagree’ about the finding in question, ‘on habeas review that does not suffice to supersede the trial *17 court’s . . . determination.’” Id. (quoting Rice v. Collins, 546 U.S. 333, 341–42 (2006)). If a petitioner can overcome the § 2254(d) bar with respect to the claims the state court did address, he must also demonstrate that he is entitled to relief without the deference required by AEDPA. See Panetti v. Quarterman, 551 U.S. 930, 953–54 (2007). Where the state court did not reach a particular issue, § 2254(d) does not apply, and we review the issue de novo. See Rompilla v. Beard, 545 U.S. 374, 390 (2005); see also Weeden v. Johnson, 854 F.3d 1063, 1071 (9th Cir. 2017) (“Because the [state court] did not reach the issue of prejudice, we address the issue de novo.”). Pursuant to Pinholster, our § 2254(d) analysis is limited to the facts in the state court record. 563 U.S. at 185. However, in narrow circumstances, when we review a claim de novo, and when a petitioner satisfied the standard for an evidentiary hearing in federal district court pursuant to § 2254(e)(2) by exercising diligence in pursuing his claims in state court, we may consider the evidence developed in federal court. See id.; see also id. at 212–13 (Sotomayor, J., dissenting); see also Schriro v. Landrigan, 550 U.S. 465, 473 n.1 (2007). — 18 — III In Claims 1 and 2, Jones alleges that his counsel provided IAC at sentencing. To prove a constitutional violation for IAC, Jones must show (1) “that counsel’s performance was deficient,” and (2) “that the deficient performance prejudiced the defense.” Strickland v. Washington, 466 U.S. 668, 687 (1984). Counsel’s performance is deficient if, considering all the circumstances, it “fell below an objective standard of reasonableness . . . under prevailing professional norms.” Id. at 688. Under this objective approach, we are required “to affirmatively entertain” the range of possible reasons counsel might have proceeded as he or she did. Pinholster, 563 U.S. at 196. To establish prejudice under Strickland, a petitioner must “show that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Strickland, 466 U.S. at 694. Under the prejudice prong, “[a] reasonable probability is one ‘sufficient to undermine confidence in the outcome,’ but is ‘less than the preponderance more-likely-than-not standard.’” Lambright v. Schriro, 490 F.3d 1103, 1121 (9th Cir. 2007) (quoting Summerlin v. Schriro, 427 F.3d 623, 640, 643 (9th Cir. 2005) (en banc)). It is therefore “not necessary for the habeas petitioner *19 to demonstrate that the newly presented mitigation evidence would necessarily overcome the aggravating circumstances.” Id. (quotation marks and citation omitted). However, “[a] reasonable probability means a ‘substantial,’ not just ‘conceivable,’ likelihood of a different result.” Shinn v. Kayer, 141 S. Ct. 517, 523 (2020) (quoting Pinholster, 563 U.S. at 189). To answer the prejudice inquiry, we must “consider all the evidence—the good and the bad,” Wong v. Belmontes, 558 U.S. 15, 26 (2009) (per curiam), and “reweigh the evidence in aggravation against the totality of available mitigating evidence,” Wiggins v. Smith, 539 U.S. 510, 534 (2003). The totality of available mitigating evidence includes evidence “both . . . adduced at trial, and the evidence adduced in the habeas proceeding[s].” Terry Williams, 529 U.S. at 397. Our review of a Strickland claim under § 2254(d) is “doubly deferential,” requiring the court to apply AEDPA deference on top of Strickland deference. Knowles v. Mirzayance, 556 U.S. 111, 123 (2009). If the state court had reached the question of Strickland prejudice, we would be required to afford the decision deference under AEDPA. However, because the state court reached only the deficient performance prong of Jones’s IAC claims, we review only that prong under § 2254(d), and we review the prejudice prong of his claims de novo. See Weeden, 854 F.3d 1063 at 1071 (“Because the [state court] *20 did not reach the issue of prejudice, we address the issue de novo.”); see also Rompilla, 545 U.S. at 390 (same); Wiggins, 539 U.S. at 534 (same). IV A In Claim 1, Jones asserts that his trial counsel was constitutionally ineffective by failing to secure a defense mental health expert. He asserts that his right to counsel was violated when his attorney failed to request a mental health expert in advance of the sentencing hearing. For the reasons below, we agree. “[C]ounsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary.” Strickland, 466 U.S. at 691. As the Supreme Court has stated, there is a “belief, long held by this society, that defendants who commit criminal acts that are attributable to a disadvantaged background or to emotional and mental problems, may be less culpable than defendants who have no such excuse.” Boyde v. California, 494 U.S. 370, 382 (1990) (quotation and emphasis omitted). Therefore, “‘[i]t is imperative that all relevant mitigating information be unearthed for consideration at the capital sentencing phase.’” Wallace v. Stewart, 184 F.3d 1112, 1117 (9th Cir. 1999) (quoting Caro v. Calderon, 165 F.3d 1223, 1227 (9th Cir. 1999) (brackets in original)). Classic mitigation evidence includes *21 mental disorders, mental impairment, family history, abuse, physical impairments, and substance abuse. Sanders v. Davis, 23 F.4th 966, 985 (9th Cir. 2022); Summerlin. 427 F.3d at 641; see also Terry Williams, 529 U.S. at 396 (noting that counsel has an “obligation to conduct a thorough investigation of the defendant’s background,” citing American Bar Association (“ABA”) Standards for Criminal Justice 4–4.1, commentary, p.4–55 (2d ed 1980)). “That investigation should include examination of mental and physical health records, school records, and criminal records.” Correll v. Ryan, 539 F.3d 938, 943 (9th Cir. 2008). Counsel’s failure to investigate and present evidence of a defendant’s mental defect constitutes deficient performance. Terry Williams, 529 U.S. at 396. In light of Terry Williams, we have also held that counsel’s performance may be deficient “if he ‘is on notice that his client may be mentally impaired,’ yet fails ‘to investigate his client’s mental condition as a mitigating factor in a penalty phase hearing.’” Caro v. Woodford, 280 F.3d 1247, 1254 (9th Cir. 2002) (quoting Hendricks v. Calderon, 70 F.3d 1032, 1043 (9th Cir. 1995)). Such performance is deficient because “[a]t the penalty phase, counsel’s duty to follow up on indicia of mental impairment is quite different from—and much broader and less contingent than—the more confined guilt-phase responsibility.” Bemore v. Chappell, 788 F.3d 1151, 1171 (9th Cir. 2015). The failure to “make even [a] cursory *22 investigation” into available means of obtaining additional funding for expert witnesses may amount to deficient performance under Strickland. See Hinton v. Alabama, 571 U.S. 263, 274 (2014). Additionally, the 1989 ABA Guidelines6 in effect at the time of Jones’s sentencing, explain that in capital cases, “[c]ounsel should secure the assistance of experts where it is necessary or appropriate for: . . .” “the sentencing phase of the trial,” and the “presentation of mitigation.” ABA Guidelines for the Appointment and Performance of Counsel in Death Penalty Cases, Guideline 11.4.1(d)(7), p. 16 (1989). The Guidelines explain that “[i]n deciding which witnesses and evidence to prepare for presentation at the sentencing phase, counsel should consider the following: . . . Expert witnesses to provide medical, psychological, sociological or other explanations for the offense(s)[.]” Id. at Guideline 11.8.3(F)(2), p. 23–24. 6 We may look to the ABA Guidelines as indicators of the prevailing norms of practice at a given time. See Strickland, 466 U.S. at 688 (“Prevailing norms of practice as reflected in American Bar Association standards and the like . . . are guides to determining what [performance] is reasonable, but they are only guides.”); see also Rompilla, 545 U.S. at 387 n.7 (using language of 1989 and 2003 ABA Guidelines to evaluate performance at 1988 trial); Florida v. Nixon, 543 U.S. 175, 191 (2004) (using 2003 ABA Guidelines to evaluate counsel’s performance at trial); but see Bobby v. Van Hook, 558 U.S. 4, 8 (2009) (“Strickland stressed, however, that American Bar Association standards and the like are only guides to what reasonableness means, not its definition. We have since regarded them as such.” (citations and quotations omitted)). — 23 — The Guidelines also note that, among the topics the defense should consider presenting at sentencing, is “[m]edical history (including mental and physical illness or injury, alcohol and drug use, birth trauma and developmental delays)” as well as “[e]xpert testimony concerning [the client’s medical history] and the resulting impact on the client, relating to the offense and to the client’s potential at the time of sentencing.” Id. at Guideline 11.8.6(B)(1)&(8), p. 25–26. “The timing of this investigation is critical.” Allen v. Woodford, 395 F.3d 979, 1001 (9th Cir. 2005) (quotation and citation omitted); see also Heishman v. Ayers, 621 F.3d 1030, 1036–37 (9th Cir. 2010). The Supreme Court has found constitutional error “where counsel waited until one week before trial to prepare for the penalty phase, thus failing to adequately investigate and put on mitigating evidence.” Allen, 395 F.3d at 1001 (citing Terry Williams, 529 U.S. at 395). “If the life investigation awaits the guilt verdict, it will be too late.” Id. (citation and quotation omitted). “[L]egal experts agree that preparation for the sentencing phase of a capital case should begin early and even inform preparation for a trial’s guilt phase[.]” Id. “Counsel’s obligation to discover and appropriately present all potentially beneficial mitigating evidence at the penalty phase should influence everything the attorney does before and during trial[.]” Id. (citation and quotation omitted). *24 Moreover, the 1989 ABA Guidelines state that “[c]ounsel should conduct independent investigations relating to the guilt/innocence phase and to the penalty phase of a capital trial[,]” and “[b]oth investigations should begin immediately upon counsel’s entry into the case and should be pursued expeditiously.” ABA Guidelines for the Appointment and Performance of Counsel in Death Penalty Cases, Guideline 11.4.1(A), p. 13 (1989) (emphasis added); see also id. at Guideline 11.8.3, p. 23 (“[P]reparation for the sentencing phase, in the form of investigation, should begin immediately upon counsel’s entry into the case.”). The state court record demonstrates that counsel’s failure to timely seek a mental health expert fell below “prevailing professional norms.” Strickland, 466 U.S. at 688. The state court record shows that counsel was on notice that Jones may have been mentally impaired, yet counsel failed to investigate Jones’s mental condition as a mitigating factor, and he failed to obtain a defense mental health expert. Counsel was in possession of medical records showing that Jones formerly attempted suicide at age twenty-two; Peggy told counsel that Jones experienced extreme moods swings, but these swings stabilized when he had been medicated with lithium; and Peggy and Randy told counsel that Jones was “often a disturbed child,” and they had to seek psychiatric help for him at age nine. This evidence *25 would have led a reasonable attorney to investigate further and obtain a defense mental health expert. See Wiggins, 539 U.S. at 527–28. An investigation into Jones’s mental health should have been pursued far in advance of when counsel requested that Jones undergo a mental health examination pursuant to Arizona Rule of Criminal Procedure 26.5. Counsel should have obtained a defense mental health expert well before the start of the guilt phase of Jones’s trial, but instead, he waited to make this request until after Jones had already been convicted on September 13, 1993. See Allen, 395 F.3d at 1001. Obtaining the court-appointed, independent expert’s short and cursory evaluation did not satisfy this duty. See Lambright, 490 F.3d at 1120–21. (“Counsel may not rely for the development and presentation of mitigating evidence on the probation officer and a court appointed psychologist. . . . The responsibility to afford effective representation is not delegable to parties who have no obligation to protect or further the interests of the defendant.” (emphasis added)). Moreover, Dr. Potts’s evaluation and opinions were limited in that he was a psychiatrist not trained in matters involving organic brain function—information that a neuropsychologist could have developed and presented. Dr. Potts had no obligation to further the interests of the defendant, even if he did present a defense-*26 favorable opinion, and his expertise and evaluation did not extend to the precise topic—organic brain function—that was essential in Jones’s case. Finally, the state court record establishes that the failure to obtain a defense expert here cannot be justified as a reasonable strategic decision. First and foremost, counsel’s failure to obtain a mental health expert was based not on strategy, but on lack of preparation, which left counsel unaware of the importance of this evidence. Counsel failed to speak adequately to Jones and Jones’s family to obtain a full picture of Jones’s mental health history. For instance, even though when Jones was interviewed for the PSR, Jones reported he was “mentally abused by his first step-father, and later physically abused by a second step-father,” and he characterized his childhood as “bad and unhappy,” when questioning Dr. Potts at sentencing, defense counsel brushed aside a mention of Randy’s physical abuse, referring to it as “clearly a mistake,” even though the information came from Jones himself. Given this lack of preparation, unsurprisingly, counsel stated that he never even considered the need for testing by a neuropsychologist until Dr. Potts suggested it to him the evening before Jones’s sentencing hearing. He attested that prior to meeting with Dr. Jack Potts, M.D. on December 7, 1993 to discuss his evaluation of Danny Jones, [he] was not aware that neurological or neuropsychological testing was necessary and — 27 — available which could determine the exact nature of injuries to Danny Jones’ brain from long term substance abuse and head injury and the resulting affect on his behavior and conduct. This failure fell below a reasonable standard of performance given the indications that Jones likely suffered from some form of mental illness. Although we need “not indulge ‘post hoc rationalization’ for counsel’s decisionmaking that contradicts the available evidence of counsel’s actions,” Harrington v. Richter, 562 U.S. 86, 109 (2011) (quoting Wiggins, 539 U.S. at 526–27), even imagining one potential strategic reason for counsel’s failure to obtain an expert—that the defense could not afford one—the failure to attempt to obtain a defense expert was neither reasonable nor informed. In the declaration he provided in the state PCR proceeding, defense counsel stated that he believed “the Mohave County Public Defender’s Office did not have sufficient monies for retaining expert witnesses,” and so he “did not ask Mr. Everett [the Mohave Public Defender] for any funding for additional necessary experts in State vs. Jones.” But according to Kenneth Everett’s affidavit provided to the state PCR court, “[i]n the last quarter of 1993, approximately Seven Thousand ($7,000.00) Dollars would have been available for experts . . . in regard to all cases that the Public Defender had in that last quarter, including the Danny Lee Jones case.” Everett also stated that during the relevant time period, counsel “perhaps could have expended *28 additional funds for experts for additional mitigation evidence,” although he did recognize that counsel needed “to be circumspect” about requesting such funds.7 But, contrary to the Supreme Court’s ruling in Hinton, counsel never even looked into requesting funding through the Public Defender’s Office. See Hinton, 571 U.S. at 274 (trial attorney’s failure to request additional funding was deficient when he mistakenly believed he had received all the funding available). In sum, the state court record demonstrates that trial counsel was constitutionally ineffective by failing to secure a defense mental health expert. Thus, pursuant to § 2254(d)(1), the Arizona Supreme Court’s contrary conclusion was an unreasonable application of Strickland and its progeny. 7 Counsel could also have gone back to the trial court for additional funding. In granting only $2,000 of counsel’s $5,000 request for funding, the court stated that: If this is all you need pretrial, you may need more at trial, and then of course the sentencing hearing if we get that far, so—but, I am willing to go $2,000 prior to trial, and then with the understanding that I am willing to listen again if you need more. Jones, 583 F.3d at 629 n.2. Although this statement may not have been specifically included in the state court record, there is no doubt the PCR court was aware of it; the same judge who sentenced Jones to death presided at his PCR hearing. — 29 — B Alternatively, Jones argues that the state PCR court’s decision was based on an unreasonable determination of the facts under § 2254(d)(2). He argues that the court employed a defective fact-finding process with respect to Claim 1 when it denied PCR counsel’s funding request for a defense neuropsychological expert, effectively preventing the factual development of this claim. He also asserts that the state court’s failure to hold a hearing on Claim 1 resulted in an unreasonable determination of the facts. We agree with both arguments. Under § 2254(d)(2), a petitioner may challenge a state court’s conclusion that is based upon an unreasonable determination of the facts. We have noted that § 2254(d)(2) challenges “come in several flavors.” Taylor v. Maddox, 366 F.3d 992, 1000 (9th Cir. 2004), overruled on other grounds by Murray v. Schriro, 745 F.3d 984, 999–1000 (9th Cir. 2014). For instance, we have stated that a petitioner may overcome the § 2254 (d)(2) bar if the fact-finding “process employed by the state court is defective.” Id. at 999 (citing Nunes v. Mueller, 350 F.3d 1045, 1055–56 (9th Cir. 2003)). “We have held repeatedly that where a state court makes factual findings without an evidentiary hearing or other opportunity for the petitioner to present evidence, the fact-finding process itself is deficient, and not entitled to deference.” Hurles v Ryan, 752 F.3d 768, 790 (9th Cir. 2014) *30 (amended) (quotations omitted); see also Perez v. Rosario, 459 F.3d 943, 950 (9th Cir. 2006) (amended) (“In many circumstances, a state court’s determination of the facts without an evidentiary hearing creates a presumption of unreasonableness.”). This rule applies with greater force where a judge bases factual findings on their own personal conduct, untested memory, or understanding of events in the place of an evidentiary hearing. See Hurles, 752 F.3d at 791 (finding it “especially troubling” when a judge’s factual findings involved her own conduct and were based on her “untested memory and understanding of the events”). Similarly, a fact-finding process may be fatally undermined “where the state courts plainly misapprehend or misstate the record in making their findings, and the misapprehension goes to a material factual issue that is central to petitioner’s claim.” Taylor, 366 F.3d at 1001; see also Wiggins, 539 U.S. at 528. And likewise, a fact-finding process may be deemed defective when the end result requires the court to make a finding on “an unconstitutionally incomplete record.” Milke v. Ryan, 711 F.3d. 998, 1007 (9th Cir. 2013). For a petitioner to prevail on these types of § 2254(d)(2) arguments, however, “we must be satisfied that any appellate court to whom the defect is pointed out would be unreasonable in holding that the state court’s fact-finding process was adequate.” Taylor, 366 F.3d at 1000. — 31 — The PCR court’s decision not to hold a hearing on Claim 1 amounted to an unreasonable determination of the facts. The court ruled on Claim 1 without holding an evidentiary hearing because it found that Dr. Potts essentially satisfied the role of a defense mental health expert. In response to PCR counsel’s argument that a defendant is entitled to his own mental health expert in capital cases, not a court-appointed independent expert, the court explained that: Dr. Potts was a very good expert. He was defense oriented. The prosecutor, I can remember, was very upset about that. . . . I’m going to deny [this claim] because I don’t think counsel was ineffective as far as Dr. Potts. The fact that the PCR court made this factual finding regarding Dr. Potts’s role without holding an evidentiary hearing or opportunity for Jones to present evidence, suggests that the PCR court’s fact-finding process was deficient. See Hurles, 752 F.3d at 790. However, the process was even more unreasonable because, even though more than six years had passed, the judge based this finding solely on his own untested memory and personal impression of Dr. Potts’s role in the sentencing hearing. See id. at 791. The judge who presided over Jones’s state PCR proceeding was the same judge who sentenced him to death, and in denying a hearing on this claim, the judge relied primarily on his personal recollection of Dr. Potts’s testimony and his memory that the prosecution was upset that Dr. Potts *32 testified favorably for the defense. There is no evidence the PCR court considered anything else in denying the request for a hearing. The PCR court “plainly misapprehend[ed]” the record in making its finding that Dr. Potts satisfied the role of a defense mental health expert. See Taylor, 366 F.3d at 1001. Dr. Potts was not a defense expert, and the fact his conclusions were favorable to the defense does not support that he filled that role. Nothing about the circumstances of Dr. Potts’s testimony suggests otherwise. Dr. Potts testified at Jones’s sentencing hearing that he regularly prepares psychological reports requested by the courts, that “at times are favorable apparently for the State” and at other times are favorable “for the defense[.]” He also explained that in other cases like Jones’s, he had found little or no mitigation for the defendant. Moreover, the limited amount of time Dr. Potts spent on his report and the level of analysis and detail that report provided do not support the conclusion that he was an advocate for the defense team. Dr. Potts submitted only a six-page report to the court. He agreed that he was under “significant time pressure” in preparing the report because he received the PSR late from Mohave County. He met with Jones for a total of four hours at the prison, and spent one and a half hours of that time administering an MMPI personality test. On the day before he testified, Dr. Potts *33 also spoke to Jones for “a couple of hours,” Peggy for about thirty minutes, and Randy for one hour. Dr. Potts also specified that it would have been helpful and valuable to have had some neurologic evaluations, not – by a neurologist, clinical exam, such as a CAT scan, possibly an MRI, possibly EEG, possibly some sophisticated neurological testing, because I think there’s very strong evidence that we have – well, there’s clear evidence that we have, I believe, of traumatic brain injury, and there’s some other evidence that I believe we may have organic neurological dysfunctions here that has gone on since he’s been about 13. So there’s some other testing that I think would be valuable to have to pin down the diagnosis. Nothing about Dr. Potts’s role in the sentencing hearing suggests that he had stepped into the shoes of a defense expert. The PCR court’s decision not to fund a defense mental health expert fatally undermined the fact-finding process, in part because that decision resulted in the court ruling on an unconstitutionally incomplete record. Without funding for a mental health expert, it was impossible for Jones to demonstrate that he had been prejudiced by counsel’s failure to obtain one during the course of Jones’s criminal proceedings. Jones could not demonstrate the inadequacy of counsel’s mitigation case without providing the mitigation evidence that could have been presented by a defense neuropsychological expert. Moreover, without funding, Jones could not show that a defense neuropsychological expert would have presented materially different evidence than that already provided by Dr. Potts. By failing to provide *34 additional funding to develop Jones’s mental health mitigation evidence, the state court, as Jones phrases it, created “its own self-fulfilling prophecy,” by preventing the development of the claim before it was even presented. We emphasize that we are not suggesting that any denial of an evidentiary hearing or denial of funding for an expert would lead to a deficient fact-finding process in state court. Our determination is expressly limited to the facts of this case, where, as described above, the court denied the evidentiary hearing based on his own recollection of a sentencing proceeding six years prior, where the state’s own expert had opined on the stand that further neurological testing was desirable. For these reasons, we conclude that any appellate court would conclude that the PCR court’s factual determination as to Dr. Potts and its fact-finding process with respect to Claim 1 were unreasonable and inadequate. See Taylor, 366 F.3d at 1000. Accordingly, Jones has satisfied the requirements of § 2254(d)(2). C Although § 2254(d) typically also applies to the prejudice prong of a petitioner’s IAC claim, here, the PCR court did not reach the issue of prejudice, and so we review the issue de novo. See, e.g., Weeden, 854 F.3d at 1071. Pinholster limits our § 2254(d) analysis to the facts in the state court record. 563 U.S. at 185. However, Pinholster does not prevent us from considering *35 evidence presented for the first time in federal district court in reviewing the merits of Jones’s claims de novo. As the district court found, Jones satisfied the standard for an evidentiary hearing pursuant to § 2254(e)(2)8. That provision permits federal district courts to hold evidentiary hearings and consider new evidence when petitioners have exercised diligence in pursuing their claims in state court. See id. (“Section 2254(e)(2) continues to have force where § 2254(d)(1) does not bar federal habeas relief.”); see id. at 212–13 (Sotomayor, J., dissenting); (Michael) Williams v. Taylor, 529 U.S. 420, 436–37 (2000). Though § 2254(e)(2) limits the discretion of district courts to conduct evidentiary hearings, Pinholster, 563 U.S. at 203 n.20, it does not impose an express limit on “evidentiary hearings for petitioners who ha[ve] been diligent in 8 Section 2254(e)(2) states that If the applicant has failed to develop the factual basis of a claim in State court proceedings, the court shall not hold an evidentiary hearing on the claim unless the applicant shows that— (A) the claim relies on— (i) a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable; or (ii) a factual predicate that could not have been previously discovered through the exercise of due diligence; and (B) the facts underlying the claim would be sufficient to establish by clear and convincing evidence that but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense. — 36 — state court.” Id. at 213 (Sotomayor, J. dissenting); see also Landrigan, 550 U.S. at 473 n.1. Here, the federal district court determined that Jones had been diligent in attempting to develop the factual basis for Claims 1 and 2 in state court, and the State does not contest that determination now. The state court record shows that Jones was diligent. His PCR counsel requested funding for a neuropsychologist and “a thorough and independent neurological assessment” to assist in the development of Claims 1 and 2, but the PCR court denied the request. Therefore, the district court did not err in its expansion of the record, and we consider the evidence developed in federal district court in conducting de novo review of Jones’s claims. To prevail on his IAC claim, Jones must demonstrate that his trial counsel: (1) performed deficiently; and (2) Jones’s defense was prejudiced by that deficient performance. Strickland, 466 U.S. at 687. He has done so. For all the reasons set forth previously in our § 2254(d)(1) analysis, Jones has demonstrated that counsel’s performance fell below an objective standard of reasonableness and below the prevailing professional norms at the time of Jones’s proceedings. Additionally, Jones has demonstrated that counsel’s failure to obtain a defense mental health expert for the penalty phase of Jones’s trial prejudiced the *37 defense. Jones has demonstrated that there is a “reasonable probability” that had such an expert been retained, “the result of the proceeding would have been different.” Id. at 694. There is a reasonable probability that had counsel secured a defense mental health expert, that expert would have uncovered (and presented at sentencing) a wealth of available mitigating mental health evidence.9 The main mitigation witness in state court was Randy, Jones’s second step-father. Randy erroneously testified that Jones enjoyed a stable home life after age seven, when Randy married Jones’s mother, and yet the trial court found that his testimony was sufficient to prove a number of non-statutory mitigating circumstances. Had counsel secured a mental health expert, that expert could have provided substantial evidence—through neuropsychological testing or otherwise—that Jones suffered from mental illness, including evidence supporting any of the diagnoses made by experts in federal district court: (1) cognitive dysfunction (organic brain damage 9 The Arizona Supreme Court concluded that Jones’s chaotic and abusive childhood and mental illness, discussed below, did not constitute mitigating factors because Jones failed to demonstrate a connection to his conduct on the day of the murders. But this rationale for discrediting Jones's mitigating evidence was contrary to clearly established Supreme Court precedent. The Supreme Court has held that a sentencer in a capital case may not “refuse to consider, as a matter of law, any relevant mitigating evidence” offered by the defendant. Eddings v. Oklahoma, 455 U.S. 104, 114 (1982) (emphasis omitted); see McKinney v. Ryan, 813 F.3d 798, 811–12 (9th Cir. 2015). — 38 — and a history of numerous closed-head injuries); (2) poly-substance abuse; (3) post-traumatic stress disorder (“PTSD”); (4) attention deficit/hyperactivity disorder (“AD/HD”); (5) mood disorder; (6) bipolar depressive disorder; and (7) a learning disorder. The experts retained in Jones’s federal habeas proceedings provided significant evidence of these conditions, demonstrating that such evidence could have been uncovered and presented at sentencing. Dr. Pablo Stewart, Chief of Psychiatric Services at the Haight Ashbury Free Clinic in San Francisco, California, testified at the federal evidentiary hearing. He estimated that he spent 130 hours working on Jones’s case, in contrast to the four hours Dr. Potts was able to spend with Jones prior to sentencing. Dr. Stewart diagnosed Jones with cognitive dysfunction, PTSD, polysubstance abuse, and mood disorder. He ultimately concluded that “[t]he circumstances surrounding Mr. Weaver’s death are a direct consequence of [Jones’s] abused and unfortunate past.” Dr. Stewart testified to a number of factors that may have contributed to Jones’s cognitive dysfunction that occurred before Jones was even born. He noted that Jones’s mother, Peggy, worked in a chrome hub cab plating factory when she was pregnant with Jones, and chrome exposure may negatively affect a baby’s birth. He testified that Peggy’s prenatal diet was also of concern: Peggy reported that during her pregnancy with Jones, her diet consisted of cigarettes, coffee, and *39 mayonnaise sandwiches. He expressed that the use of nicotine during pregnancy has been directly linked to cognitive dysfunction in children and caffeine exposure results in more difficult births. He also specified that Jones’s father beat Peggy during her pregnancy such that there was a potential for physical trauma to the fetus. And he testified to Jones’s traumatic and difficult birth: Jones was born in the breech position, with the umbilical cord wrapped around his neck, and forceps were used. He testified that any of these factors could have been potential contributors to Jones’s cognitive dysfunction. Dr. Stewart testified that Jones had suffered multiple serious head injuries over the course of his life, and he went into much greater detail than Randy had provided regarding Jones’s head injuries at the state PCR proceeding. Randy had testified that Jones fell off a roof when he was approximately thirteen, fell off a scaffolding when he was approximately fifteen, was mugged while serving in the Marines, and experienced spontaneous blackouts around the age of four. By contrast, Dr. Stewart described an incident where Jones “was about eleven (11) years old, he fell, head-first off a roof onto the metal frame of a horizontal dolly, in an attempt to retrieve a ball. His eye hit the metal bar of the dolly. He was unconscious for about five (5) to ten (10) minutes.” He also noted the fall when Jones was fifteen, but additionally, he explained that “[a]s a young adult, Danny *40 had at least three (3) car accidents where he lost consciousness.” Further, “when Danny was about five-and-a-half (5 ½) years old, Peggy found Danny regaining consciousness, lying underneath the swing set. She suspected Eland, Danny’s first step-father, had hit Danny or thrown him off the slide. Danny’s face was red and he vomited, indicating he had a concussion.” Dr. Stewart elaborated on the mugging Jones suffered while in the Marines: he was “found lying unconscious in a ditch along the highway, by a Morehead City Police Officer, who took him to the hospital. Danny had been mugged and beaten with a two-by-four.”10 Dr. Stewart discussed PTSD and explained that Jones suffered numerous traumatic experiences early in his life: he watched his first step-father hold a jigsaw to his mother’s neck and threaten to kill her, he watched that same step-father shoot a gun at his mother, and on two separate occasions, his second step-father, Randy, pointed a gun to his own head and threatened to kill himself in front of Jones. He also noted that Randy beat Jones for no reason with a belt with a buckle and engaged in other forms of severe physical discipline. On cross-examination the state challenged Dr. Stewart’s PTSD diagnosis because Dr. Stewart stated that Jones “had PTSD at the time of the murders,” but 10 Jones self-reported this incident, including to Dr. Stewart and Dr. Scialli, and corresponding medical records show treatment for a head abrasion, but the district court found there was no evidence of acute trauma. — 41 — did not state that Jones was having a flashback while committing the crimes. Dr. Stewart responded by explaining that while the media tends to show PTSD as being a person “who is thinking he’s being ambushed and he takes people hostage,” that only occurs “very, very rarely.” He explained: The much more overwhelmingly more common thing that occurs is a person having a short fuse; a person overreacting to a situation; a person finding themselves challenged by some things and then just going off; a person—and that’s PTSD. A person who drinks too much and then gets into fights, those are the more common thing. But those don’t sell movies or books. But that’s the more common presentation. So that’s why I’m saying in the case of Mr. Jones, it’s absolutely clear that he suffers from PTSD, in my opinion, and that he carries that with him throughout his entire life. Certainly on the day of these murders, that was going on. Dr. Stewart also testified that Jones’s first step-grandfather forced Jones to drink alcohol when he was only nine years old, and that it appeared the grandfather used alcohol to get Jones drunk so it would be easier to sexually abuse him. Dr. Stewart described the sexual abuse “as full contact sexual abuse, including sodomy, including oral sex, both the providing it and receiving it.” Jones became a daily marijuana user when he was in junior high, he used one gram of cocaine every weekend in high school, and he reported using LSD two hundred times. Dr. Stewart explained that the substance abuse appeared to have stemmed from Jones’s *42 genetic predisposition, and also because Jones used drugs starting at a very young age to self-medicate as a means of coping with his mental defects and past trauma. The district court did not credit or discuss Dr. Stewart’s testimony that Jones suffers from cognitive dysfunction, apparently because the court doubted Dr. Stewart’s credibility based on his purported “willingness to present an opinion on a factual issue”—specifically, Jones’s allegation that someone named “Frank” killed Tisha. The district court mistakenly stated that Dr. Stewart opined on the viability of Jones’s theory that a third party had been involved in the crime. In fact, Dr. Stewart opined that Jones had not previously mistreated or abused any child, that Jones had a “history of submissive, almost child-like behavior, against older males,” and that Jones’s “psychological profile supports the events as described by [Jones] on the night of the crimes.” Dr. Stewart thus observed that killing a child is not consistent with Jones’s psychological profile. When pressed on cross-examination whether he “believed” Jones’s version of events over the jury verdict, Dr. Stewart again affirmed his assessment that Jones’s story was consistent with Jones’s psychological profile. Dr. Stewart did not purport to contradict the jury’s findings. Dr. Alan Goldberg, a psychologist in Arizona with a speciality in neuropsychology, conducted a battery of tests that covered multiple domains of *43 cognitive functioning. Dr. Goldberg gave Jones approximately twenty-five tests and found that “when we look at the patterns across many different kinds of tests . . . we see a consistent inconsistency in performance, that is, the performance is problematic on a number of tests that all have an attention component to them.” He ultimately diagnosed Jones with a learning disability, attention deficit disorder, and “Bipolar Disorder, Depressed.” Jones submitted reports from additional experts, including Dr. David Foy, a professor of psychology at Pepperdine University. Dr. Foy diagnosed Jones with PTSD, polydrug abuse, depressive disorder, compromised cognitive emotional functioning, and various learning deficits. Dr. Foy described the numerous instances of life-threatening family violence Jones witnessed growing up, and found that on at least two occasions Jones used a baseball bat to protect himself: (1) when Jones threatened to kill Jones’s first step-father if he did not stop beating Jones’s mother, Peggy; and (2) in order to stop Jones’s first step-grandfather from continuing to sexually abuse him. Dr. Foy concluded that [t]he constant threat of sudden verbal attacks or severe physical punishment in Danny’s home environment would be expected to produce an essential state of wariness or hypervigilance . . . [and] would be expected to lead to a heightened suspiciousness and combat readiness as a systematic way of responding, even in situations which later proved to be non-life threatening. — 44 — Finally, Jones submitted a declaration from his younger sister, Carrie. She said that, as a child, Jones twice watched Randy point a gun at his own head and threaten to kill himself. She stated that contrary to Randy’s testimony during sentencing, Randy was both verbally and physically abusive to Jones, and that Jones threatened to kill Randy if he kept beating Peggy. Carrie also confirmed that Jones suffered numerous head injuries while growing up. The State called three experts to testify in response at the federal evidentiary hearing. Dr. Herron treated Jones from 2003 to 2005 for depression and anxiety, and stated that he believed that the bipolar diagnosis was reasonable, but that he detected no signs of neurological dysfunction, cognitive impairment, or PTSD, though he could not rule out those conditions. Dr. Herring, a clinical neuropsychologist, interviewed and tested Jones. Based on those results, she determined that Jones does not suffer from cognitive impairment or AD/HD. Dr. Scialli testified that based on his evaluation, he could diagnose Jones with alcohol, amphetamine, and cannabis dependence and AD/HD residual type at the time of the murders. He disputed the PTSD diagnoses of Jones’s experts, as there was no indication that Jones re-experienced the traumatic event at the time. He explained that phrases such as “cognitive dysfunction” or “cognitive impairment” are not diagnostic definitions, but are used “idiosyncratically” as “terms of art” with no *45 fixed meaning, and he asserted Jones could not be classified under any category of cognitive dysfunction as defined by the Diagnostic and Statistical Manual of Mental Disorders. He also testified that although he diagnosed Jones with AD/HD residual type, there is no link between AD/HD and violent behavior. Dr. Potts also testified and explained that he had not been tasked with providing mitigation evidence at sentencing and had not conducted the extensive testing he felt was required. Potts explained: “I was not an expert for either party. I was the Court’s expert in looking at some issues. I was not—it was clear I was not hired for mitigation, nor was I hired for aggravation.” The trial court had ordered Dr. Potts to perform an evaluation for the court pursuant to Rule 26.5 of the Arizona Rules of Criminal Procedure. In accord with this role limitation, Dr. Potts had testified at Jones’s sentencing: “My main role is working with Maricopa County Superior Court, criminal division, coordinating competency evaluations, other forensic, and working with patients. I also have clinical responsibilities. . . . [I] [p]rimarily do reports as requested by the Court.” When asked whether or not he had enough “points of data” to pull from in reaching his conclusions, he stated: [T]here’s a clear distinction between a mitigation specialist, and I’m no mitigation specialist. I may be a part of a team of mitigation, but I’m clearly not a mitigation specialist in the realm of what is dealt with now in capital cases. . . . — 46 — Mine was a cursory examination. . . . [I]nterviewing one family member certainly is not adequate, I believe, for what would be considered capital mitigation. It is below the standard of care. He stated that, prior to his testimony at sentencing, he had recommended that defense counsel seek neuropsychological testing for Jones. During cross-examination, the State tried to get Dr. Potts to admit that he only called for neurological testing, not neuropsychological testing, but Dr. Potts explained that “[s]ophisticated neurological testing would include that.” He described the reports submitted by the additional experts at the habeas proceeding as the “documents I think one would expect to see in mitigation. . . . I believe they’re very, very helpful, and I think—I know I would have liked to have had the exhaustive nature of these reports.” He stated that he found his role constrained by his court-appointed status, and therefore “did not make diagnoses,” because his “role was not to make diagnoses . . . and that’s why I would not have. I could have . . . but that was not the nature or tenor of any of this report . . . .” Defense trial counsel testified that Dr. Potts “did not act as a neutral, detached court-appointed expert. He actively assisted us in developing mitigation, planning strategy to a much larger degree than what he indicated.” He explained that he had “numerous phone conversations” with Dr. Potts, they met together the *47 night before Dr. Potts testified, and Dr. Potts “stressed to ask for the continuance for the additional testing.” Notwithstanding the defense lawyer’s testimony that Dr. Potts “did not act as a neutral, detached court-appointed expert,” and “actively assisted in developing mitigation, planning strategy,” the district court’s conclusion that Dr. Potts was a “de facto defense expert” was clearly erroneous. As a court-appointed expert, Dr. Potts’ findings were not confidential, he had “no obligation to protect or further the interests of the defendant,” Lambright, 490 F.3d at 1121, and he did not sufficiently “assist in evaluation, preparation, and presentation of the defense,” Ake, 470 U.S. at 83. Dr. Potts stated repeatedly that he was a neutral expert and found his role limited as such. We need look no further than the defense lawyer’s treatment of Dr. Potts to recognize that he was not prepared, nor thought of, as a defense expert. As we have recognized, the “duty to provide the appropriate experts with pertinent information about the defendant is key to developing an effective penalty phase presentation.” Caro, 280 F.3d at 1255 (citation omitted). A review of Dr. Potts’ entire case file shows that only 31 pages came from the defense lawyer. The defense lawyer even admitted at the evidentiary hearing in the district court that *48 because Dr. Potts was a court-appointed expert, he specifically chose not to send Dr. Potts all available mitigation evidence: THE COURT: So before Dr. Potts issued his report that turned out to be favorable to you, would you necessarily have provided all of the information you had collected on Mr. Jones? WITNESS: No, I would not, because it was – after he got started, it was apparent he wanted to help us. And at that point, then, his role, I supposed, changed. And then our relationship was a little different. Initially, I would screen what I sent him, because I honestly didn’t know what we were going to get from him. Dr. Potts’ report was dated December 3, 1993, and the defense lawyer did not get the report until two days before the sentencing hearing started on December 8, 1993. Until at most two days before the sentencing hearing, nobody considered Dr. Potts as a defense expert, and the defense lawyer had only given Dr. Potts 31 pages to review. As we held in Bean v. Calderon: “When experts request necessary information and are denied it, when testing requested by expert witnesses is not performed, and when experts are placed on the stand with virtually no preparation or foundation, a capital defendant has not received effective penalty phase assistance of counsel.” 163 F.3d at 1079. — 49 — The district court also concluded that no prejudice resulted because the State’s experts were more credible than the petitioner’s. The district court erred in doing so. It was improper for the district court to weigh the testimony of the experts against each other in order to determine who was the most credible and whether Jones had presented “evidence confirming that [he] suffers from neurological damage caused by head trauma or other factors.” We have held that a district court should not independently evaluate which expert was most believable or try to find a definitive diagnosis: The district court dismissed evidence of Correll’s brain injury, concluding that any organic brain injury played no role in Correll’s crimes. The district court’s conclusion was based on the judge’s own evaluation of two conflicting experts. But in the procedural context of this case, the district court’s role was not to evaluate evidence in order to reach a conclusive opinion as to Correll’s brain injury (or lack thereof). The district court should have decided only whether there existed a “reasonable probability” that “an objective fact-finder” in a state sentencing hearing would have concluded, based on the evidence presented, that Correll had a brain injury that impaired his judgment at the time of the crimes. Correll v. Ryan, 539 F.3d 938, 952 n.6 (9th Cir. 2008) (citation omitted). This is not to say, of course, that a district court is prohibited from making credibility determinations. However, the ultimate focus is on whether the new evidence was “sufficient to undermine confidence in the outcome” which is *50 different from “the preponderance more-likely-than-not standard.’” Lambright, 490 F.3d at 1121 (quoting Summerlin, 427 F.3d at 640). The district court also concluded that Jones’s mental conditions “do not constitute persuasive evidence in mitigation because they do not bear a relationship to Petitioner’s violent behavior.” However, if the sentencing court had decided not to consider the mitigating mental condition evidence, it would have run afoul of Eddings, which held a sentencer in a capital case may not “refuse to consider, as a matter of law, any relevant mitigating evidence” offered by the defendant. 455 U.S. at 114. The testimony provided at the federal evidentiary hearing demonstrates the types of mitigation evidence that could and should have been presented at the penalty phase of Jones’s trial. For instance, the evidence demonstrates that, had counsel retained a defense mental health expert, that expert could have provided testimony explaining the factors that contributed to Jones’s cognitive dysfunction, including: (1) prenatal chrome and nicotine exposure; (2) his mother’s malnutrition during pregnancy; (3) fetal trauma from beatings by his father; (4) a traumatic birth; (5) several severe head injuries; or (6) Jones’s substantial and extensive drug and alcohol abuse, which began when he was eight or nine years old. Any such evidence would have been significantly more probative of Jones’s mental state and *51 more persuasive in reducing Jones’s culpability than Dr. Potts’s conditional findings, compiled after far less preparation time and testing, and comprising only a six-page report. These factors illustrate how unfortunate circumstances outside of Jones’s control combined to damage his cognitive functioning and mental health at the time of his crimes. Likewise, the mental health experts’ testimony in the district court proceedings demonstrates that had trial counsel retained such an expert for sentencing, he or she could have provided evidence that Jones’s mental state was impaired by drugs and alcohol at the time of his crimes. He or she also could have offered context for his substance abuse and insight into how Jones’s long-term self-medication affected his brain. As demonstrated at the federal evidentiary hearing, any mental health expert engaged by the defense team would have attempted to explain Jones’s lifelong history of substance abuse and its physical effects on Jones’s brain. This would have included compiling a family history and hard data regarding Jones’s brain function. It also would have included information addressing how and when Jones’s substance abuse began. As the federal proceedings revealed, Jones turned to substance abuse at an extremely young age in order to self-medicate in response to the trauma he experienced from being physically and sexually abused and as a result of repeatedly witnessing *52 violence directed at his mother. A mental health expert would have relayed that Jones suffered sexual abuse from age nine until age thirteen at the hands of his step-grandfather, who introduced him to marijuana and alcohol at age nine in order to facilitate that abuse. A mental health expert could also have explained the trauma Randy inflicted on Jones by detailing how Randy physically and emotionally abused Jones, engaged in various forms of severe physical discipline, and threatened suicide in front of Jones and his family. To be sure, the expert testimony was not wholly one sided. The State’s experts disputed some of the diagnoses. For example, Dr. Scialli disagreed with Dr. Stewart’s PTSD diagnosis and noted there was no sign that Jones was re-experiencing a trauma at the time of the murders. But a conclusive diagnosis was not necessary for a sentencer to consider the wealth of evidence that Jones suffered from some form of mental illness and how that illness contributed to his commission of the crimes. Testimony explaining Jones’s history would have significantly impacted the overall presentation of Jones’s culpability with respect to his mental state, and painted a vastly different picture of Jones’s childhood and upbringing. The mitigation case actually presented to the sentencing court suggested that while Jones had undergone a traumatic early childhood, he enjoyed a largely normal *53 childhood and supportive family after the age of six. And because so little preparation had been done, Dr. Potts erroneously testified at sentencing that Jones did not suffer child abuse once Randy and Peggy married. Notably, Randy was the only mitigation witness who testified at Jones’s sentencing, and defense counsel was unaware that Randy too was an abuser. Had counsel procured a mental health expert, the mitigation case would have told the story of an individual whose entire childhood was marred by extreme physical and emotional abuse, which in turn funneled him into early onset substance abuse that exacerbated existing cognitive dysfunction. In short, the sentencing judge “heard almost nothing that would humanize [Jones] or allow [him] to accurately gauge his moral culpability.” Porter v. McCollum, 558 U.S. 30, 41 (2009). The mitigating evidence would have been powerful in painting a complete portrait of Jones’s life.11 Without it, the sentencing judge had little to counterbalance the aggravating factors. 11 See Penry v. Lynaugh, 492 U.S. 302, 319 (1989) (“‘[E]vidence about the defendant’s background and character is relevant because of the belief, long held by this society, that defendants who commit criminal acts that are attributable to a disadvantaged background . . . may be less culpable.’”), abrogated on other grounds by Atkins v. Virginia, 536 U.S. 304 (2002); Eddings, 455 U.S. at 113–15 (explaining that consideration of the offender’s life history is a “part of the process of inflicting the penalty of death” (citation omitted)); and McKinney, 813 F.3d at 821. — 54 — Indeed, the decision of life or death was given to the sentencing judge with a false picture of Danny Jones’s life. The sentencing judge had no idea any physical abuse lasted past Jones’s sixth birthday, and he had no idea that the tremendous abuse likely played a key role in leading Jones down the path of polysubstance abuse that he ultimately traveled. “This evidence adds up to a mitigation case that bears no relation to the few naked pleas for mercy actually put before the jury . . . .” Rompilla, 545 U.S. at 393.12 “We have held . . . that a defendant was prejudiced when, although counsel introduced some of the defendant’s social history, he did so in a cursory manner that was not particularly useful or compelling.” Stankewitz v. Woodford, 365 F.3d 706, 724 (9th Cir. 2004) (citation and quotation marks omitted) (“Stankewitz I”); see also Bean, 163 F.3d at 1081 (“[T]he family portrait painted at the federal habeas hearing was far different from the unfocused snapshot handed the superior court jury.”). Here, as in Stankewitz I, “[a] more complete presentation, including even a fraction of the details [Petitioner] now alleges, could have made a difference.” 365 F.3d at 724. 12 “Although, for the purposes of resolving this issue, we evaluate prejudice in the context of judge-sentencing, the result is the same.” Summerlin, 427 F.3d at 643. — 55 — As part of our prejudice analysis, we must “reweigh the evidence in aggravation against the totality of available mitigating evidence,” Wiggins, 539 U.S. at 534. Here, the sentencing court found that (1) Jones “committed the offense as consideration for the receipt, or in expectation of the receipt of anything of pecuniary value”; (2) Jones “committed the offenses in an especially heinous or depraved manner”; and (3) Jones was “convicted of one or more other homicides . . . which were committed during the commission of the offense.”13 On de novo review, we must weigh these factors against the mitigation evidence, as developed in the state court record that was available, but not presented. We also consider the mitigation evidence that was actually presented. The only mitigation witness in this case was Jones’ second step-father Randy Jones. Randy testified, second-hand, about Jones’s birth, abuse at the hands of his 13 Notably, while the Arizona Supreme Court affirmed the trial court’s finding that Jones’s crimes were for pecuniary gain, the Court also found that “Tisha, a 7-year-old child, did not present an obstacle to defendant’s goal of taking Robert’s guns.” And the Arizona Supreme Court reversed for lack of evidence on the trial court’s finding that Jones killed Tisha to eliminate her as a witness. Plainly, it was not necessary to kill a seven-year-old girl or her grandmother in order to steal guns, but these senseless murders are consistent with an outburst by someone suffering from organic brain injuries and other serious medical disorders. Thus, evidence of Jones’s cognitive dysfunction, his childhood, and his upbringing are especially relevant to his culpability and our ultimate determination that there is a reasonable probability that such evidence may have affected the sentencing court’s determination. — 56 — first stepfather and his drug abuse. Randy omitted his own physical and mental abuse of Jones, which started when Jones was six, and Jones’s sexual abuse by his step-grandfather. The available evidence that was not presented included numerous neurological disorders, including brain damage, and an extraordinary abusive childhood. The available additional mitigating evidence was powerful, Therefore, “reweigh[ing] the evidence in aggravation against the totality of available mitigating evidence,” we conclude there is at least a reasonable probability that development and presentation of mental health expert testimony would have overcome the aggravating factors and changed the result of the sentencing proceeding. Pinholster, 563 U.S. at 198 (quoting Wiggins, 539 U.S. at 534). Under Arizona law, these “mitigating circumstances [were] sufficiently substantial to call for leniency.” Ariz. Rev. Stat. § 13-703(E) (1993). Our conclusion is supported by the Strickland prejudice analysis conducted by the Supreme Court and our court in similar cases. In Porter, which involved a double homicide, the aggravating factor for which the defendant was sentenced to death was that the murder was “committed in a cold, calculated, and premeditated manner.” Porter, 558 U.S. at 32–33, 42. The mitigation defense was similar to the one presented in Jones’ case. Id. at 33–36. However, given the available mitigating evidence, including abuse, *57 troubled family history, and evidence of mental disorders, the Supreme Court concluded that Strickland prejudice existed, even affording AEDPA deference to the state court’s determination of lack of prejudice. Id. at 41–44. Terry Williams involved a homicide preceded by a prior armed robbery and burglary, and succeeded by auto thefts, and two violent assaults on elderly victims, one of whom was left in a vegetative state. Terry Williams, 529 U.S. at 368. However, the Supreme Court concluded that failure to investigate and present evidence of family abuse and mental capacity was sufficient to satisfy the Strickland prejudice standard. Id. at 395, 398. Rompilla involved a capital case in which the jury found that the murder was committed by torture, that Rompilla had a significant history of felony convictions indicating the use or threat of violence, and that the murder was committed in the course of another felony. Rompilla, 545 U.S. at 378. Nonetheless, the Supreme Court granted relief under Strickland, concluding that counsel had failed to investigate and present evidence of family history, substance abuse and the circumstances underlying the petitioner’s criminal history. Id. at 382, 390–393. In Wiggins, the Supreme Court granted relief because mitigating evidence such as severe family abuse, sexual abuse in foster care, and diminished mental capability was not presented to the jury. Wiggins, 539 U.S. at 535. — 58 — Penry involved a brutal rape and murder, where the jury found that the act “was committed deliberately and with the reasonable expectation that the death of the deceased or another would result” and that there was “a probability that the defendant would commit criminal acts of violence that would constitute a continuing threat to society.” 492 U.S. at 307, 310. Nonetheless, the Supreme Court granted relief because the defendant was unable to introduce evidence of mental retardation and family abuse as mitigating factors. Id. at 340. Our case law has yielded similar results. Sanders involved a brutal assault and homicide. Sanders, 23 F.4th at 970–71. However, despite the nature of the crime, and prior convictions, we held that failure to investigate and present evidence of social history and mental health evidence satisfied the Strickland prejudice standard. Id. at 985–86, 995. In Andrews v. Davis, 944 F.3d 1092, 1100 (9th Cir. 2019), the petitioner, who had previously been convicted of another murder, was convicted of a triple homicide and rape. However, we concluded that failure to investigate and present evidence of social history, family background, abuse, and mental health was sufficient to satisfied the Strickland prejudice standard. Id. at 1121. In Stankewitz v. Wong, 698 F.3d 1163 (9th Cir. 2012), we held that the petitioner had established Strickland prejudice when counsel failed to investigate *59 and present mitigating evidence, which included a history of family abuse, anger management issues, and substance abuse. Id. at 1168–69, 1176. Avena v. Chappell, 932 F.3d 1237, 1252 (9th 2019) involved the commission of “two brazen murders during a night of malicious criminal activity.” The defendant was also “implicated in the violent death of another inmate,” and had “assaulted a police officer” while awaiting trial for the two murders. Id. at 1252. Nonetheless, we concluded that the Strickland prejudice standard had been satisfied by failure to investigate and present evidence of childhood abuse and substance abuse. Id. at 1252–53. In Correll, which involved a triple homicide, the court found that the crimes had been especially heinous, cruel, or depraved and had been committed with the expectation of pecuniary value. Correll, 539 F.3d at 942, 959; see State v. Corell, 148 Ariz. 468, 485 (1986). We concluded that “there was a substantial amount of mitigating evidence available, which, when taken together, [was] sufficient to raise a presumption of prejudice under the Supreme Court’s standard in Wiggins.” Correll, 539 F.3d at 953–54. The evidence included family abuse, substance abuse, and evidence of mental illness. Id. at 952–54. In Douglas v. Woodford, 316 F.3d 1079 (9th Cir. 2003), the defendant had gruesomely sexually assaulted two teenage girls and then strangled them to death. *60 Id. at 1082–83. Yet we concluded that the failure to investigate and present evidence of family history and mental health issues was sufficient to establish Strickland prejudice. Id. at 1088, 1091. In Summerlin, we found Strickland prejudice even with the finding that the defendant “committed the offense in an especially heinous, cruel, or depraved manner.” Summerlin, 427 F.3d at 641 (citation omitted). We concluded that the failure to investigate and present evidence of family history and mental illness as mitigating factors constituted sufficient Strickland prejudice. Id. at 631–34. In Silva v. Woodford, 279 F.3d 825 (9th Cir. 2002), we considered a “gruesome abduction, robbery and murder.” Id. at 828. However, we concluded that the Strickland prejudice standard had been satisfied by the failure to investigate and present evidence of family history, mental illness, organic brain disorders and substance abuse. Id. at 847, 850. All of these cases involved brutal crimes, but with the additional common thread that counsel did not properly investigate and present available classic mitigating evidence. Given the relevant case law, and weighing the available mitigating evidence against the aggravating factors in this case, we conclude on de novo review that *61 Jones has demonstrated Strickland prejudice. Accordingly, we reverse the district court’s denial of relief on Claim 1. V A In Claim 2, Jones asserts that his trial counsel was constitutionally ineffective by failing to seek neurological or neuropsychological testing prior to sentencing. He asserts that the failure to do so fell below prevailing professional norms at the time. We agree. As with Claim 1, counsel’s failure to promptly seek neuropsychological testing ran contrary to his obligation to pursue reasonable investigations under Strickland, and in particular, his obligation to investigate and present evidence of a defendant’s mental defect. See Terry Williams, 529 U.S. at 396 (failure to investigate and present evidence of mental defect amounts to deficient performance). The state court record shows that counsel was on notice of numerous facts from the very beginning of the representation that Jones may have had significant brain damage. “[W]hen ‘tantalizing indications in the record’ suggest that certain mitigating evidence may be available, those leads must be pursued.” Lambright, 490 F.3d at 1117 (quoting Stankewitz I, 365 F.3d at 719–20); see also Wiggins, 539 U.S. at 527 (“In assessing the reasonableness of an *62 attorney’s investigation, however, a court must consider not only the quantum of evidence already known to counsel, but also whether the known evidence would lead a reasonable attorney to investigate further.”). Counsel specified in his declaration before the PCR court that “prior to trial and sentencing [he] was aware from interviews of Danny Jones and his mother and step-father that he had been rendered unconscious numerous times during his life from head injuries,” as well as that “he had a significant history of serious long term substance abuse.” Any reasonable attorney would understand that these details could lead to valuable, available mitigation evidence and would have pursued these leads further. However, in the state PCR proceedings, defense trial counsel provided no strategic reason for his failure to arrange for neuropsychological testing. Instead, trial counsel stated that he “was not aware that neurological or neuropsychological testing was necessary and available which could determine the exact nature of injuries to Danny Jones’ brain from long term substance abuse and head injury,” nor that testing would shine a light on “the resulting affect on his behavior and conduct.” Counsel’s failure to appreciate the importance of such testing before the sentencing phase of trial constituted deficient performance because he failed to understand the value neuropsychological testing could provide in Jones’s case, and by the time of Jones’s sentencing in 1993, counsel in capital cases was expected to *63 be versed in the role of psychiatric evidence. See ABA Guidelines for the Appointment and Performance of Counsel in Death Penalty Cases, Guideline 5.1(1)(A)(v), p. 5–6 (1989) (“Lead trial counsel assignments should be distributed to attorneys who . . . are familiar with and experienced in the utilization of expert witnesses and evidence, including, but not limited to, psychiatric and forensic evidence.”). Counsel’s request for testing (and a continuance) during Jones’s sentencing hearing came far too late. As noted by PCR counsel, the court denied these requests because it had granted funding earlier in the case for expert assistance, and “if there [had been] any follow-up questions of a psychological or neurological nature, [the court expected] that the defense would have followed them up.” The court, therefore, was placing the burden on counsel to recognize these issues and request funding and assistance earlier in the case, which counsel failed to do because he had not invested sufficient preparation time and research to be aware that such testing was available and needed. Moreover, the timing of counsel’s request for neuropsychological testing, like his request for a defense mental health expert, was in itself deficient. “[P]reparation for the sentencing phase of a capital case should begin early and even inform preparation for a trial’s guilt phase.” Allen, 395 F.3d at 1001. For this reason, the PCR court’s decision that defense *64 counsel’s performance did not fall below an objectively reasonable standard was an unreasonable application of Strickland. Jones has satisfied § 2254(d)(1). B Alternatively, Jones asserts that the state PCR court’s decision was based on an unreasonable determination of the facts under § 2254(d)(2). He argues that the court precluded Claim 2’s full factual development by denying PCR counsel’s request to fund neuropsychological testing, and he asserts that the inadequacy of the state court’s fact-finding procedures renders its rejection of this claim unreasonable. We agree. The PCR court never addressed the facts supporting Jones’s IAC claim, and it excused counsel’s failure to move timely for neuropsychological testing in a vague, inconsistent order. As with Claim 1, this had the effect of precluding Claim 2’s full factual development in a way that rendered the entire fact-finding process unreasonable. At sentencing, Dr. Potts testified that he saw indicators of brain damage, and as a result, counsel requested that the court continue the proceedings so that he could seek a neuropsychological evaluation. The court, however, ruled contrary to Dr. Pott’s recommendation, stating only that: — 65 — this case has been pending a long time, and I think the evidence is very slim, nonexistent, in fact, that the defendant has anything that requires any kind of neurological examination. So, I am ready to proceed [with sentencing]. (Emphasis added). Because the State did not call a competing expert, the only evidence in the record—Dr. Pott’s unambiguous recommendation—suggested that a neuropsychological evaluation was necessary, contrary to the sentencing court’s assessment. The sentencing court’s cursory evaluation of the record effectively foreclosed any factual development on this issue. In the PCR proceedings, the court at least granted a hearing on the issue of counsel’s ineffectiveness regarding testing, but because the court summarily denied the claim concerning the appointment of a mental health expert and denied counsel’s motion for further neuropsychological testing, the evidentiary hearing was rendered almost meaningless. The court based its denial of neuropsychological testing on the court’s own impressions and untested memory of Dr. Potts’s sentencing testimony from six years prior. The court recalled that he “thought Dr. Potts did a good job,” and “based on his testimony,” the court did not “really see any grounds for any additional psychiatric or psychological testing.” But Dr. Potts’s testimony was that additional neuropsychological testing was needed. The court paradoxically explained that “[b]ased on [Dr. Potts’s] testimony *66 and the other things that I heard during that hearing, there was no grounds in my mind for obtaining a neuropsychological examination. Not one.” The resulting decision dismissed the claim for neuropsychological testing in a single sentence: “The report and testimony of Dr. Potts who was appointed by the Court, adequately addressed defendant’s mental health issues at sentencing.” As with Claim 1, the state PCR judge made factual findings regarding the necessity of neuropsychological testing, not on the basis of evidence presented by the petitioner, but on the basis of his own personal conduct, untested memory, and understanding of events. See Hurles, 752 F.3d at 791; see also Buffalo v. Sunn, 854 F.2d 1158, 1165 (9th Cir. 1988) (finding error when the court relied on “personal knowledge” to resolve disputed issue of fact). Additionally, in making the resulting factual finding—that neuropsychological testing was not warranted—the court “plainly misapprehend[ed]” the record. See Taylor, 366 F.3d at 1001. In particular, the evidence in the record—Dr. Potts’s testimony—strongly suggested that neuropsychological testing was essential in assessing Jones’s psychological state, contrary to the court’s finding. Thus, by finding against the weight of the evidence, and proceeding to rule on the merits of Claim 2, the court employed a constitutionally defective fact-finding process and ruled on an unconstitutionally incomplete factual record. See id. at 999; see also Milke, 711 *67 F.3d at 1007 (finding the state court decision rested on an unreasonable determination of the facts where the judge relied on a distorted fact-finding process and ruled on an “unconstitutionally incomplete record”). The PCR court had an obligation to allow for reasonable fact development in reaching the merits of Claim 2; the judge did not fulfill this obligation by relying on his own untested, personal recollection of the testimony Dr. Potts presented six years earlier. For this reason, Jones has demonstrated that the PCR court’s decision was based on an unreasonable fact-finding process and determination of the facts, satisfying § 2254(d)(2). C As with Claim 1, the PCR court failed to reach the prejudice prong of Claim 2, and so we address the issue de novo. See, e.g., Weeden, 854 F.3d at 1071. As with Claim 1, because Jones was diligent in attempting to develop the factual basis of this claim in state court by requesting “a thorough and independent neurological assessment,” § 2254(e)(2) does not limit our ability to consider evidence presented for the first time in federal district court. See Pinholster, 563 U.S. at 213 (Sotomayor, J. dissenting); see also Landrigan, 550 U.S. at 473 n.1. The State does not contest that Jones was diligent in attempting to develop the factual basis for his claims in state court, or that we may consider this additional *68 evidence on appeal. And having reviewed the record, we independently conclude that the district court did not err in its diligence determination and expansion of the record. Accordingly, we consider the evidence developed in federal district court in conducting de novo review of Jones’s claims. In order for us to grant relief on Jones’s IAC claim, Jones must demonstrate that his trial counsel: (1) performed deficiently; and (2) Jones’s defense was prejudiced by that deficient performance. Strickland, 466 U.S. at 687. He has done so. Jones has demonstrated that trial counsel performed deficiently for all the reasons set forth in our § 2254(d)(1) analysis. He has demonstrated that counsel’s performance fell below an objective standard of reasonableness and below the prevailing professional norms at the time of Jones’s proceedings. Additionally, Jones has demonstrated that counsel’s failure to seek neuropsychological and neurological testing prejudiced his defense. He has demonstrated that there is a “reasonable probability” that had such testing been conducted, and had the results been presented at sentencing, “the result of the proceeding would have been different.” Strickland, 466 U.S. at 694. While Dr. Potts presented brief, conditional findings, the results of the neuropsychological and neurological tests conducted by various experts during Jones’s federal district *69 court proceedings confirmed that Jones suffered from a variety of psychological disorders stemming from birth and exacerbated by long-term drug use and trauma that affected Jones’s cognitive functioning. As explained previously, testing revealed that Jones suffered from organic brain damage, poly-substance abuse, PTSD, AD/HD, mood disorder, bipolar depressive disorder, and a learning disorder. The presentation of these results would involve presenting the contributing factors to his cognitive dysfunction, as previously described with respect to Claim 1, including that his long-term substance abuse was induced by his sexually abusive step-grandfather. At sentencing, there was no indication that Jones had suffered years of sexual abuse as a child. In combination, the testing results and the presentation of contributing factors would have dramatically affected any sentencing judge’s perception of Jones’s culpability for his crimes, even despite the existence of aggravating factors. Without repeating our prior analysis, we conclude this available mitigating evidence would have created a reasonable probability that Jones would not have received a death sentence. VI Because we have determined that Jones is entitled to relief and resentencing on the basis of Claims 1 and 2, we need not and do not reach the issue of whether the new evidence presented at the federal evidentiary hearing fundamentally *70 altered these claims such that they were unexhausted, procedurally defaulted, and excused in light of Dickens v. Ryan, 740 F.3d 1302 (9th Cir. 2014) (en banc) and Martinez v. Ryan, 566 U.S. 1 (2012). Additionally, we need not and do not reach the merits of any of Jones’s other claims. We reverse and remand to the district court with instructions to issue the writ of habeas corpus. REVERSED AND REMANDED. — 1 — FILED NOV 7 2022 Danny Jones v. Charles Ryan, No. 18-99005 MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS BENNETT, Circuit Judge, joined by CALLAHAN, R. NELSON, BADE, COLLINS, LEE, BRESS, BUMATAY, and VANDYKE, Circuit Judges, dissenting from the denial of rehearing en banc: Danny Jones, an Arizona prisoner, brutally killed three people with a baseball bat, including defenseless seven-year-old Tisha Weaver, by dragging her from under her parents’ bed, striking her multiple times with the bat, and then strangling or asphyxiating her in case the bat had not done its intended job. He received two death sentences. The Supreme Court vacated the panel’s first attempt to grant Jones habeas relief. Ryan v. Jones, 563 U.S. 932 (2011). The panel’s amended opinion grants habeas relief again. But in doing so, the panel improperly and materially lowered Strickland’s 1 highly demanding standard and failed to afford the required deference to the district court’s findings—essentially finding that no such deference was due. The first error directly conflicts with Supreme Court precedent, and the second is inconsistent with our well-established rule that district court findings and credibility determinations are subject to clear error review. Thus, we should have taken this case en banc to secure and maintain uniformity in our case law. See Fed. R. App. P. 35(a)(1). En banc review was also warranted because this case involves issues of exceptional importance. See Fed. R. App. P. 35(a)(2). Not only does the 1 Strickland v. Washington, 466 U.S. 668 (1984). — 2 — panel’s amended opinion allow courts to improperly grant sentencing relief to capital defendants who have been convicted of the most horrific crimes, but it also allows future panels to simply ignore a district court’s well-reasoned factual and credibility findings. Finally, we should have taken this case en banc so that the Supreme Court, which has already vacated our judgment once, does not grant certiorari a second time and reverse us.2 For these reasons, I respectfully dissent from our decision not to rehear this case en banc. I. BACKGROUND A. Murders Jones received two death sentences for murdering his friend, Robert Weaver, and Weaver’s seven-year-old daughter Tisha, so that he could obtain Weaver’s gun collection. Jones committed the murders in March 1992, while he was on probation for a prior felony offense. In affirming Jones’s convictions and sentence, the Arizona Supreme Court described the murders: 2 The Supreme Court routinely reverses us in capital cases, including cases based on our misapplication of Strickland. See, e.g., Shinn v. Kayer, 141 S. Ct. 517 (2020) (per curiam) (summary reversal); Cullen v. Pinholster, 563 U.S. 170 (2011); Wong v. Belmontes, 558 U.S. 15 (2009) (per curiam) (summary reversal); Woodford v. Visciotti, 537 U.S. 19 (2002) (per curiam) (summary reversal). The Court has also reversed our sister circuits for misapplying Strickland in capital cases. See, e.g., Dunn v. Reeves, 141 S. Ct. 2405 (2021) (per curiam) (summary reversal); Mays v. Hines, 141 S. Ct. 1145 (2021) (per curiam) (summary reversal); Smith v. Spisak, 558 U.S. 139 (2010); Bobby v. Van Hook, 558 U.S. 4 (2009) (per curiam) (summary reversal). — 3 — In February 1992, defendant moved to Bullhead City, Arizona, and resumed a friendship with Robert Weaver. At this time, Robert, his wife Jackie, and their 7-year-old daughter, Tisha, were living in Bullhead City with Robert’s grandmother, Katherine Gumina.[3] As of March 1992, defendant was unemployed and was planning to leave Bullhead City. On the night of March 26, 1992, defendant and Robert were talking in the garage of Ms. Gumina’s residence. Robert frequently entertained his friends in the garage, and during these times, he often discussed his gun collection. The two men were sitting on inverted buckets on the left side of the garage, and Ms. Gumina’s car was parked on the right side of the garage. Both defendant and Robert had been drinking throughout the day and had used crystal methamphetamine either that day or the day before. At approximately 8:00 p.m., Russell Dechert, a friend of Robert’s, drove to the Gumina residence and took defendant and Robert to a local bar and to watch a nearby fire. Dechert then drove defendant and Robert back to the Gumina residence at approximately 8:20 p.m. and left, telling defendant and Robert that he would return to the Gumina residence around 9:00 p.m. Although there is no clear evidence of the sequence of the homicides, the scenario posited to the jury was as follows. After Dechert left, defendant closed the garage door and struck Robert in the head at least three times with a baseball bat. Robert fell to the ground where he remained unconscious and bleeding for approximately 10 to 15 minutes. Defendant then entered the living room of the Gumina residence where Ms. Gumina was watching television and Tisha Weaver was coloring in a workbook. Defendant struck Ms. Gumina in the head at least once with the baseball bat, and she fell to the floor in the living room. Tisha apparently witnessed the attack on Ms. Gumina, ran from the living room into the master bedroom, and hid under the bed. Defendant found Tisha and dragged her out from under the bed. During 3 Ms. Gumina was seventy-four years old. See Jones v. Schriro, 450 F. Supp. 2d 1023, 1025 (D. Ariz. 2006). — 4 — the struggle, Tisha pulled a black braided bracelet off defendant’s wrist. Defendant then struck Tisha in the head at least once with the baseball bat,[4] placed a pillow over her head, and suffocated her, or strangled her, or both. Defendant next emptied a nearby gun cabinet containing Robert’s gun collection, located the keys to Ms. Gumina’s car, and loaded the guns and the bat into the car. At some point during this time, Robert regained consciousness, and, in an attempt to flee, moved between the garage door and Ms. Gumina’s car, leaving a bloody hand print smeared across the length of the garage door and blood on the side of the car. Robert then climbed on top of a work bench on the east side of the garage, leaving blood along the east wall. Defendant struck Robert at least two additional times in the head with the baseball bat, and, as Robert fell to the ground, defendant struck him in the head at least once more. A few minutes before 9:00 p.m., Dechert returned to the Gumina residence and noticed that the garage door, which previously had been open, was closed. Dechert went to the front door and knocked. Through an etched glass window in the front door, he saw the silhouette of a person locking the front door and walking into the master bedroom. Dechert then looked through a clear glass portion of the window and saw defendant walk out of the master bedroom. He heard defendant say, “I will get it,” as if he were talking to another person in the house. Defendant then opened the front door, closing it immediately behind him, walked out onto the porch, and stated that Robert and Jackie had left and would return in about 30 minutes. Dechert noticed that defendant was nervous, breathing hard, and perspiring. Although Dechert felt that something was wrong, he left the Gumina residence. As he was leaving, Dechert heard the door shut as if defendant went 4 The Arizona Supreme Court’s opinion later states: “The evidence at trial showed that defendant struck Tisha in the head with the baseball bat at least twice . . . . [D]efendant had struck Tisha with the baseball bat with sufficient force to create a wound several inches wide, extending from her left ear to her left cheek. He then struck her a second time on the back of her head. After delivering these two fatal blows, defendant then asphyxiated her, far exceeding the amount of violence necessary to cause death.” State v. Jones, 917 P.2d 200, 217–18 (Ariz. 1996). — 5 — back into the house. Shortly thereafter, defendant left the Gumina residence in Ms. Gumina’s car. At approximately 9:10 p.m., Jackie Weaver returned home from work. When she opened the garage door, she found Robert lying unconscious on the garage floor. Jackie ran inside the house and found Ms. Gumina lying on the living room floor and her daughter Tisha lying under the bed in the master bedroom. She then called the police, who on arrival determined that Tisha and Robert were dead and that Katherine Gumina was alive but unconscious. The medical examiner later concluded that Robert’s death was caused by multiple contusions and lacerations of the central nervous system caused by multiple traumatic skull injuries. The cause of Tisha’s death was the same as Robert’s, but also included possible asphyxiation. After leaving the Gumina residence, defendant picked up his clothes from a friend’s apartment where he had been staying and drove to a Bullhead City hotel. At some point before reaching the hotel, he threw the bat out the car window. Defendant parked the car at the hotel and hailed a taxi cab to drive him to Las Vegas, Nevada. [Jones was arrested in Las Vegas.]. .... The state charged defendant with two counts of premeditated first degree murder and one count of attempted premeditated first degree murder. Although Katherine Gumina ultimately died as a result of the injuries defendant inflicted, the state chose not to amend the indictment. Defendant pleaded not guilty to all of the charges. At trial, defendant testified that he killed Robert Weaver in self-defense, that he struck Katherine Gumina reflexively and without criminal intent because she startled him, and that another person killed Tisha Weaver. The jury found defendant guilty of all three counts. State v. Jones, 917 P.2d 200, 206–07 (Ariz. 1996). — 6 — B. Sentencing Proceedings Under Arizona’s then-existing death penalty statute, the trial judge held an aggravation/mitigation hearing to determine whether a death sentence was warranted. Ariz. Rev. Stat. § 13-703(B) (1993).5 The judge had to impose a death sentence if he found “one or more of the [enumerated statutory] aggravating circumstances . . . and that there [were] no mitigating circumstances sufficiently substantial to call for leniency.” Id. § 13-703(E). 1. State’s Aggravation Evidence In addition to the guilt phase trial evidence,6 the State’s most significant aggravation evidence was Jones’s presentence report (“PSR”) related to the murders and a prior 1991 presentence report (“1991 PSR”) related to Jones’s theft conviction for which he had received three years’ probation. As the sentencing judge later determined, the conflicts between the 1991 PSR and other evidence showed that Jones was “willing to lie” when it benefited him. While the PSR contained some evidence in support of aggravation, it also contained evidence favorable to Jones. He started consuming alcohol at thirteen 5 All references to Arizona’s death penalty statute are to the version in effect on December 9, 1993, when the court sentenced Jones. 6 Under the statute, the judge had to consider the guilt phase trial evidence in determining the existence or nonexistence of aggravating and mitigating circumstances. Ariz. Rev. Stat. § 13-703(C). — 7 — and became an alcoholic at seventeen. He also began using marijuana at thirteen and experimented with or used “virtually all illegal drugs he could obtain” and had become addicted to methamphetamine and cocaine by eighteen. He sustained a concussion at age nine when he fell from a roof, and at eighteen, he was “knocked unconscious by a ‘mugger,’ being in a coma for approximately three days.” 2. Jones’s Mitigation Evidence Lee Novak, the public defender who represented Jones at trial and sentencing, presented mitigation evidence, including testimony from Jones’s second stepfather, Randy Jones, and a report and testimony from Dr. Potts, a court-appointed forensic psychiatrist, who evaluated Jones.7 Through this evidence, the trial judge learned extensive information about Jones’s background. Jones’s biological father had physically abused Jones’s mother, Peggy, while she was pregnant with Jones. Jones’s birth was traumatic—the umbilical cord was wrapped around his neck, forceps were used, and Peggy’s heart stopped during delivery. Jones’s first stepfather, Richard Eland, physically and verbally abused Jones and his sister, Carrie. On at least one occasion, Richard hit Jones and locked him in a closet with a bar of soap in his mouth and then taped Jones’s mouth shut. 7 Before sentencing, the trial court had granted Novak’s request to have Jones’s mental health examined pursuant to Arizona Rule of Criminal Procedure 26.5. Am. Op. at 4. The court appointed Dr. Potts, the Chief of Forensic Psychiatry for the Correctional Health Services in Maricopa County. Id. — 8 — Richard also physically abused Peggy, and Jones had witnessed two severe incidents: one where Richard broke Peggy’s jaw and another where he broke her ribs. Jones’s personality changed around age thirteen—he started lying and cutting classes, and his grades began slipping. Jones’s step-grandfather and uncle introduced him to marijuana when he was around ten, and Jones became an alcoholic by seventeen and had used all popular drugs by that time. His “mood swings” and irritability increased, and he was expelled from high school as a senior. Jones also suffered from two or three serious head injuries growing up. Dr. Potts testified that there was “clear evidence . . . of traumatic brain injury, and there [was] some other evidence that [he] believe[d] [of] organic neurologic dysfunctions . . . that ha[ve] gone on since [Jones has] been about 13.” Dr. Potts explained that Jones was predisposed to substance abuse and possibly an affective disorder, given the history of substance abuse and other psychological problems in Jones’s family. He stated, “[T]o a reasonable degree of medical certainty . . . the defendant suffers from a psychothymic disorder, which is [a] mood disorder, possibly organic syndrome, secondary to the multiple cerebral trauma that he’s had as well as the prolonged substance abuse.” Dr. Potts’s report identified seven mitigating circumstances: — 9 — 1. The chaotic and abusive childhood that the defendant suffered; 2. His genetic loading for substance abuse and possibly an affective disorder; 3. His intoxication at the time of the offense with a concomitant decrease in an ability to conform his conduct to the law; 4. The potential for rehabilitation; 5. The likelihood that he suffers from a major mental illness- cyclothymia (an attenuated form of Bipolar Affective Disorder; i.e., manic-depressive illness); 6. The head trauma he suffered which increases the potential for neurologic sequelae contributing to his behavior; and, 7. His sense of remorse and responsibility . . . . 3. Trial Judge’s Sentencing Decision The trial judge found three aggravating circumstances for Weaver’s murder: “(1) defendant committed the murder for pecuniary gain; (2) defendant committed the murder in an especially heinous, cruel, or depraved manner; and (3) defendant was convicted of one or more other homicides that were committed during the commission of the offense.” Jones, 917 P.2d at 207 (citations omitted). For Tisha’s murder, the trial judge found the same three aggravators plus that Tisha was under age fifteen. Id. The sentencing judge’s aggravating circumstance determinations show that the sentencing judge gave great weight to the aggravating circumstances: As to Statutory Aggravating Circumstance (F)(5), the Court finds beyond a reasonable doubt that the defendant committed the offense as consideration for the receipt, or in the expectation of the receipt of anything of pecuniary value. The evidence shows that the defendant wanted to get out of Bullhead City because of pending warrants. The defendant knew of Robert Weaver’s gun collection. Defendant killed — 10 — Robert Weaver to get the guns and used them to obtain a ride to Las Vegas and to obtain money for living expenses in Las Vegas. As to the Statutory Aggravating Circumstance (F)(6), the Court finds beyond a reasonable doubt that the defendant committed the offenses in an especially heinous and depraved manner. The physical evidence as to Robert Weaver shows that the defendant initially struck Robert Weaver . . . with a baseball bat. The initial injuries were sufficient to cause a large pool of blood, but insufficient to cause death. Sometime later, in all likelihood after the defendant committed the assault within the residence, he returned to find Robert Weaver still alive. Blood smears at the scene showed that Robert Weaver attempted to run from the defendant around Katherine Gumina’s car which was parked in the garage. . . . [T]he defendant struck Robert Weaver in the head several more times. The last blow delivered, was delivered [sic] while the defendant knelt helplessly on the floor of the garage. The initial blows were all that were needed to kill the victim. The defendant, by continuing to beat Robert Weaver with a bat, inflicted gratuitous violence beyond that necessary to kill the victim. In addition, after the initial blows, the victim was completely helpless to defend himself and could only make a futile effort to flee. The defendant could have taken the guns and the car with little or no resistance from Robert Weaver. The killing was, therefore, senseless. Robert Weaver had time to contemplate his fate as he fled from the defendant. The killing, therefore, was cruel. As to the Statutory Aggravating Circumstance (F)(8), the Court finds beyond a reasonable doubt that the defendant has been convicted of one or more other homicides, as defined in ARS 13-1101, which were committed during the commission of the offense. The jury found the defendant guilty of First Degree Murder of Tisha Weaver which occurred during the same violent episode. .... As to the Statutory Aggravating Circumstance (F)(6), the Court finds beyond a reasonable doubt that the defendant committed the offenses, with regard to Tisha Weaver, in an especially heinous, cruel, — 11 — or depraved manner. The evidence showed that Katherine Gumina and Tisha Weaver had been in front of the television set in the living room of the residence. The defendant assaulted Ms. Gumina, causing Tisha to run and hide under her parents’ bed. Physical evidence from the bedroom showed that Tisha was dragged from under the bed, struck several times with a blunt instrument, and then suffocated. She had time to contemplate her fate. She knew that something terrible had happened to her grandmother. She struggled for life with the defendant. Tisha Weaver suffered great physical and emotional pain. The Court, therefore, finds that the murder was especially cruel. The Court also finds that the murder was heinous and depraved. Tisha Weaver, a seven-year old, and a helpless victim of the adult male defendant armed with a baseball bat. The murder of Tisha Weaver is senseless. She could not have stopped the defendant from stealing the guns or the car. . . . She was also beaten beyond that necessary to kill her. The trial judge found four mitigators for both murders: “(1) defendant suffers from long-term substance abuse; (2) at the time of the offense, defendant was under the influence of alcohol and drugs; (3) defendant had a chaotic and abusive childhood; and (4) defendant’s substance abuse problem may have been caused by genetic factors and aggravated by head trauma.” Jones, 917 P.2d at 207–08. In rejecting other mitigators presented by Jones, the trial court explained that Jones’s conduct did not arise “from an angry explosion or delusion caused by drug or alcohol use.” Rather, the court found that the evidence was “more consistent with the State’s theory that [Jones] committed the acts of murder so that he could steal Robert Weaver’s guns.” The court also noted that “[i]n the past the defendant *12 ha[d] shown that he [was] willing to lie if it benefit[ed] him.” Indeed, the court found that it was “obvious” that Jones and another prisoner had “manufactured [the] tale” that a third party had been involved in the murders. The trial court determined that the mitigators were not sufficiently substantial to outweigh the aggravators or to call for leniency and sentenced Jones to two death sentences for the murders. See Ariz. Rev. Stat. § 13-703(E). 8 The Arizona Supreme Court affirmed Jones’s conviction and sentence on direct review. Jones, 917 P.2d at 222. C. State Post-Conviction Review Proceedings Jones filed his state post-conviction review (“PCR”) petition, raising several claims including the ones at issue: (1) sentencing counsel was ineffective for relying on Dr. Potts rather than retaining an independent neuropsychologist and neurologist (“Claim 1”), and (2) sentencing counsel was ineffective for failing to timely seek neurological or neuropsychological testing (“Claim 2”). At an informal conference, the PCR court denied Claim 1. The court set Claim 2 for an evidentiary hearing. At the evidentiary hearing, Jones called three witnesses: Randy, Peggy, and Novak. Randy and Peggy’s testimonies were relatively short, focusing on when 8 The court gave Jones a life sentence, without the possibility of release or parole for twenty-five years, for the attempted first degree murder of Katherine Gumina. — 13 — and what they had told Novak about Jones’s life history. To the extent that they provided information about Jones’s background, it was largely cumulative of the information the court heard at sentencing. See Am. Op. at 10–11. Novak mainly testified about his experience as a criminal lawyer, the defense’s investigation into Jones’s background, and decisions that he made before and during trial and in preparation for sentencing. As to Dr. Potts, Novak testified that Potts was a court-appointed expert, Potts “was great to work with,” “did everything that we would have wanted someone that we had hired to do,” making it feel as though Potts “was part of the defense team almost.” The PCR court denied Claim 2. The Arizona Supreme Court summarily denied Jones’s petition for review. D. Federal District Court Proceedings Jones filed a federal habeas petition, and the district court granted an evidentiary hearing on Claims 1 and 2. Am. Op. at 12. At the evidentiary hearing, Jones called Novak and three mental health experts: (1) Dr. Potts; (2) Dr. Stewart, a psychiatrist; and (3) Dr. Goldberg, an attorney and neuropsychologist. Jones v. Schriro, 450 F. Supp. 2d 1023, 1030 (D. Ariz. 2006). Jones also submitted documents, including reports from Drs. Stewart and Goldberg, a report from Dr. Foy, a psychologist, and a report from Dr. Sreenivasan, a neuropsychologist. Id. at 1030–34, 1032 n.6. The State called three experts: (1) Dr. Herron, a psychiatrist *14 formerly employed by the Department of Correction; (2) Dr. Herring, a neuropsychologist; and (3) Dr. Scialli, a psychiatrist. Id. at 1034. The district court provided detailed summaries of Novak’s and the experts’ testimonies. Id. at 1030–38. The district court made the following findings, which noted many of the inconsistencies between the experts’ testimonies and within the record evidence in discounting the significance of the “new” evidence introduced in the district court evidentiary proceeding. Credibility of Experts: The district court properly evaluated the credibility of the experts. Cf. McClure v. Thompson, 323 F.3d 1233, 1243 (9th Cir. 2003) (giving “great weight” to the district court’s credibility determinations in a habeas case). The district court determined that the State’s experts, Drs. Herring and Scialli, were generally more credible than the defense experts, Drs. Stewart and Goldberg. Jones, 450 F. Supp. 2d at 1038. This was so because Drs. Herring and Scialli had previously testified for both the prosecution and criminal defendants, while Dr. Stewart had done mostly defense work, and Dr. Goldberg had never been retained by the prosecution in a capital case. Id. The district court also found Dr. Stewart less credible because he endorsed Jones’s contention that Jones did not kill Tisha, even though that fact had been rejected by the jury, the trial court, and the Arizona Supreme Court. Id. at 1033 n.7. The district court explained: In assessing Dr. Stewart’s credibility, the Court takes into account his willingness to present an opinion on a factual issue which concerns only — 15 — the guilt phase of the trial and which was resolved, with a result contrary to that reached by Dr. Stewart, by the jury, the trial court, and the Arizona Supreme Court. Id. at 1033 n.7. 9 Indeed, the sentencing judge found it “obvious” that Jones had completely “manufactured [the] tale” that a third party named Frank had killed Tisha. Cognitive impairment: The district court found that Jones’s evidence of cognitive impairment mainly due to head injuries was not persuasive: Petitioner has not presented persuasive evidence regarding either the existence or the cause of his alleged cognitive impairment. In making their diagnosis of cognitive impairment, Petitioner’s experts relied upon Petitioner’s school performance, both his grades and his scores on standardized tests; the discrepancy in his performance and verbal IQ scores; and the results of other neuropsychological tests. As discussed above, alternative explanations exist with respect to Petitioner’s declining school performance, including absenteeism, family stresses, substance abuse, and lack of motivation. Moreover, as Dr. Herring testified, Petitioner’s standardized test scores were within the average range and do not, by themselves, suggest impairment. The gap between Petitioner’s IQ scores, while notable, is not uncommon, and the fact that Petitioner scored higher on the performance subtest militates against a finding of impairment, as does the fact that Petitioner’s overall IQ is solidly in the average range. Finally, in the vast majority of instances Petitioner’s scores on neuropsychological tests were in the average range or above. The few scores that fell in the 9 I am puzzled how the panel concluded that “[t]he district court mistakenly stated that Dr. Stewart opined on the viability of Jones’s theory that a third party had been involved in the crime.” Am. Op. at 42. The district court found that Dr. Stewart “endors[ed] Petitioner’s account of the crimes,” Jones, 450 F. Supp. 2d at 1033 n.7, a conclusion supported by Dr. Stewart’s statement that: “As a result, it is my professional opinion that Danny’s psychological profile supports the events as described by Danny [Jones] on the night of the crimes, including Frank’s responsibility for Tisha Weaver’s murder.” (emphasis added). — 16 — impaired range did not implicate any particular cognitive domain, suggesting that they were aberrations and not indicative of impairment. The experts ascribed as the primary cause of Petitioner’s cognitive impairment a series of head injuries. With the exception of the 1983 “mugging,” there is no medical documentation to corroborate any of these injuries. In addition, the dates and details—and even the occurrence—of the injuries, as reported by Petitioner and his family, are inconsistent and hence difficult to credit. This difficulty is compounded by the contrast between Petitioner’s account of the 1983 incident, in which he was mugged, struck by a two-by-four, and left unconscious for three days, and the contemporaneous medical records, which indicate that Petitioner was discovered passed out or asleep from the effects of intoxication, that he responded upon being administered medication that counteracted those effects, that he suffered no neurological damage and his only injury was a small abrasion, and that if he suffered a concussion it was “resolved” upon his discharge. Id. at 1039 (footnote omitted). Post-traumatic stress disorder (“PTSD”): The district court found that Jones’s evidence that he suffered from PTSD at the time of the murders was unconvincing because, among other reasons, “none of [Jones’s] experts completed an appropriate diagnosis using all of the criteria set forth in the DSM-IV.” Id. at 1040. Drs. Foy and Stewart diagnosed Jones with PTSD. Id. at 1031, 1032 n.6. Dr. Scialli, however, testified that both Drs. Foy and Stewart’s written reports failed to discuss how Jones satisfied all the criteria required to diagnose a patient with PTSD under the DSM-IV. Dr. Stewart admitted as much,10 but insisted in his 10 Q. And can you show me in your report where you talk about that second criteria under the DSM-IV? — 17 — testimony that Jones met the remaining criteria—re-experiencing the trauma, avoidance, and hyperarousal. Id. at 1032. Dr. Scialli disagreed with the PTSD diagnosis and noted that during his examination of Jones there was no sign that Jones had re-experienced a traumatic event at the time of the murders. Id. at 1037. New sexual and physical abuse evidence: The district court found that Jones’s new allegations and evidence that he had been sexually abused by his step-grandfather from ages nine to fourteen and physically abused by Randy should be discounted “given their late disclosure, their inconsistency with other information in the record, and [Jones’s] ‘obvious motive to fabricate.’” Id. at 1047 (quoting State v. Medrano, 914 P.2d 225, 227 (Ariz. 1996)). The district court found that the alleged sexual abuse appeared for the first time in Dr. Foy’s 2002 report and was based only on Jones’s self-report. Id. at 1046. The alleged physical abuse by Randy appears in Dr. Foy’s report and Carrie’s declaration. Id. But as noted by the district court, the allegations conflicted with other information in the record. Id. at 1047. For example, Peggy testified at the PCR evidentiary hearing that Jones had a “good home life” and “normal childhood” once she married Randy; at the aggravation/mitigation hearing Dr. Potts testified that Jones had never disclosed [Dr. Stewart]. I don’t know if I have it in the report, actually. I certainly can talk about it. Q. In fact, your report doesn’t track the DSM-IV criteria, correct? [Dr. Stewart]. I believe you’re right. — 18 — that Randy was abusive; and a military medical record dated August 1983 states that Jones reported: “As far as I’m concerned [Randy] is my real dad, he’s the only one that has treated me good. He has never hit me or anything.” Attention deficit/hyperactivity disorder (“ADHD”): The district court found that the evidence supported that Jones suffered from ADHD at the time of the murders but that “[ADHD] does not serve as persuasive mitigation evidence.” Id. at 1040. The court’s finding was based on Dr. Scialli’s testimony that “the condition is unrelated to violent behavior.” Id. Mood disorder: The district court found that Jones “may suffer from a chronic, low-level mood disorder such as dysthymia,” but that such condition is not persuasive mitigation evidence because “[n]one of the experts suggested a causal relationship between the condition and [Jones’s] conduct during the crimes.” Id. The district court also discounted Jones’s evidence that he suffers from a major affective disorder, such as bipolar disorder, because he presented no evidence that he has experienced episodes of mania or hypomania. Id. Substance abuse: The district court found that Jones was addicted to alcohol, amphetamine, and cannabis at the time of the murders. Id. After setting forth its findings on the new mitigation evidence, the district court analyzed Claims 1 and 2 and determined that Jones had failed to satisfy Strickland’s prejudice prong. Id. at 1042–45. The district court found “that the *19 new information is largely inconclusive or cumulative: it ‘barely . . . alter[s] the sentencing profile presented to the sentencing judge.’” Id. at 1043 (alterations in original) (quoting Strickland, 466 U.S. at 700). It concluded that Jones had failed “to affirmatively demonstrate a reasonable probability that [the] additional information would [have] alter[ed] the trial court’s sentencing decision after it weighed the totality of the mitigation evidence against the strong aggravating circumstances proven at trial,” and “[t]herefore, [Jones was] not entitled to habeas relief on [Claims 1 and 2].” Id. at 1043. E. Panel’s Amended Opinion The panel reversed the district court and granted Jones’s habeas petition based on Claims 1 and 2. Am. Op. at 69–70. The panel acknowledged that AEDPA governed its review, and that Claims 1 and 2—both ineffective assistance of counsel claims—must be analyzed under Strickland. Id. at 15–19. The panel then determined that because the state PCR court had decided only Strickland’s deficient performance prong, AEDPA applied to that prong. Id. at 19. But because the panel determined that the state PCR court had not decided Strickland’s prejudice prong, it reviewed that piece of Jones’s claims de novo. 11 Id. The panel held that Jones had overcome AEDPA on the The panel’s conclusion that AEDPA does not apply is open to question. In 11 Harrington v. Richter, the Supreme Court stated that we must “presume[] that the *20 deficient performance prong for both claims because the PCR court’s decision was both an unreasonable application of Strickland under 28 U.S.C. § 2254(d)(1) and based on an unreasonable determination of the facts under § 2254(d)(2). Id. at 20– 34, 61–67. Thus, the panel reviewed Claims 1 and 2 de novo. Under de novo review, the panel held that Jones satisfied Strickland’s deficient performance prong for Claims 1 and 2 because his sentencing counsel was deficient by failing to secure a defense mental health expert and neurological or neuropsychological testing before sentencing. Id. at 36, 68. In analyzing de novo Strickland’s prejudice prong, the panel took Jones’s evidence at face value, while failing to appropriately credit everything on the other side of the balance— the district court’s factual and credibility findings, the overwhelming aggravating circumstances, and the State’s extensive rebuttal evidence. 12 Id. at 37–61, 68–69. state court adjudicated the claim on the merits in the absence of any indication or state-law procedural principles to the contrary.” 562 U.S. 86, 99 (2011). Among other things, the PCR court in this case found: “Testimony at the hearing showed that counsel presented the available witnesses and evidence to support mitigation. The additional witnesses and evidence suggested by petitioner would have been redundant.” Here, however, it does not matter whether AEDPA applies to the prejudice prong—the panel should have applied clear error review to the district court’s factual and credibility findings and affirmed the district court’s denial of habeas relief. See Lambert v. Blodgett, 393 F.3d 943, 964 (9th Cir. 2004). 12 The panel’s amended opinion heavily relies on other cases involving horrific crimes and similar “classic mitigating evidence” in which the courts found Strickland prejudice. Am. Op. at 20, 56–60. But the panel’s discussion simply shows that new mitigation evidence can establish prejudice, even in horrific cases. — 21 — Based on its erroneous analysis, the panel held that Jones satisfied Strickland’s prejudice prong for Claims 1 and 2 by showing a reasonable probability that he would not have received a death sentence had his new mitigation evidence been presented at sentencing. Id. at 36–37, 68–69. The panel thus granted habeas relief on Claims 1 and 2. Id. at 69–70. The panel determined that the district court could properly consider Jones’s new mitigation evidence and make credibility determinations, id. at 36, 49, and that a district court’s findings are generally reviewed for clear error, id. at 15. The panel inexplicably decided, however, that it did not have to give deference to the district court’s credibility determinations under Correll v. Ryan, 539 F.3d 938 (9th Cir. 2008). Am. Op. at 49–50. But Correll does not establish that the district court’s credibility determinations here were improperly made. To begin, Correll does not, nor could it, prohibit a district court from making credibility determinations in assessing Strickland prejudice. Indeed, before Correll we broadly stated without qualification that “[f]actual findings and credibility determinations made by the district court in the context of granting or denying [a habeas] petition are reviewed for clear error.” Lambert v. Blodgett, 393 F.3d 943, I do not dispute that proposition. Here, though, the panel erred by improperly dismissing the district court’s credibility and factual findings and essentially ignoring nearly all the evidence that cut against the mitigation evidence. None of the cases cited by the panel commit those same errors. — 22 — 964 (9th Cir. 2004). In Correll itself, the district court made an adverse credibility determination in assessing evidence relevant to Strickland prejudice, and we reviewed such determination for clear error. 539 F.3d at 942, 953 n.7. Correll simply does not hold that it is improper for a district court to make the types of credibility determinations the district court made here in assessing Strickland prejudice. Nor could it have. Strickland’s prejudice inquiry requires a court to determine whether there is a “reasonable probability” that the sentencer would not have imposed a death sentence considering all the evidence. See Wong v. Belmontes, 558 U.S. 15, 19–20, 26 (2009) (per curiam). In making that determination, a court must be able to assess the weight or probable effect of the evidence on the sentencer by, for example, making credibility determinations. How else is a court to determine, when faced with conflicting evidence, whether there is a “reasonable probability” of a different outcome? The panel first disposed of AEDPA deference. It then failed to give appropriate deference to the district court’s careful and thorough evaluation of the evidence. And finally, it made its own findings and reweighed the evidence without considering all the counterevidence, in direct contravention of Strickland. 13 13 As just one stark example, the panel’s amended opinion accepts, as a matter of fact, that Jones was sexually abused by his step-grandfather. Am. Op. at 51–52, 53. But as I detail below and as found by the district court, that mitigation evidence is weak considering the entire record: it is based wholly on a self-report — 23 — The panel, which disregarded the aggravating evidence, and the findings and determinations of the sentencing judge, the Arizona Supreme Court, and the district court, concluded that the sentencing judge “heard almost nothing that would . . . allow [him] to accurately gauge [Jones’s] moral culpability.” Am. Op. at 53 (first brackets in original) (emphasis added) (quoting Porter v. McCollum, 558 U.S. 30, 41 (2009) (per curiam)). This one statement encapsulates much that is wrong with the panel’s amended opinion. Moral culpability—the sentencing judge heard that Jones had brutally, cruelly, and senselessly killed a seven-year-old girl, her seventy-four-year-old great grandmother, and her father, all with a baseball bat, and all for financial gain. To the panel, that is “almost nothing.” The panel’s approach watered down Strickland’s demanding standard and flouted our well-established rule that district court findings are entitled to deference—both were grave errors. II. DISCUSSION “The Strickland standard is ‘highly demanding.’” Shinn v. Kayer, 141 S. Ct. 517, 523 (2020) (quoting Kimmelman v. Morrison, 477 U.S. 365, 382 (1986)). In by Jones, whom the trial court found to be a liar; there was no persuasive corroborating evidence; and such childhood evidence is given less mitigating weight when, as here, the murders are planned and deliberate. Jones, 450 F. Supp. 2d at 1047. Rather than consider, as the district court did, all that “bad” evidence, which severely weakens Jones’s mitigation evidence, the panel simply ignored it and found that Jones was sexually abused. — 24 — assessing prejudice under Strickland, “the question is whether there is a reasonable probability that, absent the errors, the sentencer—including an appellate court, to the extent it independently reweighs the evidence—would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.” Strickland, 466 U.S. at 695. “A reasonable probability means a ‘substantial,’ not just ‘conceivable,’ likelihood of a different result.” Kayer, 141 S. Ct. at 523 (additional internal quotation marks omitted) (quoting Cullen v. Pinholster, 563 U.S. 170, 189 (2011)). To answer the prejudice inquiry, we must “consider all the evidence—the good and the bad,” Wong, 558 U.S. at 26, and “reweigh the evidence in aggravation against the totality of available mitigating evidence,” Wiggins v. Smith, 539 U.S. 510, 534 (2003). And here, because the district court made findings about the evidence, including credibility findings, we must accept those findings unless they are clearly erroneous. See Lambert, 393 F.3d at 964. The panel did none of this. Rather than consider all the evidence, including the facts of the murders and the State’s rebuttal evidence, the panel accepted without question Jones’s “new” mitigation evidence. It also improperly brushed aside the district court’s well-reasoned factual and credibility determinations. And *25 it failed to seriously reweigh the aggravation evidence against the mitigation evidence. It is no wonder that the panel reached the wrong result.14 As I show below, the panel would have been compelled to affirm the denial of habeas relief on Claims 1 and 2 had it properly followed the prescribed framework. I provide a detailed analysis, not simply to show that the panel erred in assessing the weight of the evidence, but also to show first, that the panel failed to comply with Strickland’s clear requirement that a court consider all the good and the bad, and second, that the panel improperly disregarded the district court’s findings. A. Mitigation Evidence I start with the mitigating evidence. As discussed above, the trial court was presented with substantial mitigating evidence and considered four mitigating circumstances for both murders: Jones’s substance abuse, the influence of alcohol and drugs on Jones at the time of the murders, Jones’s chaotic and abusive 14 The panel in its original opinion did not discuss how, even if its factual findings in the face of conflicting evidence were correct, the “new” evidence could possibly have overcome the aggravators, like the horrific facts of Jones’s cruel and heinous murder of seven-year-old Tisha. The panel also never mentioned the trial judge’s detailed sentencing findings. And the panel’s original opinion made no attempt to reweigh the aggravators and mitigators. See Jones v. Ryan, 1 F.4th 1179 (9th Cir. 2021). The panel’s amended opinion acknowledged that it must reweigh the aggravation and mitigation evidence and claimed to have done so. Am. Op. at 55– 56. But nowhere did the panel assess the weight of the aggravation evidence, which was overwhelming. — 26 — childhood, and the fact that Jones’s substance abuse may have resulted from genetic factors and been aggravated by head trauma. Jones, 917 P.2d at 207–08. But the trial court determined that these mitigating circumstances “were not sufficiently substantial to outweigh the aggravating circumstances or to call for leniency.” Id. at 208. According to the panel, however, there is a substantial likelihood that the sentencing decision would have been different based on eight “new” categories of mitigation evidence showing: (1) cognitive impairment; (2) poly-substance abuse; (3) PTSD; (4) ADHD; (5) a mood disorder/bipolar depressive disorder; (6) impairment by drugs during the crimes; (7) sexual abuse by his step-grandfather; and (8) physical and emotional abuse by Randy. Am. Op. at 40, 51–52. But proper application of Strickland and its progeny compels a contrary result. Considering the totality of the evidence, all of this “new” mitigation evidence (some of which is not even new) is weak and would have made little, if any, difference, especially given the district court’s findings which were entitled to deferential review. (1) Cognitive impairment: The evidence of cognitive impairment is equivocal at best. The expert reports are conflicting. Dr. Stewart concluded that Jones suffers from cognitive impairment, and Dr. Goldberg diagnosed Jones with a learning disability. But Dr. Herring disagreed with those conclusions. And Dr. *27 Scialli did not diagnose Jones with cognitive impairment. Further undermining Jones’s evidence was the district court’s well-supported factual findings that Jones’s “overall IQ is solidly in the average range” and that the instances of head injuries were not credible, given the inconsistent stories and lack of medical documentation. Jones, 450 F. Supp. 2d at 1039 & n.11.15 Jones’s own mitigation evidence supported the district court’s conclusion about his IQ. Dr. Potts reported that Jones’s “cognitive abilities appeared to be consistent with his educational achievements” and estimated that his “I.Q. [was] within the normal range.” A military record from 1983 stated that Jones’s “[c]ognitive testing was normal.” And Dr. Sreenivasan’s report noted that Jones had a “weighted I.Q. average of 107” in an intellectual/psychological assessment conducted by the Arizona Department of Correction in 1992. Given the conflicting evidence, Jones’s cognitive impairment evidence would have had little mitigating weight. See State v. Dann, 207 P.3d 604, 629 (Ariz. 2009) (en banc) (giving “minimal weight” to mental health issues because of the conflicting evidence). Moreover, in Arizona, “mental health issues are entitled to little weight when there is no connection to the crime and no effect on the defendant’s ability to conform to the requirements of the law or appreciate the wrongfulness of his 15 The panel had no basis for rejecting these findings by the district court, just as it had no basis for rejecting the district court’s other well-supported findings. — 28 — conduct.” State v. Poyson, 475 P.3d 293, 298 (Ariz. 2020). And Arizona courts “will not find that a defendant’s ability to conform or appreciate the wrongfulness of his conduct was impaired when the defendant’s actions were planned and deliberate, or when the defendant seeks to cover up his crime.” Id.; see also State v. Boggs, 185 P.3d 111, 130 (Ariz. 2008) (en banc) (“Without a causal link between the murders and his troubled childhood or mental health issues, these mitigating circumstances are entitled to less weight.”). Only Dr. Stewart found that Jones’s cognitive dysfunction impaired his ability to conform his behavior to the law or appreciate the wrongfulness of his conduct. But Dr. Stewart’s opinion is substantially undermined by the district court’s adverse credibility finding and the conflicting expert opinions. His opinion is also severely weakened by evidence that Jones’s actions were planned and deliberate and that he sought to cover up the murders. See Poyson, 475 P.3d at 299 (“goal-oriented” behavior, such as taking preparatory steps or concealing crimes after the fact, “belie a claim of substantial impairment”). For example, Jones closed the garage door before striking Weaver; when a third party unexpectedly showed up at the house, Jones pretended that someone else was in the house and lied about Weaver’s whereabouts to cover up the murders; and, right after the murders, Jones retrieved his belongings from a friend’s house, ditched the car linking him to the crime scene, disposed of the murder weapon, and fled to Las *29 Vegas. Jones, 917 P.2d at 206–07. Indeed, the trial court found that the facts showed that Jones “understood the wrongfulness of his acts and took steps to avoid prosecution.” 16 In sum, Jones’s cognitive impairment evidence is weak. Given the substantial rebuttal evidence and the lack of any persuasive evidence establishing a causal link between the alleged impairment and the murders, Jones’s cognitive impairment evidence would have been given only minimal mitigating weight at best. (2) Poly-substance abuse: The evidence that Jones suffered from poly-substance abuse is not “new.” The trial court was well acquainted with Jones’s long-term substance abuse of various drugs and alcohol. Indeed, in sentencing Jones, the court found that his substance abuse was a mitigating circumstance. Jones, 917 P.2d at 207. Thus, more evidence of Jones’s substance abuse would have been cumulative and made minimal difference. See Wong, 558 U.S. at 22 (“Some of the evidence was merely cumulative of the humanizing evidence Schick actually presented; adding it to what was already there would have made little difference.”). 16 The panel did not engage with the undisputed facts as to Jones’s planned and deliberate conduct and his attempts to cover up his murders. — 30 — (3) PTSD: Like Jones’s cognitive impairment evidence, his PTSD evidence is at best inconclusive. Drs. Stewart and Foy diagnosed Jones with PTSD. But as the district court found, their opinions are unpersuasive for many reasons. Although Dr. Stewart insisted at the evidentiary hearing that Jones had met all the DSM-IV criteria for PTSD, neither his report nor Dr. Foy’s report completed an appropriate diagnosis using all the DSM-IV criteria. Jones, 450 F. Supp. 2d at 1040; see Rhoades v. Henry, 638 F.3d 1027, 1050 (9th Cir. 2011) (“The mitigating value of . . . [the PTSD diagnosis] is lessened because [the] diagnosis admittedly does not satisfy the requirements of DSM–IV for this condition.”). Indeed, Dr. Stewart explained that the second DSM-IV criteria for PTSD is re-experiencing a trauma, but he admitted that he never discussed with Jones whether Jones had re-experienced a traumatic event at the time of the murders. And Dr. Foy’s own diagnosis is unclear. His report noted “probable chronic PTSD,” (emphasis added), and he stated in his deposition that his opinion only suggested that “there’s a very high probability that [Jones] would be diagnosed by anyone with PTSD,” Jones, 450 F. Supp. 2d at 1032. And other evidence contradicted or failed to corroborate the PTSD diagnosis. Dr. Scialli disagreed with the PTSD diagnosis because during his examination of Jones there was no sign that Jones had re-experienced a traumatic event at the time of the murders. Id. at 1037. Dr. Herron, who had treated Jones *31 from 2003 to 2005, never detected any signs of PTSD. Id. at 1034. And even Dr. Potts, who was qualified to render a PTSD diagnosis, didn’t raise PTSD as a potential issue because “[he] didn’t see PTSD as a red flag” when he evaluated Jones. Id. at 1030. Jones’s PTSD evidence is weak on its own, and any mitigating value is reduced even more by the rebuttal evidence. Given this, Jones’s PTSD evidence would have made little difference. The mitigating value of Jones’s evidence is diminished even more because he presented no persuasive evidence linking his alleged PTSD to the murders. See Poyson, 475 P.3d at 298. Only Dr. Stewart offered an opinion on whether there was a causal link between Jones’s PTSD and the murders. But Dr. Stewart’s opinion is unconvincing for several reasons: his lack of credibility, his failure to complete an appropriate diagnosis using all the DSM-IV criteria, and the evidence that Jones’s actions were planned and deliberate and that he sought to cover up the murders. See id. at 298–99. (4) ADHD: The district court found that Jones’s ADHD was unrelated to his violent behavior, thus it was unpersuasive as mitigation evidence. Jones, 450 F. Supp. 2d at 1040. That finding was not clearly erroneous given Dr. Scialli’s testimony that, even if Jones had residual symptoms of ADHD, the condition doesn’t have “any relationship to the offenses. And so it’s a very minor point.” *32 Because Jones’s ADHD had no connection to his violent behavior, evidence of his ADHD would have had little mitigating weight. See Poyson, 475 P.3d at 298. (5) Mood Disorder/bipolar depressive disorder: Whether Jones suffers from a mood or depressive disorder that affected his behavior during the murders is ambiguous at best, and thus such evidence would have been given little mitigating weight. See id.; Dann, 207 P.3d at 629. Dr. Foy diagnosed Jones with “[d]epressive disorder (either bipolar or major depression), chronic.” Dr. Goldberg diagnosed Jones with bipolar disorder and depression, and Dr. Sreenivasan noted that Jones’s records “point to the presence of a cyclical mood disorder (bipolar II or Cyclothymia).” But none of these doctors opined on whether such conditions impacted Jones’s ability to conform his behavior to the law or appreciate the wrongfulness of his conduct at the time of the murders. Thus, their diagnoses would have been given minimal weight. See Poyson, 475 P.3d at 298. Only Dr. Stewart diagnosed Jones with a mood disorder not otherwise specified (NOS), meaning Jones has a mood disorder that does not fit within any of the other DSM-IV categories, and concluded that it impaired Jones’s ability to conform to the law or appreciate the wrongfulness of his actions during the murders. But as discussed above, Dr. Stewart’s opinion is severely weakened by the district court’s adverse credibility determination and the evidence showing that *33 Jones’s acts were planned and deliberate. Dr. Stewart’s mood disorder diagnosis is also weakened by the rebuttal evidence, as Dr. Scialli concluded that Jones does not suffer from a mood disorder under the DSM-IV criteria. Given the conflicting evidence of whether Jones even suffers from a mood disorder, along with the lack of any persuasive evidence connecting any mood disorder to the murders, the sentencer would have given minimal weight to Jones’s mood disorder evidence. See Poyson, 475 P.3d at 298; Dann, 207 P.3d at 629. (6) Impaired by drugs during the crimes: Contrary to the panel’s implication, see Am. Op. at 51, evidence that Jones was impaired by drugs during the murders is not “new.” The trial court knew that Jones was impaired by drugs. Indeed, the trial court found that he was under the influence of drugs and alcohol at the time of the murders and considered such fact a mitigating circumstance. Jones, 917 P.2d at 207. Thus, more evidence that Jones was impaired by drugs during the crimes would have been merely cumulative and made no difference. See Wong, 558 U.S. at 22. (7) Sexual abuse by step-grandfather: The allegations of sexual abuse are based on Jones’s self-report to Dr. Foy. Jones, 450 F. Supp. 2d at 1033. To begin, “[b]ecause of the obvious motive to fabricate, such self-serving testimony [would have been] subject to skepticism.” State v. Medrano, 914 P.2d 225, 227 (Ariz. 1996). The trial court found that Jones was not credible because “[i]n the past [he] *34 ha[d] shown that he [was] willing to lie if it benefit[ed] him.” Even Jones’s mother stated in her declaration that Jones “became a liar” and “would lie even in the face of circumstances where it was obvious he was lying.” It is therefore unlikely that the trial court would have believed Jones’s self-report with no persuasive corroboration. And there is none. While Peggy and Randy testified that Jones’s step-grandfather had introduced Jones to marijuana at a young age, which could suggest a deviant motive, such evidence does not necessarily point to sexual abuse. And during a 2001 interview with the Public Defender’s investigator, Peggy and Randy said they “never saw any indication [Jones] may have been sexually abused by anyone, nor were they aware of any sexual perpetrators in the family.” Moreover, “evidence that murders were planned or deliberate and not motivated by passion or rage decreases the mitigating effect of prior childhood abuse.” Poyson, 475 P.3d at 300. As discussed above, substantial evidence shows that Jones’s actions were planned and deliberate. Given this, along with Jones’s history of lying and the lack of any persuasive corroboration, the sentencer would have given Jones’s self-report of sexual abuse slight mitigating weight, if any. (8) Emotional and Physical Abuse by Randy: Although the trial court did not hear any allegations that Randy abused Jones, the extent of Randy’s abuse is unclear. Dr. Stewart’s report stated that Randy was “controlling” and would “hit Peggy and the children,” and Dr. Stewart testified that Randy admitted that he *35 physically abused Peggy, Jones, and Jones’s sister. Jones’s sister also reported that Randy had been physically abusive and very controlling. But there is conflicting evidence. In Jones’s military record from 1983, it states: “Relationship with step-father was described in this manner, ‘As far as I’m concerned he is my real dad, he’s the only one that has treated me good. He has never hit me or anything.’” Peggy’s declaration stated, “Randy was very strict,” but “Randy did not hit me or the children.” Randy admitted that, looking back, he believes he was verbally abusive to his children. Though the trial court didn’t know about the conflicting evidence, it knew that Jones had a chaotic and abusive childhood, which it found to be a mitigating circumstance. Jones, 917 P.2d at 207–08. Thus, the new evidence about Randy would have been more evidence supporting a mitigating circumstance that the trial court already considered in sentencing Jones. The evidence, then, would have been largely cumulative and given nominal weight. See Wong, 558 U.S. at 22; see also McDowell v. Calderon, 107 F.3d 1351, 1363 (9th Cir. 1997) (holding that there was no reasonable probability that the jury would have chosen a different sentence upon introduction of evidence of sexual and substance abuse because the jury chose a sentence of death after hearing similar evidence of defendant’s tragic childhood and severe physical abuse), amended and superseded in part by 116 F.3d 364 (9th Cir. 1997), vacated in part on other grounds by 130 F.3d 833 (9th *36 Cir. 1997) (en banc). Additionally, any mitigating effect would have been diminished given the substantial evidence showing that Jones’s actions were planned and deliberate. See Poyson, 475 P.3d at 300. In sum, the “new” mitigation evidence is far from overwhelming, and the district court found it “largely inconclusive or cumulative.” Jones, 450 F. Supp. 2d at 1043. The evidence of mental health issues and sexual abuse is equivocal and would have had little mitigating value with no persuasive evidence linking such evidence to the murders. The evidence of Randy’s emotional and physical abuse of Jones also would have made little difference, as there is conflicting evidence and, in any event, it is largely cumulative of the “chaotic and abusive childhood” evidence that the trial court knew about and considered in sentencing Jones. And any mitigating effect is diminished by the evidence that the murders were planned and deliberate. The remaining “new” mitigation evidence—Jones’s poly-substance abuse and impairment by drugs during the murders—is not even “new”; it is cumulative and would have had little effect. B. Aggravation Evidence Turning to the aggravating circumstances, the trial judge found three aggravating circumstances for each murder: (1) multiple homicides, (2) pecuniary gain, and (3) cruelty, heinousness, or depravity. Jones, 917 P.2d at 207. Each of these is entitled to substantial weight. And for Tisha’s murder, the aggravators are *37 even more substantial as the trial judge found a fourth aggravator: she was under fifteen years old. Id.17 The multiple homicides aggravator is the weightiest. The Arizona Supreme Court has “consistently given ‘extraordinary weight’ to this aggravator.” Poyson, 475 P.3d at 302. “Even when the multiple homicides aggravator is the only aggravator weighed against multiple mitigating factors, [the Arizona Supreme Court has] found the mitigation insufficient to warrant leniency.” Id. “The pecuniary gain aggravator is also especially strong and ‘weighs heavily in favor of a death sentence,’ when pecuniary gain is the ‘catalyst for the entire chain of events leading to the murders.’” Id. (citations omitted). The Arizona Supreme Court agreed with the trial court’s finding that pecuniary gain was the motive for the murders, as Jones wanted to leave Bullhead City because of impending warrants, Jones knew about Weaver’s gun collection, and Jones “murdered Tisha Weaver and Robert Weaver as part of a plan to obtain the gun collection and leave Bullhead City.” Jones, 917 P.2d at 215. The pecuniary gain aggravator is therefore especially strong here. The heinous, cruel, or depraved aggravator is also entitled to great weight given the brutal way Jones murdered Weaver and Tisha. See State v. McKinney, 17 As noted above, the panel barely engaged with the overwhelming aggravating evidence. Given the importance of this factor in a Strickland analysis, the panel’s omission materially weakens the applicable standard. — 38 — 426 P.3d 1204, 1207 (Ariz. 2018); Poyson, 475 P.3d at 302. Weaver fell to the ground after the initial blows, and he remained unconscious and bleeding for at least ten to fifteen minutes. Jones, 917 P.2d at 216. He then “regained consciousness and experienced pain and uncertainty about his fate.” Id. Even though Weaver was “helpless” and “more than likely . . . no longer physically capable” of trying to stop Jones, Jones inflicted gratuitous violence on Weaver by striking him in the head at least three more times with the bat. Id. at 217. Tisha also “experienced uncertainty about her fate,” as she hid under her parents’ bed because she was afraid and struggled with Jones for her life. Id. at 216. She was a “helpless victim,” being a seven-year-old child. Id. at 217. Still, Jones struck Tisha with the baseball bat with sufficient force to create a wound several inches wide, extending from her left ear to her left cheek. He then struck her a second time on the back of her head. After delivering these two fatal blows, [he] then asphyxiated her, far exceeding the amount of violence necessary to cause death. Id. at 218. C. Reweighing On one side, we have the mitigating circumstances that the trial court determined were insufficient to overcome the aggravators. Added to that is the “new” mitigation evidence, which is cumulative, inconclusive, and weak. On the other side, we have several aggravating circumstances that weigh heavily under Arizona law. On balance, there is simply no substantial likelihood of a different *39 result. The only difference between the sentencing profile before us and the one that was before the trial court is the addition of the deficient “new” mitigation evidence. This is precisely a case in which there is no Strickland prejudice, as the new evidence “barely . . . alter[s] the sentencing profile presented to the sentencing judge.” Strickland, 466 U.S. at 700. • * * The panel failed to consider all the evidence in evaluating the “new” mitigation evidence, and it failed to undertake a serious reweighing of the mitigation and aggravation evidence. That approach directly conflicts with Strickland and its progeny. The panel’s amended opinion improperly lowers Strickland’s “highly demanding” standard, Kayer, 141 S. Ct. at 523, as now, even questionable, weak, and cumulative mitigation evidence offered in post-conviction proceedings will be enough to overcome the weightiest of aggravating circumstances. The panel compounded this error by failing to review the district court’s factual and credibility findings for clear error, as mandated. In short, the panel established a new flawed approach for future panels to follow: in assessing Strickland prejudice, we can reweigh the evidence and make our own factual findings without regard to all the counterevidence and the district court’s findings. Because we should have fixed the panel’s exceptionally important errors, I respectfully dissent from our failure to take this case en banc. FILED NOV 7 2022 Jones v. Ryan, No. 18-99005 MOLLY C. DWYER, CLERK U.S. COURT OF APPEALS IKUTA, Circuit Judge, joined by CALLAHAN, VANDYKE, Circuit Judges, dissenting from the denial of rehearing en banc: I agree with Judge Bennett that even if the panel had been correct in conducting a de novo review of the state court’s decision, it erred in failing to defer to the district court’s factual findings. (Bennett, J., dissenting from denial of rehearing en banc). In my view, however, the panel had no business conducting such a de novo review in the first place. When a state court addresses some of the claims raised by a defendant, but not others, the unaddressed claims “must be presumed to have been adjudicated on the merits” by the state courts,” subject to rebuttal. Johnson v. Williams, 568 U.S. 289, 293 (2013). Here the state postconviction court rejected Jones’s claim that trial counsel was ineffective at sentencing without referencing Strickland v. Washington, 466 U.S. 668 (1984), or explaining its reasoning. Nothing in the state court’s opinion rebutted the presumption that it adjudicated the prejudice prong of Strickland on the merits, see Williams, 568 U.S. at 301. In this situation, “a habeas court must determine what arguments or theories supported or, as here, could have supported, the state court’s decision; and then it must ask whether it is possible fairminded jurists could disagree that those arguments or theories are inconsistent with the holding in a prior decision of this *2 Court.” Harrington v. Richter, 562 U.S. 86, 102 (2011). Under a proper deferential review, for the reasons explained by the district court, the state court could have reasonably determined that Jones did not establish prejudice under Strickland. See Jones v. Schriro, 450 F. Supp. 2d 1023, 1043 (D. Ariz. 2006), rev’d sub nom. Jones v. Ryan, 583 F.3d 626 (9th Cir. 2009). In reaching the issue of prejudice de novo, the panel mischaracterized the state court opinion and disregarded the admonitions of the Supreme Court to give such opinions proper deference, see, e.g., Harrington, 562 U.S. at 102. For these reasons, I dissent from our failure to take this case en banc.