Tommy R. Pruitt v. Ron Neal

TINDER, Circuit Judge. Tommy R. Pruitt appeals the district court’s denial of his petition for a writ of habeas corpus. A jury in Indiana state court convicted Pruitt of murdering a deputy sheriff and he was sentenced to death. His petition under 28 U.S.C. § 2254 raises four claims: (1) he is intellectually disabled and therefore categorically and constitutionally ineligible for the death penalty; (2) his trial counsel rendered ineffective assistance in violation of his Sixth Amendment rights by failing to present evidence to support his claim of intellectual disability; (3) his trial counsel rendered ineffective assistance by failing to investigate and present at the penalty phase mitigating evidence regarding his schizophrenia and its effects; and (4) the prosecutor violated Pruitt’s right to due process by reciting a poem about the death of a police officer and comparing Pruitt to notorious murderers in the closing argument of the penalty phase of trial, and appellate counsel was ineffective in failing to raise this as an error on appeal. With regard to the second ineffective-assistance-of-counsel claim ((3) above), Pruitt challenges only counsel’s presentation of evidence at the penalty phase of trial; he is not asserting ineffectiveness in failing to seek a determination of guilty but mentally ill or in otherwise presenting mental health evidence at the guilt phase of trial. We conclude that Pruitt has established that he is intellectually disabled and categorically ineligible for the death penalty and that trial counsel were ineffective in their investigation and presentation of evidence that Pruitt suffered from schizophrenia. We therefore reverse the district court’s judgment and remand for further proceedings not inconsistent with this opinion. I. BACKGROUND A. Underlying Crime These facts are taken from the Indiana Supreme Court decision affirming Pruitt’s conviction and sentence. Pruitt v. State, 834 N.E.2d 90 (Ind.2005) (Pruitt I), cert. denied, Pruitt v. Indiana, 548 U.S. 910, 126 S.Ct. 2936, 165 L.Ed.2d 962 (2006). On June 14, 2001, Morgan County Deputy Sheriff Daniel Starnes stopped Pruitt’s vehicle. Starnes obtained Pruitt’s driver’s license and registration, called that information in, and was told that a recent robbery report suggested Pruitt might be in possession of stolen weapons. When Starnes approached Pruitt’s vehicle a second time, Pruitt emerged with a handgun, and the two exchanged gunfire. Starnes was struck by five shots. He was taken to the hospital and underwent surgery. He later developed an infection and died. The state ultimately charged Pruitt with murder, attempted murder (Pruitt also shot Ryan Starnes, Deputy Starnes’s son, who was accompanying his father as part of a college internship), and related offenses. The state sought the death penalty because Starnes was a law enforcement officer killed in the line of duty. B. State and Post-Conviction Proceedings The trial court appointed attorneys William Van Der Pol, Jr. and Douglas Garner to represent Pruitt. Before trial Pruitt sought a determination of intellectual disability in a death sentence case, requesting that the court order an evaluation and dismiss the death penalty. The court or *251 dered an evaluation of Pruitt and appointed forensic psychologist Dr. George W. Schmedlen, Ph.D., J.D., for that purpose. The trial court conducted a week-long evidentiary hearing at which Pruitt presented two expert witnesses, clinical neu-ropsychologist Dr. Bryan A. Hudson, and clinical neuro-psychologist and professor of psychology Dr. Charles J. Golden, Ph.D., as well as non-expert testimony and other evidence in an effort to establish his intellectual disability. Pruitt’s experts testified that Pruitt is intellectually disabled based on his IQ test scores and their assessment of his adaptive functioning. However, Dr. Schmedlen and the state’s expert, psychologist Dr. Martin G. Groff, Ph.D., testified that Pruitt was not intellectually disabled. The trial court found that Pruitt “does not have significantly subaverage intellectual functioning,” Pruitt I, 834 N.E.2d at 103, noting that at most, “Pruitt’s functioning would be considered border-line — not [intellectually disabled],” id. at 104. The trial court also found “that the defense has failed to prove by clear and convincing evidence that Mr. Pruitt has substantial impairment of adaptive behavior.” Id. at 108. The trial court was “particularly impressed with Mr. Pruitt’s ability to function as a carpenter, obtain a commercial driver’s license, perform duties of an over the road truck driver and fill out applications for employment.” Id. at 108-09. Therefore, the court denied Pruitt’s motion to dismiss the death penalty. At the guilt phase of the trial, the jury convicted Pruitt of murder, attempted murder, and related offenses. At the penalty phase, the defense presented evidence of Pruitt’s background and intellectual disability. Drs. Hudson and Golden again testified that Pruitt is intellectually disabled, and Dr. Golden also testified about Pruitt’s mental illness. The jury found that Deputy Starnes was a law enforcement officer acting in the course of his duty at the time he was murdered and that this aggravating circumstance outweighed any and all mitigating circumstances, and recommended a sentence of' death. In imposing sentence for the convictions other than murder, the trial judge considered the “possible mitigating factors as presented in the[] proceedings” and found that none of them “constitute a mitigating circumstance.” Trial Tr. 6455. The judge also concluded that “Pruitt does not have a mental condition which would justify his actions or in any way mitigate for purposes of sentencing” and that the “aggravating factors outweigh any possible mitigating factors.” Id. at 6455-56. The judge sentenced Pruitt to a total of 115 years on the other offenses. In imposing sentence for murder, the judge found that the jury determined that the State had proven beyond a reasonable doubt the aggravating circumstance that Starnes was a law enforcement officer acting in the course of his duty, that this aggravating circumstance outweighed any and all mitigating circumstances, and that a sentence of death should be imposed. The judge accordingly sentenced Pruitt to death. Pruitt appealed his murder conviction and sentence directly to the Indiana Supreme Court; the court affirmed. Pruitt I, 834 N.E.2d 90. Although that court concluded that the trial court’s standard for assessing adaptive behavior was too restrictive and that the record did not support the trial court’s conclusion with respect to adaptive behavior, id. at 109-110, it decided that the evidence supported the trial court’s conclusion with respect to the intellectual functioning. The court explained: While some of Pruitt’s scores suggest significantly subaverage intellectual functioning, others do not. • In addition to this data, the trial court found that *252 Pruitt was able to fill out applications for employment and to have the capacity, if not the will at all times, to support himself. In light of the inconsistent IQ scores and the other evidence cited by the trial court, the trial court’s finding that Pruitt did not meet the statutory test is consistent with this record. Id. at 106. The Indiana Supreme Court therefore upheld the trial court’s conclusion that Pruitt was not intellectually disabled. Id. at 106, 110. Justice Rucker dissented, concluding that Pruitt had shown that he is intellectually disabled and therefore a death sentence was unconstitutional and statutorily impermissible. Id. at 123-26. Next, Pruitt petitioned for post-conviction relief. At the hearing on his petition, he presented testimony from newly retained experts as well as from psychologist Dr. Dennis Olvera, Ph.D., who had been retained by trial counsel to conduct IQ testing of Pruitt but had not testified at trial. Pruitt also presented evidence from his trial attorneys and appellate attorney, family, and educators. The post-conviction court denied Pruitt’s petition and motion to correct errors, finding as to the intellectual disability claim that Pruitt’s “additional evidence ... presented at the post-conviction hearing was largely cumulative of the evidence developed before and [during] trial and is entirely unpersuasive.” PCR App. 688. (The court also found the claim barred by res judicata. Id.) The Indiana Supreme Court affirmed. Pruitt v. State, 903 N.E.2d 899 (Ind.2009) (Pruitt II), reh’g denied, 907 N.E.2d 973 (Ind.2009). The court found that “Pruitt has offered no evidence undermining the correctness of the trial court’s and this Court’s findings that he is not [intellectually disabled]” and therefore concluded that his death sentence was not unconstitutional under Atkins v. Virginia, 536 U.S. 304, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002). Pruitt II, 903 N.E.2d at 938 (quotation omitted). Justice Rucker again dissented on the basis of Pruitt’s intellectual disability. Id. at 940. Pruitt petitioned for writ of habeas corpus under § 2254 in the United States District Court for the Northern District of Indiana. The district court heard oral argument and then denied the petition. Pruitt v. Wilson, No. 3:09cv380RLM, 2012 WL 4513961 (N.D.Ind. Oct. 2, 2012). Pruitt moved the court to alter or amend its decision; the motion was denied. The district court granted Pruitt a certificate of appealability. C. Pretrial and Guilt-Phase Evidence 1. Intellectual Functioning Before he turned 18, Pruitt took two Lorge-Thorndike group-administered IQ tests. In March 1973, he scored a verbal IQ of 64, and a non-verbal IQ of 65. In December 1976, he scored a verbal IQ of 64 and a non-verbal IQ of 63. These scores fall within the intellectually disabled range. However, the state’s expert Dr. Groff testified that he would give little weight to these scores; Dr. Hudson agreed that individually administered tests are better indicators of an individual’s ability. Nonetheless, the Lorge-Thorndike test is designed to measure general intelligence, and Dr. Hudson explained that the research demonstrates an “extremely high correlation” between the Lorge-Thorndike and Weschler series of intelligence tests. Both Drs. Groff and Golden agreed that since Pruitt obtained essentially the same scores on the two tests, he gave his best effort. Dr. Golden indicated that these scores suggested that Pruitt’s IQ was “fairly stable” at that point in time. Trial Tr. 1550. Both Drs. Hudson and Golden stated that the IQ scores obtained as a child would most accurately measure *253 Pruitt’s intelligence before age 18. Id. at 1355-56 (Hudson stating that “IQ is meant to be diagnosed in childhood.”); id. at 1551 (Golden stating that in assessing IQ, scores obtained before age 18 are “absolutely” more important than scores obtained after that age). Pruitt took two academic achievement tests while in school. In March of 1975, he took the Otis-Lennon School Ability Test and scored 81. Dr. Schmedlen testified that this score was inconsistent with intellectual disability. However, Dr. Schmed-len was unaware that the Otis-Lennon was known as the “Otis-Lennon School Ability Test” and that the Otis-Lennon score was not given as an IQ score. He also acknowledged that the Otis-Lennon test was not of the same quality as a Weschler or Stanford-Binet intelligence test. Dr. Hudson stated that the Otis-Lennon test was an achievement test normed by grade rather than age. He explained that Pruitt’s score was compared to the scores of children two years younger than Pruitt because he had been held back in school twice by the time he took the test. Dr. Golden similarly testified that an academic achievement test is not an appropriate substitute for an IQ test: although an achievement test “roughly correlated somewhat with IQ, it is not an IQ test at all.” Trial Tr. 1546. He stated that the Otis-Lennon score overestimated Pruitt’s IQ score by 15 to 25 points because the test was an achievement test comparing scores among students in the same grade. See id. at 1518-19. In April 1975, Pruitt took the Iowa Basic Test, another academic achievement test. His score on that test was consistent with his Otis-Lennon score; but again, his scores were normed by grade rather than age. Had his achievement test scores been adjusted for his age, Dr. Golden testified that they would fall within the range for intellectual disability. At the age of 19 while incarcerated on another matter, Pruitt took a Revised Beta intelligence test and scored 93. The test is group-administered in fifteen minutes. Trial Tr. 672. Dr. Schmedlen testified that such a score was inconsistent with having significantly subaverage intellectual functioning; instead, the score fell within the average intellectual functioning range. However, Dr. Schmedlen acknowledged that individually administered tests are considered more reliable than group administered tests. Trial Tr. 659-60. While Dr. Schmedlen testified that he believed the test was reliable, he did not actually determine whether the Revised Beta was a reliable test. Id. at 672. And he acknowledged that the test was not as reliable as the Weschler series or Stanford-Binet intelligence tests. Furthermore, Drs. Hudson and Golden testified that the Revised Beta is a non-verbal test developed by the Army in World War II for testing non-English speaking persons. It was normed on prisoners, not on the average population. Thus, Dr. Golden stated that the Revised Beta “is not an accurate test, it is not well regarded in the field, and it is not well accepted in the field as a general test of intelligence.” Id. at 1541. According to Dr. Golden, a Revised Beta score “severely overestimates” an individual’s IQ; if a Revised Beta is the only score available in a forensic file, Dr. Golden subtracts 20 to 30 points from the score to obtain what he would consider to reflect actual intellectual functioning. Id. at 1543. In April 2002, Pruitt obtained a full-scale IQ score of 76 on the Weschler Adult Intelligence Scale, Third Edition (WAIS) administered by his own expert Dr. Olv-era. Trial counsel had sought Dr. Olvera’s evaluation because they intended to pursue a determination that Pruitt is intellectually disabled and not eligible for a death sentence. The WAIS has a standard error of *254 measurement (SEM) of five points that placed Pruitt’s score within the range of 71 to 81. However, Dr. Olvera advised trial counsel that Pruitt’s full-scale IQ score of 76 did not meet the intelligence functioning prong of intellectual disability. Dr. Schmedlen stated that this score placed Pruitt in the “borderline range.” Dr. Hudson said there was a one-point error in scoring the test and that Pruitt was taking an antipsychotic medication (Trilafon) that Hudson believed enhanced Pruitt’s abilities. Dr. Golden agreed, stating that if a person with a psychiatric disorder is given antipsychotic medication, his IQ scores “may go up significantly.” Trial Tr. 1500. In addition, the medical literature supports the conclusion that an antipsychotic drug could enhance Pruitt’s performance on an IQ test because of improvements in attention and focusing. The trial court found that the 76 on the WAIS placed Pruitt above the intellectually disabled level of intellectual functioning. The Indiana Supreme Court noted that the trial court had found insufficient evidence to show what, if any, effect the Trilafon may have had on Pruitt’s performance on the test. Pruitt I, 834 N.E.2d at 105. In February 2003, Dr. Golden administered a Stanford-Binet (4th ed.) individually administered IQ test to Pruitt; he obtained a full-scale IQ score of 65. The test has an SEM of six points, placing Pruitt’s IQ score within the 59-71 range. Subav-erage intelligence is measured by the Stanford-Binet as a score below 69. Dr. Schmedlen testified that Pruitt’s score indicated that he is intellectually disabled. Trial Tr. 668. Dr. Golden testified that alternative scoring methods for the test could have produced scores of 67 or 69. Both Drs. Golden and Hudson stated that the Stanford-Binet was more sensitive and accurate than the WAIS in assessing individuals with very high or very low IQs. In July 2003, Dr. Schmedlen administered the WAIS to Pruitt; Pruitt scored a full-scale IQ of 52. That score, Dr. Schmedlen testified, was not an accurate measure of Pruitt’s intellectual functioning because he did not believe that Pruitt gave his full effort on the test. Drs. Golden and Hudson agreed that the score likely underestimated Pruitt’s IQ because of malingering. They also noted' that Pruitt was no longer being administered antipsychotic medication and had been in solitary confinement just before taking the test, both of which could have increased the likelihood of psychotic symptoms that would lower his performance on the test. As support, they indicated that Pruitt’s scores varied from the prior WAIS on subtests that were particularly sensitive to anxiety, depression, or psychosis, but showed little variance on those subtests that were more stable. Dr. Schmedlen testified at the pretrial hearing that Pruitt was not intellectually disabled, relying on IQ tests, achievement tests, an assessment of his adaptive functioning, and Pruitt’s ability to hold jobs and perform other tasks. Dr. Schmedlen believed that at most Pruitt would be in the borderline range of intellectual functioning. He took into account Pruitt’s 93 on the Revised Beta, a score that Dr. Schmedlen thought was inconsistent with intellectual disability. He also thought that Pruitt’s 81 on the Otis-Lennon test should be interpreted as an IQ score. Dr. Groff also testified at the pre-trial hearing that Pruitt was not intellectually disabled. However, Dr. Groff did not meet or test Pruitt; nor did he interview anyone about Pruitt. Based only on his review of records in the case, Dr. Groff stated that his impression was that Pruitt was “probably within the borderline intellectual functioning range.” Tr. 1463. Yet Dr. Groff conceded that he could not ad *255 dress Pruitt’s intellectual functioning “with certainty.” Id. 2. Adaptive Functioning The American Association on Mental Retardation (AAMR) (now the American Association on Intellectual and Developmental Disabilities) defined a “substantial impairment in adaptive functioning” as limitations in two or more of the adaptive skill areas: communication, self-care, home living, social, community use, self-direction, health and safety, functional academics, leisure, and work. Dr. Schmedlen administered an objective test of adaptive functioning and concluded that Pruitt met the adaptive behavior prong under the AAMR definition. Specifically, Dr. Schmedlen stated that Pruitt qualified under two adaptive skill areas: work and self-direction. According to Dr. Golden, because the test was normed against individuals with intellectual disability, it made Pruitt appear higher functioning then he actually was. Dr. Hudson testified based on his review of Pruitt’s social, educational, and work history as well as two clinical evaluations that Pruitt showed significant deficits in six adaptive skills areas: communication, home living, social skills, functional academics, health and safety, and work. In addition to the expert testimony, Pruitt presented lay testimony that supported the experts’ opinions as to his adaptive functioning. Pruitt was slow to develop as a child in terms of such abilities as walking, talking, and potty-training. He had difficulty following instructions, playing board games and participating in sports because he had trouble understanding the rales. Pruitt encountered difficulty in school beginning in the second grade. He was held back twice in the first three grades of elementary school; he was placed in special education classes; and he was socially promoted. He dropped out of school at the age of 16 while in the eighth grade. He was described as mentally slow. Other children made fun of him, calling him retarded and dumb, but he did not seem to understand that he was being teased. As an adult, he was unable to live on his own; his parents and family members took care of him. For example, they took care of his money and made doctor and dental appointments for him. The evidence established that Pruitt has held many jobs, but none of them required high intellectual functioning. In the 1980s, he worked at a fast food restaurant, a truck stop, and in construction. In 1990, Pruitt received a union carpenter’s card; he was a member of the union for almost eight years. During the 1990s, he worked briefly as a carpenter’s pre-apprentice for Carlino Construction and he worked a few years as a long-distance track driver. Pruitt also worked for several different construction companies. His father got him a job in construction, but he was hired only because the employer considered his father a valued employee. Pruitt worked primarily in demolition and clean-up work. He had obtained a journeyman’s card but he did not go through an apprenticeship program. The employer, Pete Carlino, testified that Pruitt “really didn’t have carpentry skills,” did not learn carpentry skills, and needed to “have his dad right there ... to tell him what to do continually.” Trial Tr. at 951. Pruitt did not do a good job. When Pruitt’s father retired, Pruitt was fired because he was unable to do basic work without constant supervision, and he could not communicate well with other workers. As noted, Pruitt also worked as a truck driver. In order to do so, he obtained a commercial driver’s license with several endorsements that required him to score at least an 80% on several written tests. *256 He was discharged from his first trucking job after he left one of the company’s trucks on the scales at a weigh station and walked home. Pruitt’s employment at another trucking job was terminated when he was caught selling a load of copper wire to a scrap dealer rather than delivering it to the intended destination. While working as a truck driver on another occasion, Pruitt drove his semi-truck to California and visited his cousin. He parked the semi in the middle of the street. When his cousin told Pruitt he could not park there because he was blocking the street, Pruitt responded that there were no signs saying that he could not park there. Pruitt also complained that his vehicle was a piece of junk, but it was brand new; and according to his cousin, an experienced truck driver who had ridden in the vehicle, there was nothing wrong with it. D. Penalty Phase Evidence The guilt-phase evidence was incorporated for the jury’s consideration at the penalty phase. The state’s only witness was the former Morgan County Sheriff who testified as to Deputy Starnes’ duties as a warrant officer. At the penalty phase, Pruitt’s counsel presented the expert testimony of Dr. Golden, who testified about Pruitt’s intellectual disability and neuropsychological problems. Dr. Golden had interviewed Pruitt only once, when he administered his intellectual disability testing. Counsel asked Dr. Golden to go through Pruitt’s prison records and discuss Pruitt’s prior diagnoses. Dr. Golden stated that in 1996 while Pruitt was incarcerated for another crime, the federal Bureau of Prisons (BOP) had diagnosed him with “schizotypal personality disorder ... an Axis II mental illness.” Trial Tr. 6042-43. Dr. Golden testified that “psychotic episodes should be the exception rather than the rule” for persons suffering from schizotypal personality disorder, and that if such episodes “were the main symptoms, then you would be diagnosing him as schizophrenic or something similar to that,” id. at 6045-46. Dr. Golden further testified that in August 2001, about two months after Pruitt was arrested for killing Officer Starnes, an Indiana Department of Corrections (DOC) psychiatrist diagnosed Pruitt as suffering from “schizophrenia, chronic undifferentiated type compensated residual.” Trial Tr. 6053, 6221-22 (explaining that a psychologist worked on Pruitt’s case but a psychiatrist made the actual diagnosis). Dr. Golden said that the symptomatology for that diagnosis were “actually similar to what the other psychologist was saying when he diagnosed [Pruitt] as schizotypal.” Id. at 6054. Dr. Golden explained that this diagnosis means that the provider thinks that at one time that [Pruitt] was actually schizophrenic ... but that right now most of those serious symptoms are not showing and ... [Pruitt is] somewhere now in between a personality disorder and the Axis I schizophrenia. He [is] not suggesting [Pruitt is] actively schizophrenic, but he is expressing his belief that [Pruitt] is capable of easily becoming schizophrenic at some time in the future. Id. at 6054-55. Dr. Golden said that Pruitt was prescribed Trilafon, an antipsy-chotic, explaining that “[t]hey were obviously very concerned of his ability to become psychotic.” Id. at 6056. According to Dr. Golden, the DOC psychiatrist who diagnosed Pruitt did a “detailed evaluation and screening,” seemed to have “check[ed] out [Pruitt’s] past,” id. at 6223, and must have been “pretty certain” since he prescribed medication for schizophrenia. Dr. Golden concluded that Pruitt’s mental illness was “in the schizophrenia spectrum of disorders.” Trial Tr. 6057. He *257 said that one “can have psychotic episodes if you’re schizoid or schizotypal.” Id. He opined based on Pruitt’s history that his illness was “not severe enough to be called schizophrenic, so I go with the schizotypal or the schizoid diagnosis.” Id. Dr. Golden testified about the symptoms of schizoid personality disorder and its debilitating effects, stating that Axis II personality disorders “can decompensate into a psychosis, into acute psychosis, where the effects are identical” as those of an Axis I mental illness. Id. at 6062. According to Dr. Golden, individuals with such a personality disorder will “start kind of going off the deep end,” and when they do, you have to intervene “before something bad happens to them or to somebody else.” Id. at 6063. Dr. Golden said Pruitt had decompensated for at least six months before the crime, becoming increasingly paranoid: “They cross over that line between adequate reality contact and inadequate reality contact.” Id. at 6076-77. Dr. Golden stated that “clearly ... [Pruitt], for at least six months if not longer, had been decompen-sating” and given “the degree of paranoia that he was developing ... something bad was eventually going to happen....” Id. at 6093. The jury asked Dr. Golden whether he had any firsthand knowledge of Pruitt’s thought processes, including his thinking while committing the crime. Id. at 6247-48. Dr. Golden responded that they “talked a little bit about that” and Pruitt “said to me basically what’s already in the record.” Id. at 6248-49. E. State Post-Conviction Evidence 1. Intellectual Disability In his state post-conviction proceeding, Pruitt presented evidence from two psychologists with expertise in intellectual disability, Dr. Denis Keyes and Dr. Dennis Olvera. Dr. Keyes, who had thirty-two years’ experience evaluating and working with the intellectually disabled, reviewed all of the testing since Pruitt’s childhood and conducted his own intelligence testing of Pruitt. In 2006 Dr. Keyes administered the Stanford-Binet (5th ed.) intelligence test; Pruitt obtained a full-scale IQ score of 65, representing a score between 61 and 69, which Dr. Keyes said was within the range of intellectual disability. He said ■that the testing indicated that Pruitt’s intelligence was “significantly subaverage.” However, Dr. Keyes conceded that it was safe to assume that Pruitt was aware while taking this test of the potential ramifications of his score. In Dr. Keyes’s opinion, one IQ score above 75 does not necessarily mean that a person is not intellectually disabled. Dr. Keyes emphasized the importance of clinical judgment in assessing intellectual disability, stating that persons with intellectual disability try to hide it because of the stigma associated with it. He conducted a test of Pruitt for malingering; the result suggested that Pruitt was not malingering. Dr. Keyes also said that there were signs to look for when testing to determine whether an individual might be malingering, and none of those signs appeared with Pruitt. Addressing Pruitt’s Revised Beta score of 93, Dr. Keyes explained “that test is not considered a valid intelligence indicator by any professional psychologist who knows the literature and/or the test.” PCR App. 773. Dr. Keyes administered three standardized tests to assess Pruitt’s adaptive functioning. He also interviewed Pruitt’s parents, sisters, two of his special education teachers, and family friends, and reviewed records as well as the trial testimony. Dr. Keyes concluded that Pruitt’s adaptive functioning was significantly impaired in virtually all skill areas and estimated that his overall adaptive functioning was at about the level of a child between six and *258 eight years of age. Id. at 774. Based on his examination of Pruitt, his review of Pruitt’s school records including test scores, and his own testing, Dr. Keyes opined that Pruitt is intellectually disabled and had been since birth or at least early childhood. Dr. Olvera, who had forty years’ experience working with the intellectually disabled and extensive experience working with the mentally ill, testified that he was contacted by Pruitt’s trial counsel in 2002 and administered the WAIS to Pruitt. At that time, based on Pruitt’s full-scale IQ score of 76, Dr. Olvera concluded that Pruitt did not meet the intellectual disability standard. However, Dr. Olvera had not been provided with Pruitt’s other test scores. Dr. Olvera stated that had he known of the other lower IQ scores, he would have wanted to know why his results were higher. He explained that Pruitt’s score of 76 had to be viewed in context with his other test scores. Dr. Olvera also said that when he administered the WAIS in 2002, Pruitt was taking Trila-fon, an antipsychotic drug. Olvera opined that “[t]here is a good possibility that the fact that [Pruitt] was medicated when I tested him produced the scoring pattern that was considerably higher than would have been produced had he not been medicated.” PCR Tr. 41. Dr. Olvera also testified that Pruitt’s score on the WAIS should be interpreted in light of the “Flynn Effect,” which accounts for gains in IQ scores over time. 1 ' If taken into account, Pruitt’s score was reduced to approximately 74. Dr. Olvera also testified that the Revised Beta is not an acceptable measure of intelligence in forensic cases because it is not a multifaetorial intelligence test. He agreed that the Otis-Lennon is not an acceptable measure of intelligence either, explaining because it is a group-administered test. According to Dr. Olvera, the Wechsler series of intelligence tests and the Stanford-Binet test are considered the gold standard in forensic cases'. Dr. Olv-era conducted two malingering screens; both indicated that Pruitt did not malinger. Dr. Olvera opined that Pruitt met the intellectually disabled criteria and that there was a reasonable clinical certainty that he was functioning in that range at the time of the crimes. Pruitt presented the affidavit of Mary Mann, Director of Special Education Services for the school district he had attended as a child. The affidavit established that Pruitt was returned to special education classes in the eighth grade and that this was the last grade he attended. He dropped out of school at the age of 16. Dr. Mary Jo Dare, an associate professor teaching courses on special education, testified at the post-conviction hearing. She had experience teaching children with cognitive disabilities since 1967, and she was the former Director of Special Education for Indianapolis Public Schools. Dr. Dare reviewed Pruitt’s school records related to his special education and resource class placements. She explained that when Pruitt was in school, it was not uncommon to place children with special education needs in general education classes, without adequate support, and with very little understanding on the part of teachers as to how to teach such students. She saw no rational explanation for withdrawing *259 Pruitt from special education classes and testified that the records showed that he was socially promoted. Dr. Dare concluded that Pruitt’s “school records overall, both his IQ testing as well as his functioning in society and in school, they clearly point to a student who qualifies under the special education category of mild [intellectually disabled].” PCR Tr. 329. Pruitt also presented expert testimony from Marie Dauseh, the long-time director of an organization that provides programs and services for adults with developmental disabilities. Dauseh explained that individuals with mild intellectual disability are able to find employment in a variety of fields and are able to obtain a driver’s license, even a commercial driver’s license. 2. Mental Health At the post-conviction hearing, Dr. Olv-era testified that when he conducted his clinical interview of Pruitt, he observed signs of possible psychosis and learned that Pruitt was taking anti-psychotropic medication. Dr. Hudson also had noticed during his pre-trial testing of Pruitt that some of Pruitt’s responses to questioning were not “clearly lucid.” Trial Tr. 1215. Prior to the start of trial, Dr. Olvera had recommended that trial counsel contact an expert “in dealing with psychosis, such as schizophrenia.” PCR Tr. 37. But trial counsel did not do so. Also at the post-conviction hearing, Pruitt presented testimony from three other experts regarding his mental health: forensic psychiatrist Dr. Philip Coons, clinical and forensic psychiatrist Dr. James Ballenger, and forensic clinical psychologist and neuropsychologist Dr. David R. Price. Dr. Coons had over thirty years’ experience evaluating the mentally disturbed. Based upon an evaluation of Pruitt, interviews with his family members, the reports written by Pruitt’s other post-conviction experts, and a review of Pruitt’s social history, Dr. Coons opined that Pruitt suffered from Axis I paranoid schizophrenia, which is characterized by delusions, hallucinations, and thought disorder. Dr. Coons testified that at the time of the crime, Pruitt was suffering from paranoid schizophrenia and intellectual disability, he was under the influence of extreme mental or emotional distress, and his capacity to conform his conduct to the law was substantially impaired because of mental illness. Dr. Coons stated that pri- or, to developing schizophrenia, Pruitt had a personality disorder, described as a thought disorder without the psychotic symptoms of hallucinations and delusions. According to Dr. Coons, “all of the symptoms [of] schizotypal personality disorder are included under schizophrenia.” PCR Tr. 458. Dr. Coons found no signs of malingering, although he performed no tests to guard against malingering. Dr. Ballenger had approximately thirty-five years’ experience working with the mentally ill and lectured medical students on schizophrenia. He interviewed Pruitt twice. Based on information from Pruitt, his family members, and prison records, Dr. Ballenger concluded that Pruitt suffered from paranoid schizophrenia. Dr. Ballenger also opined that Pruitt has borderline intellectual disability. According to Dr. Ballenger, his examination showed that Pruitt met “every single one of the diagnostic criteria” for schizophrenia, including delusions and hallucinations, PCR Tr. 552-56, and had “a very positive family history” for schizophrfenia. Id. at 563. Dr. Ballenger said that Pruitt had two delusional systems, one involving super beings from outer space and the other involving a giant conspiracy of police and judges somehow getting lots of money by arresting people. Id. at 553-54. Although there are relatively effective tests for malinger *260 ing, Dr. Ballenger did not test Pruitt for malingering. Nonetheless, Dr. Ballenger found no evidence of malingering in his examinations of Pruitt. Dr. Ballenger said that Pruitt “works very hard to hide that he’s schizophrenic, he works very hard to hide that he has [intellectual disability].” Id. at 567. Dr. Ballenger stated that the diagnosis of “schizotypal personality disorder” is “either the first two, three steps of schizophrenia, or is actually the beginning of schizophrenia. Schizotypal personality is just quarter-strength schizophrenia.” Id. at 587-88. He testified that Dr. Golden’s diagnosis of “a decompensated psychotic state of somebody with a schizotypal personality” was “only a hair different from what I would have said.” Id. at 589. Dr. Ballenger stated that Dr. Golden described the same behaviors that he and Drs. Coons and Price used to diagnosis Pruitt with Axis I schizophrenia; Dr. Golden just used “slightly” different terminology. Id. According to Dr. Ballenger, in the months before the crime Pruitt became increasingly paranoid, to the point where he was convinced his mother was trying to poison him. Pruitt told Dr. Ballenger that at the time of the shooting, he wanted guns because “I’m afraid of people and they make me feel safer.” Id. at 597. Pruitt believed the police “were watching him all the time” and he was “ready to go to Montana to fully withdraw.” Id. Dr. Ballenger testified that it was “grandiose psychotic logic” to “point a gun at Officer Starnes, and expect the officer to give his gun up and walk away.” Id. at 598. Dr. Ballenger said that Pruitt believed the aliens “planned the whole thing” and “were controlling both his and Officer Starnes’ mind,” and that “it couldn’t be stopped.” Id. Pruitt’s schizophrenia, Dr. Ballenger opined, “substantially interferes with his thinking, feeling, and action.” Id. at 596. Dr. Ballenger stated that Pruitt was under the influence of an extreme mental or emotional disturbance at the time of Deputy Starnes’ murder, id. at 596-99, and as a result of mental disease, his capacity to appreciate the criminality of his actions or to conform his conduct to the requirements of law was substantially impaired, id. at 599. Dr. Price diagnosed Pruitt with schizophrenia, schizotypal personality disorder, and mild intellectual disability. He testified that Pruitt “showed odd signs, ... magical thought, [and] paranoia” that progressed into the active phase with delusions and hallucinations. PCR Tr. 702. Dr. Price thought that Pruitt “had a schi-zotypal personality disorder that developed into schizophrenia.” Id. at 705. When asked whether his diagnosis of schi-zotypal personality disorder premorbid meant that Dr. Golden was right, Dr. Price answered, “Very often people that develop schizophrenia have a premorbid and actually coexisting schizotypal personality disorder.” Id. Pruitt was not malingering his mental illness or intellectual disability, Dr. Price opined. He explained that because of his intellectual ability, Pruitt would not have the sophistication to portray schizophrenia during their interaction. Dr. Price said that Pruitt’s school history, work history, and IQ test results supported his conclusion. Dr. Price believed that Pruitt was in an active phase of schizophrenia at the time of the crime and that “he had a significant psychiatric disorder that significantly impaired his ability to conform his behavior to the requirements of the law.” PCR App. 826. Dr. Price testified that Pruitt was under the influence of an extreme emotional or mental disturbance at the time of the crime and that his capacity to appreciate the criminality of his conduct or conform his conduct to the require *261 ments of the law was substantially impaired because of mental disease or defect. PCR Tr. 708-09. Letitia Haywood, a clinical therapist and licensed social worker, investigated and prepared an extensive social history of Pruitt. The parties stipulated to the admission of her social history for the purpose of showing what could have been prepared at the time of trial. Haywood’s social history concluded that Pruitt has mental disabilities, including delusional, disorganized, and disturbed thought processes. 3. Defense Theories At the post-conviction hearing, trial counsel Van Der Pol, who was primarily responsible for the mitigation part of the case, testified about the defense theory at the penalty phase. Van Der Pol described the theory as “number one” that Pruitt is intellectually disabled, which would entitle him “to great mitigating weight,” PCR Tr. 214-15; “[n]umber two” that Pruitt “was suffering a serious mental illness at or around the time of the crime,” id. at 215; and “[fjinally” that Pruitt “had serious brain damage, brain injury, [or] brain dysfunction.” Id. Based on all these things, counsel attempted to prove that Pruitt did not deserve the death penalty. Id. at 215-lb. Trial counsel Garner testified at the post-conviction hearing that the defense theory was that because of his intellectual disability, Pruitt should not be held responsible to the same extent as an individual with normal intelligence. Id. at 513. Trial counsel Van Der Pol testified that Dr. Schmedlen had reported seeing symp-tomology consistent with schizophrenia, but then for unknown reasons, Dr. Schmedlen decided that Pruitt did not have schizophrenia. PCR Tr. 180. Counsel also said that in 2003 Dr. Golden concluded that no additional neuropsychological testing was necessary or appropriate. Id. at 196. Yet counsel also recalled that Dr. Golden had testified he did not have enough, information to make a diagnosis of schizophrenia. Id. at 200. Counsel did not consider asking Dr. Golden if he believed Pruitt met the criteria for the statutory mitigating factors related to mental illness. Dr. Golden, an expert in intellectual disability, had no experience in the assessment or treatment of individuals with psychotic disorders. I. Post-Conviction Rulings After considering Pruitt’s new evidence, the post-conviction court reaffirmed its pri- or rulings on the intellectually disability issue, concluding that Pruitt is not intellectually disabled. PCR App. 682. The court found that “Pruitt’s post-conviction [mental health] evidence is substantially cumulative of the testimony and opinion presented at trial.” Id. at 665. Discussing the evidence relating to mental illness, the court noted that a DOC psychologist had “diagnosed Pruitt as suffering from schizophrenia, chronic undifferentiated residual uncompensated.” Id. at 671. The court also noted that Dr. Golden “explained that this diagnosis means the psychologist believed at one time Pruitt was schizophrenic but that at the time of diagnosis he was not actively schizophrenic” and that “Dr. Golden had concluded that a diagnosis of schizophrenia was inappropriate for Pruitt, and that he actually suffered from either schizoid or schizotypal personality disorder.” Id. The court referred to the post-conviction hearing testimony from Drs. Coons, Ballenger, and Price that Pruitt suffered from schizophrenia. Id. at 673. Based on the evidence presented at trial and post-conviction, the court concluded “that trial counsel were not ineffective in selecting experts and preparing mental health evidence for use in the penalty *262 phase or sentencing.” Id. at 674. The court did “not find that the quality of evidence presented at post-conviction was more credible or persuasive than that presented at trial”; therefore, it found that Pruitt did not show a “reasonable probability of a different sentence.” Id. Pruitt appealed. The Indiana' Supreme Court concluded that “ Truitt has offered no evidence undermining the correctness’ of the trial court’s and this Court’s findings that he is not [intellectually disabled].” Pruitt II, 903 N.E.2d at 938 (quoting PCR App. 689). Justice Rucker again dissented. Id. at 939. The court stated that it had “addressed Pruitt’s claims that he presented previously undiscovered evidence of [intellectual disability] at the PCR hearing and found the evidence to be cumulative” of that presented at trial. Id. at 938. The Indiana Supreme Court affirmed the finding that trial counsel were not deficient in their presentation of evidence of Pruitt’s mental illness. Pruitt II, 903 N.E.2d at 919, 921. The court agreed with the post-conviction court that “trial counsel presented evidence of a schizotypal personality disorder, a diagnosis different more in degree of severity than of character, than the one advanced by experts” at the post-conviction proceeding, and that a “different diagnosis does not show that trial counsel were deficient.” Id. at 921 (internal quotation marks and citation omitted). Given trial counsels’ testimony at the post-conviction hearing, the Indiana Supreme Court made the determination that counsels’ “first priority” was to show that Pruitt is intellectually disabled and their “second priority” was to prove that he was suffering from a serious mental illness around the time of the crime. See id. at 922. The court also found that “trial counsel made a deliberate strategic decision to concentrate the jury’s attention on Pruitt’s claim of [intellectual disability]” and “this strategy might well have been undermined by greater emphasis on the much weaker mental illness evidence.” Id. The court determined that trial counsel were not ineffective in selecting experts and preparing mental health evidence for use at trial. Id. at 921. II. ANALYSIS Pruitt contends that his execution is barred by the Eighth Amendment because the evidence establishes that he is intellectually disabled. He argues that the state court’s contrary decision is objectively unreasonable and based on a definition of intellectual disability that is an unreasonable application of Atkins. Pruitt’s second claim is that trial counsel rendered ineffective assistance by failing to investigate and present evidence to support his claim of intellectual disability. Third, Pruitt claims that trial counsel rendered ineffective assistance by failing to investigate and present at the penalty phase mitigating evidence regarding his schizophrenia, and that the state court’s contrary conclusion is an unreasonable application of Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). Finally, he argues that trial counsel rendered ineffective assistance in failing to raise the claim that the prosecutor violated Pruitt’s Eighth Amendment and due process rights by reciting a poem about the death of a police officer and comparing Pruitt to notorious murderers in the closing argument of the penalty phase of trial, and that appellate counsel was ineffective in failing to raise this as an error on appeal. A. Standard of Review A petitioner in state court custody is entitled to a writ of habeas corpus “only on the ground that he is in custody in violation of the Constitution or laws or *263 treaties of the United States.” 28 U.S.C. § 2254(a). Under the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), 110 Stat. 1214, a federal court may issue a writ of habeas corpus on á claim that a state court has adjudicated on the merits only if the state court’s decision “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court,” or “was 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); Cullen v. Pinholster, 563 U.S. -, 131 S.Ct. 1388, 1398, 179 L.Ed.2d 557 (2011). “AEDPA erects a formidable barrier to federal habeas relief for prisoners whose claims have been adjudicated in state court.” Burt v. Titlow, — U.S. -, 134 S.Ct. 10, 16, 187 L.Ed.2d 348 (2013). However,. AEDPA deference applies “only to those issues the state court explicitly addressed.” Makiel v. Butler, 782 F.3d 882, 896 (7th Cir.2015) (quoting Quintana v. Chandler, 723 F.3d 849, 853 (7th Cir.2013)). “If no state court has squarely addressed the merits of a habeas claim, we review the claim de novo under the pre-AEDPA standard of 28 U.S.C. § 2243.” Ruhl v. Hardy, 743 F.3d 1083, 1091 (7th Cir.2014). Thus, we “dispose of the matter as law and justice require.” Harris v. Thompson, 698 F.3d 609, 623 (7th Cir.2012) (quotation and citation omitted). A state-court decision is contrary to clearly established federal law “if it applies a rule that contradicts the governing law set forth in [a Supreme Court case], or if it confronts a set of facts that is materially indistinguishable from a decision of [the Supreme] Court but reaches a different result.” Brown v. Payton, 544 U.S. 133, 141, 125 S.Ct. 1432, 161 L.Ed.2d 334 (2005). A state-court decision unreasonably applies federal law if it “applies [the Supreme] Court’s precedents to the facts in an objectively unreasonable manner.” Id. An “unreasonable application of’ federal law means “objectively unreasonable, not merely wrong; even ‘clear error’ will not suffice.” White v. Woodall, — U.S. -, 134 S.Ct. 1697, 1702, 188 L.Ed.2d 698 (2014) (internal quotations and citation omitted). “[A] state prisoner must show that the state court’s ruling on the claim ... was so lacking in justification that there was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Harrington v. Richter, 562 U.S. 86, 103, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011). When reviewing a state court decision under § 2254(d)(1), “we presume that the court’s factual determinations are correct unless [the petitioner] rebuts the presumption by clear and convincing evidence.” Newman v. Harrington, 726 F.3d 921, 928 (7th Cir.2013) (quotation and citation omitted). “A petitioner’s challenge to a state court decision based on a factual determination under § 2254(d)(2) will not succeed unless the state court committed an ‘unreasonable error.’ ” Morgan v. Hardy, 662 F.3d 790, 798 (7th Cir.2011). Under § 2254(d)(2), a state court decision is “based on an unreasonable determination of the facts if it rests upon fact-finding that ignores the clear and convincing weight of the evidence.” Newman, 726 F.3d at 928 (internal quotation marks and citation omitted). “[E]ven if ‘[reasonable minds reviewing the record might disagree’ about the finding in question, ‘on habeas review that does not suffice to supersede the trial court’s ... determination.’ ” Wood v. Allen, 558 U.S. 290, 301, 130 S.Ct. 841, 175 L.Ed.2d 738 (2010) (quoting Rice v. Collins, 546 U.S. 333, 341- *264 42, 126 S.Ct. 969, 163 L.Ed.2d 824 (2006)) (second brackets and ellipses in original). We review de novo the district court’s denial of a § 2254 petition. Ford v. Wilson, 747 F.3d 944, 949 (7th Cir.2014). “[Although we technically hear this appeal from the district court,” id., we review the decision of the last state court to address the merits of the petitioner’s claim, id. (quoting McNary v. Lemke, 708 F.3d 905, 913 (7th Cir.2013), reviewing that decision with deference, Harris v. Hardy, 680 F.3d 942, 948 (7th Cir.2012)). B. Intellectual Disability Pruitt’s principal argument is that he is intellectually disabled and therefore is categorically and constitutionally ineligible for the death penalty. He seeks to have his death sentence vacated and his case remanded for resentencing to a sentence of years. In Atkins v. Virginia, the Supreme Court held that the Eighth Amendment prohibits the execution of the intellectually disabled. 536 U.S. at 321, 122 S.Ct. 2242. The Court did not define intellectual disability for purposes of ineligibility for the death penalty, but rather left to the states the task of developing appropriate ways for determining intellectual disability. Id. at 317, 122 S.Ct. 2242. The Court offered the states guidance by referring to the clinical definitions provided by the AAMR and the American Psychiatric Association (APA), id. at 308 n. 3, 122 S.Ct. 2242, and noting that the state statutory definitions “generally conform to the clinical definitions,” id. at 314 & 317 n. 22, 122 S.Ct. 2242. In Hall v. Florida, — U.S. -, 134 S.Ct. 1986, 188 L.Ed.2d 1007 (2014), the Supreme Court again declined to set forth a legal definition of intellectual disability. The Court held that Florida’s strict IQ test score cutoff “create[d] an unacceptable risk that persons with intellectual disability will be executed, and thus is unconstitutional.” Id. at 1990. The Court again turned to the views of the states and medical experts for guidance in determining intellectual disability, id. at 1993-94, noting that “[t]he legal determination of intellectual disability is distinct from a medical diagnosis, but it is informed by the medical community’s diagnostic framework,” id. at 2000. The Court concluded that in determining intellectual disability, sentencing courts must take into account an IQ test’s standard error of measurement. Id. at 2001. Indiana law defines an individual with intellectual disability as “an individual who, before becoming twenty-two (22) years of age, manifests: (1) significantly subaver-age intellectual functioning; and (2) substantial impairment of adaptive behavior; that is documented in a court ordered evaluative report.” Ind.Code § 35-36-9-2. In Pruitt I, 834 N.E.2d at 108, the Indiana Supreme Court decided that the statutory definition was constitutional under Atkins. Indeed, the Atkins Court had noted that Indiana’s statutory definition “generally conform[s] to the clinical definitions.” 536 U.S. at 314 & 317 n. 22, 122 S.Ct. 2242; see also McManus v. Neal, 779 F.3d 634, 650 (7th Cir.2015) (noting that Indiana’s definition of intellectual disability “is consistent with the clinical standards”); State v. McManus, 868 N.E.2d 778, 785 (Ind.2007) (noting that in determining what is “significantly subaverage intellectual functioning” the court refers to the definitions provided by the AAMR and the APA). The state court’s finding as to each prong of the statutory definition of intellectual disability is a factual determination. See McManus, 779 F.3d at 654 (“[T]he Indiana Supreme Court’s factual determination that [petitioner’s] intellectual functioning is not significantly subaverage is solidly grounded *265 in the record and thus is not objectively unreasonable.”); id. at 656 (“[T]he Indiana Supreme Court made an objectively reasonable factual determination that [petitioner] is not intellectually disabled.... ”). Intellectual functioning can be measured with an IQ test. See, e.g., Am. Psychiatric Ass’n, Diagnostic and Statistical Manual for Mental Disorders 37 (4th ed.-Text Rev. 2000) (DSM-IV) (“Intellectual functioning is typically measured with individually administered and psychometrically valid, comprehensive, culturally appropriate, psychometrically sound tests of intelligence.”); id. at 41 (stating that “general intellectual functioning” is defined by the IQ score “obtained by assessment with one or more of the standardized, individually administered intelligence tests”); American Ass’n on Intellectual & Developmental Disabilities, Definition of Intellectual Disability, http:/ /aaidd.org/intellectual-disability/definition# (last visited Apr. 8, 2015). In Indiana, “a person is considered to meet the subaverage intellectual functioning component if the person’s full-scale IQ test score is two standard deviations below the mean; i.e., an IQ between 70 and 75 or lower.” McManus, 868 N.E.2d at 785 (quoting Woods v. State, 863 N.E.2d 301, 304 (Ind.2007) (citing Atkins, 536 U.S. at 308 n. 3, 122 S.Ct. 2242)); see cf. DSM-IV, at 41 (“Significantly subaverage intellectual functioning is defined as an IQ of about 70 or below (approximately 2 standard deviations below the mean).”). The five-point range accounts for the standard margin of error in IQ testing. McManus, 779 F.3d at 646 n. 3. The Indiana Supreme Court has determined that “IQ tests are only evidence; they are not conclusive on either the subject’s IQ or the ultimate question of [intellectual disability].” Pruitt I, 834 N.E.2d at 106. Thus, the state supreme court held that “courts may consider IQ scores together with other evidence of mental capacity,” including “work history, school history, and life functioning.” McManus, 868 N.E.2d at 785. As for adaptive functioning, we recently recognized that: [t]he medical community measures adaptive behavior across three domains: conceptual, social, and practical. To satisfy this component of the definition, a person’s adaptive functioning in at least one domain must be “sufficiently impaired that ongoing support is needed in order for the person to perform adequately in one or more life settings at school, at work, at home, or in the community.” Moreover, the deficits must be caused by the person’s intellectual impairment [and must] ... appear during childhood or adolescence. McManus, 779 F.3d at 650-51 (footnote omitted; quoting Am. Psychiatric Ass’n, Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) 38 (2013) and citing DSM-5 at 33, 38). The Indiana Supreme Court rejected Pruitt’s claim of intellectual disability based solely on the intellectual functioning prong; it did not determine whether the evidence satisfied the adaptive functioning prong. However, the court concluded that the trial court applied a “too restrictive” standard for adaptive functioning and therefore that the trial court’s finding as to that prong was “not supportable.” Pruitt I, 834 N.E.2d at 110. Pruitt argues that the Indiana Supreme Court’s conclusion that he did not establish significantly subaverage intellectual functioning is based on an unreasonable determination that: 1) his IQ scores were “inconsistent”; and 2) “more importantly,” his IQ scores were not determinative given the evidence that he could “fill out applications for employment” and had “the capacity, if not the will at times, to support himself.” Pet’r-Appellant’s Br. 28 *266 (quoting Pruitt I, 834 N.E.2d at 106). We conclude that the Indiana Supreme Court’s finding that Pruitt did not meet the intellectual functioning prong was an unreasonable factual determination. Although some of Pruitt’s test scores may appear inconsistent, the reliable IQ scores were consistently within the range of significantly subaverage intellectual functioning. Beginning when he was an adolescent and long before his crimes, Pruitt achieved verbal scores of 64 on two Lorge-Thorndike- tests and nonverbal scores of 65 and 63. These scores suggest that he had significantly subaverage intellectual functioning. The state’s expert Dr. Groff and Dr. Golden agreed that Pruitt gave his best effort on these two tests. Dr. Golden stated that these scores suggested that Pruitt’s IQ was “fairly stable” at that point in time. These Lorge-Thorn-dike tests were taken when Pruitt was an adolescent; thus, the scores reflect his intellectual functioning before the age of 22, which is the critical age for the third prong of Indiana’s statutory definition — age of onset — and long before the crimes at issue and before Pruitt had any motive to malinger in order to defeat the death penalty. Both Drs. Hudson and Golden testified that the IQ scores obtained as a child would most accurately measure Pruitt’s intelligence before age 18. Trial Tr. 1355-56 (Hudson stating that “IQ is meant to be diagnosed in childhood.”), id. at 1551 (Golden stating that scores before age 18 are “absolutely” more important than scores after that age in assessing IQ). After Deputy Starnes’ murder, consistent with his scores on the Lorge-Thorn-dike tests, Pruitt achieved full-scale IQ scores of 65 on not one, but two Stanford-Binet tests. The scores on these four intelligence tests were consistent: all of them fell well within the range for intellectual disability. Dr. Hudson explained that the research demonstrates an “extremely high correlation” between the Lorge-Thorndike and Weschler, the latter of which, along with the Stanford-Binet, is considered the gold standard in forensic cases. (We note that Pruitt scored a full-scale IQ score of 52 on the WAIS in 2003, but we discount that score based on the experts’ agreement that it underestimated Pruitt’s intellectual functioning due to lack of effort.) Granted, Pruitt obtained a full-scale IQ score of 76 on the WAIS in 2002, but Dr. Olvera explained that Pruitt was taking antipsychotic medication at the time he took the test and this may have produced a higher score than he would have obtained without the medication. Dr. Hudson agreed that the drug would have increased Pruitt’s accuracy on the test and improved his score. Dr. Hudson believed that absent the drug, Pruitt’s score probably would have been closer to 70. Likewise, Dr. Golden thought that the drug could increase Pruitt’s score “significantly” because it would improve his attention and ability to focus. Given these experts’ testimony, the state court’s determination that there was insufficient evidence to show what effect, if any, the antipsychotic medication may have had on Pruitt’s IQ score “ignore[d] the clear and convincing weight of the evidence.” Taylor v. Grounds, 721 F.3d 809, 817 (7th Cir.2013) (citation omitted). Furthermore, according to every expert who expressed an opinion on the issue — Drs. Hudson, Golden, Olvera, and Keyes-Pruitt’s score of 76 was not inconsistent with significantly subaverage intellectual functioning. These experts explained that a single score has to be viewed in the context of the other test scores and other information in the record. Cf. DSM-5 at 37 (“Individual cognitive profiles based on neuropsychological testing are more useful for understanding intellectual abilities than a single IQ score.”). *267 While Pruitt scored a 93 on the Revised Beta, Drs. Golden, Hudson, and Keyes were quite critical of that test. The test is group-administered in fifteen minutes. See DSM-5 at 37 (“Invalid scores may result from the use of brief intelligence screening tests or group tests.”). According to Dr. Golden, the Revised Beta is not accurate, not well-regarded in the field, and not well-accepted as a general test of intelligence. Dr. Hudson likewise questioned the reliability of the Revised Beta. And Dr. Keyes stated that the Revised Beta “is not considered a valid intelligence indicator by any professional psychologist who knows the literature and/or the test.” PCR App. 773. Dr. Schmedlen believed that the Revised Beta test was reliable, but he acknowledged that the Revised Beta was not as rehable as the Weschler series or Stanford-Binet tests. And to the extent that the state court relied on Pruitt’s scores on the Otis-Lennon and Iowa basic academic achievement tests, all of the experts save one agreed that academic achievement tests do not test for intelligence. The one exception was Dr. Schmedlen, and he was unaware that the Otis-Lennon was known as the “Otis-Lennon School Ability Test” and that the Otis-Lennon score was not given as .an IQ score. Thus, his opinion was not well-informed. Nonetheless, he acknowledged that the Otis-Lennon test was not of the same quality as a Weschler or Stanford-Binet intelligence test. Four highly qualified experts with extensive experience with the intellectually disabled — Drs. Hudson, Golden, Olvera, and Keyes — all agreed that Pruitt’s IQ scores demonstrated significantly subaver-age intellectual functioning and that Pruitt is intellectually disabled. The only evidence in the record to counter their opinions were the opinions of Drs. Schmedlen and Groff. As noted, Dr. Schmedlen erred in thinking that Pruitt’s scores on achievement tests were equivalent to IQ scores. Dr. Groff acknowledged that he could not give an opinion about Pruitt’s intellectual functioning “with certainty,” but rather, could only give his impression. Trial Tr. 1463. On top of that, Dr. Groff never met or tested Pruitt. Dr. Groff just reviewed records and relied on Dr. Schmedlen’s report. The state courts’ reliance on these two experts went against the clear and convincing weight of the evidence. Even when viewed through AEDPA’s deferential lens, the Indiana Supreme Court’s determination that Pruitt failed to demonstrate significantly subaverage intellectual functioning based on inconsistent test scores was objectively unreasonable and ignored the clear and convincing weight of the evidence. Some of the test scores upon which the state courts relied were not IQ scores; other scores were unreliable. The record establishes that Pruitt’s reliable IQ scores consistently demonstrated significantly subaverage intellectual functioning. 2 Pruitt challenges the Indiana Supreme Court’s consideration of his ability to fill out job applications, obtain a commercial driver’s license, and work as evidence that he did not meet the intellectual functioning prong. He asserts that the court charae- *268 terized this evidence other than intelligence test scores as “the most important” evidence bearing on his intellectual functioning. The court wrote:- “More importantly, IQ tests are only evidence; they are not conclusive on either the subject’s IQ or the ultimate question of mental retardation.” Pruitt I, 834 N.E.2d at 106. Our Supreme Court made similar comments in Hall where it observed that “Courts must recognize, as does the medical community, that the IQ test is imprecise. This is not to say than an IQ test score is unhelpful. It is of considerable significance, as the medical community recognizes.” 134 S.Ct. at 2001. Nothing in Hall or Atkins directs the states to give conclusive effect to IQ scores, or precludes a state’s use of circumstantial evidence of mental capacity other than IQ scores in assessing a defendant’s intellectual functioning. Recently in McManus, we gave our nod of approval to the state court’s use of circumstantial evidence as additional support for its determination that the petitioner did not demonstrate significantly subaverage intellectual functioning, 779 F.3d at 653-54 (rejecting argument that in assessing petitioner’s intellectual functioning, state court should not have given any weight to evidence that petitioner graduated from high school, successfully worked three jobs, and cared for his disabled child), though we noted that the “court mentioned this evidence only in passing....” Id. at 654. The problem arises where, as here, the state court relies on inaccurate assumptions and select pieces of the evidence in making its factual determination. Cf. McManus, 779 F.3d at 654 (“We do not doubt that intellectually disabled people graduate from high school (with or without the assistance of special-education programming) and also hold down jobs.”). In concluding that Pruitt had not shown that he is intellectually disabled, the Indiana Supreme Court explained: While some of Pruitt’s scores suggest significantly subaverage intellectual functioning, others do not. In addition to this data, the trial court found that Pruitt was able to fill out applications for employment and to have the capacity, if not the will at all times, to support himself. In light of the inconsistent IQ scores and the other evidence cited by the trial court, the trial court’s finding that Pruitt did not meet the statutory test is consistent with this record. Pruitt I, 834 N.E.2d at 106. The other evidence cited by the trial court was Pruitt’s “functioning, work history, school history, and all other evidence presented regarding [his] intellectual functioning.” Id. Pruitt asserts that there is no evidence that activities such as filling out job applications are relevant to an intellectual disability determination. Indeed, the record contains evidence that mildly intellectually disabled individuals can obtain driver’s licenses, fill out job applications, and obtain and maintain employment. Dr. Hudson testified at trial that the nature of the application, the extent to which it was completed independently, and its accuracy would be important considerations. For example, Pruitt completed a job application in 1999, but it contained spelling errors, incomplete sentences, inaccurate punctuation, inappropriate responses to some questions, and omitted responses to other questions. It would be unreasonable to conclude that this job application demonstrates that Pruitt’s IQ is higher than that reflected by the reliable IQ scores. Even though Pruitt has had numerous jobs, they did not require high intellectual functioning: dishwasher, truck driver, carpenter, and laborer. He was unable to maintain any job for any length of time *269 without assistance. His father got him a job in construction, but he was hired only because the employer considered his father a valued employee, and Pruitt worked in demolition and clean-up work. Pruitt did not learn the carpentry skills, and he needed to have his father there constantly telling him what to do. When Pruitt’s father retired, Pruitt was fired because he was unable to do basic work without constant supervision and could not communicate well with others. Pruitt’s employment as a truck driver demonstrated similar challenges that provides additional support for the conclusion that he has significantly subaverage intellectual functioning, For example, Pruitt once parked his semi in the middle of a street and when he was told he could not park there, he responded that there were no signs saying that he could not park there. Pruitt further challenges the state court's reliance on his work history as a basis for finding that he did not satisfy the intellectual functioning prong, given the evidence that he satisfied the adaptive functioning prong, specifically in the skill area of work. Drs. Hudson, Schmedlen, and Keyes all thought that Pruitt showed substantial impairment in work. We agree with Pruitt that it is illogical and irreconcilable for the state court to find that his work history outweighs his IQ scores when all experts expressing an opinion on the issue agree that he is substantially impaired in this area of adaptive functioning. We conclude that the state court erred in finding that Pruitt’s work history supported the determination that Pruitt did not meet the intellectual functioning prong and was not intellectually disabled. Moreover, Pruitt’s school history does not support the conclusion that he failed to establish significantly subaverage intellectual functioning. Within the first three years of elementary school, Pruitt was held back twice; he was placed in special education classes; and he was socially promoted. At the age of 16, Pruitt dropped out of school without advancing past the eighth grade. As Dr. Dare explained, Pruitt’s school records “clearly point to a student who qualifies under the special education] category of mild [intellectually disabled].” PCR Tr. 329. And the record evidence established that Pruitt was unable to function on his own as an adult; he lived with and depended on his family to take care of him. The Indiana Supreme Court made an unreasonable determination of fact in concluding that Pruitt’s work history, school history, and other evidence supported the finding that Pruitt failed to establish significantly subaverage intellectual functioning. Rather, the clear and convincing weight of the evidence establishes that Pruitt suffers from significantly subaver-age intellectual functioning. Turning to the second prong of the definition of intellectual disability, Pruitt argues that he has established that his adaptive behavior is substantially impaired. We agree. The Indiana Supreme Court rejected the trial court’s use of a heightened standard to prove the adaptive-behavior prong. While the court noted that “the evidence on the adaptive behavior prong is at least conflicting,” Pruitt I, 834 N.E.2d at 110, it did not actually conclude that Pruitt failed to establish substantial impairment of adaptive behavior. Thus, we review this prong de novo. Ruhl, 743 F.3d at 1091. The state court record contains unre-butted evidence that Pruitt satisfies the adaptive behavior prong of intellectual disability. Court-appointed expert psychologist Dr. Schmedlen administered an objective test of adaptive functioning and determined that Pruitt met the adaptive behavior prong under the AAMR definí *270 tion because he was substantially impaired in two areas of adaptive functioning: work and self-direction. Defense expert psychologist Dr. Hudson testified, based on his review of Pruitt’s history and two clinical evaluations, that Pruitt showed significant deficits in six adaptive skills areas: communication, home living, social skills, functional academics, health and safety, and work. Based on standardized testing to assess adaptive functioning, interviews with Pruitt’s family, special education teachers, and family friends, and review of records and trial testimony, defense expert Dr. Keyes concluded that Pruitt’s adaptive functioning skills' were significantly impaired in virtually all areas, even estimating his overall adaptive functioning at about the level of a six- to eight-year-old child. No expert offered any contrary opinion regarding Pruitt’s adaptive functioning. In addition, Pruitt presented extensive lay testimony regarding his development, school and work history, and functioning that supported these experts’ unrebutted opinions as to his adaptive functioning. And significantly, the state does not even argue on appeal that the evidence failed to establish that Pruitt suffers from substantially impaired adaptive functioning. In sum, the Indiana Supreme .Court’s decision that Pruitt has not shown that he is intellectually disabled was based on an unreasonable determination of the facts. Pruitt has demonstrated with clear and convincing evidence that he is intellectually disabled. Therefore, he is categorically and constitutionally ineligible for the death penalty and his death sentence must be vacated. C. Ineffective Assistance of Counsel Pruitt raises three alleged errors in support of his ineffective-assistance-of-counsel claim, but we need address only one— whether trial counsel was ineffective at the penalty phase in investigating and presenting evidence that Pruitt suffered from paranoid schizophrenia. Under the familiar test of Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), Pruitt first must show that counsel’s performance was deficient, which means that it “fell below an objective standard of reasonableness.” Id. at 687-88, 104 S.Ct. 2052. He also “must show that the.deficient performance prejudiced the defense,” id. at 687, 104 S.Ct. 2052, which means “that there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694, 104 S.Ct. 2052. “Judicial scrutiny of counsel’s performance must be highly deferential.” Id. at 689, 104 S.Ct. 2052. Courts “indulge a strong presumption that counsel’s conduct falls within the wide range of reasonable professional assistance; that is, the defendant must overcome the strong presumption that, under the circumstances, the challenged action might be considered sound trial strategy.” Id. (internal quotation marks omitted). When a petitioner challenges a state court’s application of Strickland under § 2254(d), our review is “doubly deferential”: “[w]e take a highly deferential look at counsel’s performance, through the deferential lens of 2254(d).” Pinholster, 131 S.Ct. at 1403 (internal citations and quotation marks omitted). The Supreme Court has cautioned federal courts to “guard against the danger of equating unreasonableness under Strickland with unreasonableness under § 2254(d).” Richter, 562 U.S. at 105, 131 S.Ct. 770. “When § 2254(d) applies, the question is not whether counsel’s actions were reasonable,” but whether “there is any reasonable argument that counsel satisfied Strickland’s deferential standard.” *271 Id.; see Campbell v. Reardon, 780 F.3d 752, 762 (7th Cir.2015). “[C]ounsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary. In any ineffectiveness case, a particular decision not to investigate must be directly assessed for reasonableness in all the circumstances, applying a heavy measure of deference to counsel’s judgments.” Strickland, 466 U.S. at 691, 104 S.Ct. 2052. We have explained that “[rjesources are limited, and trial counsel must eventually shift from pretrial investigation to trial preparation.” Campbell, 780 F.3d at 765. “There comes a point where a defense attorney will reasonably decide that another strategy is in order, thus ‘mak[ing] particular investigations unnecessary.’ ” Pinholster, 131 S.Ct. at 1407 (quoting Strickland, 466 U.S. at 691, 104 S.Ct. 2052). “[Strategic choices made after thorough investigation of law and facts relevant to plausible options are virtually unchallengeable,” Strickland, 466 U.S. at 690, 104 S.Ct. 2052; strategic choices made after less than thorough investigation are reasonable only “to the extent that reasonable professional judgments support the limitation on investigation,” id. at 691, 104 S.Ct. 2052. Pruitt argues that trial counsel rendered ineffective assistance at the penalty phase in failing to investigate and present mitigating evidence that he suffered from paranoid schizophrenia which causes hallucinations and paranoid delusions. He asserts that two statutory mitigating circumstances applied because of his schizophrenia: he “was under the influence of extreme mental or emotional disturbance when the murder was committed,” and his capacity to “conform [his] conduct to the requirements of law was substantially impaired as a result of mental disease or defect.” Ind.Code § 35-50-2-9-(c)(2), (6). The Supreme Court has decided several cases in which it concluded that counsel’s failure to investigate and present mitigation evidence constituted ineffective assistance under Strickland. In Rompilla v. Beard, 545 U.S. 374, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005), for example, although counsel were aware of the defendant’s criminal record, they did not examine the file on his prior conviction. And although counsel knew from police reports that the defendant had been drinking heavily at the time of the crime and one of the three mental health experts had recommended investigation into his alcohol use, “counsel did not look for evidence of a history of dependence on alcohol that might have extenuating significance.” Id. at 382, 125 S.Ct. 2456. Had counsel investigated further they would have discovered that the defendant had a serious drinking problem. Id. at 391-92, 125 S.Ct. 2456. And had counsel examined the file on the defendant’s prior conviction, they would have found test results that defense mental health experts thought suggested “schizophrenia and other disorders, and test scores show[ed] a third grade level of cognition.” Id. at 391, 125 S.Ct. 2456. The Supreme Court concluded that “reasonably diligent counsel may draw a line when they have good reason to think further investigation would be a waste,” id. at 383, 125 S.Ct. 2456, but counsel “could not reasonably have ignored mitigation evidence or red flags,” id. at 391 n. 8, 125 S.Ct. 2456. As another example, in Wiggins v. Smith, 539 U.S. 510, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003), counsel failed to investigate and present mitigating evidence of the defendant’s dysfunctional background, and relied instead on the argument that the defendant was not directly responsible for the murder. Id. at 523-24, 535, 123 S.Ct. 2527. The state court denied post-conviction relief, concluding that “counsel *272 had made a deliberate, tactical decision to concentrate their effort at convincing the jury that [petitioner] was not directly responsible for the murder.” Id. at 518, 123 S.Ct. 2527 (quotation omitted). The Supreme Court said that the issue “was not whether counsel should have presented a mitigation case” but “whether the investigation supporting counsel’s decision not to introduce mitigating evidence of [petitioner’s] background was itself reasonable.” Id. at 523, 123 S.Ct. 2527. The Court concluded that the state court’s “strategic decision” rationale was a post hoc rationalization of counsel’s conduct and that counsel’s decision not to present a mitigation defense was based on an unreasonable investigation and thus that decision “was also objectively unreasonable.” Id. at 527-28, 123 S.Ct. 2527. Similar to counsel’s shortcomings in Rompilla and Wiggins, Pruitt’s trial counsel’s failure to investigate Pruitt’s mental illness fell below an objective standard of reasonableness. Trial -counsel knew that shortly after Pruitt was arrested for killing Deputy Starnes, a DOC psychiatrist had diagnosed Pruitt with “schizophrenia, chronic undifferentiated type compensated residual,” which meant that he had been schizophrenic “at one time” but was not “actively schizophrenic” at the time of the diagnosis. Counsel knew that the psychiatrist had prescribed Pruitt Trilafon, an anti-psychotic medication. Counsel also knew from Dr. Hudson’s testimony at the pretrial hearing that Dr. Hudson noted some of Pruitt’s responses to his questioning were not “clearly lucid.” And counsel was aware that Dr. Schmedlen had seen symptomology consistent with schizophrenia during his evaluations of Pruitt. If this were not enough to prompt a competent lawyer to further investigate Pruitt’s mental health, the defense’s own expert Dr. Olvera had recommended that counsel contact an expert “in dealing with psychosis, such as schizophrenia.” But trial counsel did not contact such an expert to have Pruitt evaluated, and counsel offered no reason for failing to do so. As in Rompilla^ counsel ignored the “red flags” pointing to potential mitigation evidence, and counsel acted unreasonably in failing to investigate further. The record establishes that counsel had every reason to follow through on Dr. Olv-era’s recommendation, and no good reason not to have Pruitt evaluated by an expert in psychosis. After all, counsel testified at the post-conviction hearing that the defense strategy was to prove that Pruitt did not deserve the death penalty because of his intellectual disability, his serious mental illness, and his brain damage. Since the stated strategy included showing that Pruitt had a serious mental illness, a reasonably competent attorney would have realized that investigating Pruitt’s mental health further was necessary to prove the defense. It was unreasonable for counsel to fail to contact an expert in psychosis, such as schizophrenia, particularly given that the defense’s own expert Dr. Olvera had recommended that they contact such an expert. And given the evidence already known to counsel, they had no reason to think that ‘contacting an expert in psychosis “would have been counterproductive, or that further investigation would have been fruitless.” Wiggins, 539 U.S. at 525, 123 S.Ct. 2527. The Indiana Supreme Court surmised that trial counsel made a deliberate, strategic decision to concentrate on Pruitt’s intellectual disability rather than his mental illness. Yet nothing in the record suggested to counsel that the two diagnoses— intellectual disability and schizophrenia— were mutually exclusive. Indeed, “there is substantial comorbidity between schizophrenia and intellectual disability....” *273 Owen, Michael J., et al., “Neurodevelopmental Hypothesis of Schizophrenia.” The British Journal of Psychiatry 198(3): 173-175 (Mar. 2011), available at www.ncbi.nlm.nih.gov/pmc/articles/PMC3764497 (last visited May 8, 2015). Trial counsel did not have to concentrate on one defense or the other. And counsel never suggested that they made a decision to focus on Pruitt’s intellectual disability rather than his mental illness. They did not suggest that their “first priority” was to show that Pruitt is intellectually disabled and that showing he had a serious mental illness was only secondary. Counsel testified that the strategy was to prove based on all of the things identified — intellectual disability, mental illness, and brain damage — that Pruitt did not deserve to die. This was a three-pronged- approach. Merely listing the three as “one,” “two,” and “finally,” without more did not prioritize the prongs. The state court overstated the evidence in this regard. And even if we assume that counsel made a strategic decision to focus on intellectual disability, that choice was made after a less than thorough investigation, and as a result, the decision was not fully informed and was unreasonable. See Strickland, 466 U.S. at 690-91, 104 S.Ct. 2052 (concluding that “strategic choices made after less than complete investigation are reasonable precisely to the extent that reasonable professional judgments support the limitation on investigation”). Thus, to the extent that counsel made a decision to focus the jury’s attention on Pruitt’s claim of intellectual disability rather than his mental illness, the evidence of mental illness was much weaker than the intellectual disability evidence only because counsel failed to investigate more fully Pruitt’s mental health. The state court’s determination that trial counsel were not deficient in their presentation of evidence of Pruitt’s mental health was unreasonable. While trial counsel investigated Pruitt’s mental state and presented evidence of his mental illness at the penalty phase of trial, their investigation and presentation of evidence were deficient. Because the Indiana Supreme Court did not assess whether Pruitt could satisfy Strickland’s prejudice prong, our review is de novo. Rompilla, 545 U.S. at 390, 125 S.Ct. 2456. We conclude that Pruitt can show prejudice. To establish prejudice, 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, 104 S.Ct. 2052. The likelihood of a different outcome “must be substantial, not just conceivable.” Richter, 562 U.S. at 112, 131 S.Ct. 770. “When a defendant challenges a death sentence ... the question is whether there is a reasonable probability that, absent the errors, the sentencer ... would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.” Strickland, 466 U.S. at 695, 104 S.Ct. 2052. When assessing that probability, a court should “consider ‘the totality of the available mitigation evidence — both adduced at trial, and the evidence adduced in the habeas proceeding’ — and ‘reweig[h] it against the evidence in aggravation.’ ” Porter v. McCollum, 558 U.S. 30, 41, 130 S.Ct. 447, 175 L.Ed.2d 398 (2009) (quoting Williams v. Taylor, 529 U.S. 362, 397-98, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000)). Trial counsel presented evidence of Pruitt’s mental illness and its effects on Pruitt at the penalty phase. The jury heard testimony from Dr. Golden that two months after killing Officer Starnes, a *274 DOC psychiatrist had diagnosed Pruitt with “schizophrenia, chronic undifferentiated type compensated residual,” which meant that the psychiatrist thought that at one time Pruitt had been schizophrenic, but at that time, the most serious symptoms were not present. Dr. Golden said that the symptomatology for schizophrenia was similar to that for schizotypal personality disorder. Nonetheless, the jury heard only from Dr. Golden, not the DOC psychiatrist who had diagnosed schizophrenia, and the former testified that Pruitt’s illness was “not severe enough to be” schizophrenia; it was schizotypal personality disorder instead. Although Dr. Golden testified about the symptoms and debilitating effects of schizoid personality disorder, he also told the jury that for someone with schizotypal personality disorder, “psychotic episodes should be the exception rather than the rule.” We disagree with the district court that this is simply a “battle of the experts” over the severity of Pruitt’s mental illness. See Pruitt, 2012 WL 4513961, at *30. The sentencing judge referenced the jury’s finding that “the aggravating circumstances outweighed any and all mitigating circumstances.” Trial Tr. at 6460. Indeed, in imposing sentence, the sentencing judge determined that none of the possible mitigating factors presented “constitute a mitigating circumstance” and more specifically that “Pruitt does not have a mental condition that would justify his actions or in any way mitigate for purposes of sentencing.” Id. at 6455. However, the three post-conviction experts — Drs. Coons, Bal-lenger, and Price — all testified that Pruitt met two of the statutory mitigating factors: due to his schizophrenia, he “was under the influence of extreme mental or emotional disturbance when the murder was committed” and his capacity “to conform [his] conduct to the requirements of law was substantially impaired as a result of mental disease or defect,” Ind.Code § 35 — 50—2—9(c)(2) and (6). (Drs. Ballenger and Price also testified that Pruitt’s capacity to appreciate the criminality of his conduct was substantially impaired because of mental disease or defect.) But the jury never heard any of this testimony. Because this unrebutted testimony, from not one but three highly qualified mental health experts, supported a finding that Pruitt had evidence of two statutory mitigating circumstances, whereas the sentencing judge found that Pruitt did not have a mental condition that was mitigating, there is a reasonable probability that this evidence might have affected the judge’s and jury’s assessment of Pruitt’s moral culpability, and that they might have concluded that death was not warranted. See Williams, 529 U.S. at 398, 120 S.Ct. 1495 (concluding that counsel’s failure to investigate and present mitigating evidence at sentencing prejudiced the defendant where the undiscovered mitigating evidence “might well have influenced the jury’s appraisal of [the defendant’s] culpability”). It is significant that in considering the sentence, the jury wanted to know whether Dr. Golden had any knowledge of Pruitt’s thought processes, including his thinking, at the time of the crime. Their desire to know Pruitt’s thought processes strongly suggests that evidence of what was going on in his head at the time of the crime might have made a difference in his sentence. Respondent argues that we have rejected the same basis for asserting ineffective assistance of counsel in Overstreet v. Wilson, 686 F.3d 404 (7th Cir.2012), where we held the state court reasonably determined that counsel’s presentation of mitigation evidence was not constitutionally deficient and the petitioner failed to show prejudice. Id. at 408. Overstreet does not direct the *275 conclusion that Pruitt’s ineffective assistance claim fails in this case. In Overstreet, defense counsel presented expert testimony at sentencing that the defendant had a “schizotypal personality disorder”; a second expert testified at the post-conviction hearing that he would have opined that the defendant had “schizoaf-fective disorder,” which was described as a combination of schizophrenia and depression. Id. At the post-conviction hearing, trial counsel testified as to their strategic reasons for having the first expert but not the second expert testify. Id. at 409. The petitioner argued that his counsel should have called both experts, or should have called the second expert alone “because schizotypal personality disorder is just a ‘personality disorder’ on Axis II,” whereas schizophrenia was “a more serious, Axis I ‘clinical disorder.’ ” Id. at 408. The state court doubted that a jury would have appreciated the “subtle and nuanced distinction between a schizoaffective disorder and a schizotypal personality disorder.” Id. (quoting Overstreet v. State, 877 N.E.2d 144, 156 (Ind.2007)). In addressing the ineffectiveness claim, we observed that defense counsel had presented evidence that the petitioner had a serious mental abnormality and had argued that he was not blameworthy. Id. at 408. Indeed, counsel presented the first expert’s emphatic testimony that he believed the defendant’s ability to “conform his conduct to the requirements of law” was “significantly impaired.” Id. at 409. We concluded that “[t]o undermine counsel’s choices and the state judiciary’s findings about prejudice,” id., the petitioner needed evidence “that no reasonable lawyer would have thought [the first expert] the better witness, and that jurors would be less likely to recommend death for a defendant who has schizophrenia,” id. at 409-10, and that the state court’s contrary determination was unreasonable, id. at 410. The petitioner had no such evidence; therefore he could not upset the state court’s finding that the two experts would have made essentially the same impression on the jury. Id. at 410. In contrast with Overstreet, where trial counsel presented emphatic expert testimony that the defendant satisfied two statutory mitigating circumstances, Pruitt’s case is not merely about different psychiatric terms and labels. The post-conviction experts testified that Pruitt satisfied two statutory mitigating circumstances, whereas the trial experts did not give such an opinion. And, as discussed, Pruitt’s trial counsel did not conduct a thorough investigation into Pruitt’s mental health and therefore did not make “an informed choice” about what mental health evidence to present at sentencing. The failure to contact an expert in dealing with psychosis and schizophrenia left counsel ignorant about the potential mitigation evidence. In Overstreet, we emphasized that “trial counsel made an informed choice, quite unlike the situation in Wiggins,” id. at 409, and in this case as well. Perhaps most important, in contrast with Overstreet, here the state court did not make any finding as to prejudice to which we owe AEDPA deference. We conclude that there is a reasonable probability that with the presentation of the mitigating evidence regarding Pruitt’s mental illness, Pruitt’s sentence would have been different. Although we recently recognized that “ ‘it is difficult to establish ineffective assistance [of counsel] when counsel’s overall performance indicates active and capable advocacy,’ ” Makiel, 782 F.3d at 902 (quoting Richter, 562 U.S. at 111, 131 S.Ct. 770), even “an isolated error” can establish ineffective assistance “if it is ‘sufficiently egregious and prejudicial.’ ” Id. (quoting Murray v. Carrier, 477 U.S. 478, 496, 106 S.Ct. *276 2639, 91 L.Ed.2d 397 (1986)). Counsel’s error with regard to investigation and presentation of Pruitt’s paranoid schizophrenia is such an error. The Indiana Supreme Court’s determination that trial counsel were not ineffective.for failing to investigate and present mitigating evidence of Pruitt’s mental illness was an unreasonable application of Strickland, and Pruitt has shown that he was prejudiced by counsel’s ineffectiveness. Accordingly, the district court’s judgment should be reversed and remanded for issuance of a conditional writ remanding the case for a new penalty-phase proceeding. 3 III. CONCLUSION For the foregoing reasons, we REVERSE the district court’s judgment and REMAND with instructions to grant • a conditional writ vacating Pruitt’s death sentence and remanding to the State for a new penalty-phase proceeding. Footnotes [1] Dr. Olvera explained that the gradual increase of IQ scores over long periods of time, known as the "Flynn Effect,” causes IQ test norms to become obsolete. To counter this effect, IQ tests are periodically "re-normed” (made harder) by resetting the mean, score to 100 to account for the gains in IQ scores. The WAIS was normed in 1995 and published in 1997. Pruitt took the test in 2002. [2] We reach this conclusion without consideration of the "Flynn Effect.” Dr. Olvera testified that application of the Flynn Effect was contentious in the professional community, PCR Tr. 57-58, and he has never adjusted an IQ score in the clinical setting to account for the effect. We agree that "[t]he state courts had no obligation to accept and apply the Flynn Effect in the face of conflicting expert testimony about its acceptability and reliability.” Pruitt v. Wilson, 2012 WL 4513961, at *14. And "nothing in Atkins suggests that IQ test scores must be adjusted to account for the Flynn Effect in order to be considered reliable evidence of intellectual functioning.” McManus, 779 F.3d at 653. [3] The conclusion that Pruitt established ineffective assistance of counsel under one of his theories makes it unnecessary for us to fully address his other theories. Nonetheless, we do not think that trial counsel rendered constitutionally ineffective assistance in their investigation and presentation of evidence of Pruitt’s intellectual disability or in failing to object to alleged improper prosecutorial argument’at trial. Nor do we think that appellate counsel was constitutionally ineffective in not raising an ineffective-assistance claim based on trial counsel's failure to object to the alleged prosecutorial misconduct.