ROGERS, J., delivered the opinion of the court, in which BOGGS, C.J., BATCHELDER, GILMAN, GIBBONS, SUTTON, McKEAGUE, and GRIFFIN, JJ., joined. DAUGHTREY, J. (pp. 271-73), delivered a separate opinion concurring in result only. COLE, J. (p. 273), delivered a separate opinion concurring in part and dissenting in part. MOORE, J. (pp. 273-84), delivered a separate dissenting opinion, in which MARTIN and CLAY, JJ., joined.
OPINION
ROGERS, Circuit Judge.Habeas corpus relief was properly denied in this case because William Garner validly waived his Miranda rights, notwithstanding expert testimony. — based in part on a test administered six years later — to the effect that Garner mentally could not have sufficiently understood the scope of what Miranda protects. In 1992, Garner burglarized and set fire to an apartment in Cincinnati, Ohio, killing five children who he knew were sleeping inside. After he was arrested and advised of his Miranda rights, Garner agreed to speak with police officers and confessed on tape to setting the fire. The confession was admitted at trial and Garner was eventually convicted by a jury on, among other charges, five counts of aggravated murder, and sentenced to death. The Ohio state courts affirmed Garner’s convictions and sentence on direct and collateral review, and Garner filed this 28 U.S.C. § 2254 action in federal district court, raising twenty-three grounds for relief. The district court denied Garner’s habeas petition on all twenty-three grounds, and Garner now raises four grounds on appeal, three of which relate to the Miranda waiver. Because the record shows that Garner knowingly and intelligently waived his Miranda rights before he confessed to his crimes, and because Garner’s other claims lack merit, Garner is not entitled to habeas relief.
I.
On January 25, 1992, Addie F. Mack visited a local hospital emergency room in Cincinnati, Ohio for treatment. While in the hospital waiting area, Mack called her son to update him on her status, and accidentally left her purse unattended by the *259pay telephone. Defendant Garner took the purse and removed Mack’s keys, driver’s license, and wallet. Using the address listed on Mack’s driver’s license, Garner directed a taxicab to Mack’s apartment at 1969 Knob Court. When the taxicab arrived at the apartment, Garner asked the driver, Thomas J. Tolliver, to wait while Garner went inside. Garner provided Tol-liver with Mack’s wallet as collateral for payment of the cab fare.
Using the keys found in Mack’s purse, Garner entered the apartment and noticed four girls sleeping in one bedroom and two boys sleeping in a second bedroom. The children ranged from ten to thirteen years of age. At one point, one of the girls woke up and asked Garner for a glass of water. Garner provided her with water, and the girl watched television for a short time before going back to sleep. Garner carried several items from the apartment to the taxicab, including a VCR, television set, portable telephone, and a “boom box” radio. As he brought the items to the taxicab, Garner explained to Tolliver that he was removing the items because his girlfriend “threw him out” during a fight.
After removing the stolen property, Garner returned to the apartment and set three fires, two in upstairs bedrooms and one on a couch in the living room. Although the two upstairs fires smoldered and eventually went out, the couch fire completely destroyed the contents of the living room and filled the entire apartment with heavy smoke. Mack’s oldest child awoke during the fire and was able to escape through a window. The five other children died of smoke inhalation. Upon leaving the apartment, Garner instructed Tolliver to take him to a convenience store, where Garner purchased snacks. The pair then drove to Garner’s home at 3250 Bur-net Avenue. Tolliver helped Garner carry the stolen items into Garner’s home and accepted Mack’s television set as payment for the cab fare.
During the investigation of the fire, the police located Tolliver based on information provided by two officers who had observed a person loading items into a taxicab near Mack’s apartment shortly before the fire was reported. Tolliver told the officers that he picked up Garner at the hospital emergency room, drove to 1969 Knob Court, and waited outside while Garner entered the apartment and brought several items to the taxicab. Tolliver stated that, thereafter, he drove Garner to the convenience store and then to 3250 Burnet Avenue. After police presented to Tolliver still photographs generated from surveillance video taken at the convenience store, Tolliver identified Garner based on Garner’s clothing. Tolliver also identified Garner in two photo arrays that included Garner’s photograph, and officers recovered from Tolliver Mack’s television set.
Based on the information provided by Tolliver, police obtained a search warrant and searched Garner’s Burnet Avenue residence. Officers recovered several items that matched the descriptions given by Tolliver, including a VCR, “boom box” radio, and portable telephone. Officers also recovered Mack’s keys and copies of Mack’s children’s birth certificates. During the search, officers arrested Garner and advised him of his Miranda rights. Garner was transported to police headquarters, where he was again advised of his Miranda rights and presented with a waiver form. Garner agreed to waive his Miranda rights and provided a taped statement recounting the events described above.
In the taped statement, Garner admitted finding Mack’s purse and to taking a taxicab to Mack’s apartment with the intent to “take her things.” Garner stated that he noticed the children sleeping in the apart*260ment and admitted carrying a number of items from the apartment to the taxicab. Garner confirmed having watched the couch catch fire and explained that he started the fire to cover fingerprints that he had left on the couch. Garner told officers that he believed that the children would smell the smoke and leave the apartment because one child had already been awake and because all of the children were old enough to escape.
On February 3, 1992, Garner was indicted and charged with five counts of aggravated murder, each with three death penalty specifications, one count of aggravated burglary, two counts of aggravated arson, one count of theft, and one count of receiving stolen property. Garner pleaded no contest to the theft and receipt-of-stolen-property counts, and the trial court found him guilty on those counts. On October 1, 1992, the jury convicted Garner on the remaining counts and specifications. Following a sentencing hearing, the jury recommended imposition of the death sentence. The state trial court accepted the jury’s recommendation and sentenced Garner to death on the aggravated murder counts and to consecutive terms of imprisonment on all other counts.
On direct review in state court, Garner raised twenty-three assignments of error. The Ohio Court of Appeals and Ohio Supreme Court both affirmed Garner’s convictions and sentence, State v. Garner, No. C-920864, 1994 WL 466508 (Ohio Ct.App. Aug.31, 1994); State v. Garner, 74 Ohio St.3d 49, 656 N.E.2d 623 (1995), and the United States Supreme Court denied certiorari, Garner v. Ohio, 517 U.S. 1147, 116 S.Ct. 1444, 134 L.Ed.2d 564 (1996). Thereafter, Garner filed two petitions for post-conviction relief in state court. Both petitions were denied by the state trial court, and the denials were affirmed by the state court of appeals. State v. Garner, No. C-960995, 1997 WL 778982 (Ohio Ct.App. Dee.19, 1997); State v. Garner, No. C-990659, 2000 WL 492074 (Ohio Ct.App. Apr.28, 2000). The Ohio Supreme Court declined to exercise jurisdiction to hear both cases. State v. Garner, 81 Ohio St.3d 1497, 691 N.E.2d 1058 (1998); State v. Garner, 90 Ohio St.3d 1404, 734 N.E.2d 835 (2000).
On November 18, 1998, Garner filed a petition for a writ of habeas corpus in federal district court, raising twenty-three grounds for relief. The district court ultimately denied all of Garner’s claims and dismissed the petition. The district court granted Garner a certificate of appealability on three related claims: (1) that Garner did not knowingly and intelligently waive his Miranda rights before speaking with police; (2) that Garner’s state trial counsel were ineffective for failing to investigate and argue his Miranda claims; and (3) that the state trial court erred by not providing Garner with experts to assist with his Miranda claim. After Garner filed a notice of appeal, this court issued him a certificate of appealability on a fourth claim: that the process by which the petit jury venire was selected discriminated against African-Americans.
II.
A. [1] The record indicates that Garner
The record indicates that Garner knowingly and intelligently waived his Miranda rights, and Garner is therefore not entitled to habeas relief on his Miranda claim. Notwithstanding Garner’s failure to exhaust this claim in the state courts, 28 U.S.C. § 2254 permits us to deny his application on the merits.
The totality of the circumstances in this case shows that Garner’s waiver was knowing and intelligent. Garner has the burden of establishing that, under the to-
*261tality of the circumstances, he did not knowingly and intelligently waive his rights before speaking to the police. Clark v. Mitchell, 425 F.3d 270, 283 (6th Cir.2005). “We are also mindful that in a habeas proceeding the petitioner ‘has the burden of establishing his right to federal habeas relief....”’ Caver v. Straub, 349 F.3d 340, 351 (6th Cir.2003) (quoting Romine v. Head, 253 F.3d 1349, 1357 (11th Cir.2001)). Under this inquiry, we examine “the particular facts and circumstances surrounding [the] case, including the background, experience, and conduct of the accused.” Johnson v. Zerbst, 304 U.S. 458, 464, 58 S.Ct. 1019, 82 L.Ed. 1461 (1938); see also Edwards v. Arizona, 451 U.S. 477, 482, 101 S.Ct. 1880, 68 L.Ed.2d 378 (1981). The relevant question is not whether the “criminal suspect [knew] and [understood] every possible consequence of a waiver of the Fifth Amendment privilege,” but rather whether the “suspect [knew] that he [could] choose not to talk to law enforcement officers, to talk only with counsel present, or to discontinue talking at any time.” Colorado v. Spring, 479 U.S. 564, 574, 107 S.Ct. 851, 93 L.Ed.2d 954 (1987).
Here, Garner’s conduct before and during the interrogation demonstrates that he understood his Miranda rights and the consequences of waiving those rights. Contemporaneous evidence in the record indicates that Garner appeared “perfectly normal” and “very coherent” at the time that he waived his rights and confessed to his crimes. Officers read Garner the Miranda warnings at least two times before he confessed, and Garner signed and dated a form expressly waiving his rights. It is undisputed that the police officers took care to ensure that Garner understood the warnings and waiver before he signed the form. Officer Feldhaus of the Cincinnati Police Department testified that after reading each provision of the Miranda warnings to Garner, he asked Garner if he understood the meaning of that provision. Each time that he was asked, Garner responded that he understood his rights, including the waiver provision. Further, nothing in the record indicates that Garner verbally expressed a misunderstanding to police officers or otherwise engaged in conduct indicative of a misunderstanding.
Garner’s explanation of his conduct during the commission of his crimes moreover served to confirm his capacity to understand the Miranda warnings. While transferring the stolen items from Mack’s apartment to the taxicab, Garner explained to Tolliver that his girlfriend “threw him out,” thus necessitating the removal of his personal belongings. Garner also explained to police that he started the couch fire to rid the couch of any fingerprints that he may have left. Both of these statements indicate that Garner had the capacity to understand the criminal nature of his actions and the consequences of those actions. That Garner had this capacity at the time that he committed the crimes suggests that, when questioned about those crimes on the next day, Garner also had the capacity to understand and appreciate the consequences of speaking to police about his criminal conduct. We have held, in the similar context of a challenge to the voluntariness of a confession, that a defendant’s capacity to devise a criminal scheme was evidence of capacity to admit to devising the scheme. United States v. Macklin, 900 F.2d 948, 952 (6th Cir.1990); see also United States v. Turner, 157 F.3d 552, 555 (8th Cir.1998) (holding that defendant with low IQ knowingly waived his rights and noting that, at the time that defendant was stopped by police, defendant acted “in a manner more consistent with a person attempting to avoid being caught than a person who did not know what he was doing”); United States v. Solano-Godines, No. 96-10255, 1997 *262WL 407861, at *3 (9th Cir. July 21, 1997) (holding that waiver was knowing and intelligent where defendant “understood everything else that was going on throughout the day” and made up a “clever story that he was framed”). Accordingly, all evidence in the record of Garner’s conduct during, and leading up to, the interrogation indicates that Garner had the capacity to waive his Miranda rights knowingly and intelligently.
It follows from the above that, at the time of the interrogation, police officers had no indication that Garner’s “age, experience, education, background, and intelligence” may have prevented him from understanding the Miranda warnings. See Fare v. Michael C., 442 U.S. 707, 725, 99 S.Ct. 2560, 61 L.Ed.2d 197 (1979). As discussed, Garner appeared “perfectly normal” and “very coherent” to the interrogating officers. Moreover, in a competency report prepared prior to trial, Dr. Nancy Schmidtgoessling, a clinical psychologist, stated that Garner “appeared to be of near average intelligence by observation,” “appeared to be able to understand all questions and material presented to him,” and that “his expressive language abilities were intact.” Accordingly, even if Garner’s mental capacity, background, age, and experience did somehow prevent him from actually understanding the Miranda warnings — and the evidence indicates that they did not — the officers questioning Garner had no way to discern the misunderstanding in Garner’s mind. This is of primary significance given the original purpose underlying the Miranda decision, which was to “reduce the likelihood that the suspects would fall victim to constitutionally impermissible practices of police interrogation.” New York v. Quarles, 467 U.S. 649, 656, 104 S.Ct. 2626, 81 L.Ed.2d 550 (1984). As the Seventh Circuit explained in a thoughtful opinion:
The relevant constitutional principles are aimed not at protecting people from themselves but at curbing abusive practices by public officers.... [T]he knowledge of the police is vital. If they have no reason (there was none in [Colorado v.] Connelly, see 479 U.S. [157, 161-62, 107 S.Ct. 515, 93 L.Ed.2d 473 (1986) ]) to think that the suspect doesn’t understand them, there is nothing that smacks of abusive behavior. It would seem to follow that the question is not whether if [the defendant] were more intelligent, informed, balanced, and so forth he would not have waived his Miranda rights, but whether the police believed he understood their explanation of those rights; more precisely, whether a reasonable state court judge could have found that the police believed this.
Rice v. Cooper, 148 F.3d 747, 750-51 (7th Cir.1998).1 This analysis is fully consistent *263with Moran v. Burbine, 475 U.S. 412, 106 S.Ct. 1135, 89 L.Ed.2d 410 (1986), which explained that the Miranda waiver inquiry-had two dimensions: voluntariness and comprehension. 475 U.S. at 421, 106 S.Ct. 1135. At no point did the Supreme Court say that one of the two dimensions is to be examined from the perspective of the police while the other is to be examined from the perspective of later scientific inquiry. Instead, both are to be evaluated from the “totality of the circumstances surrounding the interrogation.” Id. (quoting Fare, 442 U.S. at 725, 99 S.Ct. 2560). The underlying police-regulatory purpose of Miranda compels that these circumstances be examined, in thefi totality, primarily from the perspective of the police. Because police had no reason to believe that Garner misunderstood the warnings, and because it is undisputed that the officers were otherwise reasonable and careful in giving the warnings and obtaining the confession, there is no basis for invalidating Garner’s Miranda waiver.
*262it might be argued that officers are free to recite the standard Miranda warnings to anyone they arrest, regardless of the person's evident mental condition, and to accept the person’s waiver. But this has to be wrong, though we cannot find a case that says so. If the suspect is a small child, or if it is apparent that he cannot speak English, then attempting to extract a waiver of Miranda rights is pretty obviously an abusive practice, as it is a calculated, conscious effort to extract a decision that is not the product of a rational choice. And likewise if it is apparent that because of illness, insanity, or mental retardation the suspect is incapable of rationally waiving his Miranda rights. The significance of the principle of Connelly, the principle that the Constitution doesn't protect the suspect against himself, is that if he understands the Miranda warnings yet is moved by a crazy impulse to blurt out a confession, the confession is admissible because it is not a
*263B.
Of course, while our primary focus must remain on what the interrogating officers could have concluded about Garner's ability to understand the warnings, we may consider later-developed evidence of a defendant’s actual mental ability to understand the warnings at the time of the interrogation. This is because, if it turns out by subsequent inquiry that a defendant in his mind could not actually understand the warnings, the finder of fact may be more inclined to determine in a close case that the police should have known that the defendant could not understand. Here, however, evidence in the record of Garner’s age, experience, education, background, and intelligence does not mandate the conclusion that, even viewed from his internal perspective, Garner could not understand the Miranda warnings. At the time of the waiver, Garner was nineteen years of age. It is undisputed that he had a troubled upbringing, poor education, and that his IQ of 76 placed him in the “borderline range of intelligence.” Testimony during the mitigation hearing indicated that Garner endured physical and sexual abuse at the hands of his family members. Garner and his siblings were often left alone to fend for themselves, and Garner did not perform well in school.
Dr. Jeffrey Smalldon, a clinical psychology expert, concluded in an affidavit prepared for the penalty phase of Garner’s trial that Garner’s “borderline intelligence, functional (i.e., organic) brain impairment, abusive and socially depraved background, and long history of impulsivity raise serious questions as to whether he could or did understand the consequences of signing the ‘Waiver of Rights.’ ” Dr. Smalldon acknowledged, however, that his assessment suffered from limitations and that a “[m]ore focused assessment would provide better, and perhaps even conclusive, information on this issue.”
In her competency report, Dr. Schmidt-goessling similarly noted that Garner had a long history of hyperactivity and impul-sivity, and was “functioning in the borderline range” of intelligence. Even so, upon observing Garner, Dr. Schmidtgoessling remarked that Garner “appeared to be of near average intelligence” and “appeared *264to be able to understand all questions and material presented to him suggesting that his receptive language is intact.” Dr. Schmidtgoessling concluded that Garner was “familiar with the specifics of the allegations against him” and “was able to give a coherent, realistic account of his behavior relevant to the allegations although his account differed in a couple of major respects ... from the statement made to police.” Schmidtgoessling wrote that Garner understood the roles of the various court personnel, was able to identify his attorneys by name, and defined his attorneys’ job as “to speak up for you, argue for you, defend you.”
Additionally, Dr. Schmidtgoessling administered various psychological tests in assessing Garner’s competency to stand trial. Garner received an average score on a memory test, and a low average score on a test that measured his nonverbal problem-solving abilities. Garner scored within normal limits on a screening test for perceptual motor functions, in the superior range (90th percentile) on the simple Trail Making Test, and well below average (below the 10th percentile) on a more complex Trail Making Test. Dr. Schmidtgoessling testified that the Trail Making Test is “a special kind of test that’s very strong in detecting organic [brain] impairment.” In her report, Dr. Schmidtgoessling concluded that “there are no indications of major mental illness although the question of some sort of organic impairment remains open.”
During federal habeas proceedings, the district court granted Garner’s motion to expand the record to include an affidavit and report submitted by Dr. Caroline Ev-erington.2 Like Drs. Smalldon and Schmidtgoessling, Dr. Everington concluded that Garner’s IQ test scores placed him in the “borderline range of intellectual functioning,” and that psychological reports, social history, and school records indicated that Garner had “intellectual problems during the developmental period” and “difficulties in academic and adaptive functioning.” Dr. Everington’s report also assessed Garner’s abilities to comprehend the Miranda warnings during his interrogation, and relied heavily on her administration of the so-called Grisso test, discussed more extensively below.
The assessments of Drs. Small-don, Schmidtgoessling, and Everington indicate that Garner suffered from diminished mental capacity, a troubled upbringing, and a poor education at the time that he confessed to his crimes. These assessments do not demonstrate, however, that Garner was incapable of knowingly and intelligently waiving his Miranda rights. It is well-established, in this circuit and others, that mental capacity is one of many factors to be considered in the totality of the circumstances analysis regarding whether a Miranda waiver was knowing and intelligent. Thus, diminished mental capacity alone does not prevent a defendant from validly waiving his or her Miranda rights. See Clark, 425 F.3d at 283-84; Finley v. Rogers, 116 F.App’x 630, 636-38 (6th Cir.2004); United States v. Rojas-Tapia, 446 F.3d 1, 7-9 (1st Cir.2006); Smith v. Mullin, 379 F.3d 919, 933-34 (10th Cir.2004); Young v. Walls, 311 F.3d 846, 849 (7th Cir.2002); Turner, 157 F.3d at 555-56; Rice, 148 F.3d at 750; Henderson v. DeTella, 97 F.3d 942, 948-49 (7th Cir.1996); Corell v. Thompson, 63 F.3d 1279, 1288 (4th Cir.1995); Starr v. Lockhart, 23 F.3d 1280, 1294 (8th Cir.1994); Derrick v. Peterson, 924 F.2d 813, *265824 (9th Cir.1991); Toste v. Lopes, 861 F.2d 782, 783 (2d Cir.1988); Dunkins v. Thigpen, 854 F.2d 394, 399-400 (11th Cir.1988). Rather, that factor must be viewed alongside other factors, including evidence of the defendant’s conduct during, and leading up to, the interrogation.
Case law in other circuits is instructive in this regard. For example, in Smith v. Mullin, the Tenth Circuit held that the defendant’s Miranda waiver was knowing and intelligent despite the facts that (1) the defendant suffered from borderline mental retardation, and (2) a clinical psychologist had concluded, based on the defendant’s Grisso test scores, that the defendant could not have validly waived his rights. 379 F.3d at 932-34. The Tenth Circuit found it significant that the clinical psychologist also testified that the defendant “would understand the role of police officers and the concept of a criminal charge,” and that the Grisso test was administered years after the interrogation. Id. at 933. The court also relied on a videotape showing the defendant’s conduct during the interrogation and noted that the defendant had had previous experience with the criminal justice system. Id. at 934. In United States v. Turner, the Eighth Circuit held that the defendant’s Miranda waiver was knowing and intelligent even though the defendant’s IQ was low-average to borderline, and he was possibly intoxicated by PCP at the time of interrogation and exhibited “bizarre” behavior and possible signs of mental illness after the interrogation. 157 F.3d at 555-56. The court determined that because the defendant was cooperative during the interrogation, gave accurate information, and, when stopped by police, “acted in a manner more consistent with a person attempting to avoid being caught than a person who did not know what he was doing,” the waiver was effective. Id. at 555.
In some cases, courts have concluded that a defendant’s limited intellectual capacity contributed to the determination that a waiver was not effective. Frequently, however, those cases also feature some observable indication to police that the defendant was incapable of understanding the Miranda warnings. For example, in United States v. Garibay, 143 F.3d 534, 537-38 (9th Cir.1998), the defendant suffered from a low IQ, but also primarily spoke Spanish and thus did not possess the English skills to understand the Miranda warnings without the assistance of a Spanish-speaking officer. Additionally, an officer that questioned the defendant was forced to rephrase questions when the defendant “did not appear to understand.” Id. at 539. In Cooper v. Griffin, 455 F.2d 1142, 1144-46 (5th Cir.1972), the Fifth Circuit held that neither defendant knowingly and intelligently waived his Miranda rights where both defendants were young boys, aged fifteen and sixteen, who were severely mentally retarded.
In the instant case, Garner’s conduct, speech, and appearance at the time of interrogation indicated that his waiver was knowing and intelligent, notwithstanding his diminished mental capacity. Like the defendants in Mullin and Turner, Garner was carefully read his Miranda rights and stated clearly to officers that he understood those rights. Garner agreed to execute a written waiver form, was cooperative during the interrogation, and clearly explained the incident in Mack’s apartment. Garner also engaged in conduct prior to being arrested that was “more consistent with a person attempting to avoid being caught than a person who did not know what he was doing.” Turner, 157 F.3d at 555. Garner invented a story about having a fight with his girlfriend to explain to the taxicab driver why he was removing the items from Mack’s apart*266ment, and Garner admitted to police that his purpose in lighting the couch on fire was to ensure that he left no fingerprints behind. Finally, at no time did Garner exhibit any outwardly observable indications that he did not understand the warnings or the circumstances surrounding his interrogation. Garner was not a minor, did not have trouble understanding English, and, although his IQ level indicates that he was functioning in the borderline range of intelligence, he was not so mentally retarded that officers had reason to believe that he could not understand his rights.
The assessments of Drs. Smalldon, Schmidtgoessling, and Everington, moreover, do not establish that Garner was incapable of effectively waiving his rights notwithstanding this outward evidence. For instance, Dr. Schmidtgoessling stated in her report that, despite Garner’s borderline IQ score, Garner was “familiar with the specifics of the allegations against him” and “was able to give a coherent, realistic account of his behavior relevant to the allegations although his account differed in a couple of major respects ... from the statement made to police.” Dr. Schmidtgoessling also noted that Garner was able to understand the roles of the various court personnel, identify his attorneys by name, and define accurately the duties of his attorneys. Cf. Mullin, 379 F.3d at 933 (finding it significant that defendant “would understand the role of police officers and the concept of a criminal charge”). Although Garner performed well below average on the complex Trail Making Test, he performed in the superior range on the simple Trail Making Test, received an average score on a memory test, a low average score on a test measuring nonverbal problem solving abilities, and a score within normal limits on a screening test for perceptual motor functions. And while Dr. Smalldon concluded in his affidavit that Garner’s borderline intellectual functioning “raise[s] serious doubts” about Garner’s ability to understand the Miranda warnings, Dr. Small-don acknowledged that a “[m]ore focused assessment” was necessary. Thus, neither Dr. Smalldon’s nor Dr. Schmidtgoessling’s assessment provides conclusive evidence that Garner did not understand and appreciate his Miranda rights, and Dr. Schmidt-goessling’s assessment offers some evidence that suggests that Garner indeed had the capacity to validly waive his rights.
Finally, Dr. Everington’s assessment with regard to Garner’s understanding during the police interrogation — which relies on her administration of the so-called Grisso test to Garner six years after his confession — adds little to demonstrate that the confession was not knowing and intelligent. The Grisso test purports to “provide an index of the person’s capacity for understanding the Miranda warnings at the time of the evaluation.” Thomas Gris-so, Instruments for Assessing Understanding & Appreciation of Miranda Rights 7 (1998). The Grisso test consists of four subtests, styled “instruments,” and named as follows: Comprehension of Miranda Rights (CMR); Comprehension of Miranda Rights-Recognition (CMR-R); Comprehension of Miranda Vocabulary (CMV); and Function of Rights in Interrogation (FRI). Dr. Everington administered the Grisso test in 1998, over six years after Garner was interrogated by police.
The results of the first subtest (CMR) provide little support for concluding that Garner could not adequately understand the Miranda warnings. The first subtest “assesses the examinee’s understanding of the Miranda warnings as measured by the examinee’s paraphrased description of the warnings.” Id. at 5. The examinee is presented with each of four sentences of the *267Miranda warning and is invited to tell the examiner “what [the sentence] says in your own words.” Id. at 17. Responses for each of the four sentences are scored 2, 1, or 0, so that the examinee can get a total of 8 points by optimally paraphrasing the four sentences. The Scoring Criteria portion of the test protocol lists several examples of responses that would receive each of the three possible scores. For instance, the first sentence given to the examinee is “You do not have to make a statement and have the right to remain silent.” Id. at 19. One example of a 2-point responding paraphrase is “You can tell them everything if you want, or just not say anything.” One example of a 1-point response is “I would say it’s best to say nothing.” One example of a 0-point response is “It means if you don’t talk they lock you up” Id. at 23-25. On the CMR subtest, Garner received a score of 6 out of the possible 8 points. Dr. Everington noted that Garner had difficulty in providing “a satisfactory definition for two of the four statements of the warning.” That is, he got half-credit on two of the four questions. This subtest, at least, does not appear to provide support for a conclusion that Garner could not adequately understand Miranda warnings.3
Garner got a perfect score on the second subtest (CMR-R), which “assesses the ex-aminee’s understanding of the Miranda warnings as measured by the examinee’s ability to identify whether various interpretations provided by the examiner are the same as or different from the warning that was presented.” Id. at 5. “After each warning statement, the examiner asks the examinee to listen to three other statements, .... [and][t]he examinee simply says ‘same’ or ‘different’ after each alternative statement.” Id. On this subtest, Garner received a score of 12 out of 12 points, placing him higher than 64% of a sample of 260 adults. Id. at 85 tbl.3. Dr. Everington noted that Garner’s CMR-R score indicated that Garner “did not have difficulty in recognizing the meaning of the warning when presented in a true-false format.” This subtest, it should be noted, is the only one of the four that is objectively graded. Id. at 11.
While Garner did least well on the third subtest (CMV), much of that subtest evaluated Garner’s understanding of words more difficult than those actually used in the warnings given to him. The third subtest “assesses the examinee’s ability to define six words that appear in the version of the Miranda warnings on which the [Grisso test is] based.” Id. at 5. Those words are consult, attorney, interrogation, appoint, entitled, and right. Id. at 36-44. The examinee is asked to “tell [the examiner] in your own way what the word means.” Id. at 35. As in the first subtest, each response gets a score of 2, 1, or 0, so that there is a possible perfect score of 12. For instance, the first word is consult. According to the scoring criteria, an example of a full-credit (2-point) response is “To help to decide.” An example of a half-credit (1-point) response is “To talk confidentially.” An example of a 0-credit response is “To plan something.” Id. at 39. Dr. Everington reported that Garner had difficulty defining five of the six vocabulary words — consult, attorney, appoint, entitled, and right — and received a score of 7 out of the possible 12 points. Thus, it appears that Garner received half-credit for each of those five words. Garner’s performance was the worst on this part of the test. But three of the five words for which he received half credit (consult, at-*268tomey, and entitled) were not present in the version of the Miranda rights that police read to Garner. Police used the simpler term “lawyer” in lieu of the term “attorney,” the simpler phrase “talk to” in lieu of the term “consult,” and the simpler words “have the right to” in lieu of “entitled.”4 These differences are significant. Indeed, if Garner had responded “lawyer” (the word actually used in Garner’s warning) when asked on the Grisso test to tell what the word “attorney” means, he would have received a full two points for that word. Id. at 40. And if he had responded “Has a right to it” (essentially the words used in Garner’s warning) when asked on the Grisso test what the word “entitled” means, he would have received a full two points for that word. Id. at 43. Even the Grisso manual itself recognizes that
[w]hen local versions of the warnings are very different from those used in the measures, it is possible that the exami-nee might receive a score on the instruments that suggests poorer or better understanding than the examinee would have manifested for the version of the warnings that police officers actually provided to the examinee.
Id. at 7.5 Because the language used in the CMV subtest significantly differed from the language used in the warning that police read to Garner, Garner’s poor performance on the CMV subtest can hardly be relied upon as evidence that Garner did not understand or appreciate the Miranda warnings.6
The fourth subtest (FRI) does little more to show that Garner could not understand the Miranda warnings. That sub-test “assesses the examinee’s grasp of the significance of the Miranda rights in the context of interrogation” by using “four picture stimuli, which are accompanied by brief vignettes.” Id. at 6. “Each picture and vignette are [sic] followed by a set of standardized questions (15 in all) that assess the examinee’s grasp of the significance of’ the rights to counsel and silence, and the nature of interrogation. Id. Five of the questions relate to the nature of police interrogation, five relate to the function and significance of legal counsel, and five relate to the function and significance of the right to silence. Responses again are scored 2, 1, or 0, so that the total possible score is 30. Garner scored 24 points, including a perfect score (10) on the “nature of interrogation” portion and a perfect score (10) on the “right to counsel” portion. He received 4 out of 10 points on the “right to silence” portion.7 The sub*269jective and legally questionable nature of the grading criteria for this subtest, and particularly for the portion that Garner did poorly on, bring into question its usefulness in determining whether an examinee could understand the Miranda warnings. Shown a drawing of a suspect in a room with two police officers, the examinee is asked, “Finish this sentence. If Greg decides to tell the police about what he did, then the things that Greg says_” Id. at 48. One example of a full-credit 2-point answer is “Can turn against him later in court.” One example of a half-credit 1-point answer is “Will get him into detention.” One example of a 0-credit answer is “He will tell the policemen.” Id. at 60-61. The minor difference between the 2-point and 1-point answers is notable.8
After relating Garner’s scores on the Grisso test, Dr. Everington concluded in her report that Garner’s Grisso test results “indicate that even six years later, Garner may not have a complete understanding of [the Miranda ] warning.” In her affidavit, prepared two years after the Grisso test and eight years after the interrogation, Dr. Everington used stronger language, concluding that “Mr. Garner’s performance on this test indicates that he does not have full comprehension of Miranda warnings or his right to remain silent.” These conclusions are hardly compelling in light of the apparent problems with drawing this conclusion from Garner’s performance on the Grisso test. Other courts have had similar qualms about testimony relying on the test. For instance, the Supreme Court of Connecticut has concluded that a lower court did not abuse its discretion in excluding Grisso test results under its standard for admission of expert scientific evidence. State v. Griffin, 273 Conn. 266, 869 A.2d 640, 650-52 (2005). The court stated “we know of no case in which testimony concerning the Grisso test has been admitted into evidence over objection,” and noted cases from Florida and New York in which testimony regarding the Grisso test or similar protocol was excluded following a preliminary hearing on admissibility. Id. at 650. The Grisso test itself does not purport to “measure the validity of the waiver of Miranda rights, or ‘legal competence’ to waive Miranda rights,” see Grisso, supra, at 8, and, consequently, a poor score on one or more parts of the test does not, ipso facto, lead to a conclusion that the examinee lacks the capacity to knowingly and intelligently waive those rights, see id. (“[Tjhere is no particular degree of understanding (or score on these instruments) associated with ‘adequate’ understanding from a legal perspective.”); see also Mullin, 379 F.3d *270at 933 (rejecting claim that Miranda waiver of defendant who suffered from borderline mental retardation was unknowing and unintelligent despite low Grisso test scores).
In addition, the Grisso test purports to provide an index for capacity to understand the warnings only “at the time of the evaluation,” not at the time that the warnings were given. Grisso, supra, at 7; see also id. at 71 (“Current comprehension, even if it is valid for the present time, may or may not be representative of the individual’s comprehension at some retrospective time.”). In this case, Dr. Everington administered the test in 1998, over six years after police read the warnings to Garner. Leaving aside the obvious incentive for a defendant who has already been sentenced to death to feign misunderstanding on such a test, there is simply no way of telling whether Garner’s Grisso test scores are an accurate indicator of his ability to understand the warnings when police administered the warnings in 1992. This is so regardless of the fact that studies have indicated that Grisso test scores are generally positively correlated with age. See id. at 83 tbl.l. Cf. Mullin, 379 F.3d at 933 (noting that “the ‘Grisso test’ Dr. Hopewell administered took place years after [defendant’s] interrogation and the deterioration of his condition in jail could have affected the results”).9
It is thus not surprising that Dr. Ever-ington’s initial assessment in 1998 concluded only that Garner “may not have a complete understanding of [the Miranda ] warning” and that Garner’s deficits in intellectual functioning “could have hindered his understanding of [the] process” (emphasis added). Like the conclusions of Drs. Schmidtgoessling and Smalldon, these conclusions do not provide sufficient evidence that Garner’s waiver was not knowing and intelligent.
C.
In sum, Garner’s conduct during, and leading up to, the interrogation indicated that he understood and appreciated his Miranda rights before executing the waiver. Because nothing in the record demonstrates otherwise, Garner’s Miranda claim *271does not require habeas relief. We would reach this conclusion regardless of whether we reviewed the issue de novo, under a deferential AEDPA standard, or under a “modified AEDPA” standard.10 It is therefore not necessary for us to determine which of these standards applies in this case. See Pierre N. Leval, Judging Under the Constitution: Dicta About Dicta, 81 N.Y.U. L.Rev. 1249, 1267 (2006).
III.
None of Garner’s remaining claims warrants habeas relief. Because Garner’s substantive Miranda claim lacks merit, the Ohio state courts’ determination that Garner’s counsel were not ineffective for failing to investigate or raise that claim was not contrary to, or an unreasonable application of, Supreme Court precedent. See Dist. Ct. Op. at 49-50. Garner’s claim that the state trial court erred by not providing him with experts to assist with his Miranda claim similarly lacks merit. Garner was provided with access to mental health experts during trial. Moreover, the assistance of other experts would not have been sufficient to show that his waiver was unintelligent. Dist. Ct. Op. at 58-62. Finally, Garner’s claim that the process for selecting the petit jury venire unconstitutionally discriminated against African-Americans was procedurally defaulted and, in any event, is without merit for the reasons stated by the district court. Dist. Ct. Op. at 27-34.
IV.
For the foregoing reasons, we affirm the judgment of the district court.
. This analysis of course does not mean that the police can disregard signs or even hints that an interrogation suspect does not understand. As the Seventh Circuit explained in Rice, *263product of coercion. The police have given him his Miranda warnings in an intelligible form; it is not their fault that he is impulsive. It is different, if perhaps only by a shade, if the police question him knowing that he does not understand his rights. 148 F.3d at 750. Thus it would be impossible to read our holding today as suggesting that “a deaf defendant could give a knowing and intelligent waiver when he is given only a mumbled Miranda warning.” (Dissent at 274).
. Because the district court concluded that Dr. Everington’s assessments did not warrant further evidentiary hearings, it did not give the state the opportunity either to cross-examine Dr. Everington or to introduce expert evidence to counter her conclusions.
. The Grisso manual provides tables for clinical interpretation of an examinee’s scores. Garner’s scores would have been roughly in the bottom 30% but not in the bottom 20% of the total adult sample on the CMR subtest. Grisso, supra, at 84 tbl.2.
. As the district court noted, this difference in language also calls into question the validity of Gamer’s CMR score. For example, Garner’s CMR score depended on his ability to phrase the following warning in his own words: “You are entitled to consult with an attorney before interrogation and to have an attorney present at the time of the interrogation.” Grisso, supra, at 20. The actual warning read to Garner was phrased in simpler terms: “You have the right to talk to a lawyer for advice before we ask you any questions and have him with you during questioning.”
. Furthermore, the difference between a half-credit and a full-credit response is not always crisp. With respect to appoint, ”[t]o get a person to do the job” is a 2-point answer, while "[t]o offer them money to do the job” is a 1-point answer. Grisso, supra, at 42. In scoring definitions of right, "[y]ou are entitled to it” receives 2 points, while "[fit’s your decision” receives 1 point. Id. at 44.
. Of course, the question of how Garner compares to the rest of the population in understanding a particular set of difficultly-phrased Miranda warnings has no bearing on the question of whether he understood the constitutionally adequate warnings actually given to him.
. Garner’s total FRI score was below the mean for both adult offenders (26.31) and non-offenders (25.52). Grisso, supra, at 94 tbl. 12. His scores on the “nature of interroga*269tion” and "right to counsel” portions were above the mean for adult offenders (9.60 and 9.25) and non-offenders (9.61 and 9.07), id. at 91-92 tbl.9, 10, while his score on the "right to silence” portion was below the mean for each group (7.48 and 6.84, respectively), id. at 93 tbl.ll.
. Another question involving the right to counsel in this subtest is even more troubling. The examinee is shown a drawing of a courtroom hearing with judge, police officers, parents, the defendant’s lawyer, and the defendant. Grisso, supra, at 45. The question is: "Greg did not tell the police anything about what he did. Here in court, if he were told to talk about what he did that was wrong, will he have to talk about it?” Id. at 49. According to the Scoring Criteria, the 2-point answer is “No,” the 1-point answer is "Yes, if his lawyer says it’s best to.” Three possible 0-point answers are: “Yes,” “I don’t know,” and, remarkably, "Only if the judge tells him to.” Id. at 66. Given the conceivable situations in which a defendant would be required to talk, such as if he had been granted immunity, and given that in court it is the judge that determines the law, "Only if the judge tells him to” is not a bad answer, yet it would get no points according to the Scoring Criteria.
. Dr. Everington's administration of the test raises yet another concern with respect to Garner’s test scores. While the Grisso test manual "mentions exceptionally high estimates of interrater reliability,” that reliability was achieved by requiring raters to participate in "intensive training (16 to 32 hrs.) with additional practice sessions.” Richard Rogers, Mandy J. Jordan & Kimberly S. Harrison, A Critical Review of Published Competency-to-Confess Measures, 28 Law & Hum. Behav. 707, 712 (2004). A reliability analysis was required for three of the four Grisso subtests (CMR, CMV, FRI) because the scoring of those parts requires the rater to interpret the examinee’s response and, based on criteria set out in the scoring manual, to determine whether the response warranted 2, 1, or 0 points. See Grisso, supra, at 10-12, 17-66. Because of the subjective aspect of these parts of the test, an examinee's score could vary from rater to rater, thus requiring intensive training to establish interrater reliability. In this case, there is no evidence in the record that Dr. Everington received any training to administer the test, much less the intensive training necessary to ensure the reliability of Garner’s scores on the three subjectively-scored parts. And, as Rogers, Jordan, and Harrison have observed, "[t]he likelihood of a forensic practitioner achieving this sophisticated level of training is exceedingly remote.” Rogers, Jordan & Harrison, supra, at 712. Tellingly, Garner received a perfect score on the CMR-R subtest, the only "totally objective” part of the Grisso test. Because Dr. Ever-ington’s administration of the Grisso test brings into question the reliability of Garner’s CMR, CMV, and FRI scores, an argument can easily be made that Garner's CMR-R score most accurately represents Gamer's ability to comprehend and appreciate the Miranda warnings. See Grisso, supra, at 11 (noting that the CMR-R "requires no judgment on the part of scorers”).
. Garner did not raise in state court the substantive Miranda claim that he now raises in his federal habeas petition, but he did argue in state court on collateral review that his trial counsel were ineffective for failing to inquire into the Miranda waiver issue. Apart from the procedural default issue, it could be argued that modified AEDPA review would apply here in light of the fact that the analysis of Garner's substantive Miranda claim "bears some similarity” to the analysis of the ineffective-assistance claim adjudicated in state court. See, e.g., Filiaggi v. Bagley, 445 F.3d 851, 854 (6th Cir.2006).