Dassey v. Dittmann

Full title: Brendan DASSEY, Petitioner–Appellee, v. Michael A. DITTMANN…

Court: United States Court of Appeals, Seventh Circuit.

Date published: Dec 8, 2017

877 F.3d 297 (7th Cir. 2017)

Opinion

No. 16-3397

12-08-2017

Brendan DASSEY, Petitioner–Appellee, v. Michael A. DITTMANN, Respondent–Appellant.

Laura Nirider, Attorney, Steven A. Drizin, Northwestern University School of Law, Chicago, IL, Robert J. Dvorak, Attorney, Halling & Cayo, S.C., Milwaukee, WI, for Petitioner–Appellee. Luke N. Berg, Deputy Solicitor General, Daniel P. Lennington, Attorney, Jacob J. Wittwer, Attorney, Office of the Attorney General, Wisconsin Department of Justice, Madison, WI, Misha Tseytlin, Attorney, Ryan J. Walsh, Attorney, Kevin Michael LeRoy, Attorney, Office of the Solicitor General, Wisconsin Department of Justice, Madison, WI, for Respondent–Appellant. Marsha L. Levick, Attorney, Juvenile Law Center, Philadelphia, PA, for Amicus Curiaes.

Hamilton, Circuit Judge.

Laura Nirider, Attorney, Steven A. Drizin, Northwestern University School of Law, Chicago, IL, Robert J. Dvorak, Attorney, Halling & Cayo, S.C., Milwaukee, WI, for Petitioner–Appellee.

Luke N. Berg, Deputy Solicitor General, Daniel P. Lennington, Attorney, Jacob J. Wittwer, Attorney, Office of the Attorney General, Wisconsin Department of Justice, Madison, WI, Misha Tseytlin, Attorney, Ryan J. Walsh, Attorney, Kevin Michael LeRoy, Attorney, Office of the Solicitor General, Wisconsin Department of Justice, Madison, WI, for Respondent–Appellant.

Marsha L. Levick, Attorney, Juvenile Law Center, Philadelphia, PA, for Amicus Curiaes.

Before Wood, Chief Judge, and Easterbrook, Kanne, Rovner, Williams, Sykes, and Hamilton, Circuit Judges.^*

* Circuit Judges Flaum and Barrett did not participate in the consideration or decision of this case.

Hamilton, Circuit Judge.

Petitioner Brendan Dassey confessed on videotape to participating in the 2005 rape and murder of Teresa Halbach and the mutilation of her corpse. The Wisconsin state courts upheld Dassey's convictions for these crimes, finding that his confession was voluntary and could be used against him. The principal issue in this habeas corpus appeal is whether that finding was based on an unreasonable application of Supreme Court precedent or an unreasonable view of the facts. See 28 U.S.C. § 2254(d).

Whether Dassey's confession was voluntary or not is measured against a general standard that takes into account the totality of the circumstances. See Withrow v. Williams , 507 U.S. 680, 693–94, 113 S.Ct. 1745, 123 L.Ed.2d 407 (1993) ; Gallegos v. Colorado , 370 U.S. 49, 55, 82 S.Ct. 1209, 8 L.Ed.2d 325 (1962) ; see also Fare v. Michael C ., 442 U.S. 707, 727, 99 S.Ct. 2560, 61 L.Ed.2d 197 (1979) (admissibility of juvenile confession). Some factors would tend to support a finding that Dassey's confession was not voluntary: his youth, his limited intellectual ability, some suggestions by the interrogators, their broad assurances to a vulnerable suspect that honesty would produce leniency, and inconsistencies in Dassey's confession. Many other factors, however, point toward a finding that it was voluntary. Dassey spoke with the interrogators freely, after receiving and understanding Miranda warnings, and with his mother's consent. The interrogation took place in a comfortable setting, without any physical coercion or intimidation, without even raised voices, and over a relatively brief time. Dassey provided many of the most damning details himself in response to open–ended questions. On a number of occasions he resisted the interrogators' strong suggestions on particular details. Also, the investigators made no specific promises of leniency.

After the state courts found the confession voluntary, a federal district court and a divided panel of this court found that the state courts' decision was unreasonable and that Dassey was entitled to a writ of habeas corpus. We granted en banc review to consider the application of the deferential standards of 28 U.S.C. § 2254(d) and the implications of the panel decision for interrogations of juvenile suspects. The state courts' finding that Dassey's confession was voluntary was not beyond fair debate, but we conclude it was reasonable. We reverse the grant of Dassey's petition for a writ of habeas corpus.

Part I provides an overview of the applicable law. Part II sets forth the relevant facts about Teresa Halbach's murder, Dassey's confession, and the court proceedings. Part III applies the law to the relevant facts, keeping in mind the deference we must give under § 2254(d) to state court decisions as to which reasonable judges might differ.

I. The Applicable Law

We first discuss our standard of review under the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA) and then describe the Supreme Court's clearly established law for when a confession, particularly a confession by a sixteen–year–old like Dassey, is deemed voluntary and admissible.

A. Deference Under AEDPA

In considering habeas corpus petitions challenging state court convictions, "our review is governed (and greatly limited) by" AEDPA. Hicks v. Hepp , 871 F.3d 513, 524 (7th Cir. 2017) (citation omitted). The standards in 28 U.S.C. § 2254(d) were designed to "prevent federal habeas 'retrials' and to ensure that state–court convictions are given effect to the extent possible under law." Id. , quoting Bell v. Cone , 535 U.S. 685, 693, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002). Section 2254(d) provides that a state court conviction cannot be overturned unless the state courts' adjudication of a federal claim on the merits:

(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or

(2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.

The decision federal courts look to is the "last reasoned state–court decision" to decide the merits of the case, even if the state's supreme court then denied discretionary review. Johnson v. Williams , 568 U.S. 289, 133 S.Ct. 1088, 1094 n.1, 185 L.Ed.2d 105 (2013). In this case, we look to the Wisconsin Court of Appeals decision that Dassey's confession was voluntary.¹

1 On October 30, 2017, the Supreme Court heard argument in Wilson v. Sellers, No. 16–6855, where one question is whether federal courts in habeas cases should continue to "look through" state supreme court summary decisions on the merits to the last state court decision that provided an explanation. See generally Hittson v. Chatman, ––– U.S. ––––, 135 S.Ct. 2126, 2127, 192 L.Ed.2d 887 (2015) (Ginsburg, J., concurring in denial of certiorari). If the Court holds in Wilson that federal courts reviewing a state supreme court summary denial of review should give the state courts the benefit of any merits rationale the record could support, our review would become even more deferential, so the outcome here would not change.

The standard for legal errors under § 2254(d)(1) was meant to be difficult to satisfy. Harrington v. Richter , 562 U.S. 86, 102, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011). The issue is not whether federal judges agree with the state court decision or even whether the state court decision was correct. The issue is whether the decision was unreasonably wrong under an objective standard. Williams v. Taylor , 529 U.S. 362, 410–11, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (majority opinion of O'Connor, J.). Put another way, we ask whether the state court decision "was so lacking in justification that there was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement." Richter , 562 U.S. at 103, 131 S.Ct. 770. The existing law that applies is limited to that of the Supreme Court of the United States, which has instructed the lower federal courts to uphold a state court conviction unless the record "cannot, under any reasonable interpretation of the [Court's] controlling legal standard, support a certain ruling." Panetti v. Quarterman , 551 U.S. 930, 953, 127 S.Ct. 2842, 168 L.Ed.2d 662 (2007). Even if we were to consider the approach in past Supreme Court decisions outmoded, as the dissents suggest, a state court's decision consistent with the Supreme Court's approach could not be unreasonable under AEDPA.

As a result, federal habeas relief from state convictions is rare. It is reserved for those relatively uncommon cases in which state courts veer well outside the channels of reasonable decision–making about federal constitutional claims. AEDPA deference is not conclusive, however. Where the record shows that state courts have strayed from clearly established federal law, we can and do grant relief. E.g., Richardson v. Griffin , 866 F.3d 836 (7th Cir. 2017) ; Jones v. Calloway , 842 F.3d 454 (7th Cir. 2016) ; McManus v. Neal , 779 F.3d 634 (7th Cir. 2015) ; Shaw v. Wilson , 721 F.3d 908 (7th Cir. 2013) ; Harris v. Thompson , 698 F.3d 609 (7th Cir. 2012) ; Jones v. Basinger , 635 F.3d 1030 (7th Cir. 2011).

Review of state court factual findings under AEDPA is similarly deferential. Under § 2254(d)(2), federal courts cannot declare "state–court factual determinations ... unreasonable merely because [we] would have reached a different conclusion in the first instance." Brumfield v. Cain , ––– U.S. ––––, 135 S.Ct. 2269, 2277, 192 L.Ed.2d 356 (2015) (internal quotation marks and citation omitted). AEDPA does not permit federal courts to "supersede the trial court's ... determination" if a review of the record shows only that "[r]easonable minds ... might disagree about the finding in question." Id. (internal quotations and citations omitted). But again, "deference does not imply abandonment or abdication of judicial review, and does not by definition preclude relief." Id. (internal quotations and citations omitted).

B. The Law of Confessions

The Due Process Clause of the Fourteenth Amendment forbids the admission of an involuntary confession in evidence in a criminal prosecution. Miller v. Fenton , 474 U.S. 104, 109, 106 S.Ct. 445, 88 L.Ed.2d 405 (1985). In deciding whether a confession was voluntary, courts assess "the totality of all the surrounding circumstances—both the characteristics of the accused and the details of the interrogation." Schneckloth v. Bustamonte , 412 U.S. 218, 226, 93 S.Ct. 2041, 36 L.Ed.2d 854 (1973) ; see also Withrow v. Williams, 507 U.S. 680, 693–94, 113 S.Ct. 1745, 123 L.Ed.2d 407 (1993) (collecting relevant factors). The purpose of this test is to determine whether "the defendant's will was in fact overborne." Miller , 474 U.S. at 116, 106 S.Ct. 445.

The Supreme Court's many cases applying the voluntariness test have not distilled the doctrine into a comprehensive set of hard rules, though prohibitions on physical coercion are absolute. See Mincey v. Arizona , 437 U.S. 385, 401, 98 S.Ct. 2408, 57 L.Ed.2d 290 (1978) (statements resulted from "virtually continuous questioning of a seriously and painfully wounded man on the edge of consciousness"); Brown v. Mississippi , 297 U.S. 278, 279, 56 S.Ct. 461, 80 L.Ed. 682 (1936) (confessions extracted by "brutality and violence"). AEDPA does not "require state and federal courts to wait for some nearly identical factual pattern before a legal rule must be applied" because "even a general standard may be applied in an unreasonable manner." Panetti , 551 U.S. at 953, 127 S.Ct. 2842, quoting Carey v. Musladin , 549 U.S. 70, 81, 127 S.Ct. 649, 166 L.Ed.2d 482 (2006) (Kennedy, J., concurring in the judgment); accord, Yarborough v. Alvarado , 541 U.S. 652, 663–64, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004).

Nevertheless, applying a general standard like voluntariness "can demand a substantial element of judgment," and determining whether that judgment is reasonable "requires considering the rule's specificity." Alvarado , 541 U.S. at 664, 124 S.Ct. 2140. "The more general the rule, the more leeway courts have in reaching outcomes in case–by–case determinations." Id. (upholding state court Miranda conclusion where factors pointed in opposite directions). The state courts had such leeway here, and in the end, that leeway is decisive as we apply the test of § 2254(d)(1).

This general standard has some specific requirements to guide courts. First, a person arguing his confession was involuntary must show that the police engaged in coercive practices. See Colorado v. Connelly , 479 U.S. 157, 164–65, 107 S.Ct. 515, 93 L.Ed.2d 473 (1986). Physically abusive interrogation tactics would constitute coercion per se . Stein v. New York , 346 U.S. 156, 182, 73 S.Ct. 1077, 97 L.Ed. 1522 (1953) (physical violence is per se coercion), overruled on other grounds by Jackson v. Denno , 378 U.S. 368, 381, 84 S.Ct. 1774, 12 L.Ed.2d 908 (1964) ; Brown , 297 U.S. at 286–87, 56 S.Ct. 461 (coercion and brutality); United States v. Jenkins , 938 F.2d 934, 938 (9th Cir. 1991) (physical abuse is coercion per se ); Miller v. Fenton , 796 F.2d 598, 604 (3d Cir. 1986) (same).

Interrogation tactics short of physical force can amount to coercion. The Court has condemned tactics designed to exhaust suspects physically and mentally. Such tactics include long interrogation sessions or prolonged detention paired with repeated but relatively short questioning. Davis v. North Carolina , 384 U.S. 737, 752, 86 S.Ct. 1761, 16 L.Ed.2d 895 (1966) (finding coercive the practice of repeated interrogations over sixteen days while the suspect was being held incommunicado).

The Supreme Court has not found that police tactics not involving physical or mental exhaustion taken alone were sufficient to show involuntariness. In several cases, the Court has held that officers may deceive suspects through appeals to a suspect's conscience, by posing as a false friend, and by other means of trickery and bluff. See, e.g., Procunier v. Atchley , 400 U.S. 446, 453–54, 91 S.Ct. 485, 27 L.Ed.2d 524 (1971) (suspect was deceived into confessing to false friend to obtain insurance payout to children and stepchildren); Frazier v. Cupp , 394 U.S. 731, 739, 89 S.Ct. 1420, 22 L.Ed.2d 684 (1969) (deceiving suspect about another suspect's confession). False promises to a suspect have similarly not been seen as per se coercion, at least if they are not quite specific. See Arizona v. Fulminante , 499 U.S. 279, 285, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991) (rejecting language in Bram v. United States , 168 U.S. 532, 18 S.Ct. 183, 42 L.Ed. 568 (1897), stating that a confession could not be obtained by "any direct or implied promises," id. at 542–43, 18 S.Ct. 183, but finding promise to protect suspect from threatened violence by others rendered confession involuntary); Welsh S. White, Confessions Induced by Broken Government Promises , 43 Duke L.J. 947, 953 (1994).

False promises may be evidence of involuntariness, at least when paired with more coercive practices or especially vulnerable defendants as part of the totality of the circumstances. E.g., Lynumn v. Illinois , 372 U.S. 528, 534, 83 S.Ct. 917, 9 L.Ed.2d 922 (1963) (pre– Miranda confession found involuntary based on false promise of leniency to indigent mother with young children, combined with threats to remove her children and to terminate welfare benefits, along with other factors). But the Supreme Court allows police interrogators to tell a suspect that "a cooperative attitude" would be to his benefit. Fare v. Michael C ., 442 U.S. 707, 727, 99 S.Ct. 2560, 61 L.Ed.2d 197 (1979) (reversing finding that confession was involuntary). Supreme Court precedents do not draw bright lines on this subject.

In assessing voluntariness, courts must weigh the tactics and setting of the interrogation alongside any particular vulnerabilities of the suspect. Bustamonte , 412 U.S. at 226, 93 S.Ct. 2041. Relevant factors include the suspect's age, intelligence, and education, as well as his familiarity with the criminal justice system. Withrow , 507 U.S. at 693–94, 113 S.Ct. 1745 (collecting factors); Michael C. , 442 U.S. at 725–26, 99 S.Ct. 2560 (significant criminal justice experience); Clewis v. Texas , 386 U.S. 707, 712, 87 S.Ct. 1338, 18 L.Ed.2d 423 (1967) (limited educational attainment); Culombe v. Connecticut , 367 U.S. 568, 620, 81 S.Ct. 1860, 6 L.Ed.2d 1037 (1961) (intellectual disability); Gallegos v. Colorado , 370 U.S. 49, 53, 82 S.Ct. 1209, 8 L.Ed.2d 325 (1962) (age).

The interaction between the suspect's vulnerabilities and the police tactics may signal coercion even in the absence of physical coercion or threats. The Supreme Court has made it clear that juvenile confessions call for "special care" in evaluating voluntariness. E.g., Haley v. Ohio , 332 U.S. 596, 599, 68 S.Ct. 302, 92 L.Ed. 224 (1948) ; see also J.D.B. v. North Carolina , 564 U.S. 261, 277, 131 S.Ct. 2394, 180 L.Ed.2d 310 (2011) ; In re Gault , 387 U.S. 1, 45, 87 S.Ct. 1428, 18 L.Ed.2d 527 (1967) ; Gallegos , 370 U.S. at 54, 82 S.Ct. 1209. In juvenile cases, the law is particularly concerned with whether a friendly adult is present for or consents to the interrogation. In re Gault , 387 U.S. at 55–56, 87 S.Ct. 1428 ; Gallegos , 370 U.S. at 53–54, 82 S.Ct. 1209 ; Haley , 332 U.S. at 600, 68 S.Ct. 302. Concerns about physical exhaustion, naïveté about friendly police in the context of an adversarial police interview, and intellectual disability also take on heightened importance for assessing whether a juvenile's will was overborne.²

2 We have reservations about the use of "suggestibility" as a factor in this analysis, at least on these facts. Dassey relies heavily on the results of a Gudjonsson Suggestibility Scale test measuring him as more susceptible to fabrication than 95 people out of 100, given slight prodding by questioners. A Gudjonsson test is administered by reading a short story aloud to an examinee and then later asking leading questions about it. The more answers that change in response to mild pressure, the more suggestible the examinee is. The administration of this test for people with intellectual disabilities has been criticized because they may have good recall of their own lived experiences but poor recall of facts not relevant to their lives. Paul Willner, Assessment of capacity to participate in court proceedings: a selective critique and some recommendations, 17 Psychology, Crime & Law 117, 117 (2011). This criticism mirrors Dassey's own testimony that his recall was better for lived experiences. In any event, the State's expert forcefully contested both the administration and meaning of Dassey's Gudjonsson test at trial. We cannot draw conclusions from these disputed results.

As we detail below, Dassey's case presents different factors pointing in opposite directions. Those most important to our analysis include: his age and intellectual ability; the physical circumstances of the interrogation; the manner and actions of the police in questioning Dassey, including bluffing about what they knew and assuring him of the value of honesty; Dassey's resistance or receptiveness to suggestions by interrogators; and the extent to which he provided the most incriminating information in response to open–ended, non–leading questions.

II. The Murder, the Interrogation, and the Convictions

A. The Murder of Teresa Halbach

With the applicable law in mind, we turn to the horrifying murder of Teresa Halbach and then the circumstances of Dassey's confession. More detailed accounts are available in the panel, district court, and state court opinions. See Dassey v. Dittmann , 860 F.3d 933 (7th Cir. 2017) ; Dassey v. Dittmann , 201 F.Supp.3d 963 (E.D. Wis. 2016) ; State v. Dassey , 346 Wis. 2d 278, 827 N.W.2d 928, 2013 WL 335923 (Wis. App. 2013) (per curiam) (unpublished disposition); see also State v. Avery , 337 Wis. 2d 351, 804 N.W.2d 216 (App. 2011) (affirming convictions of Dassey's uncle).

In 2005, Teresa Halbach was a young photographer with her own business based in Calumet County, Wisconsin. On October 31, her last appointment of the day was at Avery's Auto Salvage to photograph a van for an advertisement. Halbach never returned from that appointment. A few days later during a missing–person search, her car was found at the salvage yard. Her blood stained the car's interior. A further search turned up Halbach's charred remains in a burn pit on the property, along with shell casings on the floor of Steven Avery's garage.

B. Dassey's Early Police Interviews

Police investigators spoke with a number of Avery's relatives in early November, including an hour–long interview of his sixteen–year–old nephew Brendan Dassey, who lived close by. Dassey said he had seen Halbach taking pictures at the salvage yard on the afternoon of October 31, but he resisted the suggestion that she had entered Avery's home. At that time, he provided no other useful information.

Several months later, though, investigators received word that Dassey had been crying uncontrollably and had lost about forty pounds of weight. They proceeded to interview him a total of three times on February 27, 2006. In these voluntary witness interviews, it became clear that Dassey knew much more about Teresa's murder. (Dassey was not in custody on February 27th. He signed and initialed a Miranda waiver, and his mother consented, though she did not sit in.) In those interviews, Dassey admitted that on October 31st, he had gone over to Avery's trailer around 9:00 p.m. to help with a bonfire. He told the police that he had seen parts of a human body in the fire. He also said that Avery had threatened to hurt him if he spoke to the police. When the police asked about a pair of bleach–stained jeans they had learned about from another family member, Dassey admitted that he had helped Avery clean up a spill on the garage floor late that night. But Dassey claimed to have had nothing to do with Teresa's death.

C. The March 1st Interview and Confession

1. The Circumstances of the Interview

After those interviews, investigators thought Dassey had been a witness to at least the aftermath of a terrible crime and was struggling with the horror of what he had seen. On March 1st, the investigators (Mark Wiegert and Tom Fassbender) obtained his mother's permission for another interview. They took Dassey from his high school to a local sheriff's department, where he was questioned without the presence of a friendly adult. In the car the investigators gave Dassey standard Miranda warnings about his right to remain silent, his right to an attorney, and the possibility that statements he gave could be used against him. Dassey orally acknowledged the warnings, and he initialed and signed a written Miranda waiver form. He and the officers chatted during the ride. The three took a short detour to Dassey's home to retrieve his pair of bleach–stained jeans, which were kept as evidence. When they arrived at the sheriff's department, Dassey confirmed that he understood his rights and still wanted to talk to them.

The interview took place in a so–called "soft" interview room equipped for videotaping. Dassey sat on a couch facing two officers and a camera. Over the next three hours, Dassey was repeatedly offered food, drinks, restroom breaks, and opportunities to rest. At no point in the interview did the investigators threaten Dassey or his family. Nor did they attempt to intimidate him physically. They did not even raise their voices. Neither investigator tried to prevent Dassey from leaving the room, nor did they use any sort of force to compel him to answer questions. Dassey never refused to answer questions, never asked to have counsel or his mother present, and never tried to stop the interview.

2. The First Hour of Questioning

One officer began by telling Dassey how he could help the investigation, since "this information and that information" from previous accounts needed "just a little tightening up." Sensing that Dassey "may have held back for whatever reasons," the officer assured Dassey "that Mark and I both are in your corner, we're on your side." Acknowledging Dassey's potential concern that talking to the police meant he "might get arrested and stuff like that," the investigator urged Dassey to "tell the whole truth, don't leave anything out." Talking could be in Dassey's best interest even though it "might make you look a little bad or make you look like you were more involved than you wanna be," because admitting to unfortunate facts would leave "no doubt you're telling the truth." The first investigator closed by saying that "from what I'm seeing, even if I filled" in some holes in Dassey's story, "I'm thinkin' you're all right. OK, you don't have to worry about things ... [W]e know what Steven [Avery] did ... we just need to hear the whole story from you." The other investigator went next:

Honesty here Brendan is the thing that's gonna help you. OK, no matter what you did, we can work through that. OK. We can't make any promises but we'll stand behind you no matter what you did. OK. Because you're being the good guy here .... And by you talking with us, it's, it's helping you. OK? Because the honest person is the one who's gonna get a better deal out of everything.

Supp. App. 30. After Dassey nodded in agreement, the investigator continued:

You know. Honesty is the only thing that will set you free. Right? And we know, like Tom said we know, we reviewed those tapes .... We pretty much know everything that's why we're .... talking to you again today. We really need you to be honest this time with everything, OK.... [A]s long as you be honest with us, it's OK. If you lie about it that's gonna be problems. OK. Does that sound fair?

Id . Dassey again nodded and the questioning turned to the events of October 31st.

Over the course of the next three hours, with several breaks as the investigators conferred outside the room, Dassey told an even more disturbing and incriminating story about October 31st. In the first hour, Dassey admitted that he received a telephone call from Avery, went over to Avery's garage in the six o'clock hour, and found Teresa already dead in her car. Dassey then said he helped Avery lower Teresa's bound body onto a "creeper" (used to work underneath an automobile), which he and Avery used to take her body outside and throw her onto the already–burning bonfire.

At that point, less than an hour into the interview, Dassey's story pivoted dramatically. Dassey revised his story to say that he first noticed something amiss in the four o'clock hour. Dassey volunteered that when he was out getting the mail, he heard a woman screaming inside Avery's trailer. Supp. App. 50. Dassey knocked on Avery's door, ostensibly to deliver a piece of mail, and a sweaty Avery answered the door.

Dassey said he then saw Teresa alive, naked, and handcuffed to Avery's bed. Dassey said he went inside at Avery's invitation and had a soda while Avery told him that he had raped Teresa. Dassey said that, at Avery's urging, he then raped Teresa, having intercourse against her will as she was bound to the bed, and as she protested and begged him to stop. After the rape, Dassey reported, he then watched television with Avery for a while. Supp. App. 55–65.

In Dassey's telling, he next helped Avery subdue and kill Teresa and move her to the garage. Id. at 66–76. In response to questioning and prodding, Dassey told a confusing story about these critical events. Dassey said that Avery stabbed Teresa with a large knife, that her handcuffs were removed, and that she was tied up with rope. He also said that Avery cut off some of her hair with that large knife, that he (Dassey) cut her throat with the same knife, and that at some point Avery choked or punched her. All these events reportedly happened by 6:00 or 6:30 p.m.³

3 Given the damage to Teresa's body, few of these details could have been confirmed or contradicted by the surviving physical evidence. But what did survive elsewhere does not necessarily vindicate Dassey. For example, Dassey contends that no handcuff marks were found on the head–board of Steven Avery's bed, but a thin plastic film from a substance used in rope manufacturing was found on the headboard.

The details and sequence of these events changed repeatedly, however, as investigators pressed Dassey for more details. This portion of the interrogation provides the most support for Dassey's claim that his confession was both involuntary and unreliable.⁴ For example, because the recovered remnants of Teresa's skull contained trace amounts of lead, the investigators believed that Teresa had been shot in the head. They were eager for Dassey to describe what "else was done to her head" besides cutting and punching. In this exchange, Dassey did not provide the answer they were looking for. He offered what seemed like guesses. The investigators abandoned their vague admonitions to tell the truth. They lost patience and blurted out:

4 This portion of Dassey's confession also led to another search of Steven Avery's garage that uncovered perhaps the most powerful physical evidence of the investigation: a bullet fragment with Teresa Halbach's DNA on it.

Wiegert: All right, I'm just gonna come out and ask you. Who shot her in the head?

Brendan: He did.

Fassbender: Then why didn't you tell us that?

Brendan: Cuz I couldn't think of it.

Fassbender: Now you remember it? (Brendan nods "yes") Tell us about that then.

Supp. App. 76. Dassey continued to do so over the whole course of the March 1st interview, revising upwards the number of times Teresa was shot from twice to three times, and then up to ten times.⁵ Dassey also revised the location of the shooting, first outside the garage, then inside Teresa's car, then on the floor of the garage. After this shifting exchange about the shooting, the first hour of the March 1st interview concluded with Dassey explaining how he and Avery put Teresa's body on the fire, how they moved her car, and finally how they cleaned up the stain in Avery's garage before Dassey went home.

5 Throughout the interview, however, Dassey resisted all suggestions that he personally ever shot Teresa, and he described his discomfort with guns from a young age.

3. The Second Hour of Questioning

The investigators then took a break to confer. During the break, Dassey had the opportunity to rest and to use the restroom. Before starting up again, Dassey and Wiegert had this exchange, indicating that Dassey did not understand the gravity of what he had told the investigators:

Brendan: How long is this gonna take?

Wiegert: It shouldn't take a whole lot longer.

Brendan: Do you think I can get [back to school] before one twenty–nine?

Wiegert: Um, probably not.

Brendan: Oh.

Wiegert: What's at one twenty–nine?

Brendan: Well, I have a project due in sixth hour.

Supp. App. 102.

In the second hour of questioning, the investigators sought to confirm details from the first. They had only limited success. Dassey provided more confusing details about how Teresa was killed and the status of the bonfire. But in the main, Dassey largely confirmed his account from the first hour, especially about the details of his sexual assault of Teresa. His story regarding what he saw of Teresa in the fire—hands, feet, forehead, and part of a torso—also remained mostly consistent.

Signaling that the investigators did not overwhelm his will, Dassey resisted repeated suggestions by both investigators that he and Avery used the wires and cables hanging in the garage to torture Teresa. The investigators also tested Dassey's suggestibility. They told him falsely that Teresa had a tattoo on her stomach and asked if he had seen it. Here is the exchange:

Fassbender: ... did she have any scars, marks, tattoos, stuff like that, that you can remember?

Brendan: I don't remember any tattoos.

...

Fassbender: OK. (pause) We know that Teresa had a, a tattoo on her stomach, do you remember that?

Brendan: (shakes head "no") uh uh

Fassbender: Do you disagree with me when I say that?

Brendan: No but I don't know where it was.

Fassbender: OK.

Supp. App. 150–52. In this exchange, Dassey stuck to what he thought he knew, despite being challenged and prodded by the investigators.

4. The Final Hour of Questioning

The investigators took another break, during which Dassey ate a sandwich and briefly fell asleep. The investigators returned to talk about the consequences Dassey was facing:

Fassbender: What do you think's gonna happen? What do you think should happen right now?

Brendan: I don't know.

Fassbender: You know obviously that we're police officers, OK. (Brendan nods "yes") And ... because of what you told us, we're gonna have ta arrest you. Did you kinda figure that was coming? For ... what you did we ... can't let you go right now. The law will not let us. And so you're not gonna be able to go home tonight. All right?

Brendan: Does my mom know?

Fassbender: Your mom knows.

Supp. App. 157. After briefly discussing some logistics, the exchange continued:

Fassbender: Did you kinda ... after telling us what you told us you kinda figured this was coming? (Brendan nods "yes") Yeah? (Brendan nods "yes")

Brendan: Is it only for one day or?

Wiegert: We don't know that at this time, but let me tell ya something Brendan, you did the right thing. OK. (Brendan nods "yes") By being honest, you can at least sleep at night right now ....

Fassbender: Your cooperation and help with us is gonna work in your favor. I can't say what [it's] gonna do or where [you're] gonna end up but [it's] gonna work in your favor and we appreciate

your continued cooperation. (Brendan nods "yes") ....

Id. ⁶

6 If Dassey had continued to cooperate in the case against Steven Avery, that might well have worked in his favor. At the 2010 post–conviction hearings, Dassey's lawyer and the prosecutor both indicated that the State could have advocated for more lenient punishment for Dassey if he had testified against Steven Avery. See Dkt. 19–26 at 47–48, 99–100, 158–61.

Dassey's mother Barb Janda then came into the room to speak with Brendan about his arrest and confession. Dassey, now with his head buried in his hands, asked his mother what would happen if Avery gave a different version of events, such as "I never did nothin' " to Teresa Halbach "or somethin'." His mother followed up on this point, asking whether Dassey had done anything to Teresa:

Barb Janda: Did you? Huh?

Brendan: Not really.

Barb Janda: What do you mean not really?

Brendan: They got to my head.

Barb Janda: Huh?

Brendan: ... say anything.

Barb Janda: What do you mean by that? (pause) What do you mean by that Brendan?

Supp. App. 157. Dassey was taken into custody after this interview, which he now contends was involuntary and should not have been used at his trial.

At trial, Dassey testified and denied any knowledge of or involvement in Teresa Halbach's murder. He did not try to explain what he had meant by telling his mother "not really" and "they got to my head." According to his lawyer's version of events, Brendan came home from school at 3:45 p.m. on October 31st and played video games until having dinner with his brother and mother. After the others left, Dassey claimed, he fielded a phone call from his brother's boss and then shortly after that a call from Avery. At "about sevenish," Dassey claimed, he joined Avery for the bonfire, making four or five trips around the salvage yard picking up discarded items to throw on the flames. Around nine o'clock, Dassey helped Avery clean up a spill in his garage, and after a phone call from his mother, Dassey claimed, he returned home around 9:30 or 9:45 p.m. According to his trial testimony, none of the incriminating events related in his March 1st confession ever happened.⁷

7 At trial Dassey gave no explanation for his March 1st confession beyond controverted expert testimony that he was highly suggestible and a suggestion that he had confused his own experiences on October 31st with a book he had ostensibly read "three, four years" before called Kiss the Girls. No scenes in either the book or the movie it inspired are remotely similar to the crimes Dassey described on March 1st. See generally James Patterson, Kiss the Girls (1st ed. 1995); Kiss the Girls (Paramount Pictures 1997) (fictional coast–to–coast hunt for serial killers) Also, in nearly six months after March 1st, Dassey never mentioned the book or movie to his then–counsel.

D. The State Courts' Treatment of Dassey's Confession

Before trial, Dassey moved to suppress his confession as involuntary. After briefing and a hearing, the trial judge stated detailed findings of fact in an oral ruling. Supp. App. 168–77. The judge noted Dassey's age and observed that he had "an IQ level in the low average to borderline range." The judge noted that school records showed that Dassey was in regular–track classes but had some special education help. The judge also noted Dassey's lack of a criminal record, the noncustodial nature of the February 27th and March 1st interviews (as the parties had stipulated), and Dassey's Miranda waivers from both days. The judge found that Dassey knew he could stop answering questions and knew he could leave the room at any time on February 27th, and that he repeatedly indicated his continuing interest in speaking with the police on March 1st. The judge found that both Dassey and his mother consented to the interview on March 1st. The judge also quoted several of the investigators' admonitions to tell the truth, including "honesty here is the thing that's going to help you," and "honesty is the only thing that will set you free," upon which Dassey relies so heavily now.

Throughout the interview, the judge found, the investigators had used "a normal speaking tone with no raised voices, no hectoring, or threats of any kind." "Nothing on the video–tape visually depicts Brendan Dassey as being agitated, upset, frightened, or intimidated by the questions of either investigator," and he "displayed no difficulty in understanding the questions asked of him," the judge found. Though at times "prodded to be truthful," at "no time did he ask to stop the interview or request that his mother or a lawyer be present." The admonitions, the judge found, amounted to "nothing more than a reminder to Brendan Dassey that he had a moral duty to tell the truth." The judge also found that Dassey was not coerced by the "interviewers occasionally pretending to know more than they did" because that "did not interfere with [his] power to make rational choices." And finally, the judge found that "[n]o frank promises of leniency were made by the ... interviewers to Brendan Dassey," and that he was in fact flatly told "we can't make any promises."

On the basis of these findings of fact, "given Brendan Dassey's relevant personal characteristics" and applying "a totality of the circumstances test, which I'm using here," the judge found that Dassey's admissions in the March 1st interview were voluntary statements and denied Dassey's motion to suppress. Supp. App. 177.

The March 1st confession was the most incriminating evidence at trial. The jury found Dassey guilty on all charges: participating in rape and murder, and mutilation of a corpse. In August 2007, Dassey was sentenced to life in prison. Dassey filed detailed motions for a new trial in 2009, and the same trial court held five days of hearings on those motions in January 2010, probing Dassey's claims that his attorneys rendered ineffective assistance.

A three–judge panel of the Wisconsin Court of Appeals affirmed Dassey's convictions, finding that his confession was voluntary and any ineffective assistance was not prejudicial. State v. Dassey , 346 Wis. 2d 278, 827 N.W.2d 928, 2013 WL 335923 (Wis. App. 2013). The Court of Appeals used the trial court's findings of fact to summarize the circumstances of the March 1st confession and Dassey's claim that it was involuntary. The court then cited the legal standard for such claims—the totality of the circumstances—as applied by leading Wisconsin state cases. These state cases, particularly In re Jerrell C.J. , 283 Wis.2d 145, 699 N.W.2d 110 (2005), cited and discussed several of the leading precedents on voluntariness from the United States Supreme Court. The Court of Appeals cited Jerrell C.J. for the principle that a voluntariness "analysis involves a balancing of the defendant's personal characteristics against the police pressures used to induce the statements." Wisconsin law uses a clearly erroneous standard for appellate review of trial court findings of voluntariness.

After summarizing the trial court's findings, the Court of Appeals concluded:

¶7 The court's findings are not clearly erroneous. Based on those findings, we

also conclude that Dassey has not shown coercion. As long as investigators' statements merely encourage honesty and do not promise leniency, telling a defendant that cooperating would be to his or her benefit is not coercive conduct. State v. Berggren , 2009 WI App 82, ¶31, 320 Wis. 2d 209, 769 N.W.2d 110. Nor is professing to know facts they actually did not have. See State v. Triggs , 2003 WI App 91, ¶¶15, 17, 264 Wis. 2d 861, 663 N.W.2d 396 (the use of deceptive tactic like exaggerating strength of evidence against suspect does not necessarily make confession involuntary but instead is factor to consider in totality of circumstances). The truth of the confession remained for the jury to determine.

The court went on to reject Dassey's claims that his pre–trial and trial counsel provided ineffective assistance. The Wisconsin Supreme Court denied Dassey's petition for review. Dassey did not file a petition for certiorari in the United States Supreme Court.

E. Federal Habeas Corpus Review

Dassey filed a federal habeas corpus petition in the Eastern District of Wisconsin in 2014. In a detailed opinion, the district court granted habeas relief, finding that false promises of leniency were indeed made to Dassey and that his March 1st confession was not voluntary. Dassey , 201 F.Supp.3d 963. A divided panel of our court affirmed. Dassey , 860 F.3d 933. We granted the State's petition to rehear the case en banc and now reverse with instructions to dismiss Dassey's habeas petition.

III. Applying the AEDPA Standard

A. Voluntariness Under § 2254(d)(1)

The state court decision that Dassey confessed voluntarily was not an unreasonable application of Supreme Court precedent. The state appellate court drew on fairly detailed findings of fact, which were not clearly erroneous, and provided a terse but sufficient explanation for why the trial court's decision was a reasonable application of the broad totality–of–the–circumstances test.

1. Factors Pointing in Opposite Directions

A number of relevant factors, we recognize, tend to support Dassey's claims about the March 1st confession. He was young. He was alone with the police. He was somewhat limited intellectually. The officers' questioning included general assurances of leniency if he told the truth, and Dassey may have believed they promised more than they did. At times it appeared as though Dassey simply did not grasp the gravity of his confession—after confessing to rape and murder, he asked the officers if he would be back at school that afternoon in time to turn in a project. Portions of the questioning also included leading and suggestive questions, and throughout the interrogation Dassey faced follow–up inquiries when the investigators were not satisfied with what he had told them, leading him at times to seem to guess. In addition, the confusion and contradictions in Dassey's account of the crimes of October 31st lend support to the view that his confession was the product of suggestions and/or a desire to tell the police what they wanted to hear.

At the same time, many other factors support the finding that Dassey's confession was indeed voluntary. Start with the circumstances of the interrogation. As stipulated by both sides, Dassey was not in custody when he admitted participating in the crimes of October 31st. He went with the officers voluntarily and with his mother's knowledge and consent. He was given Miranda warnings and understood them sufficiently. The interrogation was conducted during school hours and in a comfortable setting. Dassey showed no signs of physical distress. He had access to food, drinks, and restroom breaks. The interrogation was not particularly lengthy, especially with the breaks that were taken every hour.

Dassey was not subject to physical coercion or any sort of threats at all. Given the history of coercive interrogation techniques from which modern constitutional standards for confessions emerged, this is important. The investigators stayed calm and never even raised their voices. As the Wisconsin courts found, there is no sign that Dassey was intimidated.

Turning to the techniques used in the interrogation, the investigators told Dassey many times that they already knew what had happened when in fact they did not. Such deception is a common interview technique. To our knowledge, it has not led courts (and certainly not the Supreme Court) to find that a subject's incriminating answers were involuntary. See Frazier v. Cupp , 394 U.S. 731, 739, 89 S.Ct. 1420, 22 L.Ed.2d 684 (1969) (fabricating a co–conspirator's confession is relevant, but "insufficient in our view to make this otherwise voluntary confession inadmissible"). Also, most of the incriminating details in Dassey's confession were not suggested by the questioners. He volunteered them in response to open–ended questions.

When Dassey's story did not make sense, seemed incomplete, or seemed to conflict with other evidence, the questioners pressed Dassey with further questions. Those techniques are not coercive. Dassey responded to such questioning by modifying his story on some points, but he stuck to his story on others. Those passages support the view that he was not being pushed to provide a false story against his will. For example, Dassey resisted repeated suggestions that he had participated in shooting Teresa. He denied repeated suggestions that he and Avery had used wires and cables in the garage to restrain or harm her. In one telling instance, the questioners tested Dassey by falsely telling him that Teresa had a tattoo on her stomach and asking him if he had seen it. He told them no. When the questioners pushed harder, he was not willing to say he knew they were wrong, but he stuck to his recollection that he had not seen a tattoo.

Under AEDPA, the essential point here is that the totality–of–the–circumstances test gives courts considerable room for judgment in cases like this one, where the factors point in both directions. Given the many relevant facts and the substantial weight of factors supporting a finding that Dassey's confession was voluntary, the state court's decision was not an unreasonable application of Supreme Court precedent. This view is similar to Yarborough v. Alvarado , 541 U.S. 652, 664–65, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004), where the Supreme Court applied AEDPA to a state court finding that a seventeen–year–old suspect had not been in custody when he confessed to murder. The custody question was governed by a similarly general totality–of–the–circumstances standard. The Supreme Court summarized the array of factors pointing in opposite directions, in custody or not in custody. Emphasizing that the more general the rule, the more leeway courts have in reaching outcomes in case–by–case determinations, the Supreme Court found that the state court finding was not an unreasonable application of binding precedent: "These differing indications lead us to hold that the state court's application of our custody standard was reasonable. The Court of Appeals was nowhere close to the mark when it concluded otherwise." Id . at 665, 124 S.Ct. 2140.2. The Terse State Court Opinion

Dassey criticizes the Wisconsin appellate court's decision for having been too terse, addressing the confession in just two pivotal paragraphs. The relative brevity of that part of the opinion is not a reason for granting habeas relief. Given the volume of words that federal judges have devoted to this case, one might assume that the totality–of–the–circumstances test requires courts to detail at length the weight they have assigned to all factors and how the presence of one factor affects the weight or relevance of other factors.

That assumption would be incorrect. The Supreme Court itself has issued terse final determinations on voluntariness after a recitation of relevant facts. See Greenwald v. Wisconsin , 390 U.S. 519, 519–21, 88 S.Ct. 1152, 20 L.Ed.2d 77 (1968) (per curiam); Davis v. North Carolina , 384 U.S. 737, 752, 86 S.Ct. 1761, 16 L.Ed.2d 895 (1966). It has ruled on voluntariness by simply adopting the reasoning of other courts. Boulden v. Holman , 394 U.S. 478, 480–81, 89 S.Ct. 1138, 22 L.Ed.2d 433 (1969). Section 2254(d)(1) does not authorize federal courts "to impose mandatory opinion–writing standards on state courts." Johnson v. Williams , 568 U.S. 289, 300, 133 S.Ct. 1088, 185 L.Ed.2d 105 (2013). State court decisions receive significant deference even if they provide no reasons at all. Harrington v. Richter , 562 U.S. 86, 98–99, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011) ; Whatley v. Zatecky , 833 F.3d 762, 774 (7th Cir. 2016). In this case, the state appellate court endorsed detailed findings by the trial court that provide substantial support for the finding that Dassey's confession was voluntary in the eyes of the law.

3. Juveniles and Special Care

The requirement that courts take "special care" in analyzing juvenile confessions does not call for habeas relief here. The state appellate court met the requirements for analyzing juvenile confessions by considering Dassey's age, his intellectual capacity, and the voluntary absence of his mother during the interrogation. The state court noted that the officers read Dassey his Miranda rights and that Dassey later remembered his rights and agreed to talk anyway. The court assessed coercion in relation to Dassey's vulnerabilities, including his "age, intellectual limitations and high suggestibility." The court did not limit its inquiry to only whether the most abusive interrogation techniques were used. The court examined the tones and volumes of the investigators' voices, finding that the officers "used normal speaking tones, with no hectoring, threats or promises of leniency," though they did prod Dassey to be honest and sought to establish a rapport with him. The court even considered Dassey's physical comfort by noting he sat on a sofa and was offered food, drink, and restroom breaks.

4. Precedent

Dassey simply has not pointed to Supreme Court precedent that mandates relief under these circumstances. Even in cases where deferential review under AEDPA does not apply, the Supreme Court has not found a confession involuntary in circumstances like Dassey's March 1st confession.

Consider Boulden v. Holman , 394 U.S. 478, 480–81, 89 S.Ct. 1138, 22 L.Ed.2d 433 (1969). The defendant there was eighteen years old, had an I.Q. of 83, suffered from an anxiety complex, and was "susceptible to coercion." Boulden v. Holman , 385 F.2d 102, 104, 105 (5th Cir. 1967). He was interrogated for less than three hours after being told he had the "right not to make a statement, and that any statement made might be used against him." Id. at 104. He was "treated courteously and allowed to eat, smoke and to use [the] toilet facilities." Id. at 105. Though two years older than Dassey, Boulden was apparently still dependent on his parents. Id. Other facts of his interrogation were more troubling than those in this case. Boulden was interrogated from 10 p.m. until after midnight after several hours in custody. Id. at 104. Police had denied Boulden's father access to him, and after Boulden asked "whether he was supposed to have a lawyer," the police said "he would not get one until he talked." Id. The Supreme Court "determined that although the issue is a relatively close one, the conclusion ... was justified" that Boulden had confessed voluntarily. 394 U.S. at 480–81, 89 S.Ct. 1138.

In Fare v. Michael C. , 442 U.S. 707, 727, 99 S.Ct. 2560, 61 L.Ed.2d 197 (1979), the Court again ruled a juvenile confession was voluntary. Like Dassey, Michael C. was sixteen years old. He claimed that the police made promises and threats during the interrogation "in the hope of obtaining leniency for his cooperative attitude." Id. Michael C. indicated that his pleas to stop the interrogation were ignored. He also claimed he feared police coercion and pointed out that he "wept during the interrogation." Id. Despite these assertions, the Court determined that Michael C.'s claims of coercion were "without merit." Id.

Unlike Dassey, Michael C. apparently did not have a low average to borderline I.Q., and Michael C. did have significant prior experience with the criminal justice system. See id. at 726, 99 S.Ct. 2560. Though the presence of those factors may have provided room for Dassey to argue on direct appeal that Michael C. should be distinguished, they do not show that the Wisconsin courts' decision here was unreasonable within the meaning of § 2254(d)(1). As in Michael C. , the police here indicated "that a cooperative attitude would be to [the suspect's] benefit, but their remarks in this regard were far from threatening or coercive." Id. at 727, 99 S.Ct. 2560.

In reviewing these cases, we remember the Supreme Court's admonition that determining whether a confession is voluntary "requires more than a mere color–matching of cases." Reck v. Pate , 367 U.S. 433, 442, 81 S.Ct. 1541, 6 L.Ed.2d 948 (1961). But like the Court, we find these comparisons helpful after "careful evaluation of all the circumstances of the interrogation." Mincey , 437 U.S. at 401, 98 S.Ct. 2408 ; see Reck , 367 U.S. at 442, 81 S.Ct. 1541 (finding comparison to analogous cases "not inappropriate" when determining voluntariness). AEDPA "would be undermined if habeas courts introduced rules not clearly established under the guise of extensions to existing law." Alvarado , 541 U.S. at 666, 124 S.Ct. 2140. To be sure, this line between application and extension of existing law blurs "when new factual permutations arise." Id. The cases show, however, that the Supreme Court has considered and rejected claims similar to Dassey's, and Supreme Court cases do not require relief here. The Wisconsin courts did not apply the law unreasonably in finding that Dassey's confession was voluntary.

B. Factual Findings Under § 2254(d)(2)

Dassey also argues that he is entitled to relief under § 2254(d)(2) on the ground that the state courts made an unreasonable finding of fact: that the questioners made no false promises of leniency. Affirming the trial court, which found "no frank promises of leniency were made," the Wisconsin Court of Appeals determined that the investigators' statements "merely encourage[d] honesty and [did] not promise leniency." Dassey's argument that this finding was unreasonable focuses on two things: his intellectual limitations and the spots in the March 1st interrogation where he claims the investigators implied that he would not even be arrested if he told the truth. We reject this argument.

Because the Wisconsin appellate court accepted the trial court's findings of fact, we review the trial court's factual determinations directly. See Rice v. Collins , 546 U.S. 333, 339, 126 S.Ct. 969, 163 L.Ed.2d 824 (2006) (indicating that AEDPA review and deference in such a situation should extend to state trial court findings). The trial court here highlighted the key points for both sides, including the warning that the questioners could not make promises (which supports the State here) and the problematic assurance that honesty was the only thing that would set Dassey free (which helps Dassey's claim, especially in light of his limited intellect). Whether we treat the state court's decision on this point as a finding of fact or a conclusion of law, we find nothing unreasonable about it.

As noted above, the Supreme Court has not treated general assurances of leniency in exchange for cooperation or confession as coercive. To the extent precedents from other courts might be helpful in understanding a state court's factual findings, the cases signal that such general assurances are not legally relevant facts for determining whether a suspect's will was overborne and a confession was involuntary. See, e.g., United States v. Villalpando , 588 F.3d 1124, 1129 (7th Cir. 2009) ; see also United States v. Binford , 818 F.3d 261, 271–72 (6th Cir. 2016) ; United States v. Corbett , 750 F.3d 245, 253 (2d Cir. 2014) ; United States v. Jackson , 608 F.3d 100, 103 (1st Cir. 2010) ; United States v. Kontny , 238 F.3d 815, 818 (7th Cir. 2001) ; United States v. Rutledge , 900 F.2d 1127, 1130 (7th Cir. 1990) ("The policeman is not a fiduciary of the suspect. The police are allowed to play on a suspect's ignorance, his anxieties, his fears, and his uncertainties; they just are not allowed to magnify those fears, uncertainties, and so forth to the point where rational decision becomes impossible."). The state appellate court should be understood as having said that the investigators made no legally relevant false promises to Dassey.

The district court, the panel majority, and our dissenting colleagues have viewed the interrogation differently, finding psychological coercion through a form of operant conditioning, where different investigative tactics combined to convince Dassey that the police had agreed to end the interrogation and to grant him leniency in exchange for confessing. Dassey , 860 F.3d at 963, 974. As the panel explained, in its view of the interrogation, the investigators offered Dassey multiple assurances and "sounded the theme of 'truth leads to freedom' " culminating in the "direct promise, 'honesty is the only thing that will set you free.' " Id .

The state courts did not view these tactics the same way. Their view was not unreasonable. The state courts saw and read, as we have, exactly what the questioners told and asked Dassey in the interview and how he responded. AEDPA leaves room for reasonable disagreement between state and federal courts. Disagreement on a particular judgment call does not show that the state court found the facts unreasonably. Collins , 546 U.S. at 341–42, 126 S.Ct. 969.

In denying Dassey's suppression motion, the state trial court weighed the same statements that concerned the district court and the panel. The judge quoted four separate instances where investigators prodded Dassey by stating that honesty would help him, and the judge noted that these were "but a few example[s] of admonitions to be honest." The state court also recounted four quotations and other "similar statements" where investigators assured Dassey that they were behind him and in his corner. It viewed these statements as an "attempt to achieve a rapport" rather than "frank promises of leniency." These findings are reasonable and consistent with the evidence and the relevant law. Habeas review does not permit us to "use a set of debatable inferences to set aside the conclusion reached by the state court." Collins , 546 U.S. at 342, 126 S.Ct. 969.

C. Police Best Practices and the Law

The concerns expressed by our dissenting colleagues and the district court about the potential coercive effects of the police tactics here are understandable. Critics of Dassey's interrogation see evidence of fabrication through the confession's inconsistencies and lack of solid corroborating physical evidence. Some of the confession's inconsistencies are startling, particularly Dassey's shifting answers on the location of the shooting (outside the garage, on the garage floor, and in the car inside the garage), and his failure to recall consistently the order of attacks in the bedroom (stabbing, hair–cutting, and throat–slicing). Also, during the dialogue about Teresa's shooting, the investigators prodded Dassey and injected some critical facts into the discussion that corroborated evidence they already knew.

The state courts did not address these factual inconsistencies or the alleged lack of corroborating evidence, though it is not clear how they should have approached the question, if at all. United States Supreme Court precedent on this point is not unequivocal. In Blackburn v. Alabama , 361 U.S. 199, 80 S.Ct. 274, 4 L.Ed.2d 242 (1960), the Court considered the "unreliability of the confession" in determining that a mentally ill defendant's confession was not voluntary. Id. at 207, 80 S.Ct. 274. The very next year the Court indicated that "the reliability of a confession has nothing to do with its voluntariness" because extrinsic evidence that a confession is true can confound the inquiry into "whether a defendant's will has been overborne." Jackson v. Denno , 378 U.S. 368, 384–85, 84 S.Ct. 1774, 12 L.Ed.2d 908 (1964), citing Rogers v. Richmond , 365 U.S. 534, 545, 81 S.Ct. 735, 5 L.Ed.2d 760 (1961). The Court later seemed to signal another direction, writing in Colorado v. Connelly that whether a confession is reliable, as distinct from voluntary, "is a matter to be governed by the evidentiary laws of the forum ... and not by the Due Process Clause of the Fourteenth Amendment." 479 U.S. 157, 167, 107 S.Ct. 515, 93 L.Ed.2d 473 (1986).

Analysis of a confession's reliability as part of the totality of the circumstances may survive the instruction in Connelly , but it is not unreasonable to interpret Connelly as foreclosing—or at least not requiring—this line of inquiry before trial. We cannot fault the Wisconsin courts for failing to measure the inconsistency of Dassey's confession in this context. In addition, the contradictions as to some details do not necessarily undermine the reliability of the core incriminating admissions. See Dassey , 860 F.3d at 993–94 (Hamilton, J., dissenting).

The concerns about reliability echo the opinions of scholars who believe that certain interrogation tactics tend to produce false confessions. Some police departments and experts have acknowledged this criticism and have changed their interrogation practices in response. We must note, though, that some of the interrogation tactics used in this case—like the repeated challenges to explain details that seem implausible—reflect practices advocated by such reformers. See, e.g., Saul Kassin et al., Interviewing Suspects: Practice, Science, and Future Directions , 15 Legal & Criminological Psychology 39, 47 (2010) (describing as "non–coercive" the practice of investigators "challeng[ing] suspects' accounts, often by pointing out contradictions and inconsistencies"); Kassin, The Psychology of Confessions , 2008 Annual Rev. of Law & Soc. Sciences 193, 208 (favoring interrogation technique where investigators "address discrepancies that may appear in the suspect's narrative account" to determine if the suspect is fabricating).

These debates over interrogation techniques have not resulted in controlling Supreme Court precedent condemning the techniques used with Dassey. Absent a clear declaration from the Court, we may not create new constitutional restraints on habeas review. See Kernan v. Cuero , 583 U.S. ––––, 138 S.Ct. 4, 9, 199 L.Ed.2d 236 (2017) (circuit precedent does not satisfy § 2254(d)(1), "[n]or, of course, do state–court decisions, treatises, or law review articles").⁸

8 Judge Rovner's dissent cites studies of exonerated defendants showing that false confessions are more common among juveniles and mentally ill or intellectually deficient suspects. See post at 332–34; Dassey, 860 F.3d at 952–53 (panel majority). False confessions are a real phenomenon, and even one is very troubling. Yet we should not conclude from these studies of exonerated defendants that there is an epidemic of false confessions, as might be inferred by looking at studies of only demonstrably wrong convictions. The more relevant fraction uses as the denominator the number of all confessions. That number is not easy to estimate, but we can estimate a conservative lower boundary for the number of confessions to violent felonies. Bureau of Justice Statistics reports on Felony Defendants in Large Urban Counties tally violent felony convictions by guilty plea (i.e., by confessions of guilt) in just the nation's 75 largest counties. (The most recent report is Brian A. Reaves, U.S. Dep't of Justice, Bureau of Justice Statistics, Felony Defendants in Large Urban Counties, 2009—Statistical Tables (2013), https://www.bjs.gov/content/pub/pdf/fdluc09.pdf.) The dissent's statistics report 227 demonstrably false confessions from 1989 to 2016. Post at 332. From the BJS reports, we can estimate that over that period, in just those 75 largest counties, there were more than 1.5 million guilty pleas to violent felonies. The relevant fraction may thus be estimated conservatively as 227/1,500,000. For every one demonstrably false confession over those years, there were more than 6,500 guilty pleas to violent felonies in just those counties.

D. Ineffective Assistance of Counsel

Finally, Dassey has also pursued his separate claim that his original lawyer provided ineffective assistance of counsel on the theory that the lawyer was operating under an actual conflict of interest prohibited by Cuyler v. Sullivan , 446 U.S. 335, 100 S.Ct. 1708, 64 L.Ed.2d 333 (1980). On this point the state and federal courts have agreed. The Wisconsin appellate court rejected this claim. The district court also considered this claim carefully and rejected it, citing the limits placed on Sullivan claims by Mickens v. Taylor , 535 U.S. 162, 175, 176, 122 S.Ct. 1237, 152 L.Ed.2d 291 (2002). Dassey , 201 F.Supp.3d at 989.⁹ We agree for substantially the reasons set forth by the district court. Id. at 987–93. In this case there was no actual conflict of interest and no multiple or concurrent representations that could have resulted in an actual conflict of interest.

9 The panel majority did not reach the issue. 860 F.3d at 983.

Conclusion

Given the state courts' reasonable findings of fact and the absence of clearly established Supreme Court precedent that compels relief for Dassey, the district court's grant of habeas relief is REVERSED. The case is REMANDED to the district court with instructions to dismiss the petition. WOOD , Chief Judge , and ROVNER and WILLIAMS , Circuit Judges , dissenting.

Psychological coercion, questions to which the police furnished the answers, and ghoulish games of "20 Questions," in which Brendan Dassey guessed over and over again before he landed on the "correct" story (i.e. , the one the police wanted), led to the "confession" that furnished the only serious evidence supporting his murder conviction in the Wisconsin courts. Turning a blind eye to these glaring faults, the en banc majority has decided to deny Dassey's petition for a writ of habeas corpus. They justify this travesty of justice as something compelled by the Antiterrorism and Effective Death Penalty Act (AEDPA). If the writ, as limited by AEDPA, were nothing more than a dead letter, perhaps they would be correct. But it is not. Instead, as the Supreme Court wrote in Harrington v. Richter , 562 U.S. 86, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011), "[t]he writ of habeas corpus stands as a safeguard against imprisonment of those held in violation of the law." Id. at 91, 131 S.Ct. 770. It is, the Court went on to say, "a guard against extreme malfunctions in the state criminal justice systems." Id. at 102, 131 S.Ct. 770 (citation and internal quotation marks omitted).

As the district court and the panel majority recognized, we have before us just such an extreme malfunction. Dassey at the relevant time was 16 years old and had an IQ in the low 80s. His confession was coerced, and thus it should not have been admitted into evidence. And even if we were to overlook the coercion, the confession is so riddled with input from the police that its use violates due process. Dassey will spend the rest of his life in prison because of the injustice this court has decided to leave unredressed. I respectfully dissent.

I

As the Wisconsin Court of Appeals correctly noted, the question whether a confession is voluntary (i.e. , not coerced) is assessed in light of the totality of the circumstances. The age and sophistication of the person being questioned are critical factors. When the suspect is a minor, courts must review the confession and record with "special care." J.D.B. v. North Carolina , 564 U.S. 261, 280–81, 131 S.Ct. 2394, 180 L.Ed.2d 310 (2011) ; In re Gault , 387 U.S. 1, 45, 87 S.Ct. 1428, 18 L.Ed.2d 527 (1967) ; Gallegos v. Colorado , 370 U.S. 49, 53–55, 82 S.Ct. 1209, 8 L.Ed.2d 325 (1962) ; Haley v. Ohio , 332 U.S. 596, 599, 68 S.Ct. 302, 92 L.Ed. 224 (1948). Courts also must take the suspect's intellectual capacity into account. Culombe v. Connecticut , 367 U.S. 568, 620, 625, 81 S.Ct. 1860, 6 L.Ed.2d 1037 (1961) (opinion of Frankfurter, J., joined by Stewart, J.); 639 (Douglas, J., joined by Black, J., concurring); 641–42 (Brennan, J., joined by Warren, C.J., and Black, J., concurring). Dassey, as the majority concedes, was a mentally limited 16–year–old. It was thus incumbent on the state courts to evaluate his "confession" in light of those traits.

The Wisconsin courts failed to take this essential step. When asked at oral argument where one might find evidence that the state appellate court took the required special care, counsel for the state came up dry. All counsel could do was to point out a brief mention in the state court's opinion of Dassey's age and mental capabilities. But so what? The Supreme Court has never said or implied that the totality of the circumstances are beside the point as long as the state court simply jots down a fact without a hint about if or how that fact influenced the outcome. There is nothing "special" (or even meaningful) about a naked word on a page. The reader has no idea whether the state court mentioned the word meaning to indicate that it found the factor irrelevant (which would have been inconsistent with the clear Supreme Court precedent listed above), or exculpatory, or damning. Notably, even though the Wisconsin Court of Appeals gave a nod to the totality test, it made no mention of the special–care standard for juvenile confessions.

To be sure, Harrington v. Richter , 562 U.S. 86, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011), holds generally that federal courts may not draw any dispositive conclusions from a state court's silence. But by the same token, the state court's silence cannot be leveraged into any assurance that the court went the extra mile required by the U.S. Supreme Court and gave Dassey's age and limited mental ability particularized care. The majority's finding to the contrary has no support in the record. Worse, the majority writes off in a footnote Dassey's extreme suggestibility by casting doubt on the applicability of a formal test (Gudjonsson). Ante at 305 n.2. As the painstaking review of the record reflected in Judge Rovner's panel opinion reveals, even a lay–person could see readily that Dassey yielded to any suggestion the person in authority made. 860 F.3d 933 (7th Cir. 2017) ( Dassey I ). More generally, no court is entitled to pick and choose which evidence to consider when evaluating the totality of the circumstances. Clearly established U.S. Supreme Court decisions compelled the Wisconsin court to pay special attention to Dassey's age and intellectual abilities, including his high level of suggestibility. Its failure to do so is one reason why it erroneously concluded that Dassey's "confession" was not coerced.

If the Wisconsin Court of Appeals had done what it should have, it could not reasonably have concluded that Dassey's confession was either voluntary or reliable (both of which are required for the use of a confession to be consistent with due process). Nevertheless, first the state and now the en banc majority have culled a sentence here and there and have attempted to craft a coherent confession from them. The video recording of the police interrogation of Dassey, however, tells another story—one that is diametrically opposed to the state's tidy and selective summary. Among the many red flags are the following:

• Dassey's answers to questions frequently changed at the detectives' prodding.

• The officers laid a trail of crumbs (indeed, large sign–posts) to the confession they sought.

• Whenever Dassey went off–course, the investigators would shepherd him back in the desired direction—at times with the use of fatherly assurances and gestures, and frequently by questioning his honesty.

• On both February 27 and March 1 the detectives misleadingly conveyed to Dassey, whose ability to think abstractly was minimal, that his "honesty" was the "only thing that will set [him] free."

• Through subsequent questioning it became clear that "honesty" meant "what the investigators wanted to hear."

Dassey's age and mental limitations made him particularly susceptible to this psychologically manipulative interrogation. Many of the officers' tactics appear to be drawn from the "Reid Technique," which was for some time the most widely used interrogation protocol in the country. Miriam S. Gohara, A Lie for a Lie: False Confessions and the Case for Reconsidering the Legality of Deceptive Interrogation Techniques , 33 FORDHAM URB . L.J. 791, 808 (2006). The technique heavily relies on false evidence ploys and other forms of deceit. Id. at 809. It follows a nine–step approach:

[A]n interrogator confronts the suspect with assertions of guilt (Step 1), then develops "themes" that psychologically justify or excuse the crime (Step 2), interrupts all efforts at denial (Step 3), overcomes the suspect's factual, moral, and emotional objections (Step 4), ensures that the passive suspect does not withdraw (Step 5), shows sympathy and understanding and urges the suspect to cooperate (Step 6), offers a face–saving alternative construal of the alleged guilty act (Step 7), gets the suspect to recount the details of his or her crime (Step 8), and converts the latter statement into a full written confession (Step 9).

Saul M. Kassin, On the Psychology of Confessions: Does Innocence Put Innocents at Risk? , 60 AM. PSYCHOLOGIST 215, 220 (2005); see Edwin D. Driver, Confessions and the Social Psychology of Coercion , 82 HARV. L. REV. 42, 51–55 (1968) (explaining the social psychological impact of the Reid tactics). Investigators are encouraged to start by accusing the suspect while emphasizing the importance of telling the truth. FRED E. INBAU ET AL., CRIMINAL INTERROGATION AND CONFESSIONS 213 (4th ed. 2001). They learn ways to build false empathy with suspects, such as shifting the moral blame for the offense to another person or expressing understanding for the suspect's actions. Id. at 213, 241–42. Investigators are encouraged to sit physically near the suspect, maintain "soft and warm" eye contact, and speak sincerely. Id. at 214, 349. When a suspect makes an admission implying guilt, investigators are directed to make statements of reinforcement. Id. at 366. The technique builds in confirmation bias; the instructions assure investigators that while an innocent suspect will stay resolute in her denials, a guilty person will submit to the "theme" the investigator presents. Id. at 213; see Christian A. Meissner & Melissa B. Russano, The Psychology of Interrogations and False Confessions: Research and Recommendations , 1 CANADIAN J. POLICE & SECURITY SERVS . 53, 56–57 (2003).

Courts have long expressed concern about approaches such as the Reid Technique that rely on psychological coercion. Just four years after the first edition of the manual was published, INBAU ET AL ., supra , at ix, the Supreme Court in Miranda v. Arizona , 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966), "repeatedly cited and implicitly criticized" the Reid approach. Gohara, supra , at 808 n.93; Miranda , 384 U.S. at 457, 86 S.Ct. 1602 ("To be sure, this is not physical intimidation, but it is equally destructive of human dignity."). Miranda commented that the Court for decades had "recognized that coercion can be mental as well as physical, and that the blood of the accused is not the only hallmark of an unconstitutional inquisition." Id. at 448, 86 S.Ct. 1602 (quoting Blackburn v. Alabama , 361 U.S. 199, 206, 80 S.Ct. 274, 4 L.Ed.2d 242 (1960) ). Nothing in that respect has changed: the Court continues regularly to hold that psychological coercion can render a confession involuntary. Arizona v. Fulminante , 499 U.S. 279, 287–88, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991) ; Miller v. Fenton , 474 U.S. 104, 109, 106 S.Ct. 445, 88 L.Ed.2d 405 (1985) ; Schneckloth v. Bustamonte , 412 U.S. 218, 226, 93 S.Ct. 2041, 36 L.Ed.2d 854 (1973).

Following the Supreme Court's guidance, we too have repeatedly recognized that "psychological coercion alone can result in an involuntary confession ...." United States v. Lehman , 468 F.2d 93, 100 (7th Cir. 1972) (conceding that "subtle psychological ploys" can render a confession involuntary); Etherly v. Davis , 619 F.3d 654, 663 (7th Cir. 2010) (considering possible psychological coercion as part of the totality test, while noting the need to distinguish between coercion, on the one hand, and encouragement to tell the truth, on the other); United States v. Villalpando , 588 F.3d 1124, 1128 (7th Cir. 2009) ("[A] false promise of leniency may render a statement involuntary ...."); United States v. Dillon , 150 F.3d 754, 757 (7th Cir. 1998) ("A confession is voluntary if, in light of the totality of the circumstances, the confession is the product of a rational intellect and free will and not the result of physical abuse, psychological intimidation, or deceptive interrogation tactics that have overcome the defendant's free will."); Burns v. Reed , 44 F.3d 524, 527 (7th Cir. 1995) (describing the "body of due process case law, which generally proscribes the physical or psychological coercion of confessions" as "well–established, albeit heavily fact–dependent"). Outside the courtroom, our nation has long acknowledged through its international commitments that mental mistreatment can be just as bad as its physical counterpart. Convention against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment art. 1, Dec. 10, 1984, 1465 U.N.T.S. 85 (defining torture to encompass physical and mental pain and suffering).

The majority opinion downplays this reality by refusing to acknowledge anything more than mental exhaustion and false promises. But far worse than that was going on. Dassey's investigators refused to leave him alone until he gave them an "honest" answer—where "honest" meant the answer that the officers wanted to hear. One aspect, though by no means the only one, of the coercion was the false promise that "honesty" would "set him free." But there was so much more. A brief review of what went on shows that these tactics fell decisively on the "coercion" side of the line.

The majority finds some significance in the notion that the detectives' tactics were not per se coercive, but that is a red herring. These cases cannot be assessed based on one sentence, or one restroom break, or the comfort (or lack thereof) of one room. The Supreme Court has instructed that the voluntariness inquiry requires a full consideration of the compounding influence of the police techniques "as applied to this suspect." Miller , 474 U.S. at 116, 106 S.Ct. 445. Many of the factors the majority cites as evidence leaning in favor of a finding of voluntariness—the soft interview room, offers of food and drink, normal speaking tones—viewed in the context of the types of questions and answers the investigators were demanding and Dassey's conceded intellectual disabilities, were coercive. Psychological literature makes this clear. See Saul M. Kassin, The Psychology of Confession Evidence , 52 AM. PSYCHOLOGIST 221, 223–24 (1997) (criticizing the Reid Technique's maximization methods, or scare tactics, such as the false evidence ploy, in addition to its minimization methods, which "impl[y] an offer of leniency," where police lull a suspect into a "false sense of security" by expressing sympathy, blaming an accomplice, and underplaying the gravity of the situation); see also Meissner & Russano, supra , at 57–60 (discussing the "coercive" nature of the Reid interrogation techniques and particular concerns for minors and suspects with low intelligence).

The state and majority brush aside even the possibility of psychological coercion as applied to Dassey. They claim that Dassey's March 1 confession revealed certain "critical" details that were corroborated by independent evidence, some of which law enforcement never publicly disclosed. I have several responses to that argument. First, it rests on the false idea that if a confession is "accurate," that indicates that it was not coerced. See Conner v. McBride , 375 F.3d 643, 652–53 (7th Cir. 2004) (considering, under the totality test, the reliability of a confession to support a conclusion that the confession was voluntary). But coercion and reliability are two different things. A confession can be coerced yet reliable, or it can be voluntary but unreliable. Yet even if it were true that Dassey's confession revealed "critical" details, the confession would not be admissible in evidence if the totality of the circumstances demonstrated that it was not voluntary.

Just as importantly, a closer examination of the supposedly reliable facts on which the majority relies shows that they are no such thing. Without reliable facts, there is no way to draw the Conner inference (i.e. , to base a finding of voluntariness on the reliability of the facts), questionable though that link might be. This justifies a look at the reliability of Dassey's confession, even if for present purposes lack of reliability is not a stand–alone theory. A look at how some of these "key" facts emerged instills no faith in either their reliability or their knowing and voluntary quality. For ease of reference, I have summarized in the following chart how the investigators extracted the "critical" details they were looking for from Dassey. It shows that there was nothing to ensure that Dassey was offering his own independent recollection. Instead, the officers used a combination of leading questions, coaching, and refusal to accept one of Dassey's guesses as the "final" answer until it matched what they wanted to hear.

"Critical" Why It Is Not How It Was Coerced Fact Critical Halbach Dassey was fed "Tell us, and what else was shot in this fact through a did you do? Come on. the head. leading question Something with the head. after unsuccessful Brendan?" guessing. SA 73-76. After Dassey guesses cutting her hair, punching her, and cutting her throat, "All right, I'm just gonna come out and ask you. Who shot her in the head?" Dassey's This evidence Dassey testified that his jeans were corroborates jeans became stained stained Dassey's trial with bleach while he with bleach. testimony. helped his uncle clean up R. 19-21: 32-37. what looked like an automotive fluid spill. The RAV4's Dassey was fed "With, how's, the license license this fact through a plates were taken off the plates were leading question. car, who did that?" removed. SA 90; R. 19-24: 23. On February 27, the investigator also asked, "Did he tell you if he did anything with the license plates?"

Dassey The physical There was no evidence of sexually evidence does not handcuffs chafing against assaulted corroborate this the headboard. The Halbach fact. R. 19-17: 96-97; handcuffs and leg irons while she R. 19-15: 214-17. found in Avery's room was contained no fingerprints handcuffed or DNA from Dassey or to the bed. Halbach. The physical The plastic film found on evidence found the bed's spindle was on Avery's bed is polypropylene, which, not probative. according to the state's R. 19-16: 246. forensic scientist, is found in garments, in addition to plastic containers and rope manufacturing. This detail was Dassey testified that he drawn from concocted this detail from popular media. Kiss the Girls (1995), a R. 19-21: 65-67. book he read, where a woman is restrained during a sexual assault. Halbach The media widely This fact appeared in was in the publicized that news stories. back of the Halbach's blood RAV4. was found in the back of the car. RSA 70.

The RAV4's Dassey was fed "OK, what else did he do, battery the fact that he did somethin' else, cables were Avery went under you need to tell us what disconnected. the RAV4 hood he did, after that car is through a leading parked there. It's question after he extremely important. unsuccessfully (pause) Before you guys guessed. SA 92. leave that car." After Dassey responded that Avery left the gun in the car, "That's not what I'm thinkin' about. He did something to that car. He took the plates and he, I believe he did something else in that car." "I don't know." "OK. Did he, did he, did he go and look at the engine, did he raise the hood at all or anything like that?"

Halbach In addition to "Was she ever in the was shot in being fed that she garage?" the garage. was shot, Dassey was fed that she "No." was shot in the Investigators lead him, garage, after saying "Again, we have, initially denying w-we know that some she was ever in things happened in that there. SA 81-86. garage, and in that car, we know that. You need to tell us about this so we know you're tellin' us the truth." Shortly after, they ask, "Tell us where she was shot?" "In the head." "No, I mean where in the garage." After Dassey answered that she was shot in the truck and not on the garage floor, "[C]ome on, now where was she shot? Be honest here." "The truth." "In the garage."

Halbach's Dassey was fed On February 27, "Did he camera and this fact through a tell ya anything about ... phone were leading question her other possessions ... burned in a on February 27. she probably had her cell barrel. Then on March 1, phone, a camera to take he guessed that pictures." these items were burned. R. 19-24: After Dassey denied 36; SA 109-11. putting anything in the burn barrel or knowing whether she had a purse, cell phone, or camera, he was pressed about what happened to these items and guessed, "[Avery] burnt `em." The only possessions he said he saw in the burn barrel were those fed to him ("Like a cell phone, camera, purse.").

Halbach's Fed fact and On February 27, remains media reports. investigators said, "I find were R. 19-24: 5-6, 9; it quite difficult to believe burned in RSA 69. that if there was a body the bonfire in that [fire] Brendan that pit. you wouldn't have seen something like a hand, or a foot, a head, hair, something." Media had reported her remains were found there. Dassey Dassey's response "We know that Teresa resisted the seems to accept had a, a tattoo on her suggestion the suggestion stomach, do you that that she had a remember that?" Halbach tattoo. SA 151-52. had a (shakes head "no") "uh tattoo. uh." "Do you disagree with me when I say that?" "No but I don't know where it was."

The majority concedes that AEDPA does not require a "nearly identical factual pattern" to find that a decision involved an unreasonable application of law. Panetti v. Quarterman , 551 U.S. 930, 953, 127 S.Ct. 2842, 168 L.Ed.2d 662 (2007) (citation omitted). But that is in essence what the majority has demanded. In arguing that even non–AEDPA cases have found confessions voluntary under similar circumstances, the majority cites two decisions. But as it concedes, Fare v. Michael C. , 442 U.S. 707, 99 S.Ct. 2560, 61 L.Ed.2d 197 (1979), is critically different: Michael C. was of average intelligence and had many prior interactions with the criminal justice system. Id. at 726, 99 S.Ct. 2560. While Boulden v. Holman , 394 U.S. 478, 89 S.Ct. 1138, 22 L.Ed.2d 433 (1969), may superficially appear to be more similar to Dassey's case, it is of dubious relevance given the fact that it was decided (along with Michael C. ) decades before the Supreme Court instructed lower courts to recognize the unique psychological vulnerabilities of youth stemming from their incomplete neurological development. See, e.g. , Graham v. Florida , 560 U.S. 48, 68, 130 S.Ct. 2011, 176 L.Ed.2d 825 (2010) ; Roper v. Simmons , 543 U.S. 551, 569–70, 125 S.Ct. 1183, 161 L.Ed.2d 1 (2005).

The Wisconsin Court of Appeals failed reasonably to apply in any meaningful way at least three principles that the Supreme Court has clearly established: (1) special care for juvenile confessions, (2) consideration of the totality of the circumstances, and, most importantly, (3) prohibition of psychologically coercive tactics. This led to the kind of extreme malfunction in the adjudication of Dassey's case for which section 2254(d)(1) provides a remedy. By turning a blind eye to these problems, the majority has essentially read habeas corpus relief out of the books.

II

There is a second, independent, reason why the district court correctly granted Dassey's habeas corpus petition and our original panel was correct to uphold that ruling: the Wisconsin Court of Appeals made unreasonable factual determinations. See Brumfield v. Cain , ––– U.S. ––––, 135 S.Ct. 2269, 2276, 192 L.Ed.2d 356 (2015) (granting habeas corpus relief under section 2254(d)(2), without needing to reach petitioner's section 2254(d)(1) argument). The district court, whose factual assessments deserve some deference from us, found that the Wisconsin Court of Appeals erroneously concluded that investigators made no promises of "leniency." According to the district court, though no statement in particular rendered the confession involuntary, the cumulative effect of investigators' tactics overbore Dassey's free will.

The majority dismisses this concern because there was no "specific" promise of lenience. But as the district court concluded, when examining the totality of the circumstances, it is clear that Dassey was guessing at what he thought the investigators wanted to hear so that he could leave. Dassey was reassured across two days of interviews that being "honest" would allow him to go "free." Although an adult of average intelligence might recognize the Biblical allusion, see John 8:32 ("You will know the truth, and the truth will set you free."), Dassey was not an adult and not of average intelligence. Instead, he was a mentally limited teenager who did not understand abstractions. Playing their "20 Questions" game, the officers forced Dassey to try out different answers until he stumbled upon the answer they wanted—defined by them as the answer that was sufficiently truthful. And what was Dassey's response after all this? He asked if he was free to go back to school to turn in a project that was due, and when told that he could not, he indicated that he thought he would be in jail for just one day. No more conclusive evidence of his literalism and his lack of understanding is needed.

By finding no promises of lenience were made and that the confession was voluntary, the Wisconsin Court of Appeals made an unreasonable determination of fact in light of the clear and convincing weight of the evidence.

III

Under AEDPA, the role of the federal courts in reviewing Dassey's petition for habeas relief is quite limited. But AEDPA does not paralyze us in the face of a clear constitutional violation. The Due Process Clause and the right against self–incrimination demand that, in order to be admissible in evidence, a suspect's confession must be voluntary. Dassey's was not. Because the detectives used coercive interrogation tactics on an intellectually disabled juvenile, Dassey's will was overborne during his March 1 interrogation. Without this involuntary and highly unreliable confession, the case against Dassey was almost nonexistent. This court should be granting his petition for a writ of habeas corpus and giving the state an opportunity to retry him, if it so desires. I respectfully dissent.

ROVNER , Circuit Judge, and WOOD , Chief Judge , and WILLIAMS , Circuit Judge , dissenting.

I continue to believe, as I explained in the panel opinion, and as Chief Judge Wood's dissent so persuasively argues, that the state court failed to fulfill the Supreme Court's mandate to review juvenile confessions with special care, and unreasonably held that Dassey's confession was voluntary. And for all of the reasons upon which Chief Judge Wood has expounded and those set forth in the original panel opinion in Dassey v. Dittmann , 860 F.3d 933 (7th Cir. 2017), reh'g en banc granted, opinion vacated (Aug. 4, 2017), I too respectfully dissent. I write separately simply to point out the chasm between how courts have historically understood the nature of coercion and confessions and what we now know about coercion with the advent of DNA profiling and current social science research.

Although I write in the hope of encouraging courts to update their understandings of the factual nature of coercion, my conclusion about the proper outcome of Dassey's habeas petition does not depend on any change in law. Current Supreme Court precedent requires that a court view the totality of the circumstances of any interrogation, and to take special care when evaluating the confessions of juveniles. To comply with the command of the Supreme Court, therefore, a court must include within its evaluation of the totality of the circumstances the impact of coercive interrogation techniques upon the particular vulnerabilities of the individual subject to those techniques. The state court did not do so in considering Dassey's appeal. For this reason, Dassey's conviction cannot stand. Unfortunately, four members of the seven–member en banc panel of this court do not agree—a decision that I believe has worked a profound injustice. Nevertheless, I hope to convince my colleagues throughout the courts that reform of our understanding of coercion is long overdue. When conducting a totality of the circumstances review, most courts' evaluations of coercion still are based largely on outdated ideas about human psychology and rational decision–making. It is time to bring our understanding of coercion into the twenty–first century.

Half a century ago the Supreme Court held that police misrepresentations during interrogations, although relevant to a totality of the circumstances inquiry, were not in and of themselves sufficient to render an otherwise voluntary confession inadmissible. Frazier v. Cupp , 394 U.S. 731, 739, 89 S.Ct. 1420, 22 L.Ed.2d 684 (1969). In other words, police may deceive, trick, conceal, imply, and mislead in any number of ways, provided that, under a totality of the circumstances evaluation, they do not destroy a suspect's ability to make a rational choice. See id. (finding an interrogator's lie that a fellow suspect had confessed insufficient to make an otherwise voluntary confession inadmissible); Procunier v. Atchley , 400 U.S. 446, 454, 91 S.Ct. 485, 27 L.Ed.2d 524 (1971) (determining that it was not per se coercive for police to send in a cooperating insurance agent to deceive the defendant into confessing to obtain insurance payments for his children); see also United States v. Villalpando , 588 F.3d 1124, 1128 (7th Cir. 2009) ("Trickery, deceit, even impersonation do not render a confession inadmissible"); United States v. Rutledge, 900 F.2d 1127, 1131 (7th Cir. 1990) (noting that "the law permits the police to pressure and cajole, conceal material facts, and actively mislead—all up to limits").

These cases, however, were born in an era when the human intuition that told us that "innocent people do not confess to crimes" was still largely unchecked. This belief is rooted in the mind's tendency to assume that statements made to a police officer that are against one's self interest can be trusted or, to put it simply, the thought that most of us have that "I would never confess to a crime I did not commit."¹ Peer–reviewed studies confirm that jurors tend to have hard–to–dislodge beliefs that a suspect who is innocent could not be manipulated into confessing.² And, in fact, this false notion is precisely what the state implored the jurors in Dassey's trial to believe, arguing in closing that "[p]eople who are innocent don't confess." R. 19–23 at 144. We know, however, that this statement is unequivocally incorrect. Innocent people do in fact confess, and they do so with shocking regularity. As of June 7, 2016, The National Registry of Exonerations had collected data on 1,810 exonerations in the United States since 1989 (that number as of December 4, 2017 is 2,132), and that data includes 227 cases of innocent people who falsely confessed.³ This research indicates that false confessions (defined as cases in which indisputably innocent individuals confessed to crimes they did not commit) occur in approximately 25% of homicide cases.⁴

1 Saul M. Kassin et al., Police–Induced Confessions: Risk Factors and Recommendations, 34 L. & Hum. Behav. 49, 51 (2010).

2 Iris Blandón–Gitlin et al., Jurors Believe Interrogation Tactics Are Not Likely to Elicit False Confessions: Will Expert Witness Testimony Inform Them Otherwise?, 17 Psychol., Crime & L. 239, 256 (2011).

3 Samuel Gross et al., For 50 Years, You've Had "The Right to Remain Silent," The National Registry of Exonerations, False Confessions (June 12, 2016), http://www.law.umich.edu/special/exoneration/Pages/false–confessions.aspx.

4 Samuel Gross et al., Exoneration in the United States, 1989–2012: Report by the National Registry of Exonerations, 58, 60, https://www.law.umich.edu/special/exoneration/Documents/exoneration s_us_1989_2012_full_report.pdf.

In a world where we believed that "innocent people do not confess to crimes they did not commit," we were willing to tolerate a significant amount of deception by the police. Under this rubric, the thinking went, the innocent person (or at least the vast majority of healthy, sane, innocent adults of average intelligence) would not confess even in response to deception and cajoling. And so our case law developed in a factual framework in which we presumed that the trickery and deceit used by police officers would have little effect on the innocent.

If it is true that, except in extreme cases, innocent people do not confess, what difference does it make if detectives Fassbender and Wiegert made false assurances and used deception in interrogating Dassey? So what if they gave general assurances of leniency, used leading questions, fed Dassey information, lied about how much information they had, told Dassey that they were on his side, implored him that "honesty is the only thing that will set you free," suggested answers, and even went so far as to tell a confused and floundering Dassey that Teresa had been shot in the head? "Dassey was not subject to physical coercion or any sort of threats at all," the majority tells us, and "[g]iven the history of coercive interrogation techniques from which modern constitutional standards for confessions emerged, this is important."Ante at 313.But what do we do when the facts that supported our "modern constitutional standards" come from a fifty–year–old understanding of human behavior, and when what we once thought we knew about the psychology of confessions we now know not to be true? Our long–held idea that innocent people do not confess to crimes has been upended by advances in DNA profiling. We know now that in approximately 25% of homicide cases in which convicted persons have later been unequivocally exonerated by DNA evidence, the suspect falsely confessed to committing the crime.⁵ The majority points out that the number of known false confessions is low compared to the total number of guilty pleas to violent felonies. Ante at 317–18 n.8. This comparison is inappropriate for two reasons. First, the number of guilty pleas is the wrong denominator. Defendants plead guilty in all manner of situations, not only after interrogations by the police, as was the case with Dassey. Many defendants, for example, accept a plea after carefully weighing their options with a lawyer without ever having been subject to a coercive interrogation—the only type of confessions with which we are concerned in this case. Moreover, and more importantly, in the numerator, the statistics for false confessions include only those who have been exonerated based on some form of objective evidence (DNA, impossibility, the confession of another, etc.). The universe of people who falsely confess is undoubtedly larger than the subset of people who have confessed and then been fortunate enough to have been exonerated by objective, irrefutable evidence. But most importantly, as the majority concedes, even one coerced false confession is "very troubling." Ante at 317–18 n.8. Indeed any coerced false confession is an affront to due process and cannot stand.

5 Id. at 331.

Certainly human intuition makes it almost inconceivable to imagine that someone might falsely confess to the murder of one's own child. Yet in October 2004, Kevin Fox of Wilmington, Illinois did just that. He confessed to sexually assaulting his daughter, placing duct tape over her mouth, drowning her in the river, and then going home to sleep.^{6 , 7} His confession was detailed and included accounts of her moving and kicking in the water and struggling to remove the duct tape as she drowned. He quickly rescinded his confession, but spent eight months in prison until DNA testing ruled him out as a suspect and the State of Illinois dropped the charges. See generally Fox v. Hayes , 600 F.3d 819 (7th Cir. 2010). Not only did the DNA alone exclude him as a suspect, but for any who had remaining doubts, the conviction of another man six years later made it unequivocally certain that his confession had been false. In 2010, Scott Eby, who was in prison for raping a relative, confessed to the murder.⁸ At the time of the murder he had been living not far from the Fox home. While drunk and high on cocaine Eby decided to rob some houses, and when he happened upon a sleeping three–year–old Riley Fox, he abducted her, sexually assaulted her, and then drowned her to cover his crime. His DNA matched that found on the duct tape used to bind Riley. A pair of boots, which had been found at the scene, photographed, and then ignored for years, had the name "Eby" written on the tongue.

6 Bryan Smith, Kevin Fox, in True Stories of False Confessions 107 (Rob Warden et al. eds., 2009).

7 Bryan Smith, The Nightmare: A Look at the Riley Fox Case, Chi. Mag., July 3, 2006.

8 Steve Schmadeke, I'm the 'Lowest Kind of Slime,' Killer of 3–Year–Old Confessed. Court Records Outline Investigators' Path to Scott Wayne Eby, Chi. Trib., Feb. 26, 2011.

Five decades ago, when the Supreme Court issued its opinions allowing interrogator deception, there was no DNA evidence that could demonstrate with such clarity that innocent people were confessing to crimes they had not committed at a surprising rate, and therefore, only a limited body of psychological science explaining why this happens.

Even now, despite the overwhelming evidence regarding the coercive nature of constitutionally permissible interrogation techniques, we have not changed our understanding of how to view the facts surrounding coercion when evaluating the totality of the circumstances. Yet we now have a robust and growing body of rigorous, peer–reviewed, legal and psychological research demonstrating how current interrogation tactics influence people, and particularly juveniles and intellectually impaired people, to act against their own self–interest in such a seemingly irrational manner.⁹

9 See Saul M. Kassin, False Confessions, 8 WIREs Cogn Sci. e1439 (2017).

Some of the factors that induce false confessions are internal. Studies have demonstrated that personal characteristics such as youth, mental illness, cognitive disability, suggestibility, and a desire to please others may induce false confessions.¹⁰ A survey of false confession cases from 1989–2012 found that although only 8% of adult exonerees with no known mental disabilities falsely confessed to crimes, in the population of exonerees who were younger than 18 at the time of the crime, 42% of exonerated defendants confessed to crimes they had not committed, as did 75% of exonerees who were mentally ill or mentally disabled.¹¹ Overall, one sixth of the exonerees were juveniles, mentally disabled, or both, but they accounted for 59% of false confessions.¹² Indeed, youth and intellectual disability are the two most commonly cited characteristics of suspects who confess falsely.¹³ Dassey suffered under the weight of both characteristics.

10 Blandón–Gitlin et al., supra note 2, at 240.

11 Gross, Exonerations 1989–2012, supra note 4, at 60.

12 Id.

13 Samuel R. Gross et al., Exonerations in the United States 1989 through 2003, 95 J. Crim. L. & Criminology 523, 545 (2005).

In addition to the factors specific to the suspect, some of the factors that induce false confessions are externally imposed. These include "isolation, long interrogation periods, repeated accusations, deception, presenting fabricated evidence, implicit/explicit threats of punishment or promises of leniency, and minimization or maximization of the moral seriousness or legal consequences of the offence."¹⁴ "Maximization" describes the technique whereby the interrogator exaggerates the strength of the evidence and the magnitude of the charges.¹⁵ Dassey's interrogators employed maximization by constantly reminding Dassey, "We already know everything." See, e.g., R. 19–25 at 17, 19, 23, 24, 26, 28, 30, 31, 36, 37, 41, 44, 47, 48, 50, 54, 55, 60, 63, 69, 71. "Minimization" describes tactics that are designed to lull a suspect into believing that the magnitude of the charges and the seriousness of the offense will be downplayed or lessened if he confesses. ¹⁶ Studies demonstrate that minimization causes suspects to infer leniency to the same extent as if an explicit promise had been made, increasing not only the rates of true confessions (from 46% to 81% in one experiment) but also the rate of false confessions (from 6% to 18%).^{17 , 18} Although a court must exclude a confession obtained by direct promise of leniency (see, e.g., United States v. Villalpando , 588 F.3d 1124, 1128 (7th Cir. 2009) ), the research demonstrates that minimization techniques are the functional equivalent in their impact on suspects.¹⁹ The investigators in this case employed classic minimization techniques by repeatedly telling Dassey that it was not his fault that he committed the crime because his uncle, Steven Avery, had made him do it. See, e.g., R. 19–25 at 28, 47, 50, 60, 62. As Chief Judge Wood points out in her dissent, interrogators in this case, as in most police forces in the United States, used the Reid Technique to obtain Dassey's confession. This technique involves isolation, confrontation, maximization and minimization—the psychological strong–arm tactics that are known to produce coerced confessions even in adults of average intelligence.

14 Blandón–Gitlin et al., supra note 2, at 240.

15 Saul M. Kassin et al., Police Interrogations and Confessions: Communicating Promises and Threats by Pragmatic Implication, 15 L. & Hum. Behav. 233, 234–35 (1991).

16 Id. at 235.

17 Id. at 241, 248.

18 Melissa B. Russano et al., Investigating True and False Confessions Within a Novel Experimental Paradigm, 16 Psychol. Sci. 481, 484 (2005).

19 Kassin, Police Interrogations and Confessions, supra note 15, at 241, 248.

Dassey's interrogation thus combined a perfect storm of these internal and external elements. He was young, of low intellect, manipulable, without a friendly adult, and faced repeated accusations, deception, fabricated evidence, implicit and explicit promises of leniency, police officers disingenuously assuming the role of father figure, and assurances that it was not his fault.²⁰

20 The majority has reservations about the use of the Gudjonsson Suggestibility Scale and thus states that it can make no conclusions from the disputed expert testimony about the results. Ante at 305 n.2. Whatever one might make of the Gudjonsson Suggestibility Scale, the interrogation speaks for itself. Dassey is almost frantic in his desire to find the story the investigators seek. For example, in response to the question about what happened to Teresa's head, Dassey guessed at every possible injury or injustice to a head (hitting, punching, throat cutting, hair cutting) hoping to please the officers until, in frustration, they finally informed him that Teresa had been shot in the head. R. 19–25 at 60–63. In response to pressure from the investigators, he changes the locale of the crime from the house to the garage (Id. at 72–73), the color of Teresa's clothes (Id. at 20, 31–32), the location of the knife (Id. at 80–81, 121; R. 19–34 at 23–24, 27), whether Teresa was standing on the porch after school (R. 19–25 at 19–20, 27–28, 90–91), whether Avery went under the hood of Halbach's car (Id. at 77–80), when the fire occurred (Id. at 23, 32–33; R. 19–34 at 55), and whether he cut her hair (R. 19–35 at 60–61; R. 19–34 at 36–37, 65–66, 98). Even under the state's theory of the case, the naïve Dassey, who had never been in trouble with the law and had never had a sexual experience with a woman, was readily manipulated by his uncle into participating in a repulsive and heinous crime. One does not need the Gudjonsson Suggestibility Scale to conclude, under either party's theory of the case, that Dassey was highly suggestible and manipulable.

For many years, the Reid technique has been criticized by scholars and experts for increasing the rate of false confessions.²¹ As far back as Miranda , the Supreme Court warned that "[e]ven without employing brutality, the 'third degree' " used in the Reid technique "exacts a heavy toll on individual liberty and trades on the weakness of individuals," and "may even give rise to a false confession." Miranda v. Arizona , 384 U.S. 436, 455 & n.24, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966). Recently, Wicklander–Zulawski & Associates, one of the nation's largest police consulting firms, said it will stop training detectives in the method it has taught since 1984, stating that it "is not an effective way of getting truthful information."²² After a spate of high–profile false confession cases in the 1980's, Great Britain transitioned from an accusatorial and coercive Reid–like approach to an investigative model of interviewing which prohibits deception, coercion, and minimization.²³ Meta–analyses of twelve different laboratory experiments indicate that the accusatorial approach increased both true and false rates of confessions, while the information–gathering approach increased the rate of true confessions without also increasing false confessions.²⁴

21 Kassin, False Confessions, supra note 9, at 8.

22 Eli Hager, The Seismic Change In Police Interrogations: A Major Player In Law Enforcement Says It Will No Longer Use A Method Linked To False Confessions, The Marshall Project (March 7, 2017, 10:00 p.m.), https://www.themarshallproject.org/2017/03/07/the–seismic–change–in–police–interrogations.

23 Kassin, False Confessions, supra note 9, at 8.

24 Christian A. Meissner et al., Accusatorial and Information Gathering Interrogation Methods and Their Effects on True and False Confessions, A Meta–Analytic Review, 10 J. Exp. Criminology 459, 481–82 (2014).
--------

No reasonable state court, knowing what we now know about coercive interrogation techniques and viewing Dassey's interrogation in light of his age, intellectual deficits, and manipulability, could possibly have concluded that Dassey's confession was voluntarily given. Although it is my hope that our courts will, when evaluating the totality of the circumstances, engage with the more current understanding of coercion, as I noted at the start, Dassey does not need a change in our existing Supreme Court precedent or any existing law to prevail on his habeas petition. What has changed is not the law, but our understanding of the facts that illuminate what constitutes coercion under the law. Moreover, even under our current, anachronistic understanding of coercion, Dassey's confession was so obviously and transparently coercively obtained that it is unreasonable to have found otherwise. Dassey, however, need not rely on this finding either. Existing Supreme Court precedent allows for significantly deceptive and manipulative interrogation techniques, but those very techniques must then be evaluated, in a totality of the circumstances analysis, for what they are.

The requirement that confessions must be voluntary is a principle at the heart of our legal system. Although psychological and physical torture and coercion are commonplace in some countries as a means of obtaining "confessions," our system of justice rejects the notion that convictions can be obtained through such abuse. We refuse to accept such conduct as a means of obtaining information, not only because it impacts the veracity of the confession, but because it is conduct that we as human beings cannot tolerate from our government. In a case such as this one, where investigators are faced with a crime of horrific brutality and the loss of a treasured life, the impulse to coerce a confession from a suspect may be particularly strong. As judges, we are entrusted with the responsibility to protect against such abusive actions, and uphold those principles that our Constitution protects even in the darkest of times.

What occurred here was the interrogation of an intellectually impaired juvenile. Dassey was subjected to myriad psychologically coercive techniques but the state court did not review his interrogation with the special care required by Supreme Court precedent. His confession was not voluntary and his conviction should not stand, and yet an impaired teenager has been sentenced to life in prison. I view this as a profound miscarriage of justice. I respectfully dissent.