dissenting:
The only evidence against the defendant is a coerced confession. I will take up the elements of the government’s case in turn.
I. The Child’s Account
A temptation exists to refer to the child’s story as “testimony.” The government in its brief (e.g. p. 69) and the majority in its opinion (e.g. Maj. Op. at 1120) yield to this temptation. But the child’s story is not testimony. The story was not told under oath. The story was not subject to cross-examination. It was only an unsworn and untested tale.
The district court found a dozen or more facts of the child’s tale were “obviously not factual”:
-that Preston came to the child’s house and had threatened to kill him the previous day;
-that the child called 911;
-that Preston tried to stab the child with a knife;
-that the child locked Preston in the child’s bedroom with a key;
-that Preston climbed out of the window and onto the roof and jumped off;
-that the child was hiding under a blanket so Preston could not see him;
-that Preston followed the child’s tracks and fell into a water hole;
-that Preston then drove a monster truck off a cliff and the police followed him with helicopters and cars;
-that the child threw knives at Preston’s heart;
-that Preston tried to rape the child’s sister, but that he did not because kittens scratched him;
-that the child ran outside and killed a robber and almost killed Preston.
In the farrago of fantasy, no fact identifies itself as worthy of belief.
After examining the child, the forensic examiner concluded:
[He] described events that are unsupported by the subsequent forensic examination of the victim and his clothing. For example, the victim indicated that during the course of the assault Preston ejaculated in his mouth, on his lips, on his red shirt, and on his stomach. The forensic examination of the victim and the clothing he was wearing nevertheless revealed neither evidence of any semen nor the existence of any red shirt. For that matter the physical examination of the victim showed no signs of trauma or semen.
*1125It is undisputed that there exists no physical evidence of sexual contact. The forensic examiner, a key government witness, found no semen and no trauma.
II. Inconclusive DNA Evidence
The government does not argue that there exists DNA evidence of sexual contact between the child and Preston.
III. Confession Alone Is Not Enough
A defendant cannot be convicted on a confession alone. In United States v. Norris, 428 F.3d 907, 915 (9th Cir.2005), this circuit found that the state must “introduce sufficient evidence to establish that the criminal conduct at the core of the offense has occurred.” Norris, 428 F.3d at 914-15 (citing United States v. Lopez-Alvarez, 970 F.2d 583, 592 (9th Cir.1992)). Accordingly, this circuit reversed the defendant Norris’s conviction for sexual molestation because the “government produced no evidence to corroborate Norris’s confession that the core of the act, Norris’s touching of T.V.’s vulva with his penis, actually occurred.” Id.
Here, the government has produced no evidence to corroborate Preston’s confession that the “core of the act” — sexual contact — “actually occurred.” Id.
IV. Failure of Confession to Satisfy Statute
Even if we were to consider the confession alone, violating the law of this circuit, Preston’s confession fails to satisfy the statute. The government must prove beyond a reasonable doubt that Preston was guilty of “knowingly engaging] in abusive sexual contact” with the “intent to abuse, humiliate, harass, degrade, or arouse or gratify the sexual desire of any person.” 18 U.S.C. §§ 2244, 2245(3). Preston’s confession shows no such intent.
V.The Coerced Confession of The Mentally Retarded Youth
Preston’s “confession” should not be considered at all because it is involuntary.
Agent Kraus and Investigator Secatero questioned Preston on the front porch of his house. Secatero told Preston that “there’s two types of people.” The first type was a “monster,” a “sexual predator,” the “type to prey on little kids.” This type was a “cold-hearted person” for whom they “don’t show any sympathy.” The second type, Kraus and Secatero explained, was not a monster but a “guy that had a bad day” who did a “one time thing.” This second type could “move on,” because what transpired was “just a misunderstanding.” Secatero told Preston that the officers wanted to know “which kind of person” he was. He asked Preston, “Which ... person are you? Are you that type where you prey on little kids?” Preston said he was not.
After giving Preston two criminal alternatives, Secatero and Kraus proceeded to ask Preston a series of questions that forced him to choose between damaging admissions:
-“Is it because [...] you wanted to have sex [... ] or is he the one that came onto you?”
-“Is it something where you forced the issue or is it something he wanted?”
-“Did he pull away or did you pull out?”
-“Did all of your penis go in [or] just a little bit?”
-“Did you do it a lot or just that one time?”
In each case, Preston chose the less damaging admission. At times, the loaded questions were more akin to statements. To these, Preston was silent, said he didn’t know, or denied the claims. Agent Kraus said, for instance, “He pulled his pants down?” ' Preston said he didn’t know. *1126Kraus said, “Did you pull your pants down too?” Preston said, “No.” Kraus said, “You just unzipped your zipper?” Preston was silent. Kraus said, “He pulled his pants up and I assume you zipped up, too.” Preston was silent. Nearly every detail was planted by the officers.
Loaded questions are difficult for intelligent persons. For a feeble mind, they are nearly impossible. Preston is eighteen and has a brain “like a five year old.” He has an IQ of 65. Preston’s IQ places him in the range of mental retardation. Atkins v. Virginia, 536 U.S. 304, 309 n. 5, 318, 122 S.Ct. 2242, 153 L.Ed.2d 335 (2002). When asked whether he was “disabled,” Preston asked the officers to explain the word’s meaning. When explained the word’s meaning, Preston agreed that he probably was disabled. See Morgan Cloud, Words Without Meaning: The Constitution, Confessions, and Mentally Retarded Suspects, 69 U.CHI. L. REV. 495, 511-13 (2002) (finding that mentally retarded people do not understand their limitations and “feel compelled to answer a question, even if the question exceeds [their] ability to answer”).
No one can seriously argue that the paper signed by Preston was comprehended by Preston as a document of grave legal import. Kraus told Preston that the paper was merely a way to say “sorry” to [the child]: “Do you want to write any — usually what we do is we write a statement. If, like, you wanted to say sorry or something like that. You could definitely do that. And we can provide that to him.” Preston’s response was hardly assent: “I’ll just say I’m sorry for what I did, but they’re just trying to accuse me of that shit. But fuck — I mean, not like that.” Ignoring Preston’s equivocation, Kraus assured Preston that the statement was “just a kind of summary.”
The “summary” — Preston’s confession— was a brief gathering of details selected by the officers, culled to their purposes, and written by their hand. The details they selected were details they had fed him. Not once did Preston revise, correct, or counter the officers’ dictation of the statement. Agent Kraus said, for instance, “You just unzipped your zipper?” Preston was silent. Kraus said, “You pulled your penis out, and you put a condom on?” Preston was silent. Each of these details was transferred into the written statement. Kraus asked Preston, “Do you want me to put you’re sorry [...]? Are you sorry or not sorry?” Preston chose the option that kindergarten teaches as proper. When Agent Kraus finished writing the statement, he said to Preston, “I’m going to have you sign this.” Preston signed.
There is no evidence that Preston read what the officer wrote. It is difficult to describe Preston’s choice between bad alternatives, actual denials, and ambiguous silence as “a product of a rational intellect and a free will,” a necessary condition for an admissible confession. Blackburn v. Alabama, 361 U.S. 199, 208, 80 S.Ct. 274, 4 L.Ed.2d 242 (1960).
The majority asserts that coercion of a mentally impaired person “appears to turn largely on the length of the interrogation.” This reading is unsubstantiated. Nowhere does case law state such a rule. Length of an interrogation is one factor; mental deficiency is another. A short interrogation does not reduce the relevance of the defendant’s mental capacity.
The majority notes that the interrogation occurred at Preston’s house with others present. Yet Preston was distinctly vulnerable at his house. Both officers had arrived in unmarked vehicles, and both were dressed in plain clothes. Preston had no prior record and had never spoken to police before. The officers undoubtedly checked his record and knew he lacked *1127experience with law enforcement. The officers learned quickly that neither Preston’s mother nor father was home. Preston himself was eighteen years old. The only others present were two children. Questioning Preston, the officers rapidly surmised his questionable mental capacity, asking him point-blank if he was disabled. By himself on his porch, Preston was vulnerable.
VI. Unreliability of Confessions From Mentally Retarded People
Confessions from mentally retarded people are prone to be false.
Preston’s confession is unreliable. Preston was ready to sign a confession to a crime with the wrong date and did so. The police officers told Preston, at least jfifteen times, that he had been home on Friday to commit the crime; the officers later admitted that they meant to say Wednesday. Although Preston had a regular routine on Fridays of leaving his house to see his aunt, he was so suggestible that he agreed with the officers that he had been home on Friday. Preston’s statement meets the textbook case of a false confession: Preston was given a false detail of the crime, and he accepted it as an actual fact. (“One method for cheeking the authenticity of a voluntary confession, or one that seems to be the result of a mental illness, is to introduce some fictitious aspects of the crime and test whether the suspect will accept them as actual facts relating to the occurrence.” F. Inbau, J. Reid, J. Buckly, & B. Jane, Criminal Interrogations and Confessions, 349 (5th ed. 2011)).
As the officers dictated the statement for Preston to sign, Preston was either silent or made unintelligible sounds. The truth of what transpired remained unarticulated by him. As researchers have found, the “tendency [of retarded people] to mask their disabilities,” combined with the widespread ignorance of mental retardation, “make it difficult for police and others to properly interpret the responses of mentally retarded persons.” Eugene R. Milhizer, Confessions After Connelly: An Evidentiary Solution for Excluding Unreliable Confessions, 81 TEMP. L. REV. 1, 15 (2008). As a result, police detectives can “create false confessions by pressuring the suspect to accept a particular account and suggesting crime facts to him, thereby contaminating [his] narrative.” John B. Gould & Richard A. Leo, One Hundred Years Later: Wrongful Convictions After a Century of Research, 100 J.Crim. L. & Criminology 825, 849 (2010).
Confessions from mentally retarded suspects, in particular youth, are “per se untrustworthy.” See Welsh White, False Confessions and the Constitution: Safeguards Against Untrustworthy Confessions (1997), 32 Harv. C.R.-C.L.Rev. 105. In a recent study of the first 200 DNA exonerations in the U.S., 35% of the false confessors were 18 years or younger and/or had a developmental disability. Saul M. Kassin et al., Police-Induced Confessions: Risk Factors and Recommendations, 34 Law & Hum. Behav. 3, 19-22 (2010). Another study found that 69% of the exonerated persons with mental disabilities were wrongly convicted because of false confessions. Id. See also Brandon L. Garrett, The Substance of False Confessions, 62 Stan. L.Rev. 1051, 1064 (2010) (“Mentally disabled individuals and juveniles are both groups long known to be vulnerable to coercion and suggestion.”); Gould at 847 (noting that “the developmentally disabled, cognitively impaired, juveniles — all of whom tend to be unusually suggestible and compliant” — are more likely to confess falsely); Milhizer at 14 (“Certain characteristics common among mentally retarded persons make them particularly prone to confess falsely. For example, mentally retarded suspects are of*1128ten motivated by a strong desire to please authority figures, even if to do so requires them to lie and confess to a crime they did not commit. They also often lack the ability to understand the nature of police questioning.”).
The majority writes that recognizing the cognitive impairment of Preston and his susceptibility to coercion would “significantly broaden” this court’s jurisprudence. The law is more capacious than the majority admits. In Atkins, the Supreme Court determined that, due to “cognitive and behavioral impairments,” including “the diminished ability to understand and process information, to learn from experience, to engage in logical reasoning,” mentally retarded people categorically are at risk of producing false confessions. Atkins, 536 U.S. at 320, 122 S.Ct. 2242. Indeed, the Court noted the exonerations of mentally retarded people who had “unwittingly confessed to crimes they did not commit.” Id. See also Culombe v. Connecticut, 367 U.S. 568, 620, 81 S.Ct. 1860, 6 L.Ed.2d 1037 (1961) (holding involuntary the confession extracted from a “thirty-three-year-old mental defective ... with an intelligence quotient of sixty-four”); Commonwealth of the Northern Mariana Islands v. Mendiola, 976 F.2d 475 (9th Cir.1993) (finding confession involuntary on the basis that “consideration of defendant’s reduced capacity is critical because it rendered him more susceptible to subtle forms of coercion,” and citing the low intelligence of defendants in Reck v. Pate, 367 U.S. 433, 81 S.Ct. 1541, 6 L.Ed.2d 948 (1961) and Payne v. Arkansas, 356 U.S. 560, 562, 567, 78 S.Ct. 844, 2 L.Ed.2d 975 (1958)).
The majority thoughtfully asked the district court to reconsider its order approving the plethysmograph, but the majority does not ban it. I have already expressed in a concurring opinion a critique of this procedure. See United States v. Weber, 451 F.3d 552, 570-71 (9th Cir.2006). For the reasons stated in my concurrence in Weber, I would ban this procedure altogether. Psychiatric researchers have referred to my criticism with approval. See, e.g., Michael Harlow and Charles Scott, “Penile Plethysmography Testing for Convicted Sex Offenders,” 35 Jour, of the Amer. Acad, of Psych, and the Law (2007).