DAVIS, Circuit Judge:

Respondent-Appellant Byron Neil Antone appeals the district court's order of his civil commitment under the Adam Walsh Child Protection and Safety Act of 2006 ("the Walsh Act"), codified at 18 U.S.C. §§ 4247-48. Four days before he was to be released from federal prison, an official of the Federal Bureau of Prisons certified Antone as a sexually dangerous person eligible for civil commitment. Upon referral of the ensuing proceedings by the district court, a federal magistrate judge held a threeday evidentiary hearing and thereafter issued a report and recommendation that Antone should not be found to be a sexually dangerous person. The district court adopted the majority of the magistrate judge's factual and credibility determinations, but it ultimately found that the Government had satisfied its burden under the Walsh Act to prove that Antone was a sexually dangerous person, and it committed him to civil custody.

Upon our careful review of the appellate record, we conclude that the district court lacked sufficient evidence to find that Antone met the standard for civil commitment under the Walsh Act. Specifically, the Government did not present clear and convincing evidence that Antone's mental illnesses would cause him to have serious difficulty refraining from sexually violent conduct. Accordingly, we reverse.

I.

A.

Byron Neil Antone, now forty-one years old, was born in and raised on the Tohono O'odham Indian Reservation in south central Arizona.1 Until age nine or ten, Antone was raised by his mother; after that point, he resided with his grandmother and his godmother.

Antone's mother and grandmother were heavy drinkers and Antone was often neglected and verbally and physically abused as a child. At seven years old, Antone was on several occasions sexually abused by his aunt, who was a teenager at the time. By the time he was fifteen years old, he had had sexual intercourse with at least two adult women, one of whom was twenty-six.

Antone had serious behavioral issues as a child, which led to school expulsions and stints in juvenile detention. He dropped out of high school in ninth grade. He did not maintain steady employment thereafter, although he was employed seasonally as a firefighter with the United States Forestry Service and had attended specialized training classes in that field.

In 1991, when Antone was nineteen years old, he was arrested and charged with sexual misconduct with a minor, sexual abuse, and contributing to the delinquency of a minor. The arrest related to two sexual acts with a sixteen-year-old who was Antone's girlfriend at the time. The first sexual act was consensual, but the second was forcible rape. Antone pled guilty to the sexual abuse charge in the Judicial Court of the Tohono O'odham Nation ("tribal court") and served about six months in jail.

In 1997, tribal authorities charged Antone with threatening and disorderly conduct. He admitted to rubbing the buttocks of his cousin, then twenty-one years old, while she was sleeping on the couch. He was sentenced to 60 days in tribal jail.

From 1998 to 1999, Antone was charged by tribal authorities for several acts of sexual misconduct, which resulted in a consolidated plea agreement and tribal judgment entered on March 16, 1999. The consolidated tribal judgment related to four victims and spanned incidents from 1992 through 1997:

Antone pled guilty to charges related to these four incidents in the consolidated plea agreement. He was sentenced to 3,600 days in jail by the tribal court.

Almost all of the incidents described above, and certainly the June and November 1997 incidents, took place when Antone was either intoxicated from alcohol and/or high on cocaine. Indeed, Antone has a serious history of substance abuse. When he was arrested in February 1998, he was drinking 3 to 5 quarts of beer a day on average, and up to 11 quarts on some days. He was also abusing a number of drugs, including marijuana, LSD, and crack cocaine. As a result, Antone has little to no recollection of these incidents.

In November 1999, Antone was sentenced in the United States District Court for the District of Arizona on a sexual assault charge. The particular charge related to Antone's assault against C.R. in June 1997, which was also a subject of his consolidated tribal judgment. In addition, Antone admitted in the federal plea agreement to sexual misconduct as to all the incidents covered in the tribal court convictions.

According to the testimony of Antone's attorney at the time, which the magistrate judge fully credited, the federal criminal charge was actually initiated by Antone and his attorney. "The reason was to enable [Antone] to be transferred to federal custody and thereby have access to sex offense treatment at FCI-Butner, which [the attorney] believed would be designed specifically for Native Americans." J.A. 845.

The federal district court in Arizona sentenced Antone to 114 months of incarceration, with credit for time served, and 60 months of supervised release. The plea agreement reflected Antone's request to receive sex offender treatment in federal custody, and the district court included a recommendation in its judgment that Antone participate in the residential drug treatment and sex offender treatment programs.

B.

In accordance with the federal judgment and commitment order, Antone was incarcerated in the federal Bureau of Prisons system from November 1999 through February 23, 2007, when the Government initiated the instant proceeding four days before his expected release. Since then, Antone has resided in FCI-Butner, a medium security correctional institution in North Carolina, awaiting his civil commitment hearing and its resolution. As a result, Antone has been in continuous federal custody for the past fourteen years, or since he was twenty-seven years old.

During the entire period of his federal custody, Antone has not been shown to have consumed alcohol or drugs. Antone's prison record contains no sanctions or nonsanctioned incidents related to alcohol or drugs, and he testified that he has been sober for fourteen years. The Bureau of Prisons regularly administers Breathalyzer tests on inmates in recognition of the fact that it is possible to make and obtain contraband alcohol within the prison. Antone has never tested positive on those tests.

Antone has attended Alcoholics Anonymous and Narcotics Anonymous on his own initiative. He attended meetings during the first year and a half of his prison term and restarted about a year before his commitment hearing. He also completed a Drug Education Program and a non-residential substance abuse program.

Antone's behavioral problems while in prison have been minimal. He has been sanctioned for four incidents, twice for fighting without serious injury and twice for minor rule violations; the last of these sanctions occurred in 2004.2 He obtained his GED in 2001. In addition, he has maintained employment as an orderly in his housing unit. His work performance therein was characterized as "superior." J.A. 843.

Antone regularly seeks out advice and counseling from his prison's counselors and treatment specialists. In particular, he has asked his counselors how to communicate with his son, with whom he corresponds by mail, and for advice on anger management. Antone has taken classes in art, beading, meditation, and guitar. He teaches other inmates how to play the guitar.

As for sexual conduct, Antone's record indicates that he has "not engaged in sexual misconduct during his extended incarceration." J.A. 882. At the time of the evidentiary hearing, however, he had not attended sex offender therapy or treatment. Antone and his former attorney testified that he had made several requests for treatment at the early side of his incarceration period, but it was apparently not then available to him because "his release date was so far in the future."3 J.A. 830. When it became available in September 2008, after the Government filed its § 4248(a) petition, Antone did not participate in the treatment. He indicated that he did not do so because he knew that statements made during treatment "could be used against him" in the commitment proceeding. Id.

C.

On February 23, 2007, four days before Antone's expected release date, the Government filed a certification, pursuant to 18 U.S.C. § 4248(a), of Antone as a sexually dangerous person. The case was originally stayed pending an appeal relating to the constitutionality of § 4248, see United States v. Comstock, 551 F.3d 274, 276 (4th Cir.2009) (holding that Congress lacked authority to implement § 4248), rev'd and remanded, 560 U.S. 126, 130, 130 S.Ct. 1949, 176 L.Ed.2d 878 (2010) (reversing on issue of Congressional authority but remanding for due process consideration); 627 F.3d 513, 515 (4th Cir.2010) (subsequently holding that § 4248 satisfies due process clause), cert. denied, ___ U.S. ___, 131 S.Ct. 3026, 180 L.Ed.2d 865 (2011). In June 2010, Antone filed a motion for a hearing on the merits of the certification, and the district court referred the matter to a magistrate judge for an evidentiary hearing and report and recommendation.

The magistrate judge held an evidentiary hearing over the course of three days in October 2011. As will be described in further detail infra, the Government presented the testimony of Antone, as well as two expert witnesses; Antone presented the testimony of a specialist and a counselor at the correctional facility at which he resided, a United States Probation Officer from Arizona, and an expert witness. The magistrate judge admitted the testimony of all three proffered expert witnesses.

On April 30, 2012, the magistrate judge issued his M & R, in which he recommended that Antone not be found a sexually dangerous person. The Government thereafter filed a series of objections to the M & R, to which Antone responded. The Government also submitted several additional notices of supplemental authority, including Ninth Circuit case law on the tolling of supervised release during the pendency of a civil commitment proceeding.

On September 20, 2012, the district court issued its order and judgment on the instant certification. Although it accepted all of the magistrate judge's credibility determinations and findings of historical fact, it rejected the M & R's ultimate recommendation of a finding of not sexually dangerous. It found that the combination of Antone's serious mental illnesses — namely antisocial personality disorder and polysubstance dependence — would cause him to have serious difficulty in refraining from sexually violent conduct if released. It therefore committed Antone to the custody of the United States Attorney General as a sexually dangerous person. The instant appeal followed.

II.

A.

The Government seeks the commitment of Antone pursuant to 18 U.S.C. § 4248, which was enacted as part of the Adam Walsh Child Safety and Protection Act of 2006. Under § 4248, the Government may seek the civil commitment of certain individuals in the custody of the Federal Bureau of Prisons who are determined to be "sexually dangerous person[s]." 18 U.S.C. § 4248(d). The commitment process is initiated when the Attorney General or his designee files a certification attesting that an individual is sexually dangerous as defined by the Walsh Act, after which the respondent is entitled to an evidentiary hearing. "If, after the hearing, the court finds by clear and convincing evidence that the person is a sexually dangerous person, the court shall commit the person to the custody of the Attorney General." Id.

To demonstrate that an individual should be civilly committed under § 4248, the Government must prove, by clear and convincing evidence, that each one of the following criteria has been satisfied: (1) the individual has previously "engaged or attempted to engage in sexually violent conduct or child molestation" (the "prior conduct" element), 18 U.S.C. § 4247(a)(5); (2) the individual currently "suffers from a serious mental illness, abnormality, or disorder" (the "serious illness" element), id. § 4247(a)(6); and (3) as a result of such a condition, the individual "would have serious difficulty in refraining from sexually violent conduct or child molestation if released" (the "serious difficulty" or "volitional impairment" element), id. See also Comstock, 560 U.S. at 130, 130 S.Ct. 1949; United States v. Springer, 715 F.3d 535, 538 (4th Cir.2013). Antone has conceded that the Government has met its burden with regard to the prior conduct element as well as the finding of a serious mental illness. He disputes, however, the district court's conclusion as to the third element, that the Government has demonstrated a sufficient likelihood that Antone will re-offend.

We review the district court's factual findings for clear error and its legal conclusions de novo. United States v. Hall, 664 F.3d 456, 462 (4th Cir.2012). For the reasons we explain within, although the district court's ultimate mixed finding on volitional impairment is not infected with "clear error" in the traditional sense of that term, it nonetheless constitutes reversible error because it is against "the clear weight of the evidence considered as a whole." United States v. Wooden, 693 F.3d 440, 451 (4th Cir.2012). Put somewhat differently, we conclude as a matter of law that the Government failed to establish by clear and convincing evidence that Antone would, as a result of his serious illness or condition, have serious difficulty in refraining from sexually violent conduct if released.

B.

The standard set forth for civil commitment under § 4248 is clear and convincing evidence. This so-called "intermediate" standard is mandated not only by the plain language of the statute, 18 U.S.C. § 4248(d), but by constitutional due process constraints, as well. See Addington v. Texas, 441 U.S. 418, 427, 99 S.Ct. 1804, 60 L.Ed.2d 323 (1979) (observing that the clear and convincing evidence standard is required in civil commitment proceedings because "[t]he individual's interest in the outcome of a civil commitment proceeding is of such [great] weight and gravity").

When applying the clear and convincing standard, the court must identify credible supporting evidence that renders its factual determination "highly probable." Direx Israel, Ltd. v. Breakthrough Med. Corp., 952 F.2d 802, 810 n. 7 (4th Cir.1992). The court must then weigh the evidence and ask whether the totality of the record "produces in the mind of the trier of fact a firm belief or conviction, without hesitancy, as to the truth of the allegations sought to be established[.]" United States v. Heyer, 740 F.3d 284, 292, No. 12-7472, 2014 WL 185584, at *6 (4th Cir. Jan. 17, 2014) (quoting Jimenez v. DaimlerChrysler Corp., 269 F.3d 439, 450 (4th Cir.2001)); Springer, 715 F.3d at 538.

In applying the first two commitment criteria under the Walsh Act, the question is whether the Government has established with clear and convincing evidence that the respondent acted or acts in a certain manner. The third element, however, is more complicated, in that it requires the court to issue a predictive judgment: has the Government met its burden by presenting clear and convincing evidence that, in the uncertain future, the respondent will have "serious difficulty in refraining from sexually violent conduct or child molestation"? 18 U.S.C. § 4247(a)(6).

We are mindful that the Supreme Court has explained that such an inquiry "will not be demonstrable with mathematical precision." Kansas v. Crane, 534 U.S. 407, 413, 122 S.Ct. 867, 151 L.Ed.2d 856 (2002). Instead, in order to find that the third criterion is satisfied, the court must look for

Id. In other words, the Government must demonstrate that the serious illness, as it has manifested in the particular respondent, has so significantly diminished his volitional capacity such that he is distinguishable from the ordinary "dangerous but typical recidivist." Id.; see also Wooden, 693 F.3d at 460 (framing the third criterion as "the extent to which the inmate is controlled by the illness").

We now assess the instant record with this exacting standard in mind. As to the third criterion, we find that the aggregate of historical, direct, and circumstantial evidence contained therein may be best described (as the magistrate judge seemed to regard it) as in equipose, or, at most, as rising to a level of preponderance in favor of commitment. But this is simply not enough to satisfy the statutory burden of clear and convincing evidence. See Medtronic, Inc. v. Mirowski Family Ventures, LLC, 571 U.S. ___, ___, 134 S.Ct. 843, 849, ___ L.Ed.2d ___, ___, No. 12-1128, 2014 WL 223040, at *6 (2014) ("[T]he burden of proof ... [is] part of the very substance of [the plaintiff's] claim and cannot be considered a mere incident of a form of procedure.") (quoting Garrett v. Moore-McCormack Co., 317 U.S. 239, 249, 63 S.Ct. 246, 87 L.Ed. 239 (1942)). We thus have no hesitation in finding a fatal evidentiary insufficiency in the Government's presentation.

C.

The majority of the evidentiary record consists of reports and testimony presented at the three-day hearing in front of the magistrate judge. At the hearing, the Government presented testimony from Antone himself, and expert witnesses Dr. Amy Phenix, Ph.D. and Manuel E. Gutierrez, Psy.D. Antone then presented the testimony of Clement Gallop, a treatment specialist in the commitment and treatment program at FCI-Butner; Andre Taylor, a counselor at FCI-Butner; Anne Schauder, a United States Probation Officer from Arizona; and an expert witness, licensed psychologist Roy G. Daum, Psy.D. The magistrate judge found all of the witnesses credible, with a single exception related to Antone's account of certain past crimes.

Because the sole issue on appeal is whether there was sufficient evidence of Antone's future volitional impairment, we summarize the evidence only as it pertains to that issue.

1.

The Government first called respondent Antone. Antone testified that he was unable to recall the majority of his sexual assaults because he was either drunk or high at the time of the incidents. He then testified about his upbringing, substance abuse, and progress while in prison. He stated that he would always be an alcoholic and there would always be a risk that he would drink again, but that he knew to stay away from high risk places and people. He also stated that while in prison he had learned how to talk to others about his problems and to "release [his] feelings in a positive way." J.A. 221.

Subsequently, Antone presented the lay testimony of Clement Gallop and Andre Taylor. Gallop is employed as a treatment specialist in the commitment and treatment program at FCI-Butner and Taylor is a counselor at FCI-Butner. Gallop testified that he is approached by Antone on a weekly basis, and that they have discussed issues related to Antone's son and anger management in general. Taylor testified that Antone has never tested positive or been observed to have imbibed alcohol or used drugs, even though such substances are available in prison and Taylor had disciplined others for alcohol-related issues. Both Gallop and Taylor had positive impressions of their interactions with Antone.

Antone also presented the testimony of Allan Duprey and Anne Schauder. Duprey, who was Antone's attorney on the federal criminal charges, testified that the federal charges were initiated at his urging so that Antone could have access to sex offense treatment designed specifically for Native Americans. Duprey also testified that he had inquired about the availability of sex offender treatment, but was told by the Bureau of Prisons that Antone would not receive treatment until the last five years of his ten-year sentence. Schauder is a United States Probation Officer in the District of Arizona. She explained the support and supervision that her district provides to sex offenders, including the utilization of halfway houses, sex offender treatment, and polygraph tests.4

2.

The Government also presented the testimony of two expert witnesses, Dr. Amy Phenix and Dr. Manuel Gutierrez, who were admitted as experts in the field of forensic psychology without objection. Both Government experts testified that Antone met the criteria for civil commitment as a sexually dangerous person. Their conclusions were based on their review of Antone's written records. Dr. Gutierrez was unable to conduct an interview of Antone, and the portion of Dr. Phenix's report that related to an interview she conducted with Antone was excluded by the magistrate judge and the Government does not challenge that order.

Dr. Phenix diagnosed Antone with paraphilia not otherwise specified, nonconsent ("paraphilia NOS, nonconsent");5 alcohol dependence;6 and antisocial personality disorder ("APD");7 and she testified that as a result, he would have serious difficulty refraining from sexually violent conduct. She opined that the primary cause of Antone's volitional impairment was his paraphilia NOS, nonconsent, mental illness. Dr. Phenix found that Antone's paraphilia NOS, nonconsent, caused him to deviate from ordinary sexual impulses and behaviors, and then his alcohol dependence would serve as a disinhibitor and his antisocial personality disorder would reinforce his paraphilic impulses. When specifically questioned by the court, Dr. Phenix added that, even if the paraphilia diagnosis was disregarded, she would still "believe that [Antone] will go on to commit criminal sexual behavior." J.A. 420.

Dr. Phenix's conclusion on the volitional impairment prong was based on (1) the pattern and duration of Antone's offending; (2) his commission of additional offenses after his 1991 sexual abuse conviction; (3) an actuarial assessment of risk based on static risk factors; (4) the presence of dynamic risk factors; and (5) the absence of protective factors. Dr. Phenix explained at the hearing that her first methodology was to "look at the pattern and duration of his offending to see how well his behavioral controls were when he was in the community." J.A. 331. She focused on certain undisputed historical factors, emphasizing the repeated nature and aggression of Antone's assaults and that he continued to commit assaults even after his first arrest in 1991.8

Dr. Phenix viewed Antone's behavior while incarcerated only as a secondary consideration. When questioned on why she relied almost exclusively on pre-incarceration conduct, Dr. Phenix responded that "I think the best measure of his volition is prior to being in a prison where you have such strict structure and rules for your behavior[.]" J.A. 332.

The magistrate judge also heard similar testimony from Government witness Dr. Gutierrez. Dr. Gutierrez's diagnoses matched those of Dr. Phenix — paraphilia NOS, nonconsent; polysubstance (including alcohol) dependence; and antisocial personality disorder — and also included an additional diagnosis of paraphilia NOS, hebephilia. He concluded that a combination of all of the above-listed illnesses, or alternatively a sole diagnosis of APD, would "cumulative[ly]" cause Antone to have serious difficulty refraining from sexually violent conduct. J.A. 457-58.

Antone subsequently presented the testimony of his expert witness, Dr. Roy Daum, who was admitted as an expert in the field of forensic psychology over the Government's objection. After conducting a forensic evaluation of Antone in February 2011, Dr. Daum diagnosed Antone with polysubstance dependence; frotteurism; and borderline personality disorder.9 He agreed with the Government's experts that Antone met the first and second criteria of § 4248 confinement. He disagreed, however, that Antone had demonstrated that he would have serious difficulty refraining from sexually violent conduct if released.

Dr. Daum reasoned that Antone's offense conduct had not been rooted in sexual deviance, but rather stemmed from a lack of interpersonal skills and a serious substance abuse. Dr. Daum's conclusion considered as a central part of his analysis certain "dynamic" factors observed during Antone's incarceration, including the absence of evidence of any use of drugs or alcohol or any engagement in antisocial activities; the absence of records showing that Antone had a general sexual preoccupation; Antone's positive management records; and evidence of his completion of several self-help programs, learning of vocational skills, and seeking counseling while incarcerated. Of the difference between his opinion and that of Dr. Phenix and Dr. Gutierrez, he remarked the following:

J.A. 642. Finally, Dr. Daum opined that outpatient treatment of Antone during supervised release could adequately address his sex offense and substance abuse problems.

D.

On April 30, 2012, the magistrate judge issued a comprehensive M & R recommending that the district court reject the Government's certification of Antone as a sexually dangerous person. The magistrate judge concluded that the Government had met its burden with regard to the first element, in that Antone had previously engaged in sexually violent conduct. The magistrate judge also accepted the Government's contention that Antone suffered from certain serious mental illnesses within the scope of § 4247(a)(6). Specifically, the magistrate judge found evidence of polysubstance dependence, but it rejected the rest of the Government experts' diagnoses, most notably paraphilia NOS, nonconsent and antisocial personality disorder. It also rejected Dr. Daum's diagnoses of frotteurism and borderline personality disorder.

The magistrate judge ultimately concluded, however, that the Government had not presented sufficient evidence to demonstrate that Antone's polysubstance dependence would result in a serious difficulty refraining from sexually violent conduct. The magistrate judge emphasized that the Government's position on volitional impairment was "based on [a theory of] multiple diagnoses," but it had decided that the Government had not met its burden on any of those diagnoses except polysubstance dependence. As a result, the magistrate judge was not persuaded by the Government's presentation as to Antone's volitional impairment. It cited, for example, to Dr. Gutierrez's understanding that "just a substance diagnosis alone could not essentially stand by itself for civil commitment." J.A. 881.

The magistrate judge afforded near determinative weight to Antone's conduct "over the last 13 or so years," during his time in federal prison. It noted that Antone had not been shown to have consumed alcohol or drugs or to have engaged in sexual misconduct during his extended incarceration. It also pointed to his attendance in Alcoholics Anonymous and his eagerness to seek out counseling for anger management.

The magistrate judge recognized that Antone's achievements while incarcerated came about in a controlled environment where access to his vices was limited. Nevertheless, its review of the evidence — including the testimony of Dr. Daum, who had stressed the utility of dynamic factors in Antone's case-led it to conclude that over the past thirteen years, Antone "has achieved a level of sexual self-regulation" and "a measure of self-control" that significantly undercut the Government's position that he would have serious difficulty refraining if released. J.A. 882. It observed that certain evidence relied upon by the Government's expert witnesses, such as the nature, pattern, and duration of offense conduct, "is not as reliable an indicator of his behavior if released ... because of, among other reasons, the extended intervening period in which there was no manifestation of such conduct." J.A. 884-85.

The magistrate judge also considered as "significant[]" the fact that Antone would be subject to "an extended term of supervised release." J.A. 883. It noted that he would spend his first year of supervised release in a halfway house and that throughout his term, he would be subject to supervision and participation in substance abuse and sex offender treatment programs, periodic drug tests, and prohibitions against contact with children.

In light of the "paucity" of evidence that Antone would have serious difficulty refraining from sexually violent conduct if released, the magistrate judge concluded that the Government had failed to meet its burden of establishing, by clear and convincing evidence, that Antone was a sexually dangerous person under § 4248. J.A. 886.

E.

On September 24, 2012, the district court issued an order rejecting the magistrate judge's ultimate recommendation and civilly committing Antone. It accepted the M & R's findings of historical fact and witness credibility, and noted that it reviewed de novo those aspects of the M & R that were objected to by the parties.

In applying the three-prong test, the district court first accepted the magistrate judge's conclusion that the Government had established that Antone had engaged in sexually violent conduct. It also agreed with the majority of the magistrate judge's recommendations as to the diagnoses of Antone's mental illnesses. Notably, the district court found that Antone suffered from polysubstance dependence and that he did not suffer from paraphilia NOS, nonconsent. In disagreement with the magistrate judge, however, the district court found sufficient evidence of a diagnosis of antisocial personality disorder and held that these two diagnoses, as manifested in Antone, qualified as serious mental illnesses.

The district court then found that the Government had satisfied the volitional impairment requirement of § 4248. In doing so, its primary focus appeared to be Antone's admitted alcoholism. It stated:

J.A. 1115-16. It continued, "[t]he court is convinced that if respondent uses alcohol he will have serious difficulty stopping himself from sexually attacking persons he finds desirable, despite their nonconsent." J.A. 1116.

The district court looked to the combination of Antone's substance dependence and APD diagnoses to predict that his past history of sexual attacks would continue once released. "This volitional impairment has resulted in a consistent pattern of numerous violent sexual attacks in the past, and the court finds that the impairment will persist if respondent is released." Id. The court also relied on Dr. Phenix's testimony that her conclusion on the volitional impairment prong would not change without the paraphilia NOS, nonconsent diagnosis.

Finally, the court expressed concern that it would not be able to require Antone to undergo sex offender treatment. All parties — including Dr. Daum as well as Antone himself — agreed that Antone would benefit from sex offender treatment. According to the district court, however, under a recent Ninth Circuit case, United States v. Turner, a § 4248 detainee's term of supervised release is not tolled while he remains in custody awaiting a commitment hearing. 689 F.3d 1117, 1121 (9th Cir. 2012). Assuming Antone's period of supervised release actually had begun when he was due to be released from the Bureau of Prisons, supervision would have ended on February 27, 2012, but he was still civilly committed at that point. The district court thus predicted that without a tolling mechanism, Antone would not be subject to any term of supervised release under Ninth Circuit law. It also rejected as "irrelevant" the testimony of the probation officer from Arizona based on similar reasoning.

Accordingly, the district court rejected the magistrate judge's ultimate recommendation, instead finding that the Government had established that Antone was a sexually dangerous person within the meaning of 18 U.S.C. § 4247(a)(5) and (6).

F.

Because Antone has not disputed the first and second elements of the Government's § 4248 certification, the sole issue on appeal is whether the district court erred in finding that he will have serious difficulty refraining from sexually violent conduct if released. We hold that it did.

Under the clear error standard, we may not reverse the district court's holding even if we are "convinced that had we been sitting as the trier of fact, we would have weighed the evidence differently." Springer, 715 F.3d at 545 (internal citations omitted). Yet "while clear-error review is deferential, it is not toothless." Wooden, 693 F.3d at 451 (internal citations omitted). A reversal is warranted, for example, if the district court failed to "properly tak[e] into account substantial evidence to the contrary" or its "factual findings are against the clear weight of the evidence considered as a whole." Id. at 462. We may then reverse if, upon reviewing the district court's ultimate mixed findings, we are "left with the definite and firm conviction that a mistake has been committed." Id. at 451.

That is precisely what is at stake here: our review of the lower court opinion leads us to conclude that the district court's inadequate consideration of certain "substantial evidence" — namely Antone's behavior in the past fourteen years or so — constitutes reversible error. And our subsequent analysis of the evidentiary record leaves us with a definite and firm conviction that Antone's commitment should be reversed.

That Antone has "responded very well" to incarceration is not in dispute. J.A. 333. Antone has not tested positive for any substances while in prison, and he testified that he has been sober during his extended incarceration. Antone's conduct as it relates to sexual deviance is equally commendable. Not only has he not engaged in any actual sexual misconduct or hostility toward women, but, just as importantly, his record is devoid of any indication that he has even desired to manifest such misconduct.

Instead, Antone has presented significant testimony to the contrary. Two employees from the correctional facility testified on Antone's behalf, and the magistrate judge found credible their assurances that their interactions with Antone have been consistently positive and that he has demonstrated self-awareness and control on a regular basis. He has for the most part avoided conflicts with superiors or fellow inmates. Antone has completed his GED, as well as other professional programs, and he readily seeks out the prison's mental health resources. He has expressed remorse for his past acts.

Yet the district court's discussion of Antone's behavior while incarcerated is negligible at best. It failed to discuss the opinions of Gallop or Taylor, the only witnesses who have had consistent contact with Antone since his incarceration. It considered the testimony of Antone only to the extent that he admitted that he will always be an alcoholic.10 And it failed to mention the nearly ten-year period in which Antone has had zero disciplinary infractions and the nearly fifteen-year period in which Antone has had no sex-related incidents.

In fact, in the "serious difficulty" section of its opinion, the district court's analysis of Antone's conduct while incarcerated is limited to a single sentence acknowledging his "evident[]" abstinence from alcohol.11 Relying again on Antone's past history of "numerous violent sexual attacks," it concluded that his volitional impairment would persist if released.

Since upholding the constitutionality of the Walsh Act in 2010, we have disposed of more than a handful of § 4248 appeals involving the volitional impairment prong, but none of them involved a respondent who had demonstrated such positive behavior during the extended period of his incarceration. In each of those cases, the district court referred to some negative aspect of the respondent's recent (that is, during-incarceration or post-release) behavior. In United States v. Bolander, 722 F.3d 199, 204 (4th Cir.2013), for example, we affirmed the commitment of a respondent who stole pornographic materials from the treatment lab while incarcerated and collected child pornography while on supervised release. Likewise, in United States v. Wooden, 693 F.3d at 445, the respondent had written a letter to one of his previous victims, and we concluded that he had serious volitional impairment issues. See also United States v. Heyer, 740 F.3d at 287-89, 294-95, 2014 WL 185584, at *2-3, 9 (noting respondent's admission of "ongoing sexual interest in children," including showing child pornography to a teenage boy while on probation).

Even those cases in which the respondent was ultimately found not to qualify for commitment nevertheless involved some evidence of negative behavior during incarceration. In United States v. Hall, 664 F.3d at 464, the district court considered the respondent's ongoing interest in collecting pictures and drawings of children and adolescents while in custody and his report that he often masturbated to memories of his child victims, but it ultimately concluded that due to his abstention from hands-on offenses during his twenty-eight months of release, he was not sexually dangerous under § 4248; we affirmed the judgment. See also United States v. Francis, 686 F.3d 265, 271 (4th Cir.2012) (considering respondent's perceived hostility toward women and his noncompliance with supervised release, but affirming denial of government's commitment petition).

Here, Antone's behavior during the past fourteen years — indeed, during a period of time that spans the majority of his adult life — reveals no acts that conceivably come close to the sort of malfeasance present in our aforementioned precedent.12 On these facts, there is not much more that he could have done to demonstrate that he is in control of his volitional faculties and that such control is likely to persist after his release. The district court should have been aware of the uniqueness of Antone's factual record. As such, it was imperative for the court to comprehensively address why it believed Antone's recent behavior was overshadowed by his past acts. It failed to do so.

In Wooden, we recently confronted a situation in which we believed that the district court had failed to consider relevant and substantial evidence of a respondent's volitional impairment. 693 F.3d at 458-62. There, the district court had rejected the petition for civil commitment, finding that the Government had failed to demonstrate clear and convincing evidence that the respondent would have serious difficulty refraining from re-offending. Our review of the evidentiary record led us to hold otherwise. Because the district court relied on a flawed expert opinion and ignored or otherwise failed to account for a "substantial body of contradictory evidence," we found reversible error. Id. at 461.

Here, as in Wooden, we have again been "left with the definite and firm conviction that a mistake has been committed." Id. The "core" of Antone's case was his decade-long process of rehabilitation. Antone called three separate witnesses to support his position that, as a result of his efforts to obtain treatment, he had improved his ability to control his impulses. The district court's one-sentence dismissal of Antone's case in chief does not sufficiently address the valid and important evidence contained therein.

We hasten to note that it was not clearly erroneous for the district court to place significant weight on Antone's pre-incarceration acts and behavior in reaching its predictive finding. A respondent's criminal record "may well be a historical factor, but it is by no means a stale or irrelevant one. When the question is whether an inmate ... will have serious difficulty refraining from re-offending if released, consideration of the nature of his prior crimes provides a critical part of the answer." Wooden, 693 F.3d at 458. Rather, the deficiency here lies primarily in the Government's failure to muster, and the district court's failure to hold the Government to its obligation to muster, sufficient evidence of an ongoing volitional impairment in this case. The mixed finding that ensues is "against the clear weight of the evidence considered as a whole" and constitutes reversible error.

As both the magistrate judge and Dr. Daum recognized, in analyzing whether a respondent will have serious difficulty refraining from re-offending, one must look to his past and his present condition. Here, Antone has presented significant indicators that he presently "has problems, takes responsibility for them, and seeks help for them," and his pre-incarceration malfeasance cannot be the sole relevant factor of consideration. J.A. 883. We certainly do not fault the Government, as whatever evidence it had, it presented, but that evidence largely (and certainly equally) serves to bolster Antone's asserted rehabilitation and his subsequent capacity for volitional control.

The Government contends that the district court amply justified its conclusion by relying on the testimony and reports of its expert witness Dr. Phenix.13 It is true that Dr. Phenix opined that Antone met the volitional impairment prong, and we are "reluctant to set aside a finding based on the trial court's evaluation of conflicting expert testimony." Hendricks v. Cent. Reserve Life Ins. Co., 39 F.3d 507, 513 (4th Cir.1994). Yet we cannot unreservedly accept the district court's election to give determinative weight to Dr. Phenix's opinions, and for two reasons.

First, Dr. Phenix's conclusion as to whether Antone would have serious difficulty in re-offending was based on multiple diagnoses (including, most importantly, paraphilia NOS, nonconsent) that were subsequently rejected by the magistrate judge and the district court. Indeed, Dr. Phenix testified that Antone's "paraphilic disorder is primarily responsible for so many incidents of nonconsenting sexual activity and child molest[ation]," and that her other two diagnoses, substance dependence and antisocial personality disorder, served as a "contributor" or "permission giver." J.A. 318, 325. But the district court concluded that Antone did not suffer from paraphilia, the primary diagnosis supporting Dr. Phenix's conclusions. The fact that the court rejected Dr. Phenix's paraphilia diagnosis significantly minimizes the amount of persuasive force retained by her opinion as to Antone's volitional impairment.14

Second, and more fundamentally, Dr. Phenix's evaluation of Antone suffers from the same flaw as the conclusion ultimately put forth by the district court. The expert report submitted by Dr. Phenix focuses almost exclusively on events that occurred prior to 1997; indeed, she admitted as much during her testimony. Dr. Phenix explained that her decision to focus on pre-incarcerative acts stemmed from her belief that actions taken while in the outside world are more accurate predictors of future behavior upon release. That is, of course, her choice, but as it relates to our review of the evidentiary record, it will not carry the day. The district court should have at the very least explained why it found Dr. Phenix's unadorned conclusion more persuasive than that of Dr. Daum, who specifically critiqued the former's technique because it did not allow for a respondent's subsequent growth. We find that Dr. Phenix's conclusion on volitional impairment is insufficient to satisfy the Government's heightened clear and convincing evidence burden. Cf. Wooden, 693 F.3d at 457 (finding that the "many deficiencies" in an expert's testimony "leave us firmly and definitely convinced that the district court's factual findings were mistaken.").

The Government next contends that the district court's consideration of Antone's recent behavior was sufficient because it explicitly adopted the magistrate judge's factual findings and credibility determinations related to the lay witness testimony. We reject this argument. Even though the district court acknowledged its awareness of the testimony, that by itself does not indicate to us that it adequately considered its impact. If, after reading the opinion, we cannot understand how the district court came to its conclusions, then we will be unable to perform a cogent analysis on its merits.

Factfinding is "a dynamic, holistic process that presupposes for its legitimacy that the trier of fact will take into account the entire record before it." Taylor v. Maddox, 366 F.3d 992, 1007 (9th Cir.2004). When "the court's account of the evidence is not plausible in light of the record viewed in its entirety," then it is not entitled to deference upon our review. Wooden, 693 F.3d at 460. Here, the district court disposed of more than a decade of evidentiary data points in a single sentence, and we cannot find that it properly took into account all substantial evidence.

Nor can we, on the merits of the matter, find that the Government presented clear and convincing evidence that Antone will have serious difficulty refraining from re-offending if released. The Supreme Court has stated that the serious difficulty element is intended to distinguish the "dangerous sexual offender" from the "dangerous but typical recidivist convicted in an ordinary criminal case who, having been convicted and punished for one crime, proceeds through his own free choice to commit another." Kansas v. Crane, 534 U.S. at 413, 122 S.Ct. 867. Here, then, the Government must demonstrate that Antone's particular manifestation of his mental illnesses are so severe and controlling as to deprive him of his liberty for an indeterminate future.

That is not the case. Clear and convincing evidence equips a factfinder with "a firm belief or conviction, without hesitancy," of the truth of the matter asserted, and, on the record before us, we possess no such conviction about the grip strength of Antone's mental illness on his behavior. Springer, 715 F.3d at 538. We have already cited the substantial evidence in the record indicating that Antone has developed a level of general and social self-regulation; indeed, on these facts, we are hard-pressed to suggest much else that he could possibly do to undercut the notion that he would have serious difficulty in restraining from re-offending. What's more, Antone's civil commitment is based on two mental disorders that are undisputedly prevalent in the nationwide prison population.15 See, e.g., Kansas v. Crane, 534 U.S. at 412, 122 S.Ct. 867; see also Jack Vognsen & Amy Phenix, Antisocial Personality Disorder is Not Enough: A Reply to Sreeivasan, Weinburger, and Garrick, 32 J. Am. Acad. Psychiatry Law 440, 442 (2004) (J.A. 1035-37) (noting that 50 to 70 percent of the ordinary prison population suffers from antisocial personality disorder); Dept. of Justice, Bureau of Justice Statistics, Christopher J. Mumola & Jennifer C. Karberg, Drug Use and Dependence, State and Federal Prisoners, 2004 1 (2004), available at http://www.bjs.gov/content/pub/pdf/dudsfp04.pdf (finding that 45 percent of federal prisoners met DSM-IV criteria for drug dependence or abuse) (last visited January 31, 2014). We conclude that, under the clear and convincing evidence standard, the Government has failed to distinguish Antone's alleged volitional impairment from that of a "dangerous but typical recidivist." Kansas v. Crane, 534 U.S. at 413, 122 S.Ct. 867.

G.

Finally, we turn to the issue of Antone's supervised release. The district court's position seems to be that, if Antone were to contest the terms of his supervised release in front of his sentencing court in Arizona, then the sentencing court would be bound by Ninth Circuit law and hold that his supervised release expired in February 2012. See United States v. Turner, 689 F.3d at 1121 (holding that civil detention under the Walsh Act does not toll supervised release). On appeal, Antone contends that the district court erred as a matter of law because it failed to consider the possibility that he would be judicially estopped from challenging his expressly-agreed-to supervised release in the course of this civil commitment litigation.

As we hold that the evidence in this record is insufficient under the clear and convincing standard to support the district court's predictive judgment of Antone's volitional impairment, we need not wade into this legal issue at this time, nor do we find any reason to do so. Antone has given no indication that he will challenge his status as a supervised releasee under the judgment of the Arizona district court, and it is therefore premature to anticipate that Turner will even be invoked in front of the sentencing judge. Indeed, it is our understanding that the sentencing judge has accepted a recommendation by joint agreement (signed by Antone) that imposes as an additional condition of his supervised release a 365-day term in a halfway house upon his release from federal custody. Whether this modification changes the application of Turner is for the district court in Arizona (or another federal district court should Antone's supervision be transferred) to decide, not this court or the North Carolina district court.16

At oral argument before us, counsel for Antone reported that Antone is currently attending sex offender therapy.17 One can only be encouraged by Antone's commitment to self-improvement, rehabilitation, and recidivism prevention.

III.

For the reasons set forth, we conclude that the appellate record does not support the district court's determination that Antone would have serious difficulty refraining from sexually violent conduct if released. It may be that we would affirm the judgment were the Government's burden one of a mere preponderance, but it is not and we do not. The Government has not established by clear and convincing evidence that the facts and circumstances of this case establish that Antone is a sexually dangerous individual subject to commitment under § 4248. Accordingly, we reverse the judgment of the district court and remand the matter to the district court with instructions to dismiss the petition. The mandate shall issue forthwith.

REVERSED AND REMANDED WITH INSTRUCTIONS; MANDATE TO ISSUE FORTHWITH.