In the

    United States Court of Appeals
                 For the Seventh Circuit
                     ____________________
No. 19-2683
BRUCE CARNEIL WEBSTER,
                                                 Petitioner-Appellee,
                                 v.

T. J. WATSON, Warden,
                                              Respondent-Appellant.
                     ____________________

         Appeal from the United States District Court for the
         Southern District of Indiana, Terre Haute Division.
           No. 2:12-cv-86 — William T. Lawrence, Judge.
                     ____________________

   ARGUED AUGUST 5, 2020 — DECIDED SEPTEMBER 22, 2020
                ____________________

   Before KANNE, HAMILTON, and SCUDDER, Circuit Judges.
    SCUDDER, Circuit Judge. In 1996 the federal district court in
Fort Worth, Texas sentenced Bruce Webster to death for the
murder two years earlier of a 16-year-old girl. Ever since Web-
ster has sought relief from that sentence on the same ground
he advanced at trial—that he is intellectually disabled. His ef-
forts gained traction in 2009, when his lawyers came upon rec-
ords dating to 1994 from the Social Security Administration
showing that three different doctors found him intellectually
2                                                    No. 19-2683

disabled. That development sparked a renewed effort to se-
cure relief in this circuit because Webster is housed in the U.S.
Penitentiary in Terre Haute, Indiana. In 2015, sitting en banc,
we held that Webster was not barred by the limitations im-
posed on successive requests for post-conviction relief from
seeking to show that he is ineligible for the death penalty
based on newly discovered evidence. Webster v. Daniels, 

, 1139–40 (7th Cir. 2015). We remanded to allow the
district court to determine whether the Social Security records
constituted newly discovered evidence—a question turning
on whether the records were “previously existing evidence of
[Webster’s] intellectual disability that counsel did not uncover
despite diligent efforts.”

    Following extensive proceedings on remand, the district
court found that Webster’s defense counsel did not discover
the Social Security records despite reasonable diligence at the
time of trial. From there the district court held a five-day hear-
ing and determined that Webster had carried his burden of
showing by a preponderance of the evidence that he is intel-
lectually disabled. Having taken our own look at the record
evidence, we conclude that the district court’s findings con-
tain no clear error. We therefore affirm the decision to vacate
Webster’s death sentence.
                                I
    Not unlike many capital cases, Bruce Webster’s case has
traveled a long and complex path in multiple courts. Every-
thing began with Webster’s atrocious offense conduct.
    A. The Murder of Lisa Rene
    In the early 1990s Webster ran a marijuana trafficking en-
terprise with others in Pine Bluff, Arkansas. He and his
No. 19-2683                                                   3

associates bought marijuana near Dallas, Texas and trans-
ported it back to Arkansas.
    In September 1994, Webster, Orlando Hall, Demetrius
Hall, and Steven Beckley traveled from Little Rock to the Dal-
las home of Neil Rene, a local drug dealer, seeking reprisal for
a drug deal gone wrong. Neil’s 16-year-old sister, Lisa Rene,
was home alone at the time. When the men knocked, pretend-
ing to be FBI agents, Rene refused to open the door and in-
stead called the police. Before the police could respond, how-
ever, Webster and Demetrius broke through the door and
dragged her into a nearby car. Webster, Demetrius Hall, and
Beckley then drove Rene 330 miles back to Arkansas, repeat-
edly raping her on the way.
    Upon reaching Pine Bluff, the men rented a motel room
where they tied Rene to a chair and took turns raping her sev-
eral more times. Eventually they decided that Rene knew too
much and needed to die. Webster and Orlando Hall then
drove to a nearby park and dug a grave. In the early hours of
September 26, Webster, Orlando Hall, and Beckley took Rene
to the park, beat her with a shovel, stripped her naked, poured
gasoline on her body, and buried her alive.
    Arrests soon followed and resulted in Beckley and Deme-
trius Hall admitting to their roles in the murder and leading
the police to Webster. In time a federal grand jury indicted
Webster on six counts related to the kidnapping and murder
of Lisa Rene. The government sought the death penalty.
   B. Webster’s Trial, Direct Appeal, and First Motion for
      Post-Conviction Relief
  Webster proceeded to a jury trial in federal court in Fort
Worth, Texas in 1996. Demetrius Hall, Beckley, and others
4                                                   No. 19-2683

testified against him. The jury found Webster guilty of kid-
napping in which a death occurred (18 U.S.C. § 1201(a)(1)),
conspiracy to commit kidnapping (18 U.S.C. § 1201(c)), and
using and carrying a firearm during a crime of violence (18
U.S.C. § 924(c)).
    The district court then held a sentencing hearing before
the same jury. Webster’s trial counsel, Larry Moore, argued
that Webster was intellectually disabled—or, in the common
lexicon of the time, mentally retarded—and therefore statuto-
rily ineligible to receive a death sentence. See 18 U.S.C.
§ 3596(c) (“A sentence of death shall not be carried out upon a
person who is mentally retarded.”). Note the timing: Web-
ster’s sentencing occurred before the Supreme Court’s 2002
holding in Atkins v. Virginia that the execution of an intellec-
tually disabled person violates the Eighth Amendment’s pro-
hibition against cruel and unusual punishment. See 

.
    To show that Webster was intellectually disabled, Moore
relied primarily on testimony from three experts: Dr. Ray-
mond Finn, a clinical psychologist; Dr. Dennis Keyes, a pro-
fessor of special education and a certified school psychologist;
and Dr. Robert Fulbright, a clinical neuropsychologist. Each
expert offered opinions based in large part on Webster’s
scores on several intelligence tests (often called Intelligence
Quotient or I.Q. tests) used to diagnose intellectual disability.
Only one test had been administered before Rene’s murder in
1994, however. In 1992, following a report of domestic vio-
lence in the family home, Webster had taken a commonly
used I.Q. test called the Wechsler Adult Intelligence Scale or
the WAIS test. He scored a 48, well below the accepted thresh-
old for intellectual disability. At sentencing the parties agreed
No. 19-2683                                                  5

that the 1992 score may not have been accurate because, at the
time of testing, evidence showed that Webster was confused,
preoccupied, and poorly oriented.
   Dr. Finn testified that he administered the WAIS test to
Webster twice after the crime, in January 1995 and again in
June 1996. Webster scored 59 and 65, indicating that he was
“mildly retarded.” Dr. Keyes also administered a separate be-
havioral test and testified that Webster’s adaptive functioning
was that of a six- to seven-year-old. Along similar lines,
Dr. Fulbright evaluated Webster, found him unable to think
abstractly or understand complex communications, and con-
cluded that he was “mildly to moderately retarded.”
    The government strongly contested that Webster is intel-
lectually disabled. First, the government emphasized Web-
ster’s intellectual and adaptive strengths by presenting evi-
dence from lay witnesses, including police officers, school
teachers, jailers, and a fellow inmate who testified, among
other things, that Webster passed his driving test, was “street
smart,” had the capacity to quote scripture in prison, and
scored in the 43rd percentile on a seventh grade national
achievement test. The witnesses also denied that Webster was
enrolled in special education classes in grade school. One of
the government’s two expert witnesses, Dr. Richard Coons,
found Webster to be well-oriented to time, place, chronology
of events, and people, and concluded that his adaptive func-
tioning was inconsistent with intellectual disability.
    Second, the government argued that Webster had every
incentive to manipulate his intelligence test results—a con-
cept known as malingering—to avoid the death penalty. The
government’s other expert, Dr. George Parker, administered
a truncated WAIS test to Webster, resulting in a full-scale
6                                                   No. 19-2683

score of 72. Webster argued that Dr. Parker omitted important
subtests that would have lowered the final score. The bottom
line for Dr. Parker, though, was that Webster did not exhibit
an intellectual disability. He agreed with the government that
motivation can modify test results, particularly in the “foren-
sic” setting.
    The sentencing hearing concluded with a majority of ju-
rors finding that Webster was not intellectually disabled. Be-
cause the operative sentencing statute, 18 U.S.C. § 3596(c),
does not specify whether a judge or jury should make this de-
termination, the district court entered its own finding that
Webster was not intellectually disabled and sentenced him to
death. See United States v. Webster, 

, 352 (5th Cir.
1998) (rejecting Webster’s challenge to the court’s separate
finding because “[t]he statutory scheme simply does not an-
swer who decides this issue”). The Fifth Circuit affirmed the
sentence on direct review, and the Supreme Court denied cer-
tiorari. See Webster v. United States, 

(1999).
     In 2000 Webster invoked 28 U.S.C. § 2255 and began his
pursuit of post-conviction relief. Recognizing the potential
impact of the Supreme Court’s intervening decision in Atkins,
Webster amended his petition in 2002. The district court re-
jected all 16 of Webster’s grounds for relief, but granted a cer-
tificate of appealability on two issues—whether “the evidence
presented at trial was insufficient to warrant the district
court’s finding that Webster is not mentally retarded” and
whether Webster’s “alleged retardation renders him ineligible
for a death sentence.” United States v. Webster, 

,
310 (5th Cir. 2005). The Fifth Circuit denied the § 2255 motion,
reasoning that Webster’s petition was “nothing more than an
attempt to re-litigate the question” of his mental capacity.

at 314. The Supreme Court again denied review. See Webster
v. United States, 

(2006).
   C. Discovery of New Evidence
    A significant development occurred in 2009 that ulti-
mately paved the way to this appeal. One year earlier, the law
firm of Dorsey & Whitney took over from Webster’s original
post-conviction counsel. Dorsey & Whitney soon uncovered
evidence that revived Webster’s efforts to vacate his death
sentence: Social Security Administration records dating to
1993—before the murder of Lisa Rene—showing that three
doctors had diagnosed Webster as “mentally retarded.” Be-
cause these records predated the murder, Webster’s new legal
team saw the records as exceptionally important, as the evi-
dence would be immune from attack on the ground that Web-
ster manipulated the diagnosis by malingering.
   The efforts Webster’s attorney Larry Moore made to obtain
these Social Security records before trial became a major issue
in Webster’s pursuit of relief from his death sentence. For
now, though, what is important is understanding the infor-
mation revealed in the records.
    The Social Security records showed that Webster, with his
mother’s help, had applied for disability benefits in 1993 on
the basis of headaches and sinus issues. In the course of eval-
uating his application, two psychologists and one physician
examined Webster’s intellectual capacity. The records con-
tained these findings and opinions:
8                                                  No. 19-2683

       Dr. Charles Spellmann, a clinical psychologist, found
        that Webster’s “ideation was sparse,” he could not reg-
        ister and repeat three objects or complete simple calcu-
        lations, and he did not understand basic expressions.
        Dr. Spellmann observed that Webster’s functioning
        seemed limited as he was “idle around the house and
        on the streets.” Dr. Spellmann estimated that Webster’s
        I.Q. was around 69 and diagnosed him with “mental
        retardation” and “antisocial personality by history.”
       A second psychologist, Dr. Edward Hackett, adminis-
        tered the WAIS intelligence test to Webster and calcu-
        lated a score of 59. Dr. Hackett also diagnosed Webster
        with “mild mental retardation” and “antisocial person-
        ality disorder.”
       A general physician, Dr. C.M. Rittelmeyer, gave Web-
        ster a standard physical examination. In reporting his
        findings, Dr. Rittelmeyer diagnosed Webster with
        “mental retardation.”
    In the course of reviewing the Social Security records, the
Dorsey & Whitney team also came upon a letter from Web-
ster’s grade school, dated November 8, 1993, referring to rec-
ords regarding special education classes. The letter bore the
signature of the Special Education Supervisor of Watson
Chapel Schools and stated (with express reference to Webster)
that “[t]he above student’s Special Education records were de-
stroyed in 1988.” The letter then added that “[a] letter was
mailed to parents at the last known address, telling them they
could have the records if they wanted them. There was no re-
sponse to the letter.” Dorsey & Whitney believed that Web-
ster’s enrollment in special education classes would help es-
tablish an intellectual disability.
No. 19-2683                                                   9

   D. Webster’s Renewed Efforts to Vacate His Death Sen-
      tence
    Relying on the findings of intellectual disability embodied
in the newly obtained Social Security records, and pointing
again to the Supreme Court’s decision in Atkins, Webster re-
turned to federal court with renewed requests to vacate his
death sentence. He first asked for permission to file a new mo-
tion for post-conviction relief under 28 U.S.C. § 2255. The
Fifth Circuit denied the request, determining that Webster
failed to meet the stringent limitations Congress imposed on
second or successive motions for post-conviction relief. See 28
U.S.C. § 2255(h) (permitting a second or successive motion
only if it relies on either (1) newly discovered evidence show-
ing by clear and convincing evidence that no reasonable jury
would have found the applicant guilty of the offense, or (2) a
new rule of constitutional law previously unavailable that the
Supreme Court has made retroactive to cases on collateral re-
view). Applying § 2255(h), the Fifth Circuit reasoned that not
only was Atkins available to Webster at the time of his first
amended § 2255 motion, but, even more, the Social Security
records in no way undermined the jury’s finding of guilt for
the kidnapping resulting in death offense for which he re-
ceived a capital sentence. See In re: Webster, 

, 258
(5th Cir. 2010).
   Unable to pursue relief under § 2255, Webster turned to 28
U.S.C. § 2241, once again relied on Atkins, and sought anew to
vacate his sentence in the Southern District of Indiana—the
location of the U.S. Penitentiary in Terre Haute, where he was
housed on death row. See Rumsfeld v. Padilla, 

, 447
(2004) (explaining that a § 2241 petitioner “should name his
10                                                  No. 19-2683

warden as respondent and file the petition in the district of
confinement”).
    Whether Webster could pursue relief under § 2241 was it-
self a vigorously contested issue, one that touched off its own
round of litigation that culminated in our 2015 en banc deci-
sion answering in the affirmative. See 

. The issue turned on whether Webster’s Atkins claim fit
within the so-called savings clause of § 2255(e)—a question of
substantial complexity. In the end, we held that “there is no
such absolute bar to the use of the safety valve found in sec-
tion 2255(e) for new evidence that would demonstrate cate-
gorical ineligibility for the death penalty [under the Eighth
Amendment].”

legal question in Webster’s favor, we re-
manded to allow the district court to determine whether he
could meet the “new evidence” standard. We provided clear
and precise direction for the remand proceedings and that in-
struction frames much of our review in this new appeal:
      The district court must hold a hearing for the
      purpose of deciding whether the Social Security
      records were indeed unavailable to Webster and
      his counsel at the time of the trial. In considering
      that question, the district court must also evalu-
      ate trial counsel’s diligence. If the court con-
      cludes, as Webster’s lawyers urge, that the rec-
      ords were unavailable and all due diligence was
      exercised, and that Webster has them now only
      because of a fluke, then it must turn to the mer-
      its of the petition: is Webster so intellectually
      disabled that he is categorically ineligible for the
      death penalty under Atkins and Hall?
No. 19-2683                                                   11

       Both the government and Webster would be en-
       titled at a hearing to offer evidence relevant to
       that determination; if Webster is not limited to
       the record before the original trial court, there is
       no reason to impose such a limit on the govern-
       ment. Webster, as the petitioner, bears the bur-
       den of proving intellectual disability. If he gets
       this far, what this will be as a practical matter is
       a partial new sentencing hearing. At this later
       stage, the proper burden of proof is a prepon-
       derance of the evidence, not “clear and convinc-
       ing” evidence.

   So Webster’s case returned to the district court in Indiana.
                                II
   A. The District Court Proceedings on Trial Counsel’s
      Due Diligence
    In June 2018 the district court held a hearing to determine
whether Webster’s trial team was duly diligent in seeking the
Social Security records that Dorsey & Whitney later received
in 2009. The court heard testimony from two witnesses: lead
trial counsel Larry Moore and Kristen LeRoux, one of the
Dorsey & Whitney paralegals who helped to retrieve the So-
cial Security records.
    Moore testified that he had extensive experience as a pros-
ecutor and a defense attorney by the time the trial court ap-
pointed him to represent Webster in 1994. Indeed, Moore had
participated in six previous capital cases—three as a prosecu-
tor and three as defense counsel.
12                                                 No. 19-2683

    Moore explained that his strategy in Webster’s case fo-
cused on sentencing, as the government had overwhelming
evidence of Webster’s role in kidnapping and murdering Lisa
Rene. So Moore quickly determined that Webster’s intellec-
tual capacity would be the “single biggest issue” at trial and
devoted considerable energy to pursuing an intellectual disa-
bility defense to the death penalty. Moore testified that he had
Webster evaluated by five experts and “sought out every
medical record that [he] could find, every school record that
[he] could find, [and] every person that [he] could find that
had known Mr. Webster growing up.”
    Moore added that he and his team traveled to Little Rock
before trial to interview family and other potential witnesses
about Webster’s intellectual ability. Webster’s mother told
Moore that sometime in the early 1990’s (and thus before the
murder of Lisa Rene) she had applied for disability benefits
with the Social Security Administration on her son’s behalf.
Upon learning this information, and drawing on his experi-
ence in other cases, Moore thought the Social Security Admin-
istration may have records shedding light on Webster’s intel-
lectual functioning. He testified that he directed his paralegal
Kim Whitehead to request records from the agency.
    According to Moore, Whitehead then contacted Hal West
of the Social Security office in Pine Bluff, the Little Rock sub-
urb where Webster lived before his arrest. Whitehead learned
that to obtain Social Security records, she would have to sub-
mit a formal Business Records Authorization request accom-
panied by a release executed by Webster. Moore testified that
Whitehead followed those steps and advised West that the de-
fense team’s investigator J.W. Strickland would visit the Pine
Bluff office in person to pick up any records.
No. 19-2683                                                 13

    Webster introduced several documents to corroborate
Moore’s testimony. The exhibits included two handwritten
notes between Moore and Whitehead from his trial file. In one
note, written after Moore met with Webster’s mother, he di-
rected Whitehead to “call [the Pine Bluff] office to find out
what we need to get copies of those records.” Moore ended
the note by underscoring, “I need test results!” Moore’s sec-
ond note emphasized that the team “need[ed] to know how
to get Bruce’s records re: Appl. For disability benefits (any
testing?).”
    Webster further introduced a faxed letter from Whitehead
to West at the Pine Bluff Social Security office. The letter was
dated February 29, 1996 and bore a transmittal legend of
March 4. Signed by Whitehead, the letter stated that “[a]s you
requested I have enclosed a copy of a Business Records Au-
thorization and Release,” and “[i]t is my understanding that
Mr. Webster applied for disability in 1992 and some testing
was done.” Attached to the letter (and included with the fax)
was a records release signed by Webster, with Whitehead’s
letter further indicating that investigator J.W. Strickland
would visit Pine Bluff from March 1–5, 1996 to pick up any
records and to deliver the original release form.
   Moore also recalled that Strickland had in fact visited Pine
Bluff. Moore testified that he remembered receiving a call
from Strickland during jury selection, with Strickland telling
him that the Pine Bluff Social Security office reported having
no records on Webster. Moore added that he followed up by
personally calling the office in Pine Bluff to confirm that “they
didn’t have any record regarding Mr. Webster’s application
for benefits.” At that point, Moore continued, he took the
14                                                 No. 19-2683

Social Security office at its word and did not think there was
anything else he could do.
    At the hearing on diligence, Webster also called Dorsey &
Whitney paralegal Kristen LeRoux, who testified about how
the firm came to discover the existence of Social Security rec-
ords after all. According to LeRoux, Dorsey & Whitney
started representing Webster in 2008. Drawing upon a dili-
gence checklist used in a different case, a group of paralegals
(including LeRoux and Katherine Slaikeu) asked Webster to
sign a blanket records release form. They then attached the
signed release to records requests to multiple institutions, in-
cluding the Social Security office in Pine Bluff. When the
agency failed to respond, paralegal Slaikeu followed up by
phone. Only then, in February 2009, did the firm receive a
packet of records from the Pine Bluff office. Upon reviewing
the records, LeRoux immediately noticed determinations by
three doctors that Webster was “mentally retarded” and con-
veyed this information to attorneys working on the case.
    LeRoux also testified that the Social Security records con-
tained a list of exhibits purportedly included in the package
provided by the agency. Some of the referenced documents
were missing, though, so she submitted a renewed records re-
quest in October 2009. In a subsequent telephone discussion,
a member of the Pine Bluff office told LeRoux to contact the
Social Security office in Terre Haute, where Webster was de-
tained, with any further records requests. When LeRoux did
so, a representative from Terre Haute informed her, by phone
and then by letter, that the office could not fulfill her request
because Webster’s file had been “destroyed.”
  The record sheds no light on what Social Security docu-
ments were destroyed. We have come across no information
No. 19-2683                                                   15

(such as an affidavit from a records custodian) from the
agency itself on the topic. It troubles us that the agency re-
ported destroying documents in 2010, at a time when Web-
ster’s new legal team had received records dating to 1993
showing the prior findings of mental retardation. That devel-
opment alone made it predictable with certainty that Webster
would embark on a new effort to obtain relief from his death
sentence. The stakes for Webster—indeed for all—are far too
high for records to be destroyed while someone’s life is on the
line.
    In the district court, the government cross-examined Larry
Moore and LeRoux and introduced several exhibits designed
to show that Moore’s account of his diligence in trying to ob-
tain any Social Security records before Webster’s trial lacked
credibility. These exhibits included three affidavits Moore
prepared in the years since Webster’s trial—one in 2000, a sec-
ond in 2009, and a third in 2018 (in advance of the hearing on
remand in the district court). The government compared and
contrasted the information in Moore’s affidavits—focusing
specifically on his enhanced and highly particular recollection
in 2018—to discredit his account.
    On this score, the details become important. Moore pre-
pared and submitted the first affidavit, dated September 26,
2000, as part of Webster’s initial § 2255 motion in the Northern
District of Texas. The 2000 Affidavit describes the mitigation
investigation undertaken by Larry Moore prior to Webster’s
trial and focuses on the issues that arose with a particular mit-
igation expert appointed to assist with capital sentencing. The
2000 Affidavit makes no mention of Moore’s efforts to obtain
Webster’s Social Security records.
16                                                   No. 19-2683

    Moore prepared his second affidavit (actually a declara-
tion) in October 2009, shortly after Dorsey & Whitney had re-
ceived Webster’s Social Security records. In the 2009 Affidavit,
Moore described how he thought Webster’s “mental retarda-
tion was going to be a critical issue” at trial, so he “made every
effort” to secure evidence related to that issue. Moore ex-
plained that his team contacted the Social Security office in
Pine Bluff, requested records, and arranged for investigator
Strickland to retrieve them in March 1996. Moore then added
that, “While I do not currently have any direct recollection of
the response which we received [from the Social Security Ad-
ministration], it is my good faith belief that we must have
been told that no records regarding Mr. Webster existed, oth-
erwise we would have continued to pursue them by every
means available to us.” Moore further stated that he prepared
the 2009 Affidavit based on his best recollection of past events
but without the benefit of an opportunity to review his case
file from trial.
    Moore authored the third affidavit in April 2018, following
our 2015 en banc decision and in connection with the remand
proceedings in the district court. He did so after reviewing
portions of his trial file, which he received from the Dorsey &
Whitney team. Explaining that the case file had refreshed his
memory, Moore updated and expanded upon his original ac-
count of the records request to the Pine Bluff Social Security
office before trial. More specifically, he stated in the 2018 Af-
fidavit that he spoke to a representative in the Pine Bluff office
before trial and was told “there were no records in existence.”
    At the hearing in the district court, Moore emphasized the
importance of receiving (portions of) his case file to refreshing
his memory:
No. 19-2683                                                   17

   Q. How did [the case file] refresh your recollection?
   A. Well, in a number of ways. I had not looked at that file
   since I gave it up in the late 1990’s, and it was able to help
   me in terms of various dates and the things that we were
   doing. Like I did not recall that we had begin—that we had
   begun jury selection the first week of March. And by re-
   viewing the file, I was able to determine that’s exactly
   when it all occurred.
Reviewing his file proved particularly useful, Moore contin-
ued, because he had tried “[p]robably more than a hundred”
cases since he handled Webster’s case in 1996.
    The government urged the district court to find that
Moore’s enhanced and changed account of events dating to
1995 and 1996 lacked credibility. It was implausible, the gov-
ernment argued, that Moore—some 25 years later—would
come to have such a detailed recollection of past events that
conveniently aligned with Webster’s renewed request for re-
lief.
    After considering the evidence, the district court found
that Webster had carried his burden of showing that the Social
Security records obtained by Dorsey & Whitney in 2009 were
unavailable at the time of trial despite diligent efforts by Larry
Moore to obtain those records. Several factors weighed heav-
ily in the district court’s reasoning:
    First, the district court found Larry Moore’s “testimony to
be credible” after “observing [his] demeanor” during the
hearing. Webster v. Lockett, No. 2:12-cv-86, 

,
at *6 (S.D. Ind. Aug. 31, 2018). Moore is an accomplished at-
torney who, at the time of Webster’s trial, had substantial ex-
perience in criminal cases, including capital cases, on both the
18                                                 No. 19-2683

prosecution and defense side. It made sound sense, the dis-
trict court underscored, that a seasoned criminal lawyer like
Moore would have recognized the potential importance of
records from the Social Security Administration and sought
to obtain them in the ways he described. So, too, did it make
good sense, the district court continued, that Moore would
have avoided expending additional resources to obtain Social
Security records upon learning from investigator J.W. Strick-
land that a representative of the agency’s Pine Bluff office
stated that no such records existed. The district court likewise
found it credible that Moore took the added step of calling the
Pine Bluff office and receiving the same information.
   Second, the district court found that the documentary evi-
dence introduced at the hearing reinforced Moore’s account.
The court included in its order scanned copies of the two
handwritten notes from Moore as well as the faxed letter from
Whitehead to West. This contemporaneous documentary evi-
dence, in the court’s view, aligned fully with Moore’s testi-
mony of the efforts he made before trial to obtain the Social
Security records.
    Third, the district court found it important that Moore, be-
fore preparing his 2018 Affidavit and testifying at the hearing,
had received and been able to review his trial file. This oppor-
tunity was unavailable to Moore for many years and allowed
him to refresh his memory by looking at specific documents,
including his own handwritten notes.
    Having decided that Webster carried his burden on the
threshold issue of diligence, the district court scheduled a
hearing to consider the merits of his contention that he is in-
tellectually disabled and therefore ineligible for the death
penalty.
No. 19-2683                                                   19

   B. The District Court Proceedings on Intellectual Disa-
      bility
    In April 2019, after an additional discovery period, the dis-
trict court held a week-long hearing on the merits of Webster’s
Atkins claim. The court considered the transcripts and other
evidence from Webster’s trial while also, and in accord with
our remand order, not limiting either party to the trial record
alone. Webster and the government both presented new evi-
dence on intellectual disability.
    Everyone agreed that Webster shouldered the burden of
proving intellectual disability by a preponderance of evi-
dence. He had to show three “core elements,” each borrowed
from the medical community’s understanding of intellectual
disability. Moore v. Texas, 

, 1045 (2017). As the
district court recognized, Webster had to show a deficit in in-
tellectual functioning (indicated by an adjusted-for-error I.Q.
score of roughly 70 or below), a deficit in adaptive functioning
(indicated by sufficient impairment without support in either
the conceptual, social, or practical domains), and the onset of
both deficits during his developmental years. See 

n.3; Hall v. Florida, 

, 710 (2014).
    Webster called three expert witnesses—Dr. Mark Tassé,
Dr. Daniel Reschly, and Dr. John Fabian. The government re-
lied on the testimony of Dr. Robert Denney and of two Bureau
of Prisons employees who have interacted with Webster dur-
ing his period of incarceration in Terre Haute. The district
court also reviewed trial testimony from Webster’s friends
and family and documentary evidence.
   The district court heard first from Dr. Mark Tassé, a psy-
chologist who co-authored the clinical definition of
20                                                  No. 19-2683

intellectual disability promulgated by the American Associa-
tion on Intellectual and Developmental Disabilities. Dr. Tassé
did not examine Webster or opine on the question of intellec-
tual disability. Rather, Webster called Dr. Tassé for the limited
purpose of educating the district court on the criteria requisite
to establish intellectual disability.
    Beyond describing the three primary elements of intellec-
tual disability, Dr. Tassé stressed that a disability diagnosis
must be the product of clinical judgment, not of a lay person’s
observations of someone else. He also emphasized that peo-
ple with intellectual disability can exhibit strengths in partic-
ular areas, explaining that it is not unusual for someone intel-
lectually disabled to be able to read, write, drive, or work.
    Webster then called the two experts who evaluated him
following our 2015 remand and concluded he is intellectually
disabled. First up was Dr. Reschly, a retired professor of edu-
cation and psychology at Vanderbilt University, who evalu-
ated Webster in January 2019. Webster scored below the sixth-
grade level on each test that Dr. Reschly administered, save
for one subtest. Dr. Reschly concluded that Webster’s scores
“are consistent with intellectual disability.” He also evaluated
language performance, concluding that Webster’s expressive
and receptive language was very low,—“to put it bluntly,
[Webster] talks a lot; but there’s little structure or content in
his language expression.” Dr. Reschly noted that Dr. Ful-
bright, a defense expert who testified at Webster’s trial in
1996, “made virtually identical observations about Mr. Web-
ster’s language.”
    Dr. Reschly also conveyed information he received during
interviews of friends and family who knew Webster before he
entered prison. These individuals described practical ways in
No. 19-2683                                                 21

which Webster exhibited limitations—for example, being un-
able to track the plot of television programs, to follow jokes,
or to distinguish between spades and clubs in card games.
Portions of Webster’s Social Security application also stood
out to Dr. Reschly. For example, in response to a question
whether the candidate contributes around the house, Webster
wrote “make my room.” The application then asks follow-up
questions like do you need help with chores and how much
time do you spend on chores, Webster twice wrote “I don’t
know” and then added “I told you I make my room.”
    Dr. Reschly opined that Webster’s intellectual functioning
is “significantly subaverage,” he “displayed deficits in the
three domains of adaptive behavior,” and these “significant
limitations in intellectual functioning [and] in adaptive be-
havior emerged during the developmental period.” He ulti-
mately opined that “Webster is a person with intellectual dis-
ability and should be so regarded.”
    The district court then heard from Dr. Fabian, a clinical
psychologist with experience evaluating inmates’ compe-
tency, including in 70 to 80 Atkins cases. Dr. Fabian explained
that his opinions in past cases have been mixed, finding about
half of the inmates intellectually disabled and the other half
not. Based on his evaluation of Webster and analysis of Web-
ster’s history and records, Dr. Fabian found that Webster suf-
fers from serious intellectual and adaptive deficits. Foremost
emphasizing Webster’s many low I.Q. scores over many
years, Dr. Fabian then drew on observations from his exami-
nation, highlighting, for example, that Webster’s intellectual
and adaptive shortcomings manifested themselves when he
struggled to answer simple questions about daily life like
“why would you want to tell a doctor that you’re taking
22                                                 No. 19-2683

medication from another doctor,” or “why would you not
want to use a blow dryer in a bathtub?” Nor was it surprising
that Webster was unable to group photos of six people into
smaller groups based on their place of birth or job even after
Dr. Fabian explained the task to him a few times. Dr. Fabian
also testified that Webster’s exhibiting certain strengths while
incarcerated did not change his conclusion because the prison
environment is not conducive to sound analysis of adaptive
deficits. The totality of these findings led Dr. Fabian to con-
clude that Webster is intellectually disabled.
    Drs. Reschly and Fabian both noted that the severe emo-
tional, physical, and sexual abuse Webster suffered through-
out childhood is a risk factor toward the development of in-
tellectual disability. Both experts likewise reviewed (and the
district court considered) other evidence, including the Social
Security records retrieved by Dorsey & Whitney in 2009.
Those records showed that all three agency doctors who eval-
uated Webster in connection with his Social Security applica-
tion found him mentally retarded.
    Drs. Reschly and Fabian also placed emphasis on the spe-
cial education letter dating to November 1993 and included in
the Social Security file. Dr. Reschly further noted that Webster
“readily admitted that he was in special education,” and that
it would be odd for the parents of someone not in special ed-
ucation to receive such a letter. The experts likewise consid-
ered the fact that Webster failed the first grade, with Dr. Fa-
bian finding this evidence “critical,” as first grade is “the
foundation” for skills like “reading, writing, [and] spelling.”
    For its part, the government presented testimony from
clinical psychologist Dr. Robert Denney, who evaluated Web-
ster for 11-plus hours over two days following our 2015
No. 19-2683                                                  23

remand. He testified that “none” of the other cognitive testing
Webster completed is “a valid reflection of his genuine cogni-
tive abilities,” as each test result was the product of substan-
tial malingering by Webster. Malingering must have been pre-
sent, in Dr. Denney’s view, because Webster’s real-world
functioning was “strikingly inconsistent” with all prior test
results. A person with scores similar to Webster’s would not
be able to play the guitar or drive a car—“it just can’t hap-
pen.” Dr. Denney concluded that “life choices of criminal mis-
conduct” and “maladaptive personality style” describe Web-
ster—factors “not related to intellectual disability.”
    The government also called several lay witnesses. Dr. Erin
Conner, a licensed psychologist working at Terre Haute, has
observed Webster (and other death row inmates) on a
monthly basis for three years. Without offering any expert
opinion (or even being disclosed or qualified as an expert wit-
ness), Dr. Conner testified that she has never “observed any-
thing that would cause [her] concern about Mr. Webster’s in-
tellectual functioning.” On cross-examination, Dr. Conner
acknowledged that her interactions with Webster ranged
from about 30 seconds in length to 10 minutes at most. The
district court also admitted into evidence and reviewed the
deposition testimony of Dr. Jacqueline Blessinger, who met
with Webster as the lead psychologist of the Special Confine-
ment Unit at Terre Haute. Dr. Blessinger likewise testified as
a lay witness, stating that she was impressed by Webster’s ca-
pacity to express frustration, control his emotions, and speak
appropriately. Dr. Blessinger stated that she based her views
on spending about one hour with Webster over the course of
approximately ten months, but concluded that “throughout
[her] observations, there was nothing to clinically indicate a
need for further testing.”
24                                                    No. 19-2683

   Finally, the government relied on trial testimony from
Webster’s ex-girlfriends, who described him as sweet, fun to
be around, and someone able to do the “normal things cou-
ples do.” The government also highlighted certain adaptive
functioning strengths by drawing upon Dr. Reschly’s testi-
mony that Webster “certainly mastered the basic fundamental
daily living skills of eating, toileting, dressing, [and] things of
that nature,” and the fact that he played guitar in prison and
passed a driving test.
    In the end, the district court found that Webster met all
three criteria to qualify as intellectually disabled. See Webster
v. Lockett, No. 2:12-cv-86-WTL-MJD, 

, at *11
(S.D. Ind. June 18, 2019). The court rooted its finding in the
three criteria for intellectual disability.
        Onset Before Age 18. The government did not dispute
         (nor did the district court find any reason to question)
         that Webster’s deficits arose prior to age 18.
        Intellectual Functioning. The district court began with
         Webster’s intellectual functioning, giving weight to his
         many low I.Q. test scores:
No. 19-2683                                             25




    Year      Psychologist              Score
    1992 Dr. Patrick Caffey (Ar-      48 (Full scale)
         kansas Department of        56 (Verbal)
         Mental Health)              48 (Performance)
    1993 Dr. Edward Hackett (So-     59 (Full scale)
         cial Security Administra-   71 (Verbal)
         tion)                       49 (Performance)
    1995 Dr. Raymond Finn (Web-      59 (Full scale)
         ster trial expert)          59 (Verbal)
                                     60 (Performance)
    1995 Dr. Dennis Keyes (Web-      55 (Composite)
         ster trial expert)          67 (Crystalized)
                                     46 (Fluid)
    1996 Dr. Raymond Finn (Web-      65 (Full scale)
         ster trial expert)          72 (Verbal)
                                     59 (Performance)
    1996 Dr. Dennis Keyes (Web-      51 (Full scale)
         ster trial expert)          N/A (Verbal)
                                     N/A (Performance)
    1996 Dr. George Parker (Gov-     72 (Full scale)
         ernment trial expert)       77 (Verbal)
                                     67 (Performance)
    2018 Dr. Robert Denney (Gov-     61 (Full scale)
         ernment expert)             63 (Verbal)
                                     69 (Performance)
    2018 Dr. Daniel Reschly (Web-    53 (Full scale)
         ster expert)                N/A (Verbal)
                                     N/A (Performance)
26                                                     No. 19-2683

         In emphasizing these low I.Q. scores, the district court
         determined that the evidence did not support a finding
         of malingering, choosing instead to credit Dr. Re-
         schly’s testimony that it would be “extremely difficult”
         to manipulate so many test scores over the course of 27
         years. Webster, 

, at *7. Indeed, the dis-
         trict court found that Dr. Denney’s testimony—on the
         specific point of malingering and even more gener-
         ally—lacked credibility, giving it little or no weight. In
         the district court’s view Dr. Denney “ignore[d], disre-
         gard[ed], or minimize[d] evidence that called his con-
         clusions into question.”

n.13. For example, the
         article Dr. Denney cited for the proposition that malin-
         gering is easy actually reached the conclusion that
         “faking low on the WAIS-R is a difficult endeavor.”

also highlighted that Webster, in 1993
         and thus before the murder of Lisa Rene, sought bene-
         fits from the Social Security Administration due to si-
         nus symptoms and not due to any intellectual disabil-
         ity. That three separate agency doctors diagnosed
         Webster as mentally retarded at a time when he had no
         incentive to manipulate his I.Q. test results only further
         undermined Dr. Denney’s conclusion that Webster had
         exhibited malingering during prior tests.
        Adaptive Functioning. Accepting Dr. Fabian’s conten-
         tion that the prison environment is too structured to
         yield the most accurate information on adaptive abili-
         ties, the district court focused on evidence from Web-
         ster’s life outside of prison. The Social Security records,
         the district court reasoned, showed Webster is “barely
         literate.” While the government pointed to evidence
         about Webster’s strengths, including his musical
No. 19-2683                                                      27

       ability, hygiene, and conversational ability, the district
       court focused its adaptive functioning analysis on def-
       icits rather than strengths. The district court noted that
       Dr. Fabian and Dr. Reschly spent substantial time with
       Webster, with family and friends, and reviewing the
       school and Social Security records. On the other hand,
       the court gave little weight to the lay testimony of
       Dr. Conner and Dr. Blessinger because “[n]either the
       quantity nor depth of their interactions with Webster
       would lead the Court to find their testimony to be more
       persuasive than that of Drs. Fabian and Reschly.”
       Weighing the evidence, the district court found that
       Webster has adaptive functioning deficits.
The government now appeals.
                                III
    The government challenges the district court’s findings on
defense counsel’s diligence in pursuit of the Social Security
records before trial, and, in turn, Webster’s intellectual disa-
bility. The same clear error standard applies to both. We can
reverse only if we are “left with the definite and firm convic-
tion that a mistake has been committed.” Anderson v. Bessemer
City, 

, 573 (1985); see also FED. R. CIV. P. 52(a)(6).
A reviewing court may not set aside factual findings “simply
because it is convinced that it would have decided the case
differently.” 

. Rather, reversal is war-
ranted only if the district court’s finding was “so internally
inconsistent or implausible on its face that a reasonable fact-
finder would not credit it.”

see also Greyer v. Illinois
Dep’t of Corr., 

(7th Cir. 2019) (reversing a finding
of fraud because the district court never explained its reason-
ing and the record contained directly contradictory evidence).
28                                                  No. 19-2683

    Credibility determinations are entitled to “even greater
deference” because “only the trial judge can be aware of the
variations in demeanor and tone of voice.” 

; see also United States v. Austin, 

, 431 (7th
Cir. 2015) (noting that credibility findings “can virtually never
be clear error” (quoting United States v. Longstreet, 

, 837 (7th Cir. 2012))).
     A. Diligence of Webster’s Trial Counsel Larry Moore
    We begin with the government’s challenge to the finding
that Larry Moore was diligent in his pretrial pursuit of the So-
cial Security records ultimately received in 2009 by Dorsey &
Whitney. Due diligence, we have explained, means “reasona-
ble diligence,” not “the maximum feasible diligence.” Moore
v. Knight, 

, 940 (7th Cir. 2004). “[T]he duty to in-
vestigate does not force defense lawyers to scour the globe on
the off chance something will turn up; reasonably diligent
counsel may draw a line when they have good reason to think
further investigation would be a waste.” Rompilla v. Beard, 

, 383 (2005) (citing Wiggins v. Smith, 

, 525
(2003)). Where counsel travels a path of particular inquiry and
encounters information suggesting that “further investigation
would [be] fruitless,” limiting the investigation is reasonable.

. To proceed with his § 2241 petition,
Webster had to show by clear and convincing evidence that
Moore exercised reasonable diligence. See 

.
   During the 2018 hearing, Webster presented the district
court with meaningful evidence of Moore’s pretrial efforts
back in 1996. The most compelling evidence came from
Moore’s own pen—from notes in his trial file. Webster pre-
sented the district court with contemporaneous handwritten
No. 19-2683                                                 29

notes from Moore to his paralegal Kimberly Whitehead.
Those notes—the authenticity of which nobody has ques-
tioned—show that Moore recognized the importance of any
Social Security records and directed Whitehead before trial be-
gan to request any such records from the agency’s Pine Bluff
office. Moore’s trial file also contained, and the district court
reviewed, the letter Whitehead faxed to the agency requesting
any records they might have on Webster.
    At the hearing, Moore then walked the district court step-
by-step through the process he and Whitehead went through
before Strickland traveled to Pine Bluff to pick up the Social
Security records. Upon being told by the Social Security office
that no records existed, Moore ended the inquiry and turned
his attention to other aspects of Webster’s defense. Against
this evidentiary showing, the district court concluded that
Moore was duly diligent, and we see no clear error in that
finding.
    The government begs to differ, insisting that Moore’s ac-
count—root and branch—is implausible. What most troubles
the government is that Moore’s account of his own diligence
has grown in clarity and detail despite the passage of substan-
tial time—a result at odds with the workings of human
memory, at least as the government would have it. But the
district court heard and considered the government’s position
and responded with findings that are plenty reasonable and
reflect no clear error.
    The government points first to Moore’s improved recollec-
tion between the 2009 Affidavit and the 2018 Affidavit he sub-
mitted to the district court. In 2009 Moore stated that it was
his “good faith belief that we must have been told that no rec-
ords regarding Mr. Webster existed,” but in 2018 Moore
30                                                    No. 19-2683

stated that “I am certain that I was personally told that such
records did not exist, although I cannot actually recall
whether this was via phone or in person.” The government
also highlights that Moore’s inconsistencies continued even
after he spent meaningful time with Webster’s trial file. In his
2018 Affidavit, Moore could not recall whether the Social Se-
curity employee told him no records existed “via phone or in
person.” During the hearing on diligence later that year, how-
ever, Moore testified definitively that the conversation was
via phone. In the government’s view, this detail came out of
thin air.
    To be sure, Moore added to his account during the remand
proceedings. But that reality was not lost on the district court.
To the contrary, the district court listened to Moore’s account
of the newfound detail, observed the government’s cross ex-
amination, and chose in the end to credit Moore’s explana-
tion. As the district court emphasized, Moore explained that
when he executed the 2009 Affidavit he had not yet had the
opportunity to review his file. In contrast, before executing
the 2018 Affidavit, Moore noted that he “spent hours review-
ing relevant portions of [his] trial file and the trial transcript,”
and “consider[ed] this matter in depth.” Recall that between
Webster’s 1996 trial and the 2018 hearing on diligence, Moore
estimated that he tried “[p]robably more than a hundred”
cases. A number of these cases were death penalty cases, like
Webster’s. The district court plausibly reasoned that the lines
between the many cases Moore tried could blur over time,
and that reviewing the Webster case file (indeed reviewing his
own handwritten notes) could provide needed clarity. See
Webster, 

, at *5 n.2.
No. 19-2683                                                     31

    We also see no clear error in the district court’s finding that
“Moore’s additional review of the file … further helped
Moore recall the events, as he remembered that his direct con-
tact with the [Social Security Administration] took place dur-
ing voir dire.” Reviewing this finding with the knowledge that
credibility determinations can “virtually never be clear error,”

, it is plausible that a second review of
the trial file jogged Moore’s memory about voir dire schedul-
ing. At bottom, the district court heard and credited Moore’s
testimony over the government’s vigorous cross-examina-
tion. We see Moore’s account as entirely plausible, and, re-
gardless, it was up to the district court to weigh and evaluate
the government’s attacks on credibility. See

is best situated to make credibility determinations in
light of the totality of the evidence.”). And that is especially
so in light of the undeniable fact that contemporaneous rec-
ords from Moore’s case file so tightly reinforced the testimony
he provided in the district court.
    Finally, the government argues that even if we find no er-
ror in the district court’s crediting of Moore’s testimony, he
should have kept pushing to obtain the Social Security rec-
ords even after “he was initially told that the records were de-
stroyed.” But the government’s description of Moore’s testi-
mony is inaccurate. At no point did Moore state (in his 2009
Affidavit, 2018 Affidavit, or otherwise) that the Pine Bluff of-
fice told him before trial that the records had been destroyed.
To the contrary, Moore explained that the agency told him no
records existed. The two are not synonymous, for no records
existing meant there was nothing to continue requesting. And
that understanding left Moore with no reason to believe any-
thing had been destroyed. As Moore explained—and the dis-
trict court credited—he did not think “there was a basis to
32                                                  No. 19-2683

issue a court order” because he “had no good faith belief that
the records existed at that point.” A lawyer in Moore’s cir-
cumstances ought to be able to take the government at its
word when it says no records exist.
    The district court found that Moore pushed far enough,
and we are not “left with the definite and firm conviction”
that this finding reflects any clear error. 

. The documentary evidence from Moore’s trial file further
establishes that Moore focused on the possible existence of So-
cial Security records and repeatedly communicated with
Whitehead about requesting them.
   Because the district court’s findings—that Moore’s testi-
mony was credible and supported by contemporaneous doc-
umentary evidence—are not “internally inconsistent or im-
plausible,” we will not upset them. 

.
     B. Webster’s Intellectual Disability
    This brings us to the government’s challenge to the finding
that Webster is intellectually disabled. All agree that the dis-
trict court applied the proper criteria for finding intellectual
disability. The standard comes from the 11th edition of the
AMERICAN       ASSOCIATION       ON      INTELLECTUAL       AND
DEVELOPMENTAL DISABILITIES DEFINITION MANUAL (AAIDD-
11) and the 5th edition of the DIAGNOSTIC AND STATISTICAL
MANUAL OF MENTAL DISORDERS (DSM-5). See 

–49 (recognizing these exact standards).
    Recall that three criteria must be present for a person to be
intellectually disabled. Two of these criteria—onset by the age
of 18 and the presence of intellectual deficits—are not con-
tested here. Right to it, the parties dispute only whether Web-
ster suffers from adaptive deficits.
No. 19-2683                                                 33

   As a general matter, adaptive deficits are present when
someone cannot “learn basic skills and adjust behavior to
changing circumstances.” 

. The adaptive
functioning analysis considers whether a person is “suffi-
ciently impaired” such that “ongoing support is needed in or-
der for the person to perform adequately” in one of three do-
mains:
      The conceptual (academic) domain, which “involves com-
       petence in memory, language, reading, writing, math
       reasoning, acquisition of practical knowledge, problem
       solving, and judgment in novel situations, among oth-
       ers;”
      The social domain, which “involves awareness of others’
       thoughts, feelings, and experiences; empathy; interper-
       sonal communication skills; friendship abilities; and
       social judgment, among others;”
      The practical domain, which “involves learning and self-
       management across life settings, including personal
       care, job responsibilities, money management, recrea-
       tion, self-management of behavior, and school and
       work task organization, among others.”
DSM-5 at 33, 37–38; see also AAIDD-11 at 44. A deficit in
adaptive functioning requires showing sufficient impairment
in “at least one domain”—but not all three. DSM-5 at 38; see
also AAIDD-11 at 27. No matter the affected domain, the
adaptive deficit “must be directly related to the intellectual
impairments described in [the first criteria].” DSM-5 at 38.
    The DSM-5 and AAIDD-11 provide additional guidance
for assessing adaptive functioning. The AAIDD-11 empha-
sizes that demonstrating strengths in one area does not
34                                                   No. 19-2683

preclude a finding of intellectual disability. See AAIDD-11 at
47. As Dr. Fabian testified, many people with intellectual dis-
abilities “can live independently, especially with support, can
get a job, can work, [and can] find employment.” Nor is it un-
usual for people with intellectual disabilities to have unique
skills, from “musical talents” to “artistic talents [like] drawing
[or] painting.” The AAIDD puts the point even more directly,
explaining that “significant limitations in conceptual, social,
or practical adaptive skills [are] not outweighed by the poten-
tial strengths in some adaptive skills.” AAIDD at 47.
    The DSM-5 and AAIDD-11 likewise emphasize that adap-
tive skills are best measured not within the structured prison
environment but “within the context of ordinary community
environments typical of the person’s age peers and tied to the
person’s individualized needs for support.” AAIDD at 16; see
also DSM-5 at 38 (“Adaptive functioning may be difficult to
assess in a controlled setting (e.g., prisons, detention centers);
if possible, corroborative information reflecting functioning
outside those settings should be obtained.”). And the opera-
tive focus should be on “what the person typically does, ra-
ther than what the individual can do or could do.” AAIDD at
47.
     1. Intellectual Functioning
   Because adaptive deficits must be tied to intellectual im-
pairments, we owe a few words on Webster’s intellectual
functioning. This analysis focuses primarily on I.Q. scores.
See AAIDD-11 at 31 (“Although far from perfect, intellectual
functioning is currently best represented by IQ scores when
they are obtained from appropriate, standardized and indi-
vidually administered assessment instruments.”); DSM-5 at
37 (“Intellectual functioning is typically measured with
No. 19-2683                                                  35

individually administered and psychometrically valid, com-
prehensive, culturally appropriate, psychometrically sound
tests of intelligence.”).
    We see no clear error in the district court’s finding that
Webster suffers from intellectual deficits. Even a cursory re-
view of the I.Q. tests lays bare Webster’s intellectual short-
comings. “Individuals with intellectual disability have scores
of approximately two standard deviations or more below the
population mean,” which translate to “a score of 65–75.”
DSM-5 at 37. This is not a case where Webster has one or two
low I.Q. scores sitting alongside many higher scores. To the
contrary, every one of Webster’s nine I.Q. scores over 27 years
falls below 75—with many falling well below that number.
    The district court found no evidence of malingering on
any of these tests—before or after Webster’s murder of Lisa
Rene. We start with the full-scale I.Q. scores of 48 from 1992
(though recall that the parties agree this score may not have
been accurate because Webster was preoccupied) and 59 from
the 1993 Social Security records. Setting aside the 1992 score,
the 1993 score was the product of an intellectual assessment
performed in connection with the agency’s evaluation of Web-
ster’s application for disability benefits. This occurred a year
before the murder, at a time when Webster had no incentive
to malinger.
    As for the I.Q. tests administered after the murder, Web-
ster undeniably had an incentive to malinger. The govern-
ment pressed this precise point, and the district court assessed
the evidence with care. We have no reason to disturb the dis-
trict court’s decision to credit testimony that it would have
been challenging to malinger on so many additional I.Q. tests
over the course of decades. Nor will we second guess the
36                                                  No. 19-2683

decision to give little or no weight to Dr. Denney’s testimony
about malingering. As the district court explained, “Dr. Den-
ney focused on evidence that supported his conclusions,” and
some of his testimony was “demonstrably incorrect or ap-
pears to reveal biases.” Webster, 

, at *7 n.13.
For example, Dr. Denney would have rejected the results of
the Test of Malingering Memory Dr. Reschly administered to
Webster because Dr. Reschly did not administer the whole
test. As the district court found, though, this rejection was im-
proper as Dr. Reschly did administer all three portions of the
test—all of which Webster completed. See

the district court committed no error in de-
termining that Webster demonstrated deficits in intellectual
functioning. Perhaps owing to the weight of the I.Q. evidence,
the government, to its credit, conceded the same conclusion
during oral argument.
     2. Adaptive Functioning
    The government does challenge the district court’s finding
that Webster demonstrated deficiencies in each of the three
adaptive functioning domains—conceptual, social, and prac-
tical. But Webster need only exhibit a deficiency in one do-
main (not all three)—a point the government agrees with—
and the district court’s conclusion that he did so is not clearly
wrong.
   Webster’s adaptive deficits show most clearly in the con-
ceptual domain. Because this domain considers language,
reading, and writing skills, along with math reasoning and
problem solving, there is meaningful overlap between the ev-
idence of intellectual deficits (measured by intelligence tests)
and the evidence of deficits in the conceptual domain. See
No. 19-2683                                                   37

DSM-5 at 37 (“The conceptual (academic) domain involves
competency in memory, language, reading, writing, [and]
math reasoning” skills); AAIDD-11 at 44 (noting that concep-
tual skills include “language; reading and writing”).
    The record amassed on remand shows that Webster has
substantial limitations with language, reading, writing, prob-
lem-solving, and judgment. Dr. Reschly testified that Web-
ster’s speech is “imprecise, rambling, and disorganized.” On
Webster’s own account, he “struggle[ed] in school,” and had
“difficulty learning and keeping up with other students.” He
failed the first grade, and “readily admitted that he was in
special education”—a fact the district court found amply cor-
roborated by the November 8, 1993 letter from the Special Ed-
ucation Supervisor of the Watson Chapel Schools in Pine
Bluff, Arkansas. See Webster, 

, at *2.
    We likewise find no error in the district court’s reliance on
information in Webster’s 1993 application for Social Security
benefits as indicating a deficit in conceptual functioning. Web-
ster’s answers to application questions, the district court ob-
served, revealed an inability to write or reason coherently. See

Dr. Reschly saw the responses this exact way, testify-
ing that the Social Security application “related to the concep-
tual domain of adaptive behavior,” and pointed to answers
like “I sleeps look at cartoon,” and “don’t sleep to good at
night wakes up sneezing” as examples of literary deficiencies.
    On appeal, the government argues that Webster does not
have deficits in the conceptual domain because he “per-
formed well on two standardized tests of academic ability.”
Webster scored in the 43rd percentile on a grade school test
called the MAT6, and after his conviction he scored at the
freshman college level on two subjects of the Adult Basic
38                                                  No. 19-2683

Learning Exam. No question this evidence points in a differ-
ent direction. As the government points out, Webster’s own
experts acknowledged on cross examination that these test
scores, when viewed in isolation, are not consistent with a
finding of intellectual disability. But Dr. Reschly took care in
his testimony to explain that the Adult Basic Learning Exam
is unreliable and indeed “no longer available” because it em-
ploys “poor normative standards and [yields] inaccurate
grade equivalent scores.” And importantly, “the medical
community focuses the adaptive-functioning inquiry on
adaptive deficits.” See 

.
    The district court reasoned that the deficits chronicled by
Drs. Reschly and Fabian (and further corroborated by Web-
ster’s special education records, academic history, and Social
Security application) are not offset by two academic test
scores. The resultant finding that Webster suffers from deficits
in the conceptual domain does not reflect clear error.
    Whether Webster possesses deficits in the practical and so-
cial domains presents a more challenging question. But Web-
ster did not need to convince anyone one way or the other. We
will not disturb the district court’s finding that Webster exhib-
its adaptive deficits in one of the three adaptive domains, and
that finding is all that is needed for the adaptive functioning
analysis.
    Having demonstrated substantial deficits in intellectual
functioning and adaptive functioning (at the very least in the
conceptual domain) as well as an onset of the deficiencies be-
fore the age of 18, Webster has carried his burden of proving
that he is intellectually disabled and therefore constitutionally
ineligible to remain under a death sentence.
No. 19-2683                                                  39

                              IV
    There runs through the government’s position on appeal
a stream of frustration over Webster receiving relief in federal
court in Indiana after years of proceedings that had seemed to
reach finality in federal court in Texas. At one level the gov-
ernment’s position is well-founded, for there is no question
the litigation path has been long, defined by an unexpected
turn, and laden with great investments of resources by all in-
volved. At another level, though, the government’s position is
not well-received, for much of the frustration seems aimed at
registering disagreement with our 2015 en banc decision hold-
ing that Webster had made a sufficient showing to satisfy the
safety valve in 28 U.S.C. § 2255(h)(1) to pursue the prospect of
relief in the district court in Indianapolis. But now is not the
time to relitigate our en banc decision. We consider it—as we
must—law of the case for purposes of this appeal. See Dobbs
v. DePuy Orthopaedics, Inc., 

, 458 (7th Cir. 2018)
(“‘Once an appellate court either expressly or by necessary
implication decides an issue, the decision is binding upon all
subsequent proceedings in the same case’ under the law-of-
the-case doctrine.” (quoting Key v. Sullivan, 

,
1060 (7th Cir. 1991))).
    Nor do we agree that the remand proceedings in the dis-
trict court “all but swept away the nearly month-long trial and
sentencing proceedings involving 50-plus sentencing wit-
nesses and detailed jury findings on Webster’s intellectual
functioning that took place in the Fort Worth trial court.” Re-
member the reason for the remand: the government, in par-
ticular the Social Security Administration, failed to produce
documents pre-dating the murder showing that Webster was
mentally retarded. But even more, the evidence from the
40                                                  No. 19-2683

Texas proceeding was before the district court in Indiana. But
so too was substantial other evidence developed by the par-
ties on remand. If particular aspects of the Texas evidence un-
dermined Webster’s showing on remand, the government
should have so argued—not at a high level of generality (like
we see with many points on appeal), but rather at a high level
of specificity. At this stage, we cannot decide this appeal in
any way other than by focusing on evidentiary details.
    Above all else, the record shows that the district court pro-
ceeded with great care on remand. Over the course of many
years leading to the 2018 and 2019 hearings, the district court
held status conferences, sought discovery briefing, reviewed
exhibit and witness lists, solicited findings of fact and conclu-
sions of law, and ruled on motions. It then held extensive
hearings on whether Webster exercised reasonable diligence
pursuing the Social Security records before trial and, in turn,
whether, based on those records and other evidence, he car-
ried his burden of proving that he is intellectually disabled.
Both parties had full and fair opportunities to assemble their
evidence and present it to the district court. And from there
the district court made detailed findings of fact and ultimately
determinations on the questions presented. There is no way
to read the transcripts of the proceedings below and not walk
away impressed with the care taken by Judge Lawrence.
    Our role is limited. Weighty though the obligation, the
question before us is whether the evidence presented on re-
mand leaves us with the definite and firm conviction that the
district court’s findings reflect clear error. Having taken our
own detailed look at all aspects of the proceedings on remand,
we see no clear error anywhere. We therefore AFFIRM the dis-
trict court’s judgment vacating Webster’s capital sentence.