concurring in part and dissenting in part.
The irregularities of the claim-review process prompt me to agree with the majority that LINA’s denial of Cooper’s request for long-term-disability benefits was arbitrary and capricious. Yet it is precisely for that reason — that LINA has yet to give Cooper’s claim the procedural review it is due — that I cannot understand how the majority would grant the benefits itself rather than following the normal recourse: remanding the case; permitting LINA to comply with these procedures; and allowing the company to exercise its plan-given discretion over the awarding of benefits on a complete record.
First, in leapfrogging this process, the majority opinion is internally redundant, if not internally inconsistent. If the problem here is one of medical substance (that the only non-arbitrary conclusion one can draw from the record is that Cooper is disabled), then it does not seem necessary for the majority to devote most of its opinion to matters of procedure (that LINA failed in several respects to give Cooper’s claim proper review). If the medical evidence points in just one direction, why concern ourselves with LINA’s procedural mistakes in generating that record? And if LINA’s procedural mistakes are as numerous as the majority points out (they are), how can we draw one-way conclusions about the medical evidence? The more times LINA lost its way in compiling this record, the more reasons we have to doubt the accuracy of that record.
Second, while insurance companies (and perhaps federal courts) may readily handle some medical-disability claims, this is not one of them. Back-pain cases are notoriously hard to pin down: hard for a claimant to connect her complaints of pain to medical evidence; hard for a treating physician to explain how much of the diagnosis arises from objective evidence rather than from subjective complaints; and hard for a reviewing physician to verify the truthfulness of a claimant’s complaints. All of this makes it particularly important for plan administrators to use a “deliberate, principled reasoning process” in reviewing benefits claims in this area, Baker v. United Mine Workers of Am. Health & Ret. Funds, 929 F.2d 1140, 1144 (6th Cir. 1991) — -both to allow the administrator to exercise its discretion in a meaningful way and to allow a medically untrained judicia*174ry to exercise its limited review in a meaningful way. For precisely these reasons, when we find a “problem ... with the integrity of the plan’s decision-making process” in a claim of this type, Elliott v. Metro. Life Ins. Co., 473 F.3d 613, 622 (6th Cir.2006) (internal brackets omitted), we ought to be doubly reluctant to don the white coats ourselves and say that the record permits just one medical conclusion. That is particularly true here where the district court judge who reviewed this case thought that, as a matter of substance and procedure, the benefits decision was not arbitrary and capricious.
Third, consider the difficulties of saying that this record permits just one conclusion. It is one thing to say that LINA’s doctors should have done more to support their decision; it is quite another to say that this record left these same doctors with only one rational choice: find a disabling condition that precludes full-time work. No one doubts that Cooper suffered a back injury, and objective medical evidence shows as much. What the record does not show is whether Cooper’s treating physicians could point to objective medical evidence, as opposed to subjective complaints of pain from their client, that demonstrated a- disabling condition. We know this is a close case because Cooper and her doctors admitted that she could work some number of hours every day; the issue is whether it could be a full day. And Cooper’s doctors did not include the notes from their examinations (or return the phone calls from LINA’s doctors) that might have conclusively shown that objective medical evidence supported Cooper’s subjective view that she could not work full time, just part time.
All of this is important because subjective complaints of back pain by themselves do not compel an administrator to grant disability benefits. See, e.g., Yeager v. Reliance Standard Life Ins. Co., 88 F.3d 376, 382 (6th Cir.1996) (holding that administrator did not act arbitrarily in discounting claimant’s “subjective complaints[, that] are easy to make, but almost impossible to refute”); Oody v. Kimberly-Clark Corp. Pension Plan, 215 Fed.Appx. 447, 453 (6th Cir.2007) (holding that administrator did not act arbitrarily when evidence included subjective complaints and some medical evidence but not “sufficient objective medical evidence”); Nichols v. Unum Life Ins. Co. of Am., 192 Fed.Appx. 498, 504 (6th Cir.2006) (holding that administrator did not act arbitrarily in finding that treating physician’s assessment “was largely based on her acceptance of [claimant’s] descriptions of her medical conditions [neck, arm and back pain], rather than based on an objective assessment of [claimant’s] medical history”); Wical v. Int’l Paper Long-Term Disability Plan, 191 Fed.Appx. 360, 372 (6th Cir.2006) (finding “significant! ]” that a reviewing physician said a treating physician’s assessment of disability was “lacking an adequate empirical or ‘hard’ scientific basis beyond [claimant’s] subjective complaints of pain”); Bishop v. Metro. Life Ins. Co., 70 Fed.Appx. 305, 309 (6th Cir.2003) (per curiam) (upholding administrator’s denial of benefits for chronic back pain when claimant’s “symptoms seem[ed] out of proportion to object exam findings”) (internal quotation marks omitted).
On this record, we are left with more questions than answers. Did Cooper’s treating physicians base their conclusions about disability (as opposed to injury) on Cooper’s subjective complaints of pain or on their own objective findings developed through examination but not disclosed in their notes? The record, as Drs. Graulich and Sassoon fairly concluded, does not say. If Dr. Sassoon had examined Cooper, would he have arrived at the same conclusion? Would a simple conversation with Cooper’s treating physicians have con*175vinced Dr. Graulich that Cooper was impaired? Though LINA’s doctors asked Cooper for a functional capacity evaluation, which could have put the objective-medical-evidence question to rest, why wouldn’t she (or her doctors) supply one? The answer, it turns out, is not that she did not have one. More on that later.
These questions go to the heart of Cooper’s disability claim, and it is hardly unreasonable to insist that these questions be answered before an administrator, to say nothing of a court, awards benefits. See Elliott, 473 F.3d at 622 (remanding for a new decision based on flaws in the administrator’s reasoning process); Smith v. Cont’l Cas. Co., 450 F.3d 253, 265 (6th Cir.2006) (same); Yonts v. Cont’l Cas. Co., 113 Fed.Appx. 669, 672 (6th Cir.2004) (same); see also Sanford v. Harvard Indus., Inc., 262 F.3d 590, 598 (6th Cir.2001) (in non-disability ERISA case, noting that remand was proper to correct a procedural error and “expressly refraining] ... from examining the merits of the dispute”); cf. Brooking v. Hartford Life & Accident Ins. Co., 167 Fed.Appx. 544, 549 (6th Cir.2006) (awarding retroactive benefits because the “uncontested fact that [claimant] cannot maintain a seated position for more than an hour at a time” makes “work in a sedentary job ... clearly not possible”) (emphasis added).
The question, moreover, is not whether eligibility to work two to three hours a day would allow Cooper “to fulfill the ‘material duties’ of her job as a claims adjuster.” Maj. Op. at 173. Everyone takes that as a given. The question is whether the treating physicians’ two-to-three-hours determination is supported by medically conclusive evidence or merely the say-so of the doctors (based on the say-so of the patient). While this record would permit an administrator to award benefits as a matter of discretion, it does not demand it. To conclude otherwise is to contravene the well-established principle that administrators need not award benefits every time a claimant’s treating physicians say she is entitled to them. The plan after all gives reasoned discretion to the administrator, not to the claimant’s doctors. See Black & Decker Disability Plan v. Nord, 538 U.S. 822, 832-34, 123 S.Ct. 1965, 155 L.Ed.2d 1034 (2003) (holding that ERISA plans need not defer to treating physicians); McDonald v. W.-S. Life Ins. Co., 347 F.3d 161, 169 (6th Cir.2003) (“Generally, when a plan administrator chooses to rely upon the medical opinion of one doctor over that of another ... the plan administrator’s decision cannot be said to have been arbitrary and capricious.... ”).
Because “[w]e are not medical specialists,” Elliott, 473 F.3d at 622-23, it is difficult to know what to make of the medical evidence that is in the record. The majority points out, for example, that Cooper’s December 2002 magnetic resonance imaging scan showed “moderate-severe bulging of the disc” at L2-L3, Maj. Op. at 161, and that Dr. Johnson’s May 2004 exam of Cooper revealed “[c]omplete collapse of the L5-S1 disc space with a grade 1 to 2 spondylolisthesis of L5 on SI,” Maj. Op. at 162. While a thorough review of the record and several medical dictionaries might illuminate some of these statements (“spondylolisthesis,” it turns out, is the sliding of one vertebra forward relative to another one), federal judges are ill-equipped to wade through this kind of evidence in the first instance, much less to say that it supports one conclusion and one conclusion only — particularly when two doctors have told us otherwise. Add to this the fact that back-pain complaints require medical expertise to connect the dots between injury and disability, and it seems especially inappropriate to overrule the considered judgment of two medical professionals based on our review of the same *176evidence they reviewed. See Hansen v. Metro. Life Ins. Co., 192 Fed.Appx. 319, 323 (6th Cir.2006) (noting, in the context of a denial of disability benefits arising from back pain, that “illness is not to be equated with total disability”) (internal quotation marks omitted).
Fourth, this ERISA plan entrusts LINA, not the federal judiciary, with discretion to assess whether Cooper is due benefits. See Firestone Tire & Rubber Co. v. Bmch, 489 U.S. 101, 115, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989). For this reason, we normally allow administrators to hire and rely on independent medical experts in carrying out this task. See McDonald, 347 F.3d at 169. True enough, LINA erred by relying on the opinions of Drs. Graulich and Sassoon when they neither examined Cooper nor spoke with her doctors in this close case. But is it really the case that we would label LINA’s decision to deny benefits arbitrary even if the two sets of doctors had spoken, even if LINA had cured any other procedural errors and even if the evidence showed (after those conversations) that the treating physicians did no more than refiexively accept Cooper’s subjective complaints of pain? That seems doubtful. It also seems doubtful that we would award Cooper benefits based on the lack of a phone call, especially when we can never know how that conversation would have gone. While we can all agree that it would have been preferable for LINA to do this correctly the first time, the company’s errors do not entitle us to exercise plan discretion for the company, particularly when Cooper’s doctors and Cooper herself were partly to blame for the gaps in the record — (1) Cooper’s doctors because they did not explain why the objective medical evidence supported a finding of disability (as opposed to injury) and because they did not return the phone calls of Drs. Graulich and Sassoon and (2) Cooper herself because she did not produce a pre-existing functional capacity evaluation.
On remand, further investigation might reveal that the conclusions of Cooper’s treating physicians are based on objective (but unreported) findings developed through their examinations. It might reveal that Cooper’s treating physicians simply disagree with LINA’s medical experts on the severity of her condition based on the evidence in the record. Or it might reveal that Cooper’s treating physicians based their diagnoses solely on Cooper’s own testimony — testimony that LINA need not credit. In any case, the truth obscured by the ádministrative process can be revealed, and if Cooper is entitled to benefits, she will get them. In the first instance, however, “that judgment is not ours to make.” Elliott, 473 F.3d at 623.
Any doubt about the appropriate remedy, it seems to me, can be laid to rest by the motion for judicial notice. There, LINA points out, even though it sought additional objective medical evidence regarding Cooper’s medical condition during the administrative process, Cooper never produced a reliable and objective method of gauging her condition — a functional capacity evaluation — apparently because it showed that she could in fact work. As Wilkins v. Baptist Healthcare System, Inc., 150 F.3d 609 (6th Cir.1998), acknowledges, a “district court may consider evidence outside .of the administrative record only if that evidence is offered in support of a procedural challenge to the administrator’s decision,” id. at 619. See Moore v. LaFayette Life Ins. Co., 458 F.3d 416, 430 (6th Cir.2006) (“The Wilkins panel foresaw occasions in which the procedural process of gathering all pertinent information may have broken down at the administrative level and directed the courts to permit discovery in those cases.”).
*177Wilkins and Moore dealt with efforts by a claimant to introduce evidence that he could not have obtained earlier and that raised questions about the procedural propriety of the administrator’s decision. They say nothing about what happens when the shoe is on the other foot — when the claimant has refused to produce highly pertinent information that she concealed from the administrative process. Despite LINA’s repeated statements that a functional capacity evaluation would have supplied the objective medical evidence it needed, Cooper never produced the document — which stated that Cooper likely exaggerated her pain, JA 1263 (noting a “likelihood of submaximal effort given during [certain] tests”), that she could return to work immediately, id. (“Cooper DOES MEET the strength/lifting/carrying and positional demands of her job.”), and that her pain did not restrict her functional abilities, id. (“No restrictions are recommended for returning to work.”). If we can take notice of a plan administrator’s abuse of its procedural obligations under ERISA, see Wilkins, 150 F.3d at 619, surely we cannot turn our heads from a claimant’s seeming manipulation of the administrative process. That Cooper’s former employer (ACE) knew she had undergone an evaluation “at some point” does not change things because it is LINA, not ACE, that administers, the plan. It is LINA that must investigate Cooper’s eligibility for benefits, and it is LINA that the plan empowers to exercise discretion over these difficult disability determinations. Given the conclusions in Cooper’s once-withheld, now-revealed functional capacity evaluation, the “adequacy of [Cooper’s] proof is,” at the least, “reasonably debatable.” Maj. Op. at 172. On this record, the case should be remanded, and the administrator should be able to consider the functional capacity evaluation and any other evidence the parties wish to introduce. As I see this case, we have correctly identified a serious procedural mistake, then compounded the problem by committing our own procedural mistake — by awarding benefits ourselves without the assistance of a complete record and without the input of the administrator’s discretionary judgment. Two procedural wrongs do not make a substantive right. As the majority sees these issues differently, I respectfully concur in part and dissent in part.