concurring in part and dissenting in part.
In fact-specific cases like this one, the court of appeals is usually defacto the last stop on the road for the litigants. It is thus critical that we get the facts right, even if we agree on the governing legal standards. Here, although I agree with the majority’s assessment of the claims against Aetna, the administrator of the plan at issue, I must part ways with its evaluation of Brenda Mote’s claims against the long-term disability plan (“the Plan”). According to Mote’s treating physicians and other specialists, since at least 1998 she has suffered from fibromyalgia, migraines, a sleep disorder, depression, and pain throughout her body. From 1998 to 2003, she received benefits from the long-term disability plan sponsored by her former employer, Arthur Andersen. In 2003, Mote’s disability benefits were terminated because the length of Mote’s disability triggered a shift in the applicable standard for disability, from the earlier one in which she needed to show that she could not perform her own job, to the more stringent one in which she needed to show that she could not perform any work at all. Mote appealed the decision, but the Plan affirmed itself. Mote then filed this suit in federal court, alleging that the Plan’s decision to terminate her benefits was arbitrary and capricious.
Although arbitrary and capricious review ties our hands considerably, it is “not a rubber stamp.” Hackett v. Xerox Corp. Long-Term Disability Income Plan, 315 F.3d 771, 774 (7th Cir.2003). We have held that a benefits plan governed by ERISA “must weigh the evidence for and against [a benefits determination], and within reasonable limits, the reasons for rejecting evidence must be articulated.” Halpin v. W.W. Grainger, Inc., 962 F.2d 685, 695 (7th Cir.1992) (internal quotation marks omitted). Further, “ERISA requires that specific reasons for denial be communicated to the claimant and that the *612claimant be afforded an opportunity for ‘full and fair review* by the administrator.” Id. at 688.
I see two significant problems in the Plan’s consideration of Mote’s appeal, either one of which would require reversal even under arbitrary and capricious review. First, in denying Mote’s appeal on September 28, 2004, the Plan made the following statement: “Reported pain also cannot be relied upon as [a] sufficient indicator of functional impairment since perception of pain may be affected by individual tolerance, motivation or psychological factors.” Perhaps if this plan had language in it to that effect, that conclusion might be acceptable. But most plans do not, and this one is no exception. To the contrary, section VII, which includes the governing definitions for the Plan, says only that “total disability/totally disabled” means
that solely because of an illness, pregnancy or accidental bodily injury, an insured employee is unable: (1) [djuring the first 5 years of disability to perform the material duties of the employee’s own occupation; and (2) [fjrom then on, to work at any occupation for which such employee is, or may reasonably become, fitted by education, training or experience. The availability of employment will not be considered in the assessment of the employee’s disability.
Plan, sec. VII, ¶ 29. In other words, the Plan takes a functional approach to disability. It does not forbid an employee from showing functional incapacity through self-reported symptoms. The ability to “interpret” cannot mean the ability to add entirely new language to plans. In my view, Mote’s case is indistinguishable from another ERISA case involving fibromyalgia, in which we held that subjective reports of pain can suffice to show one’s complete disability. Hawkins v. First Union Corp. Long-Term Disability Plan, 326 F.3d 914, 918 (7th Cir.2003). The Plan therefore made an error of law, or behaved arbitrarily and capriciously, in its analysis. Although it was entitled to credit evidence other than Mote’s own reports of pain, it cannot begin with the premise that reported pain can never be enough.1
*613The Supreme Court’s decision in Black & Decker Disability Plan v. Nord, warns that “Plan administrators, of course, may not arbitrarily refuse to credit a claimant’s reliable evidence, including the opinions of a treating physician,” even though the administrators have no obligation “to accord special weight to the opinions of a claimant’s physician.” 538 U.S. 822, 834, 123 S.Ct. 1965, 155 L.Ed.2d 1034 (2003). In this case, even if there were some language in the Plan on which the administrators could hang their conclusion that subjective evidence is never enough to support an award of benefits, the Plan still failed to articulate its “reasons for rejecting evidence,” which is necessary “if there is to be meaningful appellate review.” Halpin, 962 F.2d at 695. It may be that even under the correct standards, a weighing of all the evidence would lead once again to a rejection of her claim. Nonetheless, Mote is entitled to have the decision, whatever it is, reached through the use of a fair process. The record makes clear that this did not happen.
Second, although surveillance evidence can be used to undermine the credibility of a doctor’s medical opinions where the diagnosis is based substantially on patient reports, whether it was used properly here depends on how it was used and for what purpose. The record indicates that the Plan’s medical reviewer, Dr. Hall, did not receive all of the surveillance information about Mote, which was gathered over days and days of observation. Instead, as the majority concedes, he received only a compilation of two hours of pre-selected footage. After reviewing this material and looking at no other new evidence, Dr. Hall withdrew his earlier conclusion that Mote was totally disabled from working in any occupation. The fact that the tapes themselves are not in the record does not somehow make Dr. Hall’s conclusion reliable. No one disputes that he never saw the vast majority of the evidence that was collected. One might just as well view a two-hour snippet of Mote sitting on a sofa, and conclude that this was all she ever did. Dr. Hall’s opinion was based on inherently unreliable evidence and thus should not have been entitled to any weight.
Properly used, surveillance evidence can provide a basis for choosing between contradictory medical evidence by rendering some of that evidence less credible. This court has noted that “[w]e can imagine an argument that even if the activity disclosed ... does not indicate a capacity to engage in full-time work, the fact that it is discrepant with the level of activity described by [the treating physician], presumably on the basis of representations made to him by [the plaintiff], fatally undermines [the plaintiffs] credibility.” Hawkins, 326 F.3d at 918. Here, the Plan contends that the surveillance evidence was used to discredit the treating physicians’ statements and not as an independent basis for terminating Mote’s benefits, but the record belies this assertion.
Mote may well have concluded that there was no need to supplement the record before this court by furnishing all of the surveillance tapes, because even the evidence that Dr. Hall viewed was generally consistent with the records from Mote’s treating physicians. One problem with the way in which the Plan used the surveillance evidence is the fact that it made assumptions that find no support in the *614record. Thus, for example, in its 2004 denial of Mote’s appeal, it described the activities viewed in surveillance as Mote’s “daily living activities,” even though the record contains no evidence that these activities were daily or even regular. As I have already noted, the record contains only the pre-edited, two-hour videotape that, in essence, constitutes a highlight reel of Mote’s most active moments during several days of surveillance. There would have been no need to plant cameras inside Mote’s home in order to collect evidence that fairly reflected her ordinary activities; a fair look at the days’ worth of footage actually obtained would have sufficed.
Most troubling to me is that when the activities observed by surveillance are put in context, their utility in assessing Mote’s level of disability appears flimsy at best. Mote’s medical records state that Mote “overdoes it. Others aware of her overdoing it and depend on it. Patient aware of need for changes.” We have recognized in the past that some disabled people manage to keep going only through superhuman efforts; in those circumstances their activities do not negate the fact that they are disabled. See Perlman v. Swiss Bank Corp. Comprehensive Disability Protection Plan, 195 F.3d 975, 983 (7th Cir.1999). In 1999, Dr. Graft noted that Mote had a goal of setting boundaries with her mother (who is also disabled) and setting limits with others generally.' The only days where Mote was observed undertaking any significant activity were the days she drove her mother to and from her mother’s doctor’s appointments (once in July 2001 and once two years later in January 2003). Her activities that day included eating at a restaurant with her mother, where she sat for one hour, and standing up for two minutes after approximately 30 minutes of sitting. The other day of surveillance came on a day where Mote was required by the Plan to undergo a functional capacity evaluation. She drove 45 minutes each way to that appointment. Although these contextual details are in the surveillance notes, there is no indication that Dr. Hall had access to the written notes or that they were incorporated into the video that he watched. Notably, Mote cancelled the second day of her evaluation because she was in pain. Although one of Mote’s treating physicians noted in her medical records that Mote could not operate a motor vehicle, this observation accompanied a new prescription for a sedative. It is entirely possible that the doctor meant to warn against operation of a motor vehicle while taking the drug, rather than to describe Mote’s ability to drive. The tape showing Mote picking up her mail and newspapers at the end of her driveway included observations that she limped to the end of the driveway, that the limp significantly worsened on the walk back to the house, and that Mote struggled twice with the newspaper, and so I am unable to see how this helps the Plan’s arguments at all.
Other courts have concluded that segments of surveillance showing light physical activities by a plaintiff do not amount to a showing that she is able to manage full-time employment. See Osbun v. Auburn Foundry, Inc., 293 F.Supp.2d 863, 870 (N.D.Ind.2003) (“[Surveillance] evidence that [the plaintiff] can perform light physical tasks for 1.5 hours over two days falls far short of demonstrating that he is capable of sustaining a job. [The defendant] produced no evidence showing how long [the plaintiff] can perform such tasks, whether he can perform them on a daily basis, or how much pain he must endure in the process.”); Crespo v. Unum Life Ins. Co. of Am., 294 F.Supp.2d 980, 996 (N.D.Ill.2003) (finding, in a claim of disability due to fibromyalgia, that the defendant’s “comparison between [the plain*615tiffs] daily activities and the requirements of a full-time job is misplaced,” as “[t]here is no evidence anywhere in the record that [the plaintiff] undertakes these activities [including taking walks and performing household chores] with the regularity and structure of a full time job”); see also id. (noting that claimants need not “become inert in order to avoid having their disability benefits denied”).
Viewed in any light, the surveillance evidence in Mote’s case is nothing like what this court faced in Shyman v. Unum Life Ins. Co., 427 F.3d 452, 456 (7th Cir.2005), which involved an allegedly bedridden man coaching basketball and baseball teams. Mote’s evidence demonstrates that she was not capable of functioning in any capacity within the workforce. Surely there is room to conclude that a person is totally disabled from working without requiring that she be bedridden and immobile during every second of the day. The Plan’s apparent assumption that only something this extreme would disable her from working is, or could be viewed by a finder of fact to be, arbitrary and capricious.
Before concluding, I note with some concern that the actual plan underlying this claim is shrouded in mystery. Bizarrely, at oral argument, defense counsel acknowledged that although he represents both Aetna and the Plan, his only direction in this case came from Aetna. Aetna, however, was dismissed from the case at the district court level, and all three judges on this panel agree that this was correct. It is odd, at best, that Aetna therefore seems to be handling the litigation on appeal and that the Plan is nowhere to be found. The evidence of the Plan was also handled carelessly. Mote attached a copy of the long-term disability insurance contract between Arthur Andersen and Aetna to her complaint, entitled “Long-Term Disability Policy,” rather than another document, entitled “Arthur Andersen LLP Group Long Term Disability Insurance Plan,” which appears later in the record. Both Mote and the defendants refer to the first document as the plan at issue in this case. In the insurance contract, obligations are imposed on “Aetna” and the “policyholder” throughout, with Arthur Andersen (Mote’s former employer) identified as the policyholder. In the latter document, which seems to be the actual plan at issue, obligations are imposed upon the “Administrator,” “Fiduciary,” and “Appeals Fiduciary,” as one would expect. Arthur Andersen, which is defunct at this point, is named as the Plan Administrator, while both Fiduciary roles are filled by Aetna.
Given these two documents, it is unclear how an employee would know which plan document she should rely upon, who was in charge of the Arthur Andersen long-term disability plan, or if the Plan was its own entity separate from Aetna. The employee would need to consult both plans, which notably have similar but not identical definitions of disability. Mote has not pressed this issue in support of any of her arguments, and so we do not need to consider its implications on the case at hand. It does help to explain, however, some of the problems in this case.
Because, in my view, Mote has raised genuine issues of fact on the question whether Aetna’s determination (or, more accurately, the Plan’s) that she could not show inability to perform any job in the economy was arbitrary and capricious, I would reverse the district court’s grant of summary judgment in favor of the Plan and remand for further proceedings. I therefore respectfully dissent.
. My colleagues believe that the Plan's letter of September 28, 2004, taken as a whole, does not rest on the premise that reported pain can never be enough, but I see nothing in the letter that qualifies the statement quoted above. Page 1 of the letter summarizes the Plan's conclusion that Mote is not entitled to relief and sets forth the definition of disability from the Plan. From the bottom of page 1 through the middle of page 2, the letter reviews Mole's medical history. It then states that
[a]vailable medical records do not include references to clinical, laboratory or radio-graphic findings of progressive or worsening organic illness, or to severe or intractable medication side effects. They also do not furnish descriptors of severity of your musculoskeletal symptoms or other subjective symptoms. Although your subjective musculoskeletal symptoms are credible, they are not accounted for by identifiable neurological or musculoskeletal pathology.
In my view, the only way to read this letter is that Mote’s reported pain is insufficient to justify relief.
The only other items to which the letter refers are the surveillance evidence, which for the reasons I outline later is insufficient to support the Plan's conclusions, and an Independent Medical Examination ("IME”) conducted by a Dr. Tuttle on September 15, 2003. Dr. Tuttle was commissioned by the defense to examine Mote, but his examination could only have been as good as the data he had. As Mote pointed out in her brief, both his background and the file he consulted were deficient. Dr. Tuttle was given only a partial record to review and “lacked the appropriate medical specialization to evaluate a fibro-myalgia claim.” Based on his one-time examination of Mote, his review of an incomplete record of her medical history, and his viewing of the selective excerpts from the surveillance tapes, Dr. Tuttle stated that he could not "see any reason why [Mote] cannot *613return to a sedentary type of position at a full 8 hours a day.” IME at 4. I fail to see how Dr. Tuttle’s opinion, given these significant limitations, could constitute a valid basis for disregarding the informed and fully documented conclusions of Mote’s treating physicians. Yet the Plan justified its decision to deny benefits based solely on this IME and the surveillance evidence.