Yassin Aref v. Loretta Lynch

BROWN, Circuit Judge: Appellants are three federal prisoners who spent several years housed in specially designated Communication Management Units (CMUs), a classification that meant family visits and communications with the outside world were curtailed. Appellants contend their designation to CMUs violated their due process rights. One appellant also alleges his continued CMU placement was in retaliation for protected speech in violation of the First Amendment. Finally, appellants seek damages under the Prison Litigation Reform Act (PLRA) for a variety of injuries allegedly arising out of their confinement in CMUs, including the denial of certain educational and professional programming, violations of their constitutional rights, and harm to familial relationships. The district court granted summary judgment for the government on each claim. Because we find the duration and atypi-cality of CMU designation sufficient to give rise to a liberty interest, we reverse the district court and remand for further proceedings to determine whether appellants were afforded sufficient process. With respect to the retaliation claim, we affirm the grant of summary judgment for the government because appellant cannot show his First Amendment rights were violated. Unlike the district court, we hold appellants have alleged harms qualifying for compensation under the PLRA because their injuries were neither mental nor emotional in nature and so do not require a showing of physical injury. We nonetheless uphold the grant of summary judgment because we find the prison official entitled to qualified immunity. I. A. Communication Management Units The CMUs at the heart of this controversy are located at two federal correctional facilities in Terre Haute, Indiana, and Marion, Illinois. They were established in 2006 and 2008, respectively. See Aref v. Holder, 774 F.Supp.2d 147, 153 (D.D.C.2011). The Bureau of Prisons (BOP) designed CMUs in response to a problem identified by the Department of Justice: a “deficiency” in the monitoring of inmate communications that allowed several inmates with terrorism-related convictions to communicate with extremist groups outside the prisons. CMUs thus house inmates who require communications monitoring beyond that which can feasibly be provided in the general population. *247 CMU inmates have access to more limited and less private communications compared to general population inmates. All visits — aside from attorney visits — must be “non-contact,” meaning a glass wall separates the inmate and visitor and communication takes place via a microphone. See 28 C.F.R § 540.205(a). All visits must be conducted in English, live-monitored, and recorded by BOP. See id. Although BOP regulations allow visitation to be restricted to four one-hour visits each month, id. BOP currently permits up to eight hours a month. CMU inmates are also restricted in the frequency and length of their written correspondence, which is subject to inspection. See id. § 540.203. Finally, except for unmonitored attorney calls, CMU inmates can telephone only immediate family members, and the calls are monitored. Id. § 540.204. Under the regulation, telephonic communication can be limited to no more than three fifteen-minute calls per month, id. but BOP currently allows inmates two fifteen-minute calls per week. Aside from these restrictions, CMUs essentially function as “self-contained general population housing unit[s].” J.A. 108. Inmates typically are not confined to their cells except at night and during security checks. They have access to common areas for up to sixteen hours a day, recreational facilities, exercise equipment, and the library. They can keep personal property in their cells, participate in religious services, receive educational and professional training, and be designated for work assignments. An inmate can be designated to a CMU for several reasons, including having a conviction offense related to international or domestic terrorism; demonstrating a propensity for using communication channels to further illegal activity outside the prison or to contact victims; abusing approved communication methods; or presenting a potential threat to prison facilities or the public as a result of unmonitored communications with persons outside the prison. See 28 C.F.R § 540.201. 1 Designation to a CMU begins when BOP becomes aware of information relevant to any of these criteria. See id. § 540.202(a). BOP’s Assistant Director evaluates and approves the designation if, after a review of the evidence, he concludes “designation ... is necessary to ensure the safety, security, and orderly operation of correctional facilities, or protection of the public.” 2 Id. § 540.202(b). Once in the CMU, the inmate receives a written Notice of Transfer (Notice) from the Warden explaining that the placement allows increased communications monitoring, the placement is non-punitive and will not affect the length of incarceration, and continued designation- will be reviewed “regularly” with both notice and an opportunity to be heard. Id. § 540.202(c). The *248 inmate also receives “an explanation of the [Assistant Director’s] decision in sufficient detail,” unless the Assistant Director determines that providing this information would jeopardize the safety of the facility or the public. Id. § 540.202(c)(4). Finally, the inmate may challenge his CMU designation through BOP’s administrative remedy program. Id. § 540.202(c)(6). In 2009 — three years after the first CMU opened — BOP instituted periodic review of prisoners, allowing for potential redesignation every six months. See id. § 524.11(a)(2). The process begins with the inmate’s Unit Team making an initial determination about whether continued CMU placement is necessary. The inmate must be given notice forty-eight hours before this review, which takes place in person. Id. § 524.11(b)(1). The recommendation considers factors like “whether the original rationale for CMU designation has been mitigated” and “whether the inmate no longer presents a risk.” J.A. 689. The Warden then receives the transfer recommendation for his review. If he agrees, the recommendation is sent to the Counter-Terrorism Unit (CTU) for its independent assessment — which is then forwarded to the Assistant Director 3 for a final decision. The inmate is informed in writing of the decision and (at least theoretically) provided an explanation for the result. There is no limitation on the duration of a prisoner’s CMU placement. B. The Plaintiffs (1) Yassin Aref. Aref is an Iraqi refugee convicted of helping a terrorist organization prepare to launch a missile attack oh American soil by helping to finance the missile’s purchase. United States v. Aref, 285 Fed.Appx. 784, 790 (2d Cir. 2008). He is serving a fifteen-year sentence for money laundering, providing material support for terrorism, conspiracy, and making a false statement to the FBI. Aref, 774 F.Supp.2d at 154. He was initially classified as a “low security” inmate with no disciplinary record, but he was transferred to the Terre Haute CMU in May 2007. Id. Within a day, he received a one-page Notice stating his designation was because of his terrorism-related conviction and because his “offense conduct included significant communication, association, and assistance to Jaish-e-Mohammed (JeM),” a designated terrorist organization. Id. at 154-55. Aref appealed, arguing he had never made contact with any JeM members; he had instead unknowingly been communicating with an individual cooperating with the government. The Regional Director denied the appeal. After eighteen months, Aref was transferred to the Marion CMU. In September 2010, three years after Arefs initial designation, his Unit Team and the Warden recommended him for transfer. This request was denied after the CTU received confidential law enforcement information from the Joint Terrorism Task Force. He was notified about the denial, but the notification provided no explanation. He was again recommended for transfer in March 2011, and this time the CTU agreed. Since April 2011, he has been housed in Marion Prison’s general population. (2) Kifah Jayyousi. In 2008, Jayyousi was sentenced to a 152-month term for conspiracy to murder, kidnap, and maim in a foreign country and conspiracy to provide material support to terrorism. He and his co-conspirators were found to have communicated in code and posed as a charitable organization to further these goals. See United States v. Jayyousi, 657 F.3d 1085, 1091-92 (11th Cir. 2011). Although *249 he was originally classified as a “low security” prisoner, he was transferred to the Terre Haute CMU in June 2008. Upon arrival, he received a Notice pointing to his terrorism-related conviction and offense conduct — which involved communication and association with al-Qaida — as the basis for his transfer. He appealed administratively, arguing this information was inaccurate; BOP denied his appeal without responding to his factual challenges. Jayyousi was first considered for redes-ignation in December 2009, but his Unit Team recommended against it because of the severity of his offense. In October 2010, he was transferred to the Marion CMU. His Unit Team and Warden recommended him for transfer in 2011 based on good conduct. Leslie Smith' — then-Chief of BOP’s CTU — disagreed because of a sermon Jayyousi gave as part of a Muslim prayer meeting in which he participated in 2008 while at Terre Haute’s CMU. Although Jayyousi received a disciplinary charge for that incident, he was cleared of any wrongdoing years before Smith considered this request. In March 2013, Jayyousi was again recommended for transfer, which was approved by the Regional Director without explanation. He continues to be housed in Marion Prison’s general population. (S) Daniel McGowan. McGowan was a member of the Earth Liberation Front, a domestic terrorist organization. Aref, 774 F.Supp.2d at 155. He was sentenced to a seven year term in 2007 for two counts of arson. Id. McGowan was also originally classified as a “low security” prisoner with no prison disciplinary record. Nonetheless, he was transferred to the Marion CMU in August 2008. He received his Notice ten days later, which cited his offense conduct as involving arson and the “destruction of an energy facility,” as well as communicating in code and teaching others how to commit arson. See id. McGowan appealed, challenging the factual assertions in his Notice as demonstrably false — pointing out he had never been accused or convicted of any crime relating to the destruction of an energy facility. BOP did not respond directly to McGowan’s challenge, denied his. appeal, and directed him to his pre-sen-tence report, which contained no mention of any energy facility. McGowan was first recommended for transfer in 2010, which the Regional Director denied without explanation. In July 2010, he was again recommended for transfer, which the Regional Director granted without explanation. A few months later, BOP officials determined that McGowan was attempting to circumvent the communication monitoring controls imposed on the general population; he was thus redesignated to the CMU in 2011. He remained in the CMU until his release from prison in December 2012; he was fully released from BOP supervision in June 2013. C. Procedural History On April 1, 2010, seven plaintiffs filed suit against BOP, alleging a variety of claims related to their CMU placement: violation of their procedural due process rights due to inadequate notice and lack of opportunity to be heard; violation of their substantive due process and First Amendment rights to “family integrity”; violation of the Eighth Amendment’s prohibition on cruel and unusual punishment; retaliatory transfer into the CMU in violation of the First Amendment; and unlawful discrimination on the basis of religion in violation of the First and Fifth Amendments. See Aref v. Holder, 953 F.Supp.2d 133, 138 (D.D.C. 2013). Plaintiffs sought declaratory and injunctive relief, transfer out of the CMUs, and an order requiring they be allowed the same communication privileges *250 as other prisoners. See Aref, 774 F.Supp.2d at 157. The district court dismissed all but the procedural due process and First Amendment retaliation claims. See id. at 161-71. In November 2012, Aref, Jayyousi, and McGowan filed an amended complaint adding a retaliation claim against defendants in their official capacities and against Leslie Smith in his individual capacity. Aref, 953 F.Supp.2d at 138. At the motion to dismiss stage, the district court found the PLRA barred plaintiffs’ individual-capacity claims and dismissed McGowan’s equitable claims as moot because he had been released from BOP custody. See id. at 142-44,147-49. 4 Defendants then filed motions for summary judgment on the remaining claims: Jayyousi and Arefs official-capacity due process claim and Jayyousi’s First Amendment retaliation claim. In March 2015, the district court granted summary judgment in favor of defendants, finding plaintiffs lacked any liberty interest sufficient to trigger due process protections and that Jayyousi’s First Amendment rights were not violated. See Aref v. Holder, No. 10-cv-0539, 2015 WL 3749621 at *1, *8-*9 (D.D.C. Mar. 15, 2015). Plaintiffs timely appealed. 5 II. We review the district court’s grant of summary judgment de novo. See Pharm. Research & Mfrs. of Am. v. Fed. Trade Comm’n, 790 F.3d 198, 204 (D.C. Cir. 2015). In doing so, we must, “view the evidence in the light most favorable to the non-moving party, draw all reasonable inferences in his favor, and eschew making credibility determinations or weighing the evidence.” Baumann v. District of Columbia, 795 F.3d 209, 215 (D.C. Cir. 2015). We also review the district court’s dismissal of appellants’ individual-capacity claims de novo. See Kimberlin v. U.S. Dep’t of Justice, 318 F.3d 228, 231 (D.C. Cir. 2003). III. Since this lawsuit’s inception, the government has urged at least some if not all of plaintiff-appellants’ arguments are moot because they were removed from the CMUs years ago. The parties agree McGowan’s official-capacity claims are mooted by his full release from BOP custody, see Aref, 953 F.Supp.2d at 142-43, so we consider only whether Aref and Jayyousi’s transfer into general population moots their claims. The mootness doctrine ensures compliance with Article Ill’s case and controversy requirement by “limit[ing] federal courts to deciding actual, ongoing controversies.” Am. Bar Ass’n v. Fed. Trade Comm’n, 636 F.3d 641, 645 (D.C. Cir. 2011). Accordingly, mootness must be assessed at “all stages” of the litigation to ensure a live controversy remains. 21st Century Telesis Joint Venture v. FCC, 318 F.3d 192, 198 (D.C. Cir. 2003). A case is moot if our decision “will neither presently affect the parties’ rights nor have a more-than-speculative chance of affecting them in the future.” Am. Bar Ass’n, 636 F.3d at 645. *251 The government argues that, because it has been years since any appellant was housed in a CMU, the appellants cannot identify any current injury for which this court can provide effective relief. While “[n]ormally[ ] a prisoner’s transfer or release from a prison moots any claim he might have for equitable relief arising out of the conditions of his confinement in that prison,” Scott v. District of Columbia, 139 F.3d 940, 941 (D.C. Cir. 1998), appellants point to the likelihood of redesignation from general population to a CMU. See, e.g., Aref, 114 F.Supp.2d at 158 (“McGowan was designated to a CMU, transferred back into the general population and then redesignated to a CMU.”). Appellants have also challenged BOP’s reliance on flawed information used to justify their CMU designations, which remains in their prison files. See Rezaq v. Nalley, 677 F.3d 1001, 1009 (10th Cir. 2012) (holding prisoners’ claims not mooted by transfer out of maximum security facility because, “[ejven though the new transfer policies may provide adequate process, the case is not moot if the BOP made decisions under the old policies that have ongoing, long-term consequences for the plaintiffs that could be mitigated by an award of prospective relief’). We need not decide that issue, however, because a defendant’s voluntary cessation of allegedly unlawful conduct can moot a case only if (i) “there is no reasonable expectation ... that the alleged violation will recur,” and (ii) “interim relief or events have completely and irrevocably eradicated the effects of the alleged violation.” Am. Bar Ass’n, 636 F.3d at 648. The government bears the “heavy” burden of showing it is “absolutely clear that the allegedly wrongful behavior could not reasonably be expected to recur.” Friends of the Earth, Inc. v. Laidlaw Envtl. Servs., Inc., 528 U.S. 167, 189, 120 S.Ct. 693, 145 L.Ed.2d 610 (2000) (emphasis added). We agree with the district court that the government has not met this high bar. Moreover, as the district court observed, appellants are challenging the procedure used for designation — so even if new information would be needed to return them to the unit, they have not “obtained all the relief’ they seek in their complaint with respect to the designation process. Schmidt v. United States, 749 F.3d 1064, 1068 (D.C. Cir. 2014). We therefore conclude the voluntary cessation exception applies and proceed to consider appellants’ claims on the merits. 6 *252 IV. Having found jurisdiction, we turn now to appellants’ due process claim. The Fifth Amendment ensures no individual is “deprived of life, liberty, or property, without due process of law.” U.S. Const. amend. V. Appellants challenge as inadequate the procedures used to designate them to the CMUs, claiming their transfer and lengthy placement in the units deprived them of their liberty in violation of the Constitution. Outside the penal context, we simply would evaluate the procedures under the now familiar Mathews v. Eld/ridge balancing test: first identifying the liberty interest at stake, then considering the risk of erroneous deprivation under existing procedures, and finally weighing the government’s interest against the burdens any additional process would entail. See 424 U.S. 319, 335, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976); Lepelletier v. Fed. Deposit Ins. Corp., 164 F.3d 37, 45-46 (D.C. Cir. 1999). This first step is complicated, however, by appellants’ incarceration, which “brings about the necessary withdrawal or limitation of many privileges and rights,” including the protections of due process. Jones v. N.C. Prisoners’ Labor Union, Inc., 433 U.S. 119, 125, 97 S.Ct. 2532, 53 L.Ed.2d 629 (1977). Evaluating due process claims in this context thus requires us to consider two competing — but significant — realities. First, we must “giv[e] appropriate deference to the decisions of prison administrators and appropriate recognition to the peculiar and restrictive circumstances of penal confinement.” Id. Prison officials face the unenviable task of ensuring the safety and security of large populations of people convicted of crimes and frequently are confronted with novel challenges in doing so. We therefore afford them “broad administrative and discretionary authority over the institutions they manage.” Hewitt v. Helms, 459 U.S. 460, 467, 103 S.Ct. 864, 74 L.Ed.2d 675 (1983). At the same time, prisoners are “not wholly stripped of constitutional protections” once they pass through the prison gates. Wolff v. McDonnell, 418 U.S. 539, 555, 94 S.Ct. 2963, 41 L.Ed.2d 935 (1974). Guarantees of due process may contract, but they are not eliminated entirely. A. Liberty Interest With these broad principles in mind, the Supreme Court in Sandin v. Conner articulated a new test for identifying liberty interests in confinement conditions. See 515 U.S. 472, 115 S.Ct. 2293, 132 L.Ed.2d 418 (1995). Courts previously had looked to state law to determine whether a liberty interest existed; under Sandin, the inquiry now focuses on the nature of the deprivation and its duration. Specifically, the Court held a liberty interest exists only if the conditions amount to an “atypical and significant hardship on the inmate in relation to the ordinary incidents of prison life.” Id. at 484, 115 S.Ct. 2293. Absent a liberty interest, an inmate is not entitled to any process. In Sandin, a prisoner challenged the procedure used to place him in disciplinary segregation for thirty days. The Court held this placement did not amount to a liberty interest, noting “disciplinary segregation, with insignificant exceptions, mirrored those conditions imposed upon inmates in administrative segregation and protective custody.” Id. at 486, 115 S.Ct. 2293. The Court observed inmates in that prison’s general population also had “sig *253 nificant amounts” of “lockdown time.” Id. Finally, the Court found the inmate’s confinement would not “inevitably affect” the length of his sentence. Id. at 487, 115 S.Ct. 2293. Thus, “a comparison between inmates inside and outside disciplinary segregation” demonstrated his placement there for 30 days “did not work a major disruption in his environment.” Id. at 486, 115 S.Ct. 2293. (1) Precedent Applying Sandin The Sandin Court did not define the baseline from which to measure what is “atypical and significant” in a particular prison system, so lower court assessments have diverged. See Wilkinson v. Austin, 545 U.S. 209, 223, 125 S.Ct. 2384, 162 L.Ed.2d 174 (2005). In Hatch v. District of Columbia, our circuit adopted a multifactor approach to determining the appropriate baseline. See 184 F.3d 846, 856-58 (D.C. Cir. 1999). But because the district court and both parties rely on several instructive cases from other circuits, we find it helpful to briefly survey the current state of the law before turning to our own standard. The Third, Sixth, and Tenth Circuits all generally look to administrative confinement as the baseline. 7 See, e.g., Griffin v. Vaughn, 112 F.3d 703, 706-08 (3d Cir. 1997) (finding no liberty interest for inmate who, suspected of raping a prison guard, was placed in administrative confinement for fifteen months because inmates can reasonably expect to be placed in administrative confinement during their sentence); Jones v. Baker, 155 F.3d 810, 812-13 (6th Cir. 1998) (finding no liberty interest for inmate placed in administrative segregation for thirty months pending investigation for murder of a prison guard as segregation during investigation is not atypical and was justified); Gaines v. Stenseng, 292 F.3d 1222, 1224-26 (10th Cir. 2002) (remanding to district court to compare conditions in disciplinary segregation to those in administrative segregation). The Fifth Circuit, on the other hand, has held disciplinary segregation can never implicate a liberty interest unless it “inevitably” lengthens a prisoner’s sentence, see Carson v. Johnson, 112 F.3d 818, 821 (5th Cir. 1997), and that administrative segregation — being an ordinary incident of prison life — is essentially incapable of creating a liberty interest, see Orellana v. Kyle, 65 F.3d 29, 31-32 (5th Cir. 1995). 8 The Seventh Circuit also has adopted a high standard, holding the baseline is not just the conditions of confinement within that particular prison, but those at the harshest facility in the state’s most restrictive pris *254 on. See Wagner v. Hanks, 128 F.3d 1173, 1175 (7th Cir. 1997). By contrast, the Fourth Circuit looks to the general population as the baseline. See Beverati v. Smith, 120 F.3d 500, 504 (4th Cir. 1997). And the Second Circuit requires a fact-specific determination that compares the duration and conditions of segregation with conditions in both administrative confinement and the general population. See, e.g., Arce v. Walker, 139 F.3d 329, 336 (2d Cir. 1998); Brooks v. DiFasi, 112 F.3d 46, 48-19 (2d Cir. 1997). As a result, the Second Circuit has found confinements as short as 180 and 305 days create a liberty interest under Sandin. See Colon v. Howard, 215 F.3d 227, 230-31 (2d Cir. 2000) (305 days); Kalwasinski v. Morse, 201 F.3d 103, 106 (2d Cir. 1999) (180 days). In sum, divergences in the baseline often lead to divergences in outcome. We are therefore cautious about relying too heavily on out-of-circuit precedent in evaluating appellants’ claims, except to note that courts are generally hesitant to find a liberty interest in the confinement context. Our circuit laid out its approach to the comparative baseline in Hatch. The Hatch court examined Sandin’s language and motivations to conclude a liberty interest arises only when the deprivation “imposes an ‘atypical and significant’ hardship on an inmate in relation to the most restrictive confinement conditions that prison officials ... routinely impose on inmates serving similar sentences.” 184 F.3d at 856 (emphasis added). Because administrative segregation is most routinely imposed, the court held it constitutes the proper baseline. Id. In doing so, though, the court took pains to emphasize this comparison “does not end our analysis.” Id. We must look “not only to the nature of the deprivation ... but also to its length” in evaluating atypicality and significance. Id. Since San-din noted the thirty-day disciplinary segregation at issue “was within the range of confinement to be normally expected for one serving an indeterminate term of [thirty] years to life,” 515 U.S. at 487, 115 S.Ct. 2293 (emphasis added), Hatch held atypicality also depends “in part on the length of the sentence the prisoner is serving.” 184 F.3d at 856. Applying this standard, the Hatch court remanded to the district court for further fact-finding to determine whether the inmate’s segregation for twenty-nine weeks amounted to a liberty interest. Id. at 858. Specifically, the district court was to compare the conditions faced by the inmate (who was segregated due to a disciplinary infraction) to the usual conditions of administrative segregation. Id. And even if the district court concluded those conditions were “no more restrictive” than administrative segregation, it was still required to determine whether confinement for twenty-nine weeks was “atypical” compared to the length of administrative segregation routinely imposed on similarly situated prisoners. Id. Though our circuit may be unique in considering the duration of confinement relative to similarly situated prisoners, duration itself is widely regarded as a crucial element of the Sandin analysis. See, e.g., Wilkinson, 545 U.S. at 223-24, 125 S.Ct. 2384 (considering indefinite duration of confinement and infrequency of review when finding a liberty interest in placement at a particularly harsh supermax prison); Harden-Bey v. Rutter, 524 F.3d 789, 793 (6th Cir. 2008) (“[M]ost (if not all) of our sister circuits have considered the nature of the more-restrictive confinement and its duration in determining whether it imposes an ‘atypical and significant hardship.’ ”). Duration is significant precisely because “especially harsh conditions endured for a brief interval and somewhat harsh conditions endured for a prolonged interval might both be atypical.” Sealey v. *255 Giltner, 197 F.3d 578, 586 (2d Cir. 1999); see also Hutto v. Finney, 437 U.S. 678, 686-87, 98 S.Ct. 2565, 57 L.Ed.2d 522 (1978) (“[T]he length of confinement cannot be ignored in deciding whether the confinement meets constitutional standards. A filthy, overcrowded cell and a diet of ‘grue’ might be tolerable for a few days and intolerably cruel for weeks or months.”). Indeed, we have remanded a case for the sole purpose of determining whether “the duration of plaintiffs administrative segregation ... impose[d] an atypical and significant hardship,” even when it was “apparent that the conditions of plaintiffs restraint” could not be considered atypical. Brown v. District of Columbia, 66 F.Supp.2d 41, 45-46 (D.D.C. 1999), on remand from Brown v. Plaut, 131 F.3d 163 (D.C. Cir. 1997). We conclude, then, that the proper methodology for evaluating deprivation claims under Sandin is to consider (i) the conditions of confinement relative to administrative segregation, (ii) the duration of that confinement generally, and (iii) the duration relative to length of administrative segregation routinely imposed on prisoners serving similar sentences. We also emphasize that a liberty interest can potentially arise under less-severe conditions when the deprivation is prolonged or indefinite. Having shown how our circuit’s baseline differs from that of our sister circuits, we now note another important distinction between this case and the usual penal due process case. Like Sandin, the vast majority of penal due process cases involve punitive deprivations, ie., confinement or privilege restriction for disciplinary purposes or while pending the outcome of an investigation. See, e.g., Skinner v. Cunningham, 430 F.3d 483, 487 (1st Cir. 2005) (“Skinner was a prisoner serving a sentence for murder who had just killed another inmate. It made perfect sense to isolate him pending further investigation.”); Thomas v. Ramos, 130 F.3d 754, 761 (7th Cir. 1997) (“Both temporary confinement and investigative status have been determined to be discretionary segregation and do not implicate a liberty interest.”); Bazzetta v. McGinnis, 430 F.3d 795, 804-05 (6th Cir. 2005) (finding no liberty interest for prisoners subjected to a permanent ban on visitation after two violations of the prison’s drug abuse policy). While an inmate can be designated to a CMU for abusing the prison’s communication system, most were transferred there to ensure prison officials could effectively monitor their communica tions — not for any punitive purpose. In this way, CMU designation is more analogous to transferring an individual to a harsher prison based on gang status, for instance, than it is to disciplinary segregation. We do not think this similarity ends the inquiry, however. We recognize the Court held in several pre-Sandin cases that “transfer of an inmate to less amenable and more restrictive quarters for non-punitive reasons is well within the terms of confinement ordinarily contemplated by a prison sentence.” Hewitt, 459 U.S. at 468, 103 S.Ct. 864. In Meachum v. Fano, for example, the Court found no liberty interest even when the transfer would “place the prisoner in substantially more burdensome conditions [than] he had been experiencing” because such transfers “are made for a variety of reasons and often involve no more than informed predictions as to what would best serve institutional security or the safety and welfare of the inmate.” 427 U.S. 215, 225, 96 S.Ct. 2532, 49 L.Ed.2d 451 (1976); see also id. at 228, 96 S.Ct. 2532 (noting it does not matter if the transfer is “for whatever reason or for no reason at all”). Circuit courts have also consistently held that, “generally speaking, *256 a prisoner has no liberty interest in his custodial classification.” Hernandez v. Velasquez, 522 F.3d 556, 562 (5th Cir. 2008). It follows then that a classification like gang status' — and any deprivations that flow from it — cannot de facto constitute a liberty interest. See, e.g., id. at 563-64 (concluding lockdown, to prevent gang-violence should be expected as an ordinary incidence of prison life); Adams v. Small, 542 Fed.Appx. 567, 568 (9th Cir. 2013) (holding no liberty interest in classification status as a gang member); Perez v. Fed. Bureau of Prisons, 229 Fed.Appx. 55, 58 (3d Cir. 2007) (“Because changes in security classifications and limits on telephone usage are ordinary incidents of prison confinement,” no liberty interest existed). This line of pre-Sandin precedent undermines appellants’ arguments. But, most recently, the Supreme Court acknowledged that, while “the Constitution itself does not give rise to a liberty interest in avoiding transfer to more adverse conditions of confinement,” a lesser liberty interest “in avoiding particular conditions of confinement may arise” if Sandin’s requirements are met. Wilkinson, 545 U.S. at 221-22, 125 S.Ct. 2384 (emphasis added). In Wilkinson, inmates who were assigned to Ohio’s Supermax Prison (OSP) on the basis of either their convictions (e.g., organized crime) or their engagement in specific conduct (e.g., leading a prison gang) challenged their transfer as violating due process. The Court concluded these inmates had a liberty interest in avoiding transfer to OSP because OSP prohibited almost all human contact and because placement there was indefinite, subject only to annual review, and disquali-fled otherwise eligible inmates from parole consideration. See id. at 223-24, 125 S.Ct. 2384. A district court in our circuit also recently found a prisoner plausibly alleged harsh and atypical conditions because “he [had] been segregated from the general population for over six years” after he was formally classified as a “terrorist inmate.” Royer v. Fed. Bureau of Prisons, 933 F.Supp.2d 170, 190. (D.D.C. 2013) 9 .° The court concluded that, even if the conditions alleged were “no more restrictive” than administrative segregation, the complaint should survive because the conditions were permanent and intended to last for the remainder of his twenty-year sentence— another sixteen-and-a-half years. Id. In doing so, the court distinguished Meachum as relating to the location of an inmate’s confinement rather than to the atypical conditions of that confinement. See id. at 191. We agree. Although appellants’ deprivations are more akin to transfer based on a non-punitive classification than disciplinary segregation, the Sandin framework still guides our analysis of whether these particular conditions can be considered “atypical and significant.” 10 (2) Appellants’ Due Process Claim Having examined this legal backdrop, we turn now to appellants’ specific claims. Whether a liberty interest exists here is admittedly a close call. All parties agree CMUs are less extreme in terms, of deprivation than administrative segregation. Inmates in administrative segregation must remain in their cells for twenty-three hours a day; they are unable to hold jobs *257 or access most educational opportunities. Their possessions are also limited, and they can exercise only one hour a day, five days a week. By contrast, CMU inmates are allowed in common spaces with other CMU inmates for sixteen hours a day. They have access to educational and professional opportunities, can keep as many possessions as inmates in the general population, and have no added restrictions on exercise. Communication deprivations in administrative segregation are also harsher: those inmates can make only one fifteen-minute phone call per month and are limited to four hours of non-contact visits per month. CMU inmates can make two fifteen-minute calls per week and are allowed two four-hour non-contact visits per month. We therefore conclude CMU confinement involves significantly less deprivation than administrative segregation. On the other hand, CMU designation is indefinite — lasting years in appellants’ case — and atypical because even though several thousand inmates could be designated to CMUs based on their commitment offenses, only a handful are placed under these restrictions. The main tension, then, is how atypicality, indefiniteness, and the harshness of the depravations should be weighed. We find three factors significant. Although CMU designation seems analogous to a classification, it is exercised selectively; the duration is indefinite and could be permanent; the deprivations — while not extreme- — necessarily increase in severity over time. An inmate placed in administrative segregation may be wholly unable to communicate with his family or the outside world, but that restriction will generally only last for a few weeks. Inmates housed in CMUs, by contrast, may spend years denied contact with their loved ones and with diminished ability to communicate with them. The harms of these deprivations are heightened over time, as children grow older and relationships with the outside become more difficult to maintain. Cf. Wilkerson v. Stalder, 639 F.Supp.2d 654, 684 (M.D. La. 2007) (“With each passing day its effects are exponentially increased, just as surely as a single drop of water repeated endlessly will eventually bore through the hardest of stones.”). Admittedly, Sandin’ s metric seems more difficult to apply where the transfer involves non-punitive classification rather than disciplinary segregation; and, as Wilkinson acknowledges, the difficulty of establishing an appropriate Sandin baseline has led to widely disparate conclusions about what constitutes an atypical and significant hardship. 545 U.S. at 223, 125 S.Ct. 2384. However, as Wilkinson makes clear, Sandin did not eliminate liberty interests created by prison regulations; instead, it focused the inquiry on the condition itself. Sandin determines whether this lesser interest receives protection, but is silent as to its weight in the Mathews balance. Id. What we think pushes CMU designation over the Sandin threshold is its selectivity and duration, not its severity, and BOP’s recognition that some process — however de minimis — is due. Thus, because we find the designation meets Sandin’s requirements, we must consider the sufficiency of BOP’s response. As a final note, we address the relevance of appellants’ contention that CMUs are viewed as an unusual designation reserved primarily for Muslim individuals convicted of terrorism-related offenses — giving rise to a stigma analogous to sex-offender classification. Appellants rely on Neal v. Shimoda, 131 F.3d 818 (9th Cir. 1997), and Chambers v. Colorado Department of Corrections, 205 F.3d 1237 (10th Cir. 2000), to support their claim. But, in those cases, sex-offender classification affected the length of the inmates’ sentences. In Neal, *258 parole eligibility was contingent on successful completion of a lengthy treatment program, 131 F.3d at 825, and, in Chambers, full good time credits were not available to sex offenders, 205 F.3d at 1239. In contrast, CMU designation is not based on any formal status as a “terrorist” and not every CMU inmate is associated with terrorist activities. Additionally, CMU designation has no bearing on the length of an inmate’s sentence. Thus, we do not find stigma to be relevant in this context. 11 B. Process Due We must next examine the question whether the assignment process used by the government is adequate. The district court never reached this question because it concluded no constitutional liberty interest existed. Aref, 2015 WL 3749621, at *9. Although both sides partially briefed the issue, appellants assert the deficiencies detailed in their briefs were “but a small piece of the voluminous and painstakingly detailed evidence [they] provided to the District Court to demonstrate the risk of erroneous deprivation of liberty.” Appellant Reply Br. 22. We therefore remand this issue for resolution on a further record. We note, however, that appellants are challenging fundamentally predictive judgments in an area where administrators are given broad discretion and the government’s legitimate interests in maintaining CMUs must be accorded substantial weight. Because the cardinal principle in due process analysis is flexibility — i.e., attention to relevant context and consideration of competing interests — only minimal process is likely due. See Hewitt v. Helms, 459 U.S. at 472, 103 S.Ct. 864. V. We turn next to appellant Jayyousi’s First Amendment retaliation claim against Leslie Smith, then-Chief of BOP’s CTU, in his official capacity. Jayyousi alleges that Smith retaliated against him by denying his transfer out of the CMU in 2011 because of a sermon he gave as part of a Muslim prayer meeting in August 2008 while housed in the Terre Haute CMU. The government counters Jayyousi’s language could reasonably have been viewed as an attempt to radicalize fellow Muslims, amounting to a potential security threat. To prevail on his retaliation claim, Jayyousi must show: “(1) he engaged in conduct protected under the First Amendment; (2) the defendant took some retaliatory action sufficient to deter a person of ordinary firmness in plaintiffs position from speaking again; and (3) a causal link between the exercise of a constitutional right and the adverse action taken against him.” Banks v. York, 515 F.Supp.2d 89, 111 (D.D.C. 2007). Because Jayyousi’s claim fails at the first prong, we need not reach the final two inquiries. While constitutional protections do not disappear at the prison gate, it is well established that “a prison inmate retains [only] those First Amendment rights that are not inconsistent with his status as *259 a prisoner or with the legitimate penological objectives of the corrections system.” Pell v. Procunier, 417 U.S. 817, 822, 94 S.Ct. 2800, 41 L.Ed.2d 495 (1974). To evaluate whether Jayyousi’s conduct was protected by the First Amendment, we look to the factors laid out by the Supreme Court in Turner v. Safley: (i) whether there was a “valid, rational connection between the prison [action] and the legitimate governmental interest put forward to justify it;” (ii) whether “alternative means of exercising the right ... remain open to prison inmates;” (iii) “the impact accommodation of the asserted constitutional right will have on guards and other inmates, and on the allocation of prison resources generally;” and (iv) whether any “ready alternative” existed. 482 U.S. 78, 89-90, 107 S.Ct. 2254, 96 L.Ed.2d 64 (1987). 12 These factors, taken together, allow us to assess whether the challenged conduct was “reasonably related to legitimate penological interests.” Id. at 89, 107 S.Ct. 2254. Our circuit has cast this “reasonable relation” test as “very similar,” if not identical, to rational basis review. Amatel v. Reno, 156 F.3d 192, 198-99 (D.C. Cir. 1998). This flexible standard “ensures the ability of corrections officials to anticipate security problems and to adopt innovative solutions to the intractable problems of prison administration.” O’Lone v. Estate of Shabazz, 482 U.S. 342, 349, 107 S.Ct. 2400, 96 L.Ed.2d 282 (1987). As an initial note, appellants make much of what, in their eyes, amounted to “excessive deference” on the part of the district court to Smith’s justifications for his actions. See Appellant Br. 36 (“When applying this already deferential Turner standard, the court is not meant to also defer to a defendant’s assertion that there is, in fact, a valid, rational connection between his actions and the legitimate governmental interest.”). Appellants further emphasize the legal standard for summary judgment, which requires courts to draw all reasonable inferences in their favor. See id. at 39-40. The interplay between Turner and the summary judgment standard is admittedly murky. But the Supreme Court in Beard v. Banks provided some guidance: We must distinguish between evidence of disputed facts and disputed matters of professional judgment. In respect to the latter, our inferences must accord deference to the views of prison authorities. Unless a prisoner can point to sufficient evidence regarding such issues of judgment to allow him to prevail on the merits, he cannot prevail at the summary judgment stage. 548 U.S. 521, 530, 126 S.Ct. 2572, 165 L.Ed.2d 697 (2006). But the Court also cautioned that Turner “requires prison authorities to show more than a formalistic logical connection between a regulation and a penological objective.” Id. at 536,126 S.Ct. 2572. We agree with the Eighth Circuit that “[a] ‘reasonableness’ standard is *260 not toothless.” Salaam v. Lockhart, 905 F.2d 1168, 1171 (8th Cir. 1990). With these principles in mind, we turn now to the first Turner factor. This factor, which “looms especially large,” asks whether the prison’s actions bear a rational connection to any legitimate penological interest. Amatel, 156 F.3d at 196. We agree with the district court that Smith could rationally have interpreted Jayyousi’s language during the prayer meeting as an attempt to “radicalize” other prisoners, thereby constituting a continued security risk. Although appellants claim Smith exaggerated the contents of the remarks, several portions rationally could have been considered troubling, particularly when Jayyousi stated “you are here because you are Muslim, not because you are a criminal” and cautioned “it is not U.S. versus Jayyousi; it is U.S. versus Islam.” J.A. 835. Jayyousi also asserted the CMU was created from evil, and that the suffering faced by Muslim inmates is “why we martyr.” J.A. 836. Prison staff were concerned about the sermon at the time it was given, as evidenced by the several emails and follow-ups that ensued. J.A. 1292, 1296, 1298, 1300, 1302. That Jayyousi was cleared of any wrongdoing through the prison disciplinary process does not render it unreasonable for Smith, as the head of BOP’s CTU, to consider the content of Jayyousi’s statements in evaluating his CMU placement — especially the portions that indicated Jayyousi may have been continuing some of the same actions that led to his incarceration. The first factor weighs in favor of the government. The second Turner factor asks whether prisoners have any alternative means of exercising the right at stake. The right at issue “must be viewed sensibly and expansively.” Thornburgh v. Abbott, 490 U.S. 401, 417, 109 S.Ct. 1874, 104 L.Ed.2d 459 (1989). Turner and O’Lone are particularly instructive. In Turner, the Court “did not require that prisoners be afforded other means of communicating with inmates at other institutions, nor did ... O’Lone require that there be alternative means of attending the [Muslim] religious ceremony.” Id. Instead, the Turner Court held “it was sufficient if other means of expression ... remained available” and, in O’Lone, it was sufficient “if prisoners were permitted to participate in other Muslim religious ceremonies.” Id. at 418, 109 S.Ct. 1874. Here, the second factor is easily satisfied; Jayyousi had other means of communicating his dissatisfaction still available to him, and he was not prohibited from giving similar sermons in the future. The third Turner factor looks to the impact accommodation of the asserted right will have on guards and other inmates in the prison. It is unclear, in this context, how BOP could have otherwise accommodated Jayyousi, as he was allowed' to pray and free to lead similar prayer meetings in the future. Appellants’ view of “accommodation” would require Smith to entirely disregard the content of Jayyousi’s sermon when evaluating whether he should remain in the CMU. But the government is not required to disregard potentially relevant information when making that sort of security assessment. Appellants additionally argue that Jayyousi’s sermon did not affect allocation of prison resources, but the number of emails generated and the need to conduct the disciplinary hearing undermine that assertion. Finally, under the fourth factor, we must consider whether Smith had any ready alternatives. Again we find the government’s position persuasive: the CMU exists precisely because inmates who present security risks require heightened monitoring. Given that charge, Smith had no real alternative but to consider all information available about Jayyousi — including *261 the language used during the prayer meeting. And once Smith determined continued monitoring was necessary, the only option available to the government (except, perhaps prolonged confinement in administrative segregation) was to keep Jayyousi in the CMU. At bottom, appellants are challenging a “disputed matter of professional judgment” rather than disputed matters of fact. We do not require government officials to be perfect in their judgment, merely reasonable. Because all four Turner factors uniformly indicate Smith’s actions here were reasonably related to a legitimate security interest, we affirm the district court’s grant of summary judgment on this claim. VI. Finally, we turn to Jayyousi and McGowan’s claims against Smith in his individual capacity. Appellants seek “compensatory and punitive damages” for injuries they suffered during their purportedly retaliatory placements in the CMUs. These injuries include the denial of job-related programming, the stigma of being designated to a “terrorist” unit, the prolonged deprivation of First Amendment rights to political speech, and the undue damage to familial relationships caused by the CMUs’ unique communication restrictions. But before we address the availability of damages in this context, we must answer a threshold jurisdictional question: whether these individual-capacity claims survive Smith’s death in March 2015. 13 A. Mootness Both sides agree state law determines whether a Bivens action survives the death of a party. See Haggard v. Stevens, No. 2:09-cv-1144, 2010 WL 3658809, at *3-*6 (S.D. Ohio, Sept. 14, 2010) (undertaking an exhaustive survey of law in this area and concluding questions of survivor-ship are overwhelmingly decided by looking to state law), affd 683 F.3d 714 (6th Cir. 2012). They do not agree, however, about which state’s law should govern. We have several options: West Virginia law (where Smith was domiciled and worked), Indiana law (where the Terre Haute CMU is located), or Illinois law (where the Marion CMU is located). The starting point for assessing which state’s law should apply is the law of the forum state. See Haggard v. Stevens, 683 F.3d 714, 718 (6th Cir. 2012); see also Malone v. Carr. Corp. of Am., 553 F.3d 540, 542 (7th Cir. 2009) (“[I]t is a familiar principle that federal courts use the whole law of the forum state, including that state’s choice-of-law rules.”). The District of Columbia employs the “governmental interest” test to determine which state’s law to apply. See Raflo v. United States, 157 F.Supp.2d 1, 5 (D.D.C. 2001). This test involves a two-step inquiry: we begin by “identifying the governmental policies underlying the applicable law” and then “determin[e] which state’s policy would be most advanced by having its law applied to the facts of this case.” Id. Courts use four factors to determine which state’s policy is most advanced by application of its laws: “(1) the place where the injury occurred; (2) the place where the conduct causing the injury oc *262 curred; (3) the domicile ... of the parties; and (4) the place where the relationship is centered.” Id. We need not determine which state’s law applies with respect to survivorship, however, because the result is the same under all three — appellants’ claims are not extinguished by Smith’s death. See Ind. Code § 34-9-3-l(a); 755 Ill. Comp. Stat. 5/27-6; W. Va. Code § 55-7-8a(a). We proceed to the merits. B. Damages Under The PLRA Jayyousi and McGowan contend they are entitled to compensation under the PLRA for a variety of injuries: loss of educational opportunity in the form of release preparation programming, reputa-tional harm, violation of their First Amendment rights, and lasting harm to their familial relationships. The PLRA — in a provision entitled “Limitation on Recovery” — states: “No Federal civil action may be brought by a prisoner confined in a jail, prison, or other correctional facility, for mental or emotional injury suffered while in custody without a prior showing of physical injury.” 42 U.S.C. § 1997e(e) (emphasis added). We are thus faced with an issue of first impression for this circuit: whether injuries that are allegedly neither mental nor emotional are compensable under the PLRA without a prior showing of physical injury. Our circuit has addressed Section 1997e(e) once before in Davis v. District of Columbia, 158 F.3d 1342 (D.C. Cir. 1998). The government and district court both contend Davis controls this issue. We disagree. In Davis, the plaintiff sought compensatory and punitive damages for an alleged violation of his privacy after a prison official opened his sealed medical records and disclosed their contents without his consent. See id. at 1345. While the Davis court assumed without deciding that this intrusion on his privacy amounted to a constitutional violation, it was careful to point out that Davis “alleged resulting emotional and mental distress, but no other injury” when it held his claims for compensatory and punitive damages were foreclosed by the PLRA. 14 Id. (emphasis added). Our circuit has therefore never squarely addressed whether actual injuries that are neither mental nor emotional are precluded under the PLRA absent a showing of physical injury. Language in Davis and from other circuits confirms this distinction. See, e.g., id. at 1349 (“[Section] 1997e(e) precludes claims for emotional injury without any prior physical injury, regardless of the statutory or constitutional basis of the legal wrong” (emphasis added)); Cassidy v. Ind. Dep’t of Corr., 199 F.3d 374, 375-77 (7th Cir. 2000) (dismissing claims for mental and emotional harm stemming from an underlying constitutional violation but allowing plaintiff to pursue claims for loss of opportunity, loss of participation in prison activities, loss of access to prison programs and services, and loss of freedom of movement and social context stemming from the same violation). Circuits have split over the applicability of Section 1997e(e) to claims involving constitutional violations but no physical injury. A majority has held that Section 1997e(e) precludes compensatory damages for any claim that does not include physical harm. 15 In doing so, these courts focus on *263 the type of injury asserted. See, e.g., Thompson v. Carter, 284 F.3d 411, 417-18 (2d Cir. 2002) (holding a “plaintiff cannot recover damages for mental or emotional injury for a constitutional violation in the absence of a showing of actual physical injury”); Brooks v. Warden, 800 F.3d 1295, 1298 (11th Cir. 2015) (“Because [plaintiff] has not alleged any physical injury resulting from his hospital stay, under the [PLRA], he cannot recover compensatory or punitive damages” for his Eighth Amendment claim). These cases necessarily imply a constitutional violation, absent physical harm, is necessarily a type of “mental or emotional injury.” See, e.g., Allah v. Al-Hafeez, 226 F.3d 247, 250 (3d Cir. 2000) (considering plaintiffs claim of a First Amendment violation and concluding “the only actual injury that could form the basis for the award he seeks would be mental and/or emotional,” thus barring his claim); Searles v. Van Bebber, 251 F.3d 869, 876 (10th Cir. 2001) (“The statute limits the remedies available, regardless of the rights asserted, if the only injuries are mental or emotional.”). Several circuits have taken the opposite approach. These courts view alleged constitutional violations as a type of intangible harm wholly apart from mental or emotional injury. See, e.g., King v. Zamiara, 788 F.3d 207, 213 (6th Cir. 2015), cert. denied, — U.S.-, 136 S.Ct. 794, 193 L.Ed.2d 710 (2016) (“The statute provides that a prisoner may not bring a civil action for mental or emotional injury.... It says nothing about claims brought to redress constitutional injuries, which are distinct from mental and emotional injuries.”); Robinson v. Page, 170 F.3d 747, 749 (7th Cir. 1999) (“If the suit contains separate claims, neither involving physical injury, and in one the prisoner claims damages for mental or emotional suffering and in the other damages for some other type of injury, the first claim is barred by the statute but the second is unaffected.”). Both approaches involve some degree of slicing-and-dicing claims: one by injury pled and one by relief requested. This is best illustrated by example. Take a prisoner who has alleged a credible violation of his First Amendment right to free exercise but made no showing of physical harm. Rather than dismiss his entire action, the majority view of Section 1997e(e) would bar his claim for compensatory damages but allow his claims for injunctive relief and punitive damages to proceed. The minority view, on the other hand, would look to the type of injury alleged— if, say, the prisoner claimed mental anguish in addition to the substantive constitutional violation, then the first claim would be barred while the second would be eligible for compensatory damages. For the reasons laid out below, we are convinced this narrower reading of the PLRA is the proper one. While Section 1997e(e) “may well present the highest concentration of poor drafting in the smallest number of words in the entire United States Code,” John Boston, The Prison Litigation Reform Act: The New Face of Court Stripping, 67 BROOK. L. Rev. 429, 434 (2001), we nonetheless begin with the statute’s plain language. See, e.g., Consumer Prod. Safety Comm’n v. GTE Sylvania, Inc., 447 U.S. 102, 108, 100 S.Ct. 2051, 64 L.Ed.2d 766 (1980). The government’s preferred interpretation would render the phrase “mental and emotional injury” superfluous. Had Congress intended to graft a physical-injury requirement onto . every single claim, the statute could simply have provided: “No Federal civil action may be brought by a prisoner ... for any injury suffered while in custody without a prior showing of physical injury.” See Zamiara, 788 F.3d at 213; Robinson, 170 F.3d at 749; Amaker v. *264 Haponik, No. 98 CIV. 2663, 1999 WL 76798, at *7 (S.D.N.Y. Feb. 17, 1999) (“If Congress had intended to apply § 1997e(e)’s restriction to all federal- civil suits by prisoners, it could easily have done so simply by dropping the qualifying language ‘for mental or emotional injury.’ ”). The “mental and emotional” language is significant precisely because prisoners can allege types of intangible injury that fall outside that ambit. Courts that advocate the opposite interpretation of Section 1997e(e) claim the provision’s “clear mandate” is a focus on the type of injury pled rather than the nature of the underlying right. See, e.g., Searles, 251 F.3d at 876. The Tenth Circuit, for instance, cautioned the statute’s plain language forecloses “divore[ing]” the underlying substantive violation from the resulting injury. Id. But our reading does no such thing; the focus remains on the type of injury alleged, with an understanding that plaintiffs can allege intangible harms that are neither mental nor emotional, ie., not every non-physical injury is by default a mental or emotional injury. In the PLRA context, many of our sister circuits have awarded compensatory damages for non-mental and non-emotional injuries. See, e.g., Rowe v. Shake, 196 F.3d 778, 781 (7th Cir. 1999) (“A deprivation of First Amendment rights standing alone is a cognizable injury.”); Cassidy, 199 F.3d at 375-77 (allowing claims of loss of access to prison programs and services); Brooks v. Andolina, 826 F.2d 1266, 1269-70 (3d Cir. 1987) (finding a prisoner entitled to compensatory damages for his unconstitutional placement in punitive segregation including for the loss of visiting, phone, and library privileges). Analogous Supreme Court and circuit precedent supports the view that there can be real harms separate and apart from mental or emotional injury. For instance, in Carey v. Piphus, the Court held that a plaintiff is eligible to recover damages under Section 1983 if he can demonstrate “some actual, if intangible, injury” caused by a constitutional violation. 435 U.S. 247, 264, 98 S.Ct. 1042, 55 L.Ed.2d 252 (1978). Similarly, this court in Hobson v. Wilson instructed “intangible interests must be compensated” as long as they can be shown with “sufficient certainty.” 737 F.2d 1, 62 (D.C. Cir. 1984). The Hobson court even enumerated some examples of “First Amendment compensable rights” separate from any common-law mental or emotional harm, such as the restriction of an inmate’s access to books. Id. Indeed, courts frequently allow plaintiffs in Section 1983 actions to recover damages for constitutional violations that fall outside the domain of common-law injuries. See, e.g., Simmons v. Cook, 154 F.3d 805, 808-09 (8th Cir. 1998) (affirming an award of compensatory damages for Eighth Amendment claim of paraplegic prisoners unconstitutionally placed in solitary confinement); Ricciuti v. N.Y.C. Transit Auth., 124 F.3d 123, 130 (2d Cir. 1997) (finding damages to be an appropriate remedy for the harm caused by fabrication of evidence). Courts have also consistently treated the loss of liberty as an independently cognizable injury, separate from any mental or emotional harm. See, e.g., Heck v. Humphrey, 512 U.S. 477, 484, 114 S.Ct. 2364, 129 L.Ed.2d 383 (1994) (“[A] successful malicious prosecution plaintiff may recover, in addition to general damages, compensation for ... loss of time and deprivation of the society.”); Dellums v. Powell, 566 F.2d 216, 277 (D.C. Cir. 1977) (affirming damage award for, among other relief, compensation for “the duration of loss of liberty” in a Fourth Amendment case); Kerman v. City of New York, 374 F.3d 93, 128 (2d Cir. 2004) (holding a plaintiff is “entitled to be compensated for [his] loss of liberty” “independently of his *265 claims of physical, mental, emotional, or economic injury”). We therefore conclude there exists a universe of injuries that are neither mental nor emotional and for which plaintiffs can recover compensatory damages under the PLRA. Our holding also comports with the purpose of the PLRA, as expressed in its legislative history. The Act’s passage was precipitated by an increase in prison litigation, much of it frivolous. 16 Examples cited by PLRA proponents ranged from due process cases alleging injuries like a defective haircut to, most famously, an inmate filing a cruel and unusual punishment claim after he was given chunky rather than creamy peanut butter. See 141 Cong. Rec. 27,042 (1995) (statement of Sen. Bob Dole). But members of Congress also made it clear that the PLRA was not meant to bar serious, potentially meritorious claims. See id. at 26,553 (statement of Sen. Jon Kyi) (“Prisoners still have the right to seek legal redress for meritorious claims.... ”); see also id. at 27,044 (statement of Sen. Strom Thurmond) (“This amendment will allow meritorious claims to be filed, but gives the judge broader discretion to prevent frivolous and malicious lawsuits filed by prison inmates.”). Indeed, we find it hard to believe that Congress intended to afford virtual immunity to prison officials even when they commit blatant constitutional violations, as long as no physical blow is dealt. 17 It is especially difficult to see how violations of inmates’ First Amendment rights could ever be vindicated, given the unlikelihood of physical harm in that context. Against that backdrop, and a legislative record indicating an intention to still allow awards for meritorious claims, we believe our reading of Section 1997e(e) best aligns with the purposes of the PLRA. We note also that the PLRA contains several other mechanisms to curb the filing of frivolous suits — making it even less likely Congress intended the physical-injury requirement to bar claims for every serious but non-mental or emotional harm. See 42 U.S.C. § 1997e(a), (d) (requiring the exhaustion of administrative remedies and capping attorneys’ fees for successful claims at 150 percent of damages); 28 U.S.C. § 1915(b)(1), (g) (compelling personal payment of initial filing fees and *266 imposing a limitation on filing in forma pauperis after having three suits previously dismissed). And, of course, courts have always had the power to weed out claims that lack merit at earlier pleading stages, preserving judicial resources. Finally, we point out one additional limitation relevant to appellants’ claims here. Even when a party pleads an injury that is neither mental nor emotional, that harm still must “be shown with sufficient certainty to avoid damages based either on pure speculation or the so-called inherent value of the rights violated.” Hobson, 737 F.2d at 62. When “no value [can] reasonably be placed on the particular injury demonstrated,” then the plaintiff is entitled only to nominal damages. Id. at 63; see also Carey, 435 U.S. at 264, 98 S.Ct. 1042 (noting plaintiffs could recover compensatory damages “for racial discrimination, the denial of voting rights and the denial of Fourth Amendment rights” assuming they could prove “some actual, if intangible, injury”); Ker-man, 374 F.3d at 130 (“The present case does not involve ... an attempt to vindicate an abstract societal interest. Rather, it involves an anything-but-abstract physical detention. And although a given person’s loss of time may be difficult to evaluate in terms of dollars, his loss of liberty is not just ‘virtually certain’ to occur; it is inseparable from the detention itself.”). Having concluded a prisoner may recover compensatory damages under the PLRA if he can show an actual injury— separate from any mental or emotional harm — for which damages can be reasonably ascertained, we note the vast majority of circuits — including the majority of those that have adopted the broader application of Section 1997e(e) — agree the provision does not limit the availability of punitive damages provided a proper showing is made. See, e.g., Searles, 251 F.3d at 879-80; Allah, 226 F.3d at 252-53; Cassidy, 199 F.3d at 376-77; Carter, 284 F.3d at 418. Our circuit has uniquely held that punitive damages are unavailable to plaintiffs who plead only mental or emotional injury, without a showing of physical harm. See Davis, 158 F.3d at 1348 (“Amicus argues that because punitive damages are awarded to punish the tortfeasor rather than to compensate the victim, they are not embraced by § 1997e(e). But § 1997e(e) draws no such distinction. It simply prevents suits ‘for’ mental injury without prior physical injury.”). We have no occasion to reconsider that holding here. Instead, we construe Davis narrowly and hold appellants who allege actual harms that are neither mental nor emotional are entitled to punitive damages if they can show the defendant’s conduct was “motivated by evil motive or intent” or “involve[d] reckless or callous indifference to the federally protected rights of others.” Wade, 461 U.S. at 56, 103 S.Ct. 1625. Finally, every circuit, regardless of its interpretation of Section 1997e(e), agrees that nominal damages are available in this context. See Carter, 284 F.3d at 418 (listing cases). Our court declined to reach this issue in Davis because it found the plaintiff failed to sufficiently plead nominal damages. See Davis, 158 F.3d at 1349. The district court here also concluded these appellants waived their claim to nominal damages by failing to “specifically plead” them in their complaint. Aref, 953 F.Supp.2d at 149. In doing so, the court below relied entirely on our language in Davis. There, the court felt it could not “strain! ] to find inferences that [were] not available on the face of the complaint or the briefs submitted” because “Davis never sought nominal damages” nor did his “submissions to the court ever mention a claim to nominal relief.” Davis, 158 F.3d at 1349. But appellants here made a specific request for nominal damages in their opposition to the government’s motion to dis *267 miss. Brief for Plaintiffs at 86, Aref, et al. v. Holder, et al., No. 10-cv-539 (D.D.C. Mar. 19, 2013), EOF No. 102. Moreover, the plaintiff in Davis only requested compensatory and punitive-damages; his pleadings did not contain any catch-all prayer for relief. See Complaint, Davis v. District of Columbia, No. 97-cv-00092 (D.D.C. Jan. 14, 1997), ECF No. 1. By contrast; these appellants included a broad prayer for relief in their complaint. See Aref, 953 F.Supp.2d at 149 (“Plaintiffs respectfully request the Court ... [o]rder such other relief as this Court deems just and proper.”). We therefore find the reasoning in Davis inapt. The Federal Rules of Civil Procedure even indicate “[e]very other final judgment should grant the relief to which each party is entitled, even if the party has not demanded that relief in its pleadings.” Fed R. Civ. P. 54. Thus, we conclude appellants here have made out a sufficient claim for nominal damages. And we join our sister circuits in holding that an inmate who cannot make out “sufficiently certain” claims for compensatory damages is entitled to nominal damages, provided he proves an injury occurred. See Memphis Comm. Sch. Dis. v. Stachura, 477 U.S. 299, 308 n.11, 106 S.Ct. 2537, 91 L.Ed.2d 249 (1986) (“[NJominal damages ... are the appropriate means of ‘vindicating’ rights whose deprivation has not caused actual, provable injury.”). Overall, then, we conclude appellants are eligible to seek compensatory, punitive, and nominal damages under Section 1997e(e). They have asserted the following injuries: the disadvantage of being denied essential reintegration programming provided by BOP; the stigma of being designated to facilities known as “terrorist units;” the prolonged deprivation of First Amendment rights to political speech and activity; and the undue damage to primary family relationships. The district court held these harms to be too speculative and abstract to provide a basis for compensatory or punitive damages. See Aref, 953 F.Supp.2d at 148M:9. We need not evaluate these injuries, however, because we conclude infra that Smith is entitled to qualified immunity. We similarly decline to remand to the district court to consider — in the first instance — whether nominal damages would be appropriate. C. Qualified Immunity Having concluded the PLRA does not bar appellants’ claims, we turn at last to the question whether Smith is entitled to qualified immunity. Appellants can prevail only if they show “(1) that the official violated a statutory or constitutional right, and (2) that the right was ‘clearly established’ at the time of the challenged conduct.” Ashcroft v. al-Kidd, 563 U.S. 731, 735, 131 S.Ct. 2074, 179 L.Ed.2d 1149 (2011). For a right to have been “clearly established,” it must have been “clear to a reasonable officer that his conduct was unlawful in the situation he confronted.” Saucier v. Katz, 533 U.S. 194, 202, 121 S.Ct. 2151, 150 L.Ed.2d 272 (2001). The Supreme Court has cautioned us not to define the right at too high a level of generality; instead, we must examine the right in its “particularized” context. See Reichle v. Howards, — U.S. -, 132 S.Ct. 2088, 2094, 182 L.Ed.2d 985 (2012). Jayyousi and McGowan allege their continued confinement to the CMU (and, in McGowan’s case, his redesignation) was retaliatory action taken in violation of the First Amendment. At the outset, we note “the right in question [here] is not the general right to be free from retaliation for one’s speech, but the more specific right to be free” from retaliation in this particular, penological context. Id. . *268 Especially in the context of prison security, we cannot — and do not — require public officials to be perfect in their assessments. Indeed, qualified immunity is intended to allow “government officials breathing room to make reasonable but mistaken judgments about open legal questions.” Lane v. Franks, — U.S. -, 134 S.Ct. 2369, 2381, 189 L.Ed.2d 312 (2014); see also Malley v. Briggs, 475 U.S. 335, 341, 106 S.Ct. 1092, 89 L.Ed.2d 271 (1986) (observing that qualified immunity “provides ample protection to all but the plainly incompetent or those who knowingly violate the law”). We evaluate appellants’ arguments against this backdrop. With respect to Jayyousi’s claim, as previously discussed, appellants have failed to make out an adequate ease that his First Amendment rights were violated — much less that it would have .been clear to any reasonable officer in Smith’s position that denying Jayyousi’s transfer request on the basis of his sermon given during the prayer meeting would violate the First Amendment. McGowan’s claims are similarly untenable. Appellants point first to McGowan’s initial placement in the CMU in 2008, contending it was retaliation for protected political speech. But Smith had several reasons for recommending the placement; most notably, that McGowan’s conviction involved domestic terrorist activity and that he continued to communicate with individuals outside the prison involved in extreme environmental advocacy. While the First Amendment may protect this sort of speech and association generally, those protections are less robust in the prison context. See Pell, 417 U.S. at 822 (“[L]awful incarceration brings about the necessary withdrawal or limitation of many privileges and rights, a retraction justified by the considerations underlying our penal system.”). Moreover, placement in the CMU did not force McGowan to give up all methods of communication; it merely limited the frequency and amount. Even assuming McGowan could make out a First Amendment violation (an unlikely prospect), he certainly cannot show Smith violated any clearly established right when he recommended designation to the CMU. Appellants then point to Smith’s decision to redesignate McGowan to the CMU in 2011, alleging it was illegal retaliation for McGowan’s exercise of protected speech. But — after being returned to the general population — McGowan asked his wife to have his attorney send him law-enforcement sensitive documents, in an apparent attempt to circumvent communications monitoring. It was reasonable for an officer in Smith’s position to consider this attempted end-run around the prison’s monitoring systems when deciding whether redesignation would be prudent. Yet again, even if McGowan could make out a First Amendment violation here, he cannot meet the high bar of showing an official in Smith’s position would have known his actions violated a clearly established right. Smith is therefore entitled to qualified immunity on all of appellants’ individual-capacity claims. 18 VII. In sum, we hold appellants’ claims were not mooted by their transfer out of the CMU and they have a liberty interest in avoiding transfer into the CMU. We therefore reverse the grant of summary judgment on this claim and remand to the district court to determine whether the *269 government’s procedures comport with due process as applied to appellants. With respect to Jayyousi’s First Amendment retaliation claim, we hold he failed to establish any constitutional violation and so cannot prevail. Appellants’ individual-capacity claims against Lieutenant Smith survive his death and, moreover, constitute claims for actual though intangible harms that are neither mental nor emotional. Although these claims are eligible to be brought under the PLRA, they too fail because Smith is entitled to qualified immunity. For the foregoing reasons, we affirm the district court’s grant of summary judgment on both these claims. So ordered. Footnotes [1] It bears noting that both CMUs were opened before BOP established any written designation criteria. In April 2010, BOP published a proposed rule for public notice-and-comment. See 80 Fed. Reg. 3168 (Jan. 22, 2015). The final rule entered into effect on February 23, 2015, almost a decade after the first CMU opened. Id. [2] Before codification of the CMU regulations, the ultimate decisionmaker was BOP's Regional Director. The process otherwise has remained essentially the same. Initial consideration begins when an entity (institutional or otherwise) refers a prisoner to BOP's Counter-Terrorism Unit (CTU). The CTU creates a "designation packet” that includes a summary of the supporting information, a recommendation for or against, and a proposed Notice of Transfer. The packet is sent to the Office of General Counsel to be reviewed for legal sufficiency and then to the Correctional Programs Division. Previously, the Regional Director would distribute the packet to several administrators, allowing each to comment before making his final decision. Now the Assistant Director makes his assessment and decision independently. [3] Previously, the decision-maker was the Regional Director. [4] This case was originally assigned to Judge Urbina of the district court. It was transferred to Judge Rothstein on November 5, 2012, who ruled on this motion to dismiss and the subsequent summary judgment motions. See Aref v. Holder, No. 10-cv-0539, 2015 WL 3749621 at *2 n.1 (D.D.C. Mar. 15, 2015). [5] Leslie Smith passed away on March 16, 2015. The government did not file its notice of death until December 22, 2015 — seven months later and after this appeal had begun. See Aref v. Lynch, Dkt. #1554923, at 45. [6] The government also argues voluntary cessation only applies if the cessation came about "because of” the litigation — an argument the district court says the government waived. See Aref, 2015 WL 3749621, at *4 n.3. The government claims the Ninth Circuit has "implied” this requirement. Pub. Utils. Comm’n of Cal. v. FERC, 100 F.3d 1451, 1460 (9th Cir. 1996). But neither this circuit nor the Supreme Court consistently has required a finding that the cessation was undertaken because of the litigation. See, e.g., Friends of the Earth, Inc., 528 U.S. at 193-94, 120 S.Ct. 693 (assessing voluntary cessation without any indication the plant’s shutdown, years after the case was filed, was a response to the litigation); Am. Iron & Steel Inst. v. ERA, 115 F.3d 979, 1006-07 (D.C. Cir. 1997) (analyzing as voluntary cessation an EPA policy change announced before litigation began). A defendant who ceased the challenged conduct for reasons unrelated to the litigation may have an easier time showing the challenged conduct is unlikely to reoccur, but "the cessation of an ongoing activity pending a lawsuit may [also] well imply an intent to renew the activity once the court has dropped out.” Clarke v. United States, 915 F.2d 699, 705-06 (D.C. Cir. 1990). We are therefore unpersuaded by the government’s argument that appellants must prove their transfers were "because of” this litigation. And even if so, circumstantial evidence indicates the transfers may have been motivated at least in part by the pending litigation: not a single prisoner was transferred back into general population during the first three years of the CMU, until a then-named plaintiff was transferred out on the eve *252 of this litigation. All other named plaintiffs subsequently were transferred out during the pendency of this litigation. [7] The Tenth Circuit has acknowledged some inconsistency in its application: "[wjhen considering whether the conditions, duration or restrictions of confinement are atypical as compared with other inmates, this court has inconsistently used comparisons either with inmates in the same segregation or those in the general prison population.” Jordan v. Fed. Bureau of Prisons, 191 Fed.Appx. 639, 650 (10th Cir. 2006). Notably, regardless of which baseline it has used, the circuit "has never held the conditions, duration or restrictions of the detentions presented on appeal created a liberty interest.” Hill v. Fleming, 173 Fed.Appx. 664, 670 (10th Cir. 2006). The Tenth Circuit also uniquely considers whether the prison action is "reasonably related to legitimate penological interests.” Jordan, 191 Fed.Appx. at 652-53 (finding no liberty interest in five year detention in administrative segregation because "it was commensurate with ongoing security concerns and a pending investigation”). [8] The Fifth Circuit has found a liberty interest in a few extraordinary cases involving solitary confinement that spans decades. See, e.g., Wilkerson v. Goodwin, 774 F.3d 845, 855 (5th Cir. 2014) (finding liberty interest for prisoner kept in solitary confinement for thirty-nine years given the exceptional duration and restrictive conditions of confinement). [9] The case was never resolved on its merits as it was ultimately dismissed as moot after the motion to dismiss stage. [10] The Tenth Circuit has also used its version of the Sandin analysis to evaluate whether inmates — transferred on the basis of their terrorism-related offenses — had a liberty interest in avoiding designation to the Administrative Maximum Prison (ADX). See Rezaq, 677 F.3d at 1013. [11] Amicus Curiae Seton Hall Center for Social Justice also alleges a liberty interest can be found under the Supreme Court's "stigma plus” test. To prevail on this claim, appellants must show the government is “the source of the defamatory allegations” and the resulting stigma involved "some tangible change of status vis-á-vis the government.” Doe v. U.S. Dep’t of Justice, 753 F.2d 1092, 1108-09 (D.C. Cir. 1985). Most important here, the reputation-tarnishing statement must be false. See Vega v. Lantz, 596 F.3d 77, 81-82 (2d Cir. 2010) (finding no viable "stigma plus” argument for plaintiff's sex-offender classification because he had in fact been convicted of a sex offense). All three of these appellants were convicted of terrorism-related activity. They therefore cannot satisfy this test’s defamation requirement. [12] The district court considered only the first Turner factor, finding "it [made] little sense” to inquire into the subsequent factors after finding for the government on the first. See Aref, 2015 WL 3749621, at *11. While the first factor is widely recognized as the most important, precedent indicates all four factors must be weighed. See, e.g., Lindell v. Frank, 377 F.3d 655, 657 (7th Cir. 2004) (stating “[flhere are four factors that courts must consider in determining whether a prison regulation is constitutional”); Jacklovich v. Simmons, 392 F.3d 420, 427 (10th Cir. 2004) ("The district court erred in not considering the remaining three Turner factors in the context of summary judgment.”). At least one court has even held a regulation violated the First Amendment because all but the first factor cut against the prison. See Lindell, 377 F.3d at 658-60. The district court’s failure to consider the last three factors can be remedied on appeal. [13] In its briefing, the government acknowledged "[t]here is ... no defendant to respond to the individual-capacity claims, and government counsel does not represent any party with respect to those claims.” Appellee Br. 1- . 2. However, the government decided to respond to appellants' claims against Smith as amicus curiae since “the United States has an interest in the proper resolution of constitutional claims against its employees.” Id. at 2. The government is entitled to file as amicus without the consent of the parties or leave of court. See 29 C.F.R. § 18.24. [14] But because the Davis court found other forms of relief (specifically injunctive and declaratory) were still available under Section 1997e(e), the court upheld the constitutionality of the statute under rational basis review. See id. at 1346, 1349. [15] Like Davis, these circuits sidestep concerns about Section 1997e(e) unconstitutionally foreclosing any relief by holding injunc-tive and declaratory relief remain available regardless of whether the plaintiff can show physical injury. [16] Notably, although there was a large increase in the absolute number of cases filed from 1975 to 1994 (6,606 cases to 39,065 cases, respectively), Congressional debate failed to account for the exponential growth in the prison population in -the interim. The rate of inmate filings actually dropped by approximately seventeen percent during that time period. Jennifer Winslow, Comment, The Prison Litigation Reform Act’s Physical Injury Requirement Bars Meritorious Lawsuits: Was It Meant To?, 49 UCLA L. Rev. 1655, 1662-63 (2002). [17] As noted above, circuits that have adopted the broader reading of Section 1997e(e)’s bar support the provision’s constitutionality by emphasizing the availability of alternative forms of relief: namely, injunctive or declaratory relief, punitive damages, and nominal damages. We need not consider the constitutionality question here but simply note the illusory nature, in practice, of such relief. Injunctive relief is commonly moot by the time a case is heard and cannot provide relief for past harms. Punitive damages are never awarded as a matter of right, and the standard is understandably high — requiring evil motive or reckless indifference to the rights of others. See Smith v. Wade, 461 U.S. 30, 51-52, 103 S.Ct. 1625, 75 L.Ed.2d 632 (1983). Finally, nominal damages do little to deter repetition of the illegal conduct and do not provide any compensation for actual harms suffered. Cf. Butz v. Economou, 438 U.S. 478, 506, 98 S.Ct. 2894, 57 L.Ed.2d 895 (1978) (“In situations of abuse, an action for damages against the responsible individual can be an important means of vindicating constitutional guarantees.”); see Doe v. District of Columbia, 697 F.2d 1115, 1124 (D.C. Cir. 1983). [18] Because we hold Smith is entitled to qualified immunity, and therefore uphold the district court's order dismissing all claims against him, we need not consider the parties' arguments regarding substitution óf a representative for the deceased.