concurring in the judgment.
In Tapia v. United States, — U.S. -, 131 S.Ct. 2382, 180 L.Ed.2d 357 (2011), the Supreme Court held that 18 U.S.C. § 3582(a) prohibits the federal courts from “impos[ing]” any portion of a prison sentence or “lengthenfing]” it in order to rehabilitate a criminal defendant. Id. at 2393. The majority opinion limits Tapia to the latter portion of its holding— that Tapia prohibits district courts only fi’om “identifiabl[y]” lengthening a prison sentence for rehabilitative purposes. I respectfully disagree. In my view, the majority opinion confuses the question of whether there was a Tapia error with the question of whether the error was prejudicial and therefore remediable upon appellate review. In so doing, it minimizes the shift in penological attitudes enshrined in § 3582(a) and recognized in Tapia as binding on the federal courts. Although I ultimately agree with the majority to affirm Krul’s within-Guidelines sentence, the district court should be affirmed not because it did not err, but because Krul has failed to demonstrate that the district court’s error affected his substantial rights. Therefore, I concur in the judgment only.
I.
Under the majority’s approach, a federal prison sentence violates Tapia only if an appellate court can isolate a discrete portion of the sentence as having been imposed solely for impermissible rehabilitative reasons, regardless of whether the district court’s sentencing colloquy reveals that rehabilitation was one of the factors that shaped its determination of whether the defendant should be sent to prison and how long he should stay there. Admittedly, this approach has the potential merit of being relatively simple to administer on appeal. But it has the distinct disadvantage of being incompatible with what Ta-pia commands.
18 U.S.C. § 3553(a)(2)(D) generally advises sentencing courts that rehabilitation is one possible consideration that may factor into its determination of an appropriate sentence. But the statute that specifically governs which considerations may guide a district court when it imposes a sentence of imprisonment (as opposed to supervised release, probation, or fines) provides that a district court “shall consider the factors set forth in section 3553(a) to the extent that they are applicable, recognizing that imprisonment is not an appropriate means of promoting correction and rehabilitation.” 18 U.S.C. § 3582(a) (emphasis added). In Tapia, the Supreme Court re*377solved this potentially conflicting statutory language by holding that § 3582(a) precludes a district court from using a prison sentence to promote a defendant’s rehabilitation: “ ‘[W]hen sentencing an offender to prison, the court shall consider all the purposes of punishment except rehabilitation — because imprisonment is not an appropriate means of pursuing that goal.’” United States v. Deen, 706 F.3d 760, 765 (6th Cir.2013) (quoting Tapia, 131 S.Ct. at 2389); United States v. Walker, 649 F.3d 511, 513 (6th Cir.2011).
Tapia directs that no portion of a prison sentence may be imposed for the purpose of rehabilitating the defendant, regardless of whether the prison sentence also serves other, legitimate penological ends. As the Court put it, § 3582(a) contains a succinct command to district courts: “Do not think about prison as a way to rehabilitate an offender.” Tapia, 131 S.Ct. at 2390. A district court therefore commits a Tapia error where it “considers rehabilitation when crafting a sentence of imprisonment.” United States v. Vandergrift, 754 F.3d 1303, 1310 (11th Cir.2014); see also United States v. Bennett, 698 F.3d 194, 199 (4th Cir.2012) (“Tapia ... drew a line’ between the consideration of a defendant’s rehabilitative needs when determining the fact or length of imprisonment — which is improper — and the consideration of the same when recommending treatment options or the location of confinement— which is altogether sound.”).
This is not to say, however, that a sentencing court errs every time that it mentions a defendant’s ability to be rehabilitated. Tapia itself recognized that it may be appropriate for a sentencing court to observe that a prison sentence will have the beneficial side-effect of affording an opportunity for a defendant to rehabilitate, so long as the court does not send the defendant to prison or keep him there for the purpose of giving the defendant the opportunity to do so. See id., 131 S.Ct. at 2392; United States v. Blackmon, 662 F.3d 981, 986 (8th Cir.2011). Thus, the “dividing line” separating impermissible from permissible conduct is “whether a sentencing court’s reference to rehabilitative needs was causally related to the length of the sentence” or was instead simply an observation about one of several “consequences and opportunities” attendant upon the fact that the defendant would be going to prison. United States v. Del Valle-Rodriguez, 761 F.3d 171, 174 (1st Cir.2014). The district court may hope that a defendant uses his time in prison wisely in order to better integrate with society upon his release, but it cannot send him to prison or prolong his stay there for the purpose of making it happen. See United States v. Grant, 664 F.3d 276, 282 (9th Cir.2011) (“When a judge imposes prison, he may wisely believe that it will have rehabilitative benefits, but those benefits cannot be the reason for imposing it.”).
Thus, the majority opinion is incorrect that Tapia is violated only if an appellate court can identify a discrete portion of the sentence that has been tacked on for rehabilitative purposes. Tapia’s dictate that a district court may not consider rehabilitation either when “imposing]” any portion of a prison sentence or when “lengthening]” it is comprehensive: it means that rehabilitative considerations may not factor into a prison sentence at all. See Tapia, 131 S.Ct. at 2390 (“Do not think about prison as a way to rehabilitate an offender.”). Under the majority’s contrary approach, a district court is permitted to impose a sentence for rehabilitative purposes so long as there are also other legitimate purposes driving the sentence. That may be perfectly acceptable as a matter of policy, but it is incorrect as a matter of law. Congress has chosen to preclude federal courts from considering rehabilitation *378when fashioning prison sentences, and Ta-pia cannot be read as permitting district courts to flout its direction.
After all, although it is possible to draw quite a fine line between Tapice compliant and Tapm-violative practices, Tapia’s insistence that rehabilitation be taken off the table when determining whether or how long to send a defendant to prison is not a mere technicality. Instead, as Tapia recognized, § 3582(a)’s prohibition of using incarceration for rehabilitative ends represents a fundamental shift in penological theory. Congress recalibrated its approach to rehabilitation in part because of the unfairly disparate effects of using indeterminate prison sentences as a behavioral tool to keep an offender in prison, see Tapia, 131 S.Ct. at 2387; in part due to a growing consensus that efforts at rehabilitation had largely failed, see id.; and in part out of a desire to punish offenders for what they had done rather than for who they were. Mistretta v. United States, 488 U.S. 361, 363, 109 S.Ct. 647, 102 L.Ed.2d 714 (1989). By enacting the current version of § 3582(a), Congress made the conscious decision to alter federal sentencing practices, choosing to allow only the imposition of a set term of imprisonment based on the offender’s deeds instead of consigning the offender’s release date to judicial officers with “very broad discretion” to determine whether the offender was in their view worthy of reentry into society. Id. A federal court may send a defendant to prison and keep him there in order to punish, deter, or incapacitate him, see 18 U.S.C. § 3553(a)(2)(A)-(C), but it cannot use the blunt instrument of institutional incarceration to try to. mold him into a socially acceptable citizen.
Thus, although the majority downplays the importance of Tapia and narrowly reads the sentencing transcript in this case in an attempt to avoid its implications, the statutory prohibition against sending a federal defendant to prison in order for him to become a “better” person is not simply a matter of semantics. It involves questions foundational to the federal peno-logical enterprise — issues that have been debated literally for centuries. See id. at 365-66; Jessica M. Eaglin, Against Neorehabilitation, 66 SMU L.Rev. 189, 214-18 (2013) (discussing some potentially disparate effects of rehabilitative emphases); Edward L. Rubin, The Inevitability of Rehabilitation, 19 Law & Ineq. 343, 367-76 (2001) (noting disputes over rehabilitation theory); Immanuel Kant, The Metaphysics of Morals 105 (Mary Gregor trans., Cambridge Univ. Press 1996) (“Punishment by a court ... can never be inflicted merely as a means to promote some other good for the criminal himself or for civil society. It must always be inflicted upon him only because he has committed a crime. For a human being can never be treated merely as a means.... He must previously have been found punishable before any thought can be given to drawing from his punishment something of use for himself or his fellow citizens.”) (emphases deleted).
II.
In the present case, the district court began the relevant discussion by observing, “So that brings us to today, and that brings us to the question of the sentence .... ” The district court explained that “this sentence” needed to both deter Krul from future misconduct “and provide you with some educational, correctional, and medical treatment at the same time.” According to the district court, the appropriate sentence “requires that you be taken out of the community for a while and you be given some programmatic opportunities .... ” After explaining, “I don’t like the direction” that Krul’s recidivist path was taking him, the district court an*379nounced that “[t]he sentence of this Court will be 63 months in the custody of the Federal Bureau of Prisons,” observing that “6[3] months should be able to give the Federal Bureau of Prisons an opportunity” to provide Krul with rehabilitative programs “that will deal with your ability to interact sociably with other people, will give you the ability to hold your anger in check, and will give you an understanding that laws must be obeyed.”
The import of the district court’s statements is obvious: the district court believed that the appropriate term of incarceration needed to provide Krul with rehabilitative programs “at the same time” that it would function to deter him from wrongful conduct and concluded that a 63-month prison term was long enough to give the federal prison system an opportunity to do so. Because the transcript reflects that the district court considered rehabilitative goals when fashioning Krul’s prison sentence, its conduct fell afoul of Tapia. See Walker, 649 F.3d at 513 (although Tapia does not preclude a district court from observing that the length of the sentence that it has imposed for other (legitimate) reasons may present an opportunity for rehabilitation, it precludes a district court from using rehabilitation as a factor when determining how lengthy of a sentence to impose in the first place).
The majority, however, interprets the district court’s initial use of the word “sentence” as referring not only to the chosen period of incarceration but to the “overall sentence,” including the subsequent period of supervised release — making the entire discussion Tapia-compliant. This reading of the transcript does not withstand principled scrutiny. The district court itself explained that “[t]he sentence of this Court will be 63 months” in prison. At that point, obviously, the district court used the term “the sentence” to refer to the contemplated term of incarceration, not to both the prison term and the following term of supervised release. It strains credulity to assert that the district court used the same term — “the sentence” — a mere two-and-a-half paragraphs (or roughly sixty seconds) earlier to mean something entirely different. The majority’s position is that two nearly contemporaneous uses of the identical term “clearly” have different meanings, but the far less strained interpretation is that the district court used the same word twice to mean the same thing.
The district court’s post-prison-sentence discussion of supervised release is also significant. It is difficult to conclude that the district court was speaking of supervised release when it explained to Krul that his “sentence” required that “you be taken out of the community for a while and you be given some programmatic opportunities.” A person on supervised release, obviously, is in the community, not out of it. And, as examples of the “programmatic opportunities” that the district court had in mind, the district court identified some specific rehabilitation programs that Krul had already started while in prison. Which makes sense: the district court had just indicated that Krul needed to receive rehabilitative programming “at the same time” that he was serving the portion of the sentence that would function to deter future wrongful conduct — i.e., the term of incarceration. In short, every indication suggests that the district court fashioned Krul’s prison sentence, not just his supervised release, with rehabilitation in mind. Indeed, as is typical in federal sentencing practice, the district court did not even mention supervised release until after it had finished determining the requisite term of incarceration.
Regardless of whether we disagree with Congress’s sentencing policy as expressed *380in § 3582(a), it is our duty to fairly enforce the law as written. After doing so, I conclude that at sentencing the district court violated § 3582(a) as construed by Tapia.
III.
Nevertheless, I agree with my colleagues to affirm-Krul’s sentence. After all, the question whether there was an ■ error at sentencing is different than whether defendant was prejudiced by the error. In my view, the disposition of this case depends upon whether Krul can demonstrate that the district court’s error affected his substantial rights.
Much of this portion of the analysis is controlled by the standard of review. As Krul observes, this court has previously treated Tapia claims as a substantive reasonableness issue that garners abuse-of-discretion review. See, e.g., Deen, 706 F.3d at 762. But the case upon which Deen relied for the proposition that “[b]as-ing a sentence on impermissible factors is an example of substantive reasonableness” is United States v. Webb, 403 F.3d 373, 385 (6th Cir.2005), which predated the Supreme Court’s subsequent elucidation in Gall v. United States, 552 U.S. 38, 128 S.Ct. 586, 169 L.Ed.2d 445 (2007), of the proper standards under which to review a federal sentence. See United States v. Herrera-Zuniga, 571 F.3d 568, 579-80 (6th Cir.2009) (observing that Gall undercuts the language cited from Webb); United States v. Bolds, 511 F.3d 568, 581 (6th Cir.2007) (explaining sentencing review after Gall).
Gall clarified that procedural reasonableness review applies to the process by which the district court arrives at its chosen sentence, whereas substantive reasonableness review requires a bird’s-eye assessment of the reasonableness of the 'sentence’s length, based on “the totality of the circumstances.” 552 U.S. at 51, 128 S.Ct. 586. Gall, in fact, identifies “failing to consider the § 3553(a) factors” as an example of procedural — not substantive — unreasonableness, and it seems apparent that an opposite error of exactly the same kind — namely, improperly considering an extra § 3553(a) factor (i.e., rehabilitation) — is likewise an error that implicates a sentence’s procedural, rather than substantive, reasonableness. Id. This court suggested as much in United States v. Vonner, 516 F.3d 382 (6th Cir.2008) (en banc), where we noted that the defendant’s challenge to “the reasonableness of the length of his sentence” was a substantive-reasonableness challenge, whereas a challenge to a sentencing issue “that became apparent as soon as the court finished announcing its proposed sentence and that counsel nonetheless declined the court’s invitation to address” was an allegation of procedural error that garnered only plain-error review. See id. at 386, 389. Cf. United States v. Tristan-Madrigal, 601 F.3d 629, 632-33 (6th Cir.2010) (“The essence of a substantive-reasonableness claim is whether the length of the sentence is greater than necessary to achieve the sentencing goals set forth in 18 U.S.C. § 3553(a).” (internal quotation marks omitted)); United States v. Rigas, 583 F.3d 108, 123 (2d Cir.2009) (defining substantive-reasonableness review as “providing] a backstop for those few cases that, although procedurally correct, would nonetheless damage the administration of justice because the sentence imposed was shockingly high, shockingly low, or otherwise unsupportable as a matter of law”).
A Tapia error likewise involves a claim that the district court’s sentencing process was flawed, is obvious at the time of sentencing, and could be remediated with a contemporaneous objection. See Von*381ner, 516 F.3d at 386. Gall and Vonner therefore strongly suggest that an alleged Tapia error is a claim of procedural unreasonableness that, if unpreserved, is reviewed only for plain error.
In Henderson v. United States, — U.S. -, 133 S.Ct. 1121, 185 L.Ed.2d 85 (2013), which was decided soon after Deen, the Supreme Court cast serious doubt on the continuing validity of this circuit’s characterization of a Tapia error as implicating á sentence’s substantive reasonableness. Henderson, after all, was predicated on the assumption that an unpreserved Tapia error garners only plain-error review. There, the defendant had been sentenced before the Supreme Court decided Tapia, and his counsel had not objected at sentencing to the district court’s consideration of rehabilitation at sentencing. The Supreme Court observed that, “since Henderson’s counsel had not 'objected [to the Tapia error] in the trial court, the Court of Appeals could not correct the error unless [Federal Rule of Criminal Procedure] 52(b) applied. The Rule, however, applies only if the error was ‘plain.’ ” Id. at 1125. Henderson went on to conclude that an error is “plain” as long as it is plainly wrong at the time of appellate review, even if it was not plainly incorrect at the time of the district court’s decision. Id. at 1124-25.1
Moreover, every other circuit that has decided the issue has chosen to review unpreserved Tapia claims for plain error, consistent with Henderson. See, e.g., United States v. Del Valle-Rodriguez, 761 F.3d 171, 174 (1st Cir.2014); United States v. Vandergrift, 754 F.3d 1303, 1307 (11th Cir.2014); United States v. Wooley, 740 F.3d 359, 368 (5th Cir.2014); United States v. Lifshitz, 714 F.3d 146, 150 (2d Cir.2013); United States v. Pickar, 666 F.3d 1167, 1169 (8th Cir.2012); United States v. Bennett, 698 F.3d 194, 200 (4th Cir.2012); United States v. Grant, 664 F.3d 276, 279 (9th Cir.2011); United States v. Cordery, 656 F.3d 1103, 1105 (10th Cir.2011). See also United States v. Godoy, 706 F.3d 493, 496 (D.C.Cir.2013) (declining to decide the question); United States v. Gallagher, 526 Fed.Appx. 658, 661 (7th Cir.2013) (assuming, without deciding, that plain error applied).
To the extent that Deen — relying upon this court’s pre-Gall and pr e-Henderson practice — suggested a result different than that reached by every other circuit that has considered the issue, Henderson has placed its approach in serious peril. After all, if this court’s pr e-Henderson approach was correct, then Henderson was “a fool’s errand.” Cf. United States v. Blewett, 746 F.3d 647, 658 (6th Cir.2013) (en banc) (noting similar effects in the Fair Sentencing Act context). This court’s pr e-Henderson approach therefore enjoys the dubious distinction of being both alone among the circuits and implicitly rejected by the Supreme Court-.
Given this authority, I conclude that plain-error review should apply to Krul’s unpreserved Tapia challenge. Beyond being mandated by pertinent case law, this approach makes practical sense. The specter of plain-error review for an unpre-served Tapia challenge incentivizes a criminal defendant to make a contemporaneous Tapia objection at sentencing, thereby giving the district court an opportunity to ensure that impermissible rehabilitative goals are not influencing its sentencing decision. See Vonner, 516 F.3d at 386.
*382On de novo review, a Tapia error results in a procedurally unreasonable sentence. But where a defendant has not raised the Tapia issue to the sentencing court’s attention, plain-error review requires him to show “(1) error (2) that was obvious or clear, (3) that affected [his] substantial rights and (4) that affected the fairness, integrity, or public reputation of the judicial proceedings.” Id. (internal quotation marks omitted). Because “[a] sentencing error affects a defendant’s substantial rights when there is a reasonable probability that, but for the error, the defendant would have received a more favorable sentence,” United States v. Ferguson, 681 F.3d 826, 833 (6th Cir.2012) (internal alteration and citation omitted), subjecting unpreserved Tapia claims to plain-error review means in practical terms that a defendant must identify some reason to believe that the district court would have given him a different sentence if it had not impermissibly considered rehabilitative goals.
Although the majority in this case affirms without reaching the issue of plain error, the result would be the same even if it had. “[T]he burden of establishing entitlement to relief for plain error is on the defendant claiming it,” United States v. Dominguez Benitez, 542 U.S. 74, 82, 124 S.Ct. 2333, 159 L.Ed.2d 157 (2004), and Krul has failed to show a reasonable probability “that his sentence would have been different but for the court’s consideration of rehabilitation.” Vandergrift, 754 F.3d at 1312. Krul has only pointed to the district court’s statements and stated that they violated Tapia. He then requests that, because there is no way of knowing what would have happened if the district court had taken rehabilitation entirely off the table, his case should be remanded for resentencing.
But relying on an absence of evidence is not the same as demonstrating a “reasonable probability” that the result at sentencing would have been different. “[B]e-cause the district court never expressed an intention to lengthen [Krul’s] sentence for rehabilitative purposes ... we would have to speculate whether he would have received a lighter sentence absent the court’s eomment[s] about the defendant’s [rehabilitative needs].” Pickar, 666 F.3d at 1169 (internal quotation marks omitted). Krul has therefore failed to demonstrate that his substantial rights were affected by the alleged error. Id. Cf. United States v. Tolbert, 668 F.3d 798, 803 (6th Cir.2012) (“[T]he district court ... appears to have selected a within-Guidelines sentence primarily due to the seriousness of the crime, the need for deterrence, and the need to protect the public.”); Vandergrift, 754 F.3d at 1312 (citation omitted) (Tapia error did not affect substantial rights where rehabilitation concerns did not “dr[i]ve the district court’s sentencing decision”); Bennett, 698 F.3d at 201 (Tapia error did not affect defendant’s substantial rights where the sentencing court stated that it “will impose a sentence that provides ample time for [intensive substance abuse treatment]” and that the sentence was intended to both “reflect the serious nature of the breach of trust” and “provide ample time for substance abuse treatment”). See also United States v. Culbertson, 712 F.3d 235, 243-44 (5th Cir.2013) (reversible Tapia error where the district court repeatedly referred to the defendant’s rehabilitation needs in imposing a sentence three times the Guideline range); Wooley, 740 F.3d at 365 (reversible Tapia error where sentence was crafted explicitly for the “purpose[ ]” of affording the defendant rehabilitative treatment); Cordery, 656 F.3d at 1105 (reversible Tapia error where district court expressly calculated the length of the *383sentence to permit the defendant to participate in rehabilitation programs).
For these reasons, although I disagree with the majority’s holding that the district court did not commit a Tapia error, I agree that Krul is not entitled to reversal of his within-Guidelines sentence, having failed to satisfy his burden under plain-error review. I therefore concur in the judgment only.
. Tapia itself also seems to have assumed that Rule 52(b)'s plain-error analysis would control the ability of a defendant to rely upon an unpreserved allegation of Tapia error. See id., 131 S.Ct. at 2393.