concurring in part and dissenting in part:

Weighty state and personal liberty interests are placed at issue by an extremely important question raised at oral argument in this case: does the Supreme Court’s recent decision in Lawrence v. Texas, 539 U.S. 558 , 123 S.Ct. 2472 , 156 L.Ed.2d 508 (2003), which created a substantive due process right to private consensual sex, have any bearing on the precision with which the government must act when criminalizing an otherwise consensual sexual act on the ground that the sex partner is too retarded to consent to sexual contact? Although the panel reaches the issue, procedural obstacles and prudential considerations militate against doing so. Serious constitutional questions such as this one should not be decided after 90 seconds of oral argument presentation, on appeal from denial of federal habeas relief. I respectfully dissent, principally from the majority’s decision to reach this issue but, because they have determined to decide it, from the majority’s treatment on the merits of Anderson’s Lawrence argument as well.

I.

The Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), Pub.L. 104-132, 110 Stat. 1214, provides, in relevant part:

28 U.S.C. § 2254 .

The majority is apparently proceeding on the assumption — although it does not say so expressly — that Anderson’s Lawrence claim “was adjudicated on the merits in State court proceedings” as part of the general vagueness claim. Nowhere does the majority indicate that Anderson’s Law rence-based vagueness claim is unexhaust-ed.

That being the case, the question posed by AEDPA is: was the state court adjudication of Anderson’s vagueness claim “contrary to,” or “an unreasonable application *1034 of,” clearly established Supreme Court law at the time of the relevant state court decision? See, e.g., Lockyer v. Andrade, 538 U.S. 63, 71-72 , 123 S.Ct. 1166 , 155 L.Ed.2d 144 (2003) (“ ‘clearly established Federal law’ under § 2254(d)(1) is the governing legal principle or- principles set forth by the Supreme Court at the time the state court renders its decision”); Williams v. Taylor, 529 U.S. 362, 405 , 120 S.Ct. 1495 , 146 L.Ed.2d 389 (2000).

Of course, Lawrence v. Texas was not “clearly established Federal law” on November 18, 1998, the day the Oregon Supreme Court summarily affirmed the denial of Anderson’s state habeas petition. Rather, Bowers v. Hardwick, 478 U.S. 186 , 106 S.Ct. 2841 , 92 L.Ed.2d 140 (1986), was the law in November 1998, and that case limited the contours of the substantive due process right that Anderson seeks to invoke here to “family, marriage, [and] procreation,” none of which is implicated in this case. Id. at 191 , 106 S.Ct. 2841 (“any claim that [the prior substantive due process] cases nevertheless stand for the proposition that any kind of private sexual conduct between consenting adults is constitutionally insulated from state proscription is unsupportable”).

As is apparent, no case decided after November 1998, Lawrence included, has any bearing on the question whether Anderson is entitled to habeas relief under 28 U.S.C. § 2254 (d)(1) based on the proceedings thus far conducted in state court. The majority therefore has no justification for deciding the Lawrence question on the merits if it is part of the exhausted vagueness issue, given the procedural posture of this case; 1

II.

The other possibility is that the majority does not view the Lawrence variant as part of the exhausted vagueness issue but is deciding it anyway. That approach also has little to commend it. There are three vastly preferable alternatives.

First, the majority could have simply ruled the Lawrence claim unexhausted under. AEDPA and dechned to decide it. As the majority rejects the Lawrence claim, it is technically proper to forgo requiring exhaustion, as AEDPA so permits. See § 2254(b)(3). But surely, where, as here, the legal question is both novel and important, and has not been briefed in any court, both prudence and fairness dictate against such a precipitous course. Application of § 2254(b)(3) should be reserved for circumstances in Which the result on the unexhausted claim is a forgone conclusion, not open to any dispute. Absent that circumstancé, deciding an issue not litigated in the state courts is in tension with comity principles, as the state court may view the issue differently than we do.

Second, Anderson never raised a separate Lawrence claim before the district court. Thus, if the majority thought the Lawrence issue, although unexhausted, was properly raised in this appeal, a remand to allow the district court first to address the question would have been appropriate. Again, such a tack was not required. Appellate courts are entitled to decide pure questions of law not raised *1035 below, if doing so does not prejudice the opposing party. See, e.g., Ellis v. City of San Diego, 176 F.Sd 1183, 1191 n. 6 (9th Cir.1999). Nonetheless, reaching the merits of this complicated and important issue was not the more prudent path.

Third — and this would be my preferred approach — the court could remand the case to the district court with directions to stay proceedings while Anderson exhausts the Lawrence-based vagueness claim.

The Lawrence issue was raised for the first time at oral argument. We cannot fairly blame Anderson for tardiness, however. The Supreme Court’s decision in Laiorence was announced less than two weeks prior to oral argument, well after the close of briefing, and Lawrence expressly overruled prior precedent. 123 S.Ct. at 2484 (“Bowers v. Hardwick should be and now is overruled.”).

AEDPA requires a habeas petitioner to exhaust each of his federal claims in state court, making no exception for claims precipitated by new law decided by the Supreme Court after the relevant state court adjudication on the merits. 28 U.S.C. §§ 2254 (b) & (c); see also, e.g., Gatlin v. Madding, 189 F.3d 882, 887 (9th Cir.1999) (“It is not sufficient to raise only the facts supporting the claim; rather, the constitutional claim inherent in those facts must be brought to the attention of the state court.”) (internal quotation marks and citations omitted).

Anderson did not attempt to invoke before any of the Oregon courts a vagueness argument premised on a federal constitutional right of consenting, unmarried adults to engage privately in sexual conduct. It would therefore be proper, if one views the Lawrence claim as unexhausted, to remand the Lawrence-based claim to the district court so that leave to exhaust the claim through an Oregon post-conviction procedure might be granted.

Kelly v. Small, 315 F.3d 1063, 1070 (9th Cir.2003), held that a district court should ordinarily both permit a prisoner who filed a mixed petition, one raising both exhausted and unexhausted claims, to dismiss the unexhausted claims, and entertain a motion for a stay pending presentation of the unexhausted claims to an appropriate state court. 2 I have found no case authorizing or prohibiting a district court from employing this procedure when the Supreme Court decision giving rise to the claim had not yet issued at the time the state courts were adjudicating his claims.

Tyler v. Cain, 533 U.S. 656 , 121 S.Ct. 2478 , 150 L.Ed.2d 632 (2001), supports, albeit indirectly, the conclusion that new law announced while a case is pending on habeas can be pertinent to deciding the habeas petition. In Tyler , putting to one side Justice O’Connor’s concurrence, id. at 668-70 , 121 S.Ct. 2478 , four Supreme Court Justices interpreted 28 U.S.C. § 2244 (b)(2)(A) to prevent a prisoner raising a successive habeas petition from relying on a new rule of law unless the Court “held” the rule to be retroactive in a prior habeas case. Id. at 664-68 , 121 S.Ct. 2478 (‘We cannot decide today whether Cage is retroactive to cases on collateral review, because that decision would not help Tyler in this case. Any statement on Cage’s *1036 retroactivity would be dictum-”). If that is so, the only way for the Court initially to hold a new rule to be retroactive, such that future habeas petitioners would be able to take advantage of it on a successive petition, would be on a first petition. Even on a first petition, however, the Court would be unable to “hold” a new rule to be retroactive unless the rule was previously unavailable.

Section § 2254(e)(2) also envisions the existence of a habeas claim that relies on “a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable” but, unlike § 2244(b)(2)(A), is not limited to successive petitions. 3 Section 2254(e)(2) provides that if, on a first federal petition, an applicant raises issues based on a retroactively applicable new rule, the federal court should not hold an evidentiary hearing unless “the facts underlying the claim would be sufficient to establish by clear and convincing evidence that but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” § 2254(e)(2)(B). The necessary negative implication of that provision is that federal habeas petitioners may raise on federal habeas issues based on such new rules of constitutional law, although they must generally do so on the basis of the existing state court record.

Federal habeas relief can therefore sometimes be granted on the basis of claims that rely on new Supreme Court decisions, as long as the claim was exhausted in state court. A habeas petitioner raising a claim that relies on new Supreme Court law, announced while the federal habeas is pending and raising purely legal issues, should therefore be permitted to return to state court to exhaust his claim through an appropriate state court post-conviction procedure. 4

That Anderson’s Lawrence-based claim only originated after his case left the district court’s jurisdiction and entered ours does not change this conclusion. AED-PA’s exhaustion requirement entitles a state to pass on a prisoner’s federal claims before the federal courts do so. A state has no interest, however, in seeing that the prisoner’s ability to invoke new Supreme Court law in a first habeas petition expires upon the filing of a notice of appeal from a federal district court’s ruling, rather than at the time that his first petition has been conclusively ruled on by the federal courts. To require instead that Anderson raise his Lauyrence claim in a second habeas petition would, because of the severe limitations on second petitions, see, e.g., Tyler, *1037 533 U.S. at 667-68 , 121 S.Ct. 2478 , result in precluding entirely consideration of issues that, under the normal rules governing review within the federal courts, we would consider even if first raised on appeal because only purely legal issues are involved. See Ellis, 176 F.3d at 1191 n. 6

I would therefore remand to the district court with instructions to stay proceedings while the Lawrence-based vagueness claim is exhausted in state court.

III.

Because the majority nonetheless reaches the Lawrence issue on the merits, I do as well.

The only analysis the majority offers as to why Lawrence does not impact the vagueness inquiry in this case is the following assertion:

Ante at 1032-33.

It is true that Lawrence did not involve a case where consent was in doubt. Lawrence explicitly said as much: “[The present case] does not involve persons who might be injured or coerced or who are situated in relationships were consent might not easily be refused.” 123 S.Ct. at 2484. But that distinction does not conclusively decide this case. The Supreme Court was simply making clear what could just as well have gone unsaid — that the Court expresses no view as to issues not before it. On habeas, the question remains — or would remain, had the state court had an opportunity to address Lawrence 5 — whether declining to acknowledge that Anderson’s case implicates the right established in Lawrence is “unreasonable,” § 2254(d)(1), given that the Supreme Court has not directly addressed the current circumstances. See Williams, 529 U.S. at 407-08 , 120 S.Ct. 1495 (noting that a “state-court decision that correctly identifies the governing legal rule but applies it unreasonably to the facts of a particular prisoner’s case certainly would qualify as a decision ‘involv[ing] an unreasonable application of ... clearly established Federal law,’ ” and equating an unreasonable extension of a legal principle to a new context with an unreasonable refusal to extend an established principle to a new context).

What follows is my explanation as to why the sexual liberty interest set out in Lawrence v. Texas does indeed implicate the validity of Anderson’s conviction.

A. Background

As the facts provided by the majority are sparse, I spell out the factual background in a bit more detail:

At the time of the sexual encounter for which Anderson was prosecuted, JH was 26 years old and lived with her mother. She is 4'7" tall, has the emotional maturity of a 6 — 8-year-old child, and is moderately mentally retarded. She has “a little” vision in one eye, and “fairly good” vision in the other. She also has a hearing deficiency but does not always wear a hearing aid. It is difficult to understand her speech.

*1038 As an adult, JH has had at least two boyfriends with whom she had sexual relationships. After both relationships had ended, with one former boyfriend she “continued a sexual relationship ... a couple of times a year.” She has also received counseling at the Benton County Mental Health Center, where she learned about using condoms. Prosecution experts testified that, “She understands that a man is on top and the woman is on the bottom and the penis is inserted into the vagina,” and that she understands the health and pregnancy risks of not using condoms, but that she does not understand dating norms, such as the circumstances in which kissing or holding hands is appropriate dating behavior.

Others also testified as to JH’s sexual behavior. Anderson’s wife testified that in June 1993, roughly one year after the sexual encounter for which Anderson was prosecuted, she witnessed JH display to a mutual friend six hickeys that she had received on her neck. JH exhibited happiness about the marks. Also, Tina Pahre, an acquaintance of Anderson’s, testified that in 1992 she had frequently seen JH at Pahre’s apartment complex. JH’s apparent purpose was to visit a man named Jim, who “[s]he seemed to really, really like.” According to Pahr, even when Jim was not there, JH would visit other men at the apartment complex, “follow them up to their apartment ... [e]ven if they didn’t want her there ... and flirt with them.” This behavior continued for a month.

Additionally, Anderson testified at trial about a 1989 consensual sexual encounter with JH, for which he was not prosecuted.

The 1992 sexual encounter for which Anderson was prosecuted began when Anderson saw JH as he was driving home from work. JH was walking home from a bar where she had been dancing, an activity in which she regularly engaged, and drinking. Anderson claims that she accepted a ride with him and that they proceeded to a secluded parking lot. According to him, they conversed about them previous intimate relationships. While he suspected that she might be deaf, he maintained that he did not realize that she was mentally retarded. Anderson claims that JH initiated the consensual sexual encounter that followed.

JH’s testimony, which mostly consisted of her affirmations of the prosecutor’s leading questions, was that Anderson grabbed her by the arm and told her to get into the car. She also testified that Anderson was drinking, that Anderson removed her pants despite her protest, and that he forcibly raped and sodomized her. The medical examiner testified that JH suffered no bruising or injuries anywhere on her body.

Anderson was charged with five counts of criminal conduct: (1) kidnapping, (2) forcible rape, (3) rape of “a person who was incapable of consent by reason of mental defect,” (4) forcible sodomy, and (5) sodomy of “a person who was incapable of consent by reason of mental defect.” The jury acquitted him of Counts I, II, and IV. Thus, the jury rejected JH’s version of the incident, as it concluded there was no forcible rape or sodomy. Instead, the basis for Anderson’s conviction was non-forcible sexual activity with a person legally incapable of consent.

The statutory scheme under which Anderson was convicted provides:

OR. REV. STAT. § 163.375. Additionally,

OR. REV. STAT. § 163.305(3). As the trial judge made clear in Anderson’s case, the applicable intent requirement, knowledge, extends only to knowingly having sexual intercourse. A defendant need not have known that his sexual partner was incapable of consent by reason of a mental defect. See State v. Phelps, 141 Or.App. 555 , 920 P.2d 1098, 1100 (Ct.App.1996).

There is, however, an affirmative defense:

OR. REV. STAT. § 163.325(3). The defendant bears the burden of proof as to this defense, see Phelps, 920 P.2d at 1100 , which is limited to not knowing the facts and conditions responsible for the incapacity to consent. It is not a defense that the defendant was unaware of the incapacity to consent as a result of those facts or conditions. Thus, if Anderson knew that JH was retarded, the defense fails, whatever his evaluation of her capacity to consent.

In short, as instructed, the jury did not need to find that Anderson’s sexual conduct with JH was in fact non-consensual, nor did it need to find that Anderson knew JH was incapable of consent. It simply had to find that JH was incapable of consenting, and that he had not proven that he was unaware of her mental retardation.

B. Analysis

Anderson contends that the phrase “incapable of appraising the nature of the conduct” in the definition of “mentally defective” is unconstitutionally vague. At trial, the jury instructions simply repeated this statutory language without elaboration. 6 Thus, any uncertainty we perceive when applying the statute now is the same uncertainty the jury would have perceived when deciding Anderson’s guilt.

The degree of permissible vagueness depends on various characteristics of the statute:

Village of Hoffman Estates v. Flipside, Hoffman Estates, Inc., 455 U.S. 489, 498-99 , 102 S.Ct. 1186 , 71 L.Ed.2d 362 (1982) (footnotes omitted).

In Anderson’s case, a relatively high standard of precision is required. This is a criminal conviction for personal, as opposed to economic, conduct. Additionally, as noted, there is no scienter requirement as to the ability of the victim to consent.

Further, the affirmative defense shifts the burden of proof to Anderson, requiring him to prove his lack of knowledge by a preponderance of the evidence, as opposed to merely showing that a reasonable doubt exists. Cf. United States v. Mazurie, 419 U.S. 544 , 550 n. 9, 95 S.Ct. 710 , 42 L.Ed.2d 706 (1975) (noting, in the course of its vagueness analysis, the due process problems that arise when the burden of proof is shifted to a criminal defendant). Moreover, even aside from the shift in the burden of proof, the scienter requirement in the affirmative defense does not significantly reduce the vagueness problem here. Anderson could know of JH’s mental retardation, thus ruling out the affirmative , defense, without knowing that she was “incapable of appraising the nature of [her] conduct.” There are degrees of mental retardation, of course, and many retarded people do engage in consensual sexual activity. See generally R.K. MO-NAT, SEXUALITY AND THE MENTALLY RETARDED: A CLINICAL AND THERAPEUTIC GUIDEBOOK (1982).

The final consideration that governs the degree of vagueness allowed in Anderson’s prosecution is whether constitutionally protected conduct is implicated. As I demonstrate below, section 163.305(3)’s definition of “mentally defective,” which ultimately determines whether or not JH can legally consent to sex, implicates the sexual liberty interest fleshed out in Lawrence v. Texas, and therefore does impinge directly on constitutionally protected conduct.

1. Substantive Due Process

The Supreme Court in Lawrence struck down Texas’s same-sex sodomy law. 123 S.Ct. at 2484. In doing so, the Court affirmed the existence of a substantive due process right to engage in private, consensual sexual conduct. If the application of section 163.305(3) to Anderson’s case implicates the protected conduct discussed in Lawrence , that statutory provision should be held to a heightened degree of certainty-

The question, then, is whether the sexual liberty interest outlined in Lawrence regulates the manner in which a state drafts and applies its statutory rape law as applied to adult victims. I believe it does.

Before proceeding, it is crucial to note the interrelationship between JH’s sexual liberty interest and Anderson’s: If JH has, in certain circumstances, a constitutionally protected right to consent to sex and she does in fact consent, 7 then there is no constitutionally legitimate basis, under Lawrence , to preclude Anderson from having sex with her in those circumstances. 8 Anderson therefore need not satisfy the doctrinal requirements of jus tertii or third-party standing (although I believe that he could). He asserts his own right to engage in private consensual conduct, not JH’s.

*1041 As written and as read to the jury at Anderson’s trial, the language of section 163.305(3) is susceptible to at least two constructions that could, for reasons I survey below, impermissibly limit JH’s ability to consent to sex: (1) the construction, ultimately proffered by the prosecution in this case after some wavering, that JH is never able legally to consent to sex; and (2) a construction that invites law enforcement, including police officers, prosecutors, judges, and juries, to impose its own sexual mores upon JH when deciding whether her consent was valid. Because section 163.305(3)’s indefiniteness is responsible for the availability of at least one construction, section 163.305(3) clearly implicated constitutionally protected conduct in this case.

The first reason why section 163.305(3), as written and as read to the jury at Anderson’s trial, could implicate constitutionally protected conduct is that it is susceptible to a construction that JH will never -be able legally to consent to sex. The jury could have easily interpreted the phrase “incapable of appraising the nature of the conduct” to mean that either JH is always capable of consenting to sex or she never is. This binary view of mentally retarded individuals generally and JH in particular might well be an unconstitutional imposition on their sexual liberty. Despite the lack of a consistent clinical definition of what constitutes a “valid” consent, see infra note 10, there is clear consensus among experts in the field of mental retardation that mentally retarded individuals experience sexual desire and can meaningfully consent to sex in some situations. 9 The prosecution expert witnesses conceded as much at trial.

Moreover, JH has evidently engaged in voluntary sexual intercourse on a number of occasions in a manner that offended neither her mother nor the state. Given the general clinical belief that mentally retarded individuals desire and can “ethically” consent to sex, as well as JH’s own desire and demonstrated capacity to understand and appreciate sexual contact, it could well be unconstitutional for Oregon law to hold that JH can never legally consent to sex. Section 163.305(3) is readily susceptible to this type of potentially unconstitutional application.

To conclusively demonstrate unconstitutionality on this ground, however, would require a more lengthy analysis than I have performed here, and would then still fall short of' showing compliance with § 2254(d)(l)’s requirement that a contrary result would involve an “unreasonable application” of clearly established Supreme Court law. I need not engage in that endeavor, however, because I am convinced that section 163.305(3) implicates constitutionally protected conduct for a different reason.

Specifically, the statutory provision alternatively invites those applying the law to invoke their own sexual mores and override JH’s sexual choice when deciding whether JH is capable of consent in a particular instance. During the pretrial hearing, the prosecutor first interpreted the statutory scheme in precisely this contextual manner. The prosecution’s initial attempt at interpreting the statute was to propose a rule that JH’s mother could *1042 provide consent on JH’s behalf. Realizing the problem with that interpretation, the prosecution suggested that JH’s consent would be legally valid in the context of a “boyfriend-girlfriend” relationship. Finally, the prosecution arrived at the following reading of the statute: “I guess, what we would argue is that she is incapable of consent under all circumstances. It’s just that in some circumstances we would not charge a crime and those were circumstances that were condoned by her mother, condoned by her in the sense of relationship and that gave her an individual right to life — life like the rest of us lead.”

Expert testimony at trial also suggested an invitation to apply one’s own moral framework to JH’s sexual choice. In explaining why JH’s consent was not valid, the prosecution’s non-medical expert on sexually abused, mentally retarded individuals testified that whereas JH sees “sex” as merely a physical act, “If you ask, you know, anyone else what sex was or what intercourse is you see an entire picture. You see the candles, the wine, the dating, you know, whatever else goes on. With her sex is just one quick spur of the moment thing.”

That the state may not burden a particular sexual choice out of distaste or disagreement is the central holding of Lawrence, 123 S.Ct. at 2478 (“When sexuality finds overt expression in intimate conduct with another person ... [t]he liberty protected by the Constitution allows homosexual persons the right to make this choice.”). Taking Anderson’s version of the facts as true (the jury having rejected the only other version of the facts available), JH’s sexual choice was clearly demonstrated and uncoerced.

The line of cases concerning the analogous right of the mentally disabled to refuse medical treatment support the conclusion that the state is not free simply to ignore JH’s particular sexual choice. In Youngberg v. Romeo, 457 U.S. 307 , 102 S.Ct. 2452 , 73 L.Ed.2d 28 (1982), Nicholas Romeo, a severely mentally disabled individual with an IQ below 10 who had been involuntarily confined and adjudged an incompetent, challenged certain conditions of his confinement, including the administration of bodily restraints. Id. at 311-12 , 102 S.Ct. 2452 . After recognizing Romeo’s constitutional liberty interest in refusing the bodily restraints, the Court discussed two considerations justifying the liberty deprivation: the state interest in maintaining the safety and security of the institution, and the entitlement of the state to rely on the judgment of qualified health professionals when administering involuntary treatment. See id. at 321-23 , 102 S.Ct. 2452 ; see also Washington v. Harper, 494 U.S. 210, 223-23 , 110 S.Ct. 1028 , 108 L.Ed.2d 178 (1990) (discussing only these two considerations as the legitimate state interests in the involuntary administration of antipsychotic drugs to a mentally ill prisoner).

Notably, the Court has discussed no other legitimate state interest in disregarding the desires of the mentally disabled to refuse medical treatment, and neither of the two considerations that the Court did discuss is applicable here. JH has not been involuntarily confined and poses no safety threat to herself or those around her. More important, Oregon’s statutory rape scheme places the consent determination in the hands of a jury, not a psychiatrist, clinician, or parent. 10 Thus, those legitimate state interests that have been *1043 considered sufficient to override the constitutionally protected choices of mentally disabled individuals are not applicable here.

So, while the state surely has a very strong, legitimate interest in ensuring that the consent of a mentally disabled individual is knowledgeable and truly voluntary, and in disregarding that consent in situations where the alleged victim does not understand either the circumstances and consequences of sexual conduct or the extent of her ability to refuse sex, the state has no legitimate interest in imposing sexual mores on retarded individuals or their consensual partners. 11 When considering how to construct a regime that both respects the sexual choice of the mentally retarded and protects them from predation, others have recognized this distinction between an appropriate voluntariness inquiry and inappropriate moralizing. See, e.g., State v. Sullivan, 298 N.W.2d 267, 271 (Iowa 1980); People v. Cratsley, 86 N.Y.2d 81 , 629 N.Y.S.2d 992 , 653 N.E.2d 1162, 1165 (1995); Deborah W. Denno, Sexuality, Rape, and Mental Retardation, 1997 U. ILL. L. REV. 315, 321 (1997).

2. Vagueness

As part of a criminal statute that targets constitutionally protected personal conduct without an effective mitigating scienter requirement, section 163.305 is precisely the type of law that must convey its interdic-tions with a heightened clarity to be valid. See Village of Hoffman Estates, 455 U.S. at 498-99 , 102 S.Ct. 1186 . A law’s vagueness is evaluated in light of the two policies underlying the doctrine, providing fair warning to the defendant and establishing standards to guide law enforcement, the latter being more important than the former. See Kolender v. Lawson, 461 U.S. 352, 357-58 , 103 S.Ct. 1855 , 75 L.Ed.2d 903 (1983). I focus on this second element, noting, however, that the lack of guidance to law enforcement and the lack of notice to Anderson are by no means unrelated matters. 12

*1044 That section 163.305 does not establish meaningful standards to guide law enforcement rang true in a very literal sense during Anderson’s trial: the trial court failed to supply any explanation to the jury as to what “incapable of appraising the nature of the conduct” means. Thus, jurors may well have interpreted section 163.305(3) either as a blanket disqualification of any attempt by JH to consent to sex, or as an invitation to apply their own moral views to JH’s conduct. The former possibility flows directly from the statutory language “incapable of appraising the nature of the conduct.” The latter does as well. A natural “human response,” to being asked about the “nature ” of particular sexual conduct is to focus on one’s own sense of the morality or wisdom of that conduct. See Cratsley, 629 N.Y.S.2d 992 , 653 N.E.2d at 1165 . That section 163.305(3) is susceptible to the second, contextual interpretation renders it a paradigmatic violation of the prohibition against vague criminal laws. 13

The contextual interpretation of section 163.305 squarely conflicts with the precepts underlying the vagueness doctrine. The open-ended character of section 163.305, which potentially allows those applying it to invoke their own personal moral standards, “impermissibly delegates basic policy matters to policemen, judges, and juries for resolution on an ad hoc and subjective basis, with the attendant dangers of arbitrary and discriminatory applications.” Grayned, 408 U.S. at 108-09, 92 S.Ct. 2294 . Subjecting Anderson to the prevailing social morality on sexual conduct is particularly dangerous, because of “the discredited but long-held view of mentally retarded females as either asexual or hypersexual — perceptions fueled, in part, by fear of their procreation.” Denno, supra, at 321.

Even before Lawrence , when governmental regulation of private adult sexual conduct was not subject to heightened scrutiny, the Iowa Supreme Court held a statute criminalizing sexual activity with a person who “lacks the mental capacity to know the right and wrong of conduct in sexual matters” unconstitutionally vague. See Sullivan, 298 N.W.2d at 272 (reasoning that “[ijmpossible uncertainty over the so-called general mores” renders a moral “right and wrong” test “an unfit tool for determining the mental competency of a person to consent to a sex act”). As the jury, due to lack of instruction, could have interpreted section 163.305 as a morality-based “right and wrong” test, Anderson’s conviction violated due process.

*1045 I have no doubt that the statutory scheme could be applied in a more restrained manner that comports with constitutional directives. A meaningful scienter requirement is the most obvious adjustment.

Another way to bring the statute into line with constitutional norms, albeit one rejected in Callender, is to construe the statute as applying only when the victim is not able to comprehend the physical aspects of the sexual act and its physical consequences — pregnancy, for example, and disease. That test for capacity to consent would be quite clear, and appears to be the majority interpretation for statutes of this sort. See Elizabeth J. Reed, Criminal Law and the Capacity of Mentally Retarded Persons to Consent to Sexual Activity, 83 VA. L. REV. 799, 813-14 & n. 110 (1997). Under it, Anderson could not have been convicted, as the evidence established that JH did have that degree of understanding.

The path Oregon instead chose to pursue after Anderson’s conviction — tying capacity to consent with the “ability to contemplate and assess the ‘right or wrong’ and the ‘moral quality’ ” of the sexual act, see Callender, 47 P.3d at 520 — is problematic, see Sullivan, 298 N.W.2d at 271 (rejecting, on vagueness grounds, precisely this interpretation of the Iowa statute), but not unsalvageable. With improved jury instructions, the Callender approach might survive a vagueness challenge. Idaho, for instance, also requires that the putative victim be able to assess the moral quality of the act, but makes clear in its detailed jury instructions that it is the mental capacity for moral reasoning of the consenting individual, not the morality of the act itself, that is legally relevant. See State v. Soura, 118 Idaho 232 , 796 P.2d 109 , 113 n. 1 (1990). 14

In any event, the manner in which the statute was applied was unconstitutional both as to Anderson and as to JH.

I respectfully dissent.

. By contrast, Anderson’s Lawrence claim is not barred by the doctrine of Teague v. Lane, 489 U.S. 288 , 109 S.Ct. 1060 , 103 L.Ed.2d 334 (1989). Lawrence falls quite cleanly into the first Teague exception, because, by striking down Texas’ sodomy law, it placed "certain kinds of primary, private individual conduct beyond the-power of the criminal lawmaking authority to proscribe.’’ Id. at 307 , 109 S.Ct. 1060 . Section 2254(d) and Teague , however, are independent barriers to habeas relief. See Horn v. Banks, 536 U.S. 266, 272 , 122 S.Ct. 2147 , 153 L.Ed.2d 301 (2002).

. An aspect of the doctrine applied in Kelly v. Small is currently undergoing Supreme Court review in Pliler v. Ford, - U.S. -, 124 S.Ct. 981 , 157 L.Ed.2d 811 (2004) (granting writ of certiorari to review Ford v. Hubbard, 330 F.3d 1086 (9th Cir.2003)). I proceed under the assumption that the most basic form of the procedure fleshed out in Kelly — a permissive stay by a district court to allow for exhaustion of unexhausted claims in a mixed petition- — -will remain in effect, that procedure having been expressly recognized as appropriate in Stewart v. Martinez-Villareal, 523 U.S. 637, 644-45 , 118 S.Ct. 1618 , 140 L.Ed.2d 849 (1998).

. I note that Justice O'Connor's controlling concurrence in Tyler suggests that this is precisely the type of case involving a claim that relies on “a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court.” As Justice Breyer stated in his dissent:

. Importantly, not every new Supreme Court decision will warrant reexhaustion under the procedure I propose. Section 2244(b)(2)(A) and § 2254(e)(2)(A) explicitly require that the claim invoking the new Supreme Court decision be "previously unavailable.” While groundbreaking cases, overruling prior precedents, such as Lawrence will generally give rise to claims that were "previously unavailable,” many new Supreme Court decisions will not.

. The need for this counterfactual assumption confirms that we should have allowed the state courts to address the question in the first instance. Asking whether a state court decision is an "unreasonable” application of clearly established Supreme Court law when there was no chance for the state courts to address the new law distorts AEDPA’s federalism premises. The alternative, deciding on the merits, without § 2254(d)(1) deference, an unexhausted, nonfrivolous question is similarly at odds with comity principles.

. In State v. Callender, 181 Or.App. 636 , 47 P.3d 514 (2002), decided after Anderson's trial, the Oregon Court of Appeals provided some clarification as to what section 163.305(3) means. The state appellate court stated, "being capable of appraising the nature of a person’s conduct requires more than a mere understanding of the physical aspects of the conduct. Instead, it includes an ability to contemplate and assess the 'right or wrong' and the 'moral quality’ of the conduct." Id. at 520. As this limiting construction came ten years after the fact, and, more importantly, was not available to the jury at Anderson’s trial, it does not alter the vagueness analysis. At the same time, that a state appellate court interpreted the statute to include concepts of "right or wrong” indicates that the jury could have done so as well, as I later discuss.

. Again, the jury never found that JH did not consent to sex with Anderson.

. Of course, there may be a different basis, unrelated to lack of consent, for forbidding a given sexual act with a consenting adult — its public location, for example. Anderson’s prosecution, however, was grounded only in lack of consent.

. See, e.g., K.R. Held, Ethical Aspects of Sexuality of Persons with Mental Retardation, 10 SEXUALITY & DISABILITY 237, 237-55 (1992); MONAT, supra, at 58-59 (1982); L. Heshusius, Research on Perceptions of Sexuality by Persons Labelled Mentally Retarded 51-52, in MENTAL HANDICAP and SÉXUALITY: ISSUES AND PERSPECTIVES (Ann Craft ed., 1987). But cf. R. Binder, Sex Between Psychiatric Inpatients, PSYCHIATRIC Q. 121 (1985) (offering a dissenting view, but conceding that sexual relations may be therapeutic in some cases).

. Moreover, while deference to a health professional makes sense in the context of medical treatment, it may be less appropriate in the morally-infused context of sexual consent. See, e.g., John M. Niederbuhl & C. Donald Morris, Sexual Knowledge and the Capability of Persons with Dual Diagnoses to Consent to Sexual Contact, 11 SEXUALITY & DISABILITY 295, 304 (1993) (noting the lack of consis *1043 tent professional standards for evaluating capacity to consent).

. The Supreme Court has recognized an analogous principle in the First Amendment context:

R.A.V. v. City of St. Paul, 505 U.S. 377, 383-84 , 112 S.Ct. 2538 , 120 L.Ed.2d 305 (1992). Similarly, that a government may legitimately regulate the sexual conduct of "persons who might be injured or coerced or who are situated in relationships were consent might not easily be refused,” Lawrence, 123 S.Ct. at 2484 , does not mean that it may do so by imposing moral judgment concerning appropriate sexual behavior not applicable to others.

. It is likely that Anderson did not receive fair warning that his conduct was illegal. At the time of his sexual encounter with JH, the Oregon courts had not yet read any substance into section 163.305(3)’s definition of "mentally defective.” Thus, one might well have looked to New Jersey law, which employs an identical "appraising the nature of the conduct” standard. See, e.g., State v. Olivio, 123 N.J. 550 , 589 A.2d 597, 605 (1991) (holding that a mental defect exists when "at the time of the sexual activity, the mental defect ren *1044 dered him or her unable to comprehend the distinctively sexual nature of the conduct, or incapable of understanding or exercising the right to refuse to engage in such conduct with another”). JH’s understanding of the physical mechanics of sex and her trial testimony that she in fact said "No” and "Stop” suggest that she would not have been considered "mentally defective” under New Jersey law. Alternatively, for the reasons pertaining to legislative history provided in State v. Callender, 47 P.3d at 519-20 , one may have looked to Iowa law for guidance concerning the phrase "appraising the nature of the conduct.” Iowa law would have revealed, however, that the legislation upon which the Oregon statute had supposedly been modeled was partially struck down by the Iowa Supreme Court on federal constitutional vagueness grounds for impermissibly moralizing sexual conduct, and that only a "nature and consequences” standard survived. See State v. Sullivan, 298 N.W.2d 267, 272-73 (Iowa 1980). Because JH testified at trial that she understood the pregnancy and STD consequences of intercourse, she would likely have been capable of sexual consent under the Iowa regime. Thus, it is likely that Anderson did not receive fair warning that his conduct was illegal.

. As noted, supra at 1041-42, the first interpretation may well also unconstitutionally impinge on JH's right to engage in sexual behavior, but I need not so decide for purposes of this dissent.

. I note, however, that the Callender decision, as written, does not conclusively commit Oregon to one particular definition of capacity to consent, but instead is open-ended, adding to rather than detracting from the vagueness of the statutory scheme. 47 P.3d at 520 ("[B]eing capable of appraising the nature of the person's conduct requires more than a mere understanding of the physical aspects of the conduct. Instead, it includes an ability to contemplate and assess the 'right or wrong' and the 'moral quality’ of the conduct.”) (emphasis added).