Linno Llenos died engaging in an act known as autoerotic asphyxiation. His widow and beneficiary, LeTran Tran, filed a claim with Minnesota Life Insurance Company, seeking the proceeds from Llenos's ERISA-governed life insurance policies. Minnesota Life paid most of her claims but denied coverage under Llenos's Accidental Death & Dismemberment policy riders. Minnesota Life determined Llenos's death was not accidental and fell under a policy exclusion for deaths resulting from "intentionally self-inflicted injury." The district court reversed, ruling that Llenos's death qualified as an accidental death and did not result from an intentionally self-inflicted injury.

Because a reasonable person would interpret Llenos's cause of death, autoerotic asphyxiation, to be an "intentionally self-inflicted injury," we reverse.

I. Background

The facts are not in dispute. In August 2016, while home alone in Wilmette, Illinois, Llenos hung a noose from a ceiling beam in his basement, stood up on a stool with the noose around his neck, and stepped off. Llenos died as a result. When Tran came home, she found her husband's body hanging in the basement and immediately called police. Though his death was initially reported a suicide, the medical examiner subsequently concluded from sexual paraphernalia on Llenos's body that he died performing autoerotic asphyxiation.

Autoerotic asphyxiation is a sexual practice by which a person purposefully restricts blood flow to the brain to induce a feeling of euphoria. "Asphyxiophilia" as defined in the DSM-5 is a subset of sexual masochism disorder, by which an "individual engages in the practice of achieving sexual arousal related to restriction of breathing." AMERICAN PSYCHIATRIC ASSOCIATION, DIAGNOSTIC AND STATISTICAL MANUAL OF MENTAL DISORDERS 694 (5th ed. 2013). The pleasurable feeling experienced during autoerotic asphyxiation derives from cerebral hypoxia, or brain cell death from deprivation of oxygen. Acute to severe hypoxia can lead to loss of consciousness in ten to twenty seconds, permanent brain damage in three minutes, and death in four to five minutes.

Llenos was covered by two life insurance policies, a Basic Insurance Policy and a Supplemental Insurance Policy. These provided $ 517,000 in coverage. Each policy also included Accidental Death & Dismemberment ("AD&D") policy riders. The Basic Insurance Policy provided an additional *382$ 10,000 of AD&D coverage, and the Supplemental Insurance Policy provided an additional $ 50,000 of AD&D coverage.

After her husband's death, Tran filed a claim with Minnesota Life, which paid the $ 517,000 but denied Tran's claim for the additional $ 60,000 in AD&D coverage based on two provisions (with identical text) in the policy riders. Minnesota Life concluded Llenos's death was not "accidental" under the AD&D riders. The insurer also took the position that Llenos's death fell under an exclusion for intentionally self-inflicted injury, which states:

In no event will we pay the accidental death or dismemberment benefit where an insured's death or dismemberment results from or is caused directly by any of the following: ... intentionally self-inflicted injury or any attempt at self-inflicted injury, whether sane or insane..."

(emphasis added). Tran appealed the decision internally at Minnesota Life and again was denied.

Tran then brought an action under the Employee Retirement Income Security Act ("ERISA"), 29 U.S.C. § 1132(a)(1)(B), seeking the AD&D coverage payouts. After reviewing the stipulated facts from both parties, the district court awarded judgment in favor of Tran under FED. R. CIV. P. 52(a). The court ruled that Minnesota Life had conceded the death was accidental, and the only issue in dispute was whether autoerotic asphyxiation qualified as an "injury" under the policy's language. After reviewing precedent on autoerotic asphyxiation from other circuits, the court determined that reasonable minds could disagree about whether Llenos's intentional inducement of cerebral hypoxia was a self-inflicted injury within the meaning of the AD&D rider language. Because all policy ambiguities must be construed in favor of coverage, the district court ruled that the exclusion for intentional injuries did not apply to autoerotic asphyxiation and entered judgment in favor of Tran. Minnesota Life filed this appeal.

II. Discussion

Challenges to ERISA benefit determinations under 29 U.S.C. § 1132(a)(1)(B) are reviewed de novo when, like here, the plan does not grant discretionary authority to the plan fiduciary. Firestone Tire & Rubber Co. v. Bruch , 489 U.S. 101, 115, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989) ; Cheney v. Standard Ins. Co. , 831 F.3d 445, 449 (7th Cir. 2016). We apply federal common law to interpret policy terms. Schultz v. Aviall, Inc. Long Term Disability Plan , 670 F.3d 834, 838 (7th Cir. 2012). The federal common law of insurance contracts requires "that Plan terms be interpreted in an 'ordinary and popular sense, as [they] would [be understood by] a person of average intelligence and experience.' " Sellers v. Zurich American Ins. Co. , 627 F.3d 627, 632 (7th Cir. 2010) (quoting Cannon v. Wittek Cos. Intern. , 60 F.3d 1282, 1284 (7th Cir. 1995) ). Where terms are ambiguous, courts construe them in favor of coverage. Santaella v. Metropolitan Life Ins. Co. , 123 F.3d 456, 461 (7th Cir. 1997).

Minnesota Life first challenges the district court's finding that the insurer waived its position that Llenos's death was not "accidental" under the language of the AD&D riders. But because the riders stipulate an accidental death is still excluded if it "result[ed] from or was caused directly by ... intentionally self-inflicted injury," and that is dispositive of this case, we address only the exclusions. To determine whether Llenos's death is excluded from AD&D coverage, we must determine first whether autoerotic asphyxiation is an "injury," and second, whether that injury was "intentionally self-inflicted."

*383A. Autoerotic Asphyxiation As "Injury"

We interpret the meaning of "injury" as a layperson would commonly understand the word. Sellers , 627 F.3d at 632. The district court's analysis of whether autoerotic asphyxiation (and the accompanying cerebral hypoxia ) is an injury relied on three cases: a 1997 case from our court, Santaella v. Metropolitan Life Ins. Co. ; Padfield v. AIG Life Ins. Co. , 290 F.3d 1121 (9th Cir. 2002) ; and Critchlow v. First UNUM Life Ins., America , 378 F.3d 246 (2nd Cir. 2004). In Santaella , we held that an accidental overdose death did not result from intentionally self-inflicted injury because there was no evidence the woman intended to injure herself when she took too much of a legal prescription painkiller. 123 F.3d at 465. Here, the district court applied Santaella to the question of injury because it was "cited approvingly by the Ninth Circuit in Padfield ... ." Tran v. Minnesota Life Ins. Co. , No. 17-cv-450, 2018 WL 1156326, at *7 (N.D. Ill. Mar. 5, 2018).

But Padfield did not rely on Santaella for its determination that autoerotic asphyxiation is not an injury. Padfield cited Santaella in support only of its "intentionally self-inflicted" analysis. 290 F.3d at 1129-30. Indeed, Santaella sheds little light on the question of whether autoerotic asphyxiation is an injury, because the opinion did not explore the issue in any depth. In Santaella , we simply stated the facts did not show that the insured meant to injure herself. 123 F.3d at 465. The dose the insured ingested was relatively low to be fatal, and she was unaware of other medical conditions that made her particularly susceptible to an overdose. Id. There was no evidence "she was aware of the risk of serious injury or death" when she ingested the painkiller. Id. The entire discussion of injury amounted to only two paragraphs and was largely dependent on the case's facts. For these reasons, we do not find Santaella instructive on whether autoerotic asphyxiation is an injury.1

We turn next to the other two cases the district court relied on, Padfield and Critchlow . Both dealt with deaths by autoerotic asphyxiation, and both addressed policy exclusions for intentionally self-inflicted injury. This court has never adopted the reasoning used in Padfield and Critchlow , and we decline to do so here.2 We find both cases grounded on a *384false premise: that the act of strangling oneself is severable into distinct phases and distinct injuries. In Padfield , for example, the Ninth Circuit reasoned that what killed the insured was not the autoerotic asphyxiation, but the continued asphyxiation that occurred after he blacked out. Padfield , 290 F.3d at 1129. The same reasoning was applied in Critchlow , in which the Second Circuit found that Critchlow's death "was not caused by 'partial' strangulation but by the total loss of oxygen for a sustained period." 378 F.3d at 260.

We reject such reasoning because it artificially separates one continuous act into two or more parts. The insured in Padfield did not strangle himself in a nonlethal manner, then involuntarily shift into a different form of lethal strangulation. He pulled a necktie tightly around his neck to cut off oxygen to his brain; as the self-strangulation continued, he gradually lost consciousness and eventually died. Padfield , 290 F.3d at 1123-24. The insured in Critchlow died under very similar circumstances. Critchlow , 378 F.3d at 250. For both men, there was no intervening cause, and no break in the chain of causation: one act of autoerotic asphyxiation caused the hypoxia that killed them. The same reasoning applies here: Llenos placed a noose around his neck and stepped off a stool, strangling himself. The resulting hypoxia caused his euphoria, his black out, and his death-all the result of one intentionally inflicted injury.

Even if we accept the Ninth Circuit's premise that Llenos's autoerotic asphyxiation injury could be viewed in different "stages" of strangulation, the partial strangulation he sought to inflict is still an "injury" as the term is commonly understood, and thus falls within the exclusion. See, e.g. , MAMSI Life & Health Ins. Co. v. Callaway , 375 Md. 261, 825 A.2d 995, 1007 (2003) ("We conclude that a layperson would understand partial strangulation to be an injury as that term is commonly used."). The Ninth Circuit disagreed, holding that no "persons of average intelligence and experience" would consider partial strangulation to be an injury. Padfield , 290 F.3d at 1129. To the contrary, we find that an ordinary person would consider choking oneself by hanging from a noose to be an injury, even if that strangulation is only "partial." For example, if Llenos had partially strangled another person, there would be no debate he had inflicted an injury. The criminal codes within our circuit confirm this: in Wisconsin, Indiana, and Illinois, partial strangulation not resulting in death is a prosecutable offense. See, e.g. , WIS. STAT. § 940.235 (2007), ("Strangulation and Suffocation"); IND. CODE§ 35-42-2-9 (2017) ("Strangulation"); People v. James , 2017 IL 160148-U, ¶ 21, 2017 WL 6722835 (affirming a finding of "great bodily harm" when the strangulation victim survived, but during strangulation "could not breathe, lost consciousness, and suffered bruising to the neck."). Partial strangulation, even when not intended to cause death, is an injury. See, e.g. , Critchlow , 378 F.3d at 265 (Van Graafeiland, J., dissenting) ("Partial strangulation is an injury. A suicidal motive is not required.").

The dissent asserts we have ignored the sexual nature and pleasurable aim of autoerotic asphyxiation. Even acknowledging both, we fail to see their relevance. That Llenos performed the act on himself and enjoyed the accompanying euphoria does not make partial strangulation less of an *385injury. Compare this with someone who engages in nonsuicidal self-injury, such as by cutting or burning himself. See AMERICAN PSYCHIATRIC ASSOCIATION, DIAGNOSTIC AND STATISTICAL MANUAL OF MENTAL DISORDERS 803 (5th ed. 2013) (DSM-5) (defining "Nonsuicidal Self-Injury" as when an individual has "engaged in intentional self-inflicted damage to the surface of his or her body of a sort likely to induce bleeding, bruising, or pain ... with the expectation that the injury will lead to only minor or moderate physical harm (i.e., there is no suicidal intent)."). The individual has still suffered an injury, regardless of the desired side effects. Autoerotic asphyxiation is no different. See Callaway , 825 A.2d at 1007 (Md. 2003) ("That the injured party also derived pleasure from the self-inflicted injury [of autoerotic asphyxiation ] does not mean there was no injury.").

Nor do we find relevant the popularity of autoerotic asphyxiation. The dissent contends no one would practice autoerotic asphyxiation if it were commonly understood to be an injury. This ignores scientific and psychological evidence to the contrary, including the DSM-5 which has an entire section that deals exclusively with sexual masochism disorder, defined as "sexual arousal from the act of being humiliated, beaten, bound, or otherwise made to suffer ..." DSM-5 at 694 (emphasis added). The same section defines autoerotic asphyxiation as a subset of sexual masochism disorder. Some people enjoy harming themselves. That harm is still an injury, regardless of its popularity or the pleasure some people may derive from it.

We also disagree with the Second Circuit's determination in Critchlow that oxygen deprivation, not partial strangulation, was the injury that killed the insured. See Critchlow , 378 F.3d at 260 ("Critchlow's death was not caused by 'partial' strangulation but by the total loss of oxygen for a sustained period."). Again, this improperly parses the causal chain of events: strangulation (the "injury") causes hypoxia, which leads to euphoria, then blackout, and eventually death. The Second Circuit's decision in Critchlow obscures the actual cause of death in autoerotic asphyxiation fatalities. For Llenos, as for the insured in Critchlow , there was no intervening cause or break in the chain of causation: absent the strangulation injury, Llenos never would have experienced hypoxia (and euphoria), lost consciousness, and died. Autoerotic asphyxiation was the ultimate and the proximate cause of Llenos's death. According to the language of the exclusion in the AD&D riders, then, the act of autoerotic asphyxiation was the "injury" that killed Llenos.

B. "Intentionally Self-Inflicted"

Having determined autoerotic asphyxiation is an injury, the remainder of our inquiry is straightforward. We decide whether an act was accidentally or intentionally done-as required by the "intentionally self-inflicted" exclusion in the AD&D riders-by applying the subjective/objective test we adopted in Santaella . 123 F.3d at 462-63. For an injury, we examine whether the injured individual had a subjective expectation of injuring himself, and if that cannot be determined, whether an expectation of injury was objectively reasonable.3 Id. Here, we need *386not reach the objective step in the analysis, because Llenos's subjective intent was clear.4 Llenos intentionally performed autoerotic asphyxiation. Because that act itself is an injury, Llenos's death falls under the policy exclusion for intentionally self-inflicted injuries.

This holding does not conflict with our holding in Santaella , as the dissent and the Ninth Circuit suggests. See Padfield , 290 F.3d at 1130 ("This case is analytically identical to Santaella ."). In Santaella , we concluded there was no record evidence to indicate the insured had intended to injure herself by taking the prescription painkiller. Santaella , 123 F.3d at 465 ; see also Padfield , 290 F.3d at 1131 (Leavy, J., dissenting) (distinguishing the overdose in Santaella from autoerotic asphyxiation ). That differs from here: Llenos intentionally strangled himself so he could experience hypoxiainduced euphoria. That strangulation itself, partial or otherwise, was an injury that he intentionally inflicted on himself, unlike the insured in Santaella .

Strangling oneself to cut off oxygen to one's brain is an injury, full stop. When that injury kills, it is "an intentionally self-inflicted injury which resulted in death," regardless of whether it was done recreationally or with an intent to survive. Padfield , 290 F.3d at 1131 (Leavy, J., dissenting); Callaway , 825 A.2d at 1007. Under the plain and ordinary meaning of Llenos's AD&D riders, his death is excluded from coverage.5

III.

This opinion does not purport to establish a per se rule on insurance coverage for autoerotic asphyxiation. Interpretations of insurance policies are rarely amenable to per se rules because the policy language and factual circumstances involved in a death can vary, sometimes greatly. See, e.g. , Todd v. AIG Life Ins. Co. , 47 F.3d 1448, 1453 (5th Cir. 1995) (declining to establish a per se rule on whether autoerotic asphyxiation deaths are accidental); Cozzie v. Metropolitan Life Ins. Co. , 140 F.3d 1104, 1110 (7th Cir. 1998) (declining to establish a per se rule on whether drunk driving deaths are accidental). But under the language of this policy's exclusion for AD&D coverage, Llenos died from an "intentionally self-inflicted injury." Even assuming Llenos's death were accidental, Tran is not entitled to AD&D coverage and an additional $ 60,000 payment.

For these reasons, we REVERSE the judgment.

I would affirm the decision of the district court. The district court found that reasonable people could conclude that Linno Llenos' ("Llenos") cerebral hypoxia was not an intentional injury under the terms of Minnesota Life Insurance Company's *387Accidental Death & Dismemberment policy riders ("Rider") and his death was an unexpected and unforeseen accident. Because reasonable people could conclude that his death was an accident, and ambiguities in the plan must be construed in favor of coverage, I believe that the district court properly determined that Llenos' death is covered by the Rider.

As the majority notes, the facts surrounding Llenos' death are not in dispute. During an act of manual-stimulation, Llenos engaged in self-strangulation. He hung himself by the neck and was unable to free himself before succumbing to cerebral hypoxia and ultimately death.

The district court's determination of whether the Rider's terms are ambiguous is subject to de novo review and construed in favor of the insured. Cheney v. Standard Ins. Co. , 831 F.3d 445, 450 (7th Cir. 2016). Findings of fact "must not be set aside unless clearly erroneous[.]" Fed. R. Civ. P. 52(a)(6). " '[A] finding is clearly erroneous when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.' " Madden v. United States Dep't of Veterans Affairs , 873 F.3d 971, 973 (7th Cir. 2017) (quoting United States v. U.S. Gypsum Co. , 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948) ).

The majority focuses its analysis on two prongs: whether autoerotic asphyxiation is an injury and, if so, whether the injury was intentionally self-inflicted. It concludes that because Llenos intentionally engaged in the sexual act which led to his death, even though his death was an accident, he is not entitled to coverage under the Rider. This analysis separates the manual-stimulation from the self-strangulation and erroneously concludes that autoerotic asphyxiation is an injury.

The majority cleaves the act into two separate actions: (1) the act of masturbation, and (2) the act of self-strangulation. This confuses the analysis; it erroneously divides one global process into two distinct acts and the majority focuses solely on the strangulation aspect while ignoring the contemporaneous masturbatory act. So, the question should be: could reasonably intelligent people conclude autoerotic asphyxiation is not intentionally injurious behavior? I believe they can.

When examining whether or not autoerotic asphyxiation is an injury, our analysis in Santaella and our sister circuits' perspective on the question is illuminating. Santaella v. Metro. Life Ins. Co. , 123 F.3d 456 (7th Cir. 1997).

In Santaella , this Court adopted the Fifth Circuit's methodology for analyzing whether a death under an accidental death policy was accidental: the court must determine "(1) that the deceased had a subjective expectation of survival, and (2) that such expectation was objectively reasonable which it is if death is not substantially certain to result from the insured's conduct." Santaella , 123 F.3d at 463 (quoting Todd v. AIG Life Ins. Co. , 47 F.3d 1448, 1456 (5th Cir. 1995) ). The same analysis should be imported to the question of whether or not the act at bar-autoerotic asphyxiation-is injurious; that is to say, did the practitioner reasonably expect not to injure himself and was that expectation objectively reasonable.

In Santaella , like here, there was no factual dispute. The decedent intentionally took a mild prescription pain killer and suffered an overdose. The medical examiner ruled out natural causes, suicide, homicide, and unknown causes and concluded that the decedent's death was the result of an overdose from propoxyphene at a level less than one-third the typically lethal blood level. This Court concluded that because *388the decedent had subjective expectation of survival and the objectively reasonably person would not think death a substantial certainty, the death was an accident.

Here, there is evidence that Llenos intended to weather the masturbatory episode unscathed. During the course of the investigation, Llenos' wife informed police that he was not suicidal, that the family's finances were secure, and she did not think he committed suicide. The report further states that there were prophylactic measures in place to mitigate the risk of injury during the act, specifically, a towel wrapped around his neck, his foot resting on a step stool, and a possible release mechanism. Lastly, the coroner found rubber rings around Llenos' genitals and noted that his "pubic hair was shaved in a semi-circular pattern consistent with prior use[.]" The record, limited thought it may be, indicated Llenos had a history of engaging in autoerotic asphyxiation and doing so without injury, leading one to the belief that the act, as it was intended to be performed, was not injurious.

Because the first prong of the analysis is satisfied, we should look to whether such expectation was objectively reasonable-that is to say injury was not substantially certain. For injury to be the substantially certain result of autoerotic asphyxiation, the objectively reasonable person would have to expect that the injury was the likely outcome from the act. See Santaella , 123 F.3d at 462 (quoting Wickman v. Nw. Nat. Ins. Co. , 908 F.2d 1077, 1088 (1st Cir. 1990) ); see also Med. Protective Co. of Fort Wayne, Indiana v. Am. Int'l Specialty Lines Ins. Co. , 911 F.3d 438, 449 (7th Cir. 2018), reh'g denied (Jan. 29, 2019) (briefly discussing substantial certainty in the known loss context as "virtually inevitable"). This objective prong was also discussed by our sister circuits.1

In Padfield v. AIG Life Ins. Co. , 290 F.3d 1121, 1127 (9th Cir. 2002), and Critchlow v. First UNUM Life Ins. Co. of Am. , 378 F.3d 246, 258 (2d Cir. 2004), the Ninth and Second Circuits acknowledged that asphyxiophilia and autoerotic asphyxiation are widely practiced sexual acts. If it were objectively reasonable to conclude that an injury, as commonly understood, was the likely result of autoerotic asphyxiation there would be few if any repeat practitioners. Moreover, autoerotic asphyxiation has permeated popular culture and has become a commonplace punchline. Accordingly, I believe the objective prong of the analysis is satisfied.

Because Llenos had a subjective expectation of escaping unscathed and the objectively reasonable person would not think an injury was a substantial certainty, his death should not be deemed the result of an intentional injury. Therefore, the Rider's exclusion would not apply.

The majority's position incorrectly separates the masturbation from the asphyxiation and as a result creates a rule where it will always be excluded from coverage under an accidental death & dismemberment policy that includes an intentional injury exclusion. Llenos' conduct was undoubtedly risky but was not inherently injurious. The determination that autoerotic asphyxiation is an injury ignores that fact that when done correctly it can and does have a recreational purpose with no lasting health consequences.

By contrast, in the case of the skydiver or bungee jumper whose equipment malfunctions, the injury is the sudden stop when the thrill seeker crashes back to earth, but the act-the jump-was an inten *389tional one. But, the subjective and objective intent was to survive. Like the thrill seeker or extreme athlete who steps out of an airplane or purposefully exposes himself to outrageous conditions, the asphyxiophiliant is not necessarily acting injuriously.

The majority focuses its inquiry on the injury aspect of Llenos' self-strangulation incorrectly concluding that any amount of asphyxiation is injurious and therefore autoerotic asphyxiation must be excluded under the Rider. But, as the district court points out, "reasonable minds could differ on whether the term 'injury' as used in the [Rider] includes Llenos' induction of cerebral hypoxia under the facts of this case."

Because reasonable minds can differ about whether or not autoerotic asphyxiation would be an intentionally inflicted injury, it creates an ambiguity in coverage, and ambiguities must be resolved in favor of coverage. Minnesota Life is in the best position to remedy this ambiguity by expressly excluding coverage for such inherently dangerous activities, as other insurance providers have done. See Johnson v. Am. United Life Ins. Co. , 716 F.3d 813, 817 (4th Cir. 2013) (noting a policy exclusion for "hang-gliding, bungee jumping, automobile racing, motorcycle racing, skydiving, rock climbing, or mountain climbing."); Kovach v. Zurich Am. Ins. Co. , 587 F.3d 323, 336 (6th Cir. 2009) (noting policy exclusions for "skydiving, parasailing, hangglinding [sic], bungee-jumping, or any similar activity.").

In conclusion, I would affirm the decision of the district court. The district court properly concluded that cerebral hypoxia was not an intentional injury and Llenos' death was an unforeseen accident. Autoerotic asphyxiation is an inherently dangerous sexual practice. The matter at hand shows that even seasoned practitioners can succumb to cerebral hypoxia and die. Therefore, I respectfully DISSENT.