Dwight Samuel O'Connor was indicted on three counts of first-degree sexual assault against P.A.B. - one count of penile-vaginal penetration, one count of digital-vaginal penetration, and one count of fellatio. A jury acquitted O'Connor of digital penetration and fellatio but did not reach a verdict on the charge of penile penetration. The State then retried O'Connor. At the second trial, a jury convicted O'Connor of the remaining count.
On appeal, O'Connor argues that the trial court erred in precluding him from introducing evidence of the acquittals from his first trial.
Initially, the court ruled that O'Connor's two prior acquittals would be admissible if the State introduced testimony regarding the conduct underlying those two counts. But at trial, the State did not elicit testimony from P.A.B. regarding the conduct for which O'Connor had been acquitted, nor did the State rely on this conduct as substantive evidence of O'Connor's guilt. Rather, the evidence pertaining to the digital penetration and the fellatio was elicited by O'Connor's attorney during his cross-examination of P.A.B., solely for the purpose of impeaching P.A.B.'s testimony.
Accordingly, the trial court instructed the jury that any testimony about those other alleged acts of penetration was relevant only to assessing the credibility of the witnesses - and not as substantive evidence of O'Connor's guilt. And the court declined to admit evidence of O'Connor's prior acquittals.
As we explain in this opinion, we conclude that the trial court's ruling was not an abuse of discretion, given the way this issue was litigated.
O'Connor also argues that the State presented insufficient evidence to support his conviction. Viewing the evidence in the light most favorable to the jury's verdict, we conclude that the evidence was sufficient to support O'Connor's conviction.
Finally, O'Connor argues that the sentencing judge erred in declining to refer his case to the statewide three-judge sentencing panel *229based on the non-statutory mitigating factor of extraordinary potential for rehabilitation. For the reasons explained in this opinion, we conclude that a remand is required so that the sentencing judge can re-assess whether referral to the three-judge panel is warranted under the totality of the circumstances.
Factual background and procedural history
Because O'Connor challenges the sufficiency of the evidence to support his conviction, we present the following background facts in the light most favorable to upholding the jury's verdict.1
In July 2011, after a night of drinking and socializing, P.A.B. took a taxi to Penland Mobile Home Park in Anchorage. At the time, P.A.B. was homeless, and she would often sleep at a friend's house.
When P.A.B. arrived at the residence where she was supposed to sleep that night, P.A.B. discovered that the door was locked and she could not get inside. P.A.B. walked back to the road and decided to hitchhike to the apartment of another friend.
P.A.B. eventually saw a white truck drive past her. This truck turned around and pulled up next to her. The driver - later identified as O'Connor - offered P.A.B. a ride, and P.A.B. got into the truck.
Instead of driving P.A.B. to her friend's apartment, O'Connor took her to a fenced-off construction yard. O'Connor unlocked the gate, parked the truck, and guided P.A.B. to a small camper trailer. P.A.B. went inside because she assumed that O'Connor was taking her to get another drink.
The next thing P.A.B. remembered was O'Connor lying naked on top of her, with her pants removed. P.A.B. tried to kick O'Connor off of her, but she was unable to do so. According to P.A.B.'s later testimony, O'Connor pulled P.A.B.'s hair back and put his hands around her neck. O'Connor repeatedly yelled at P.A.B. that he could not "come" in her vagina and that he wanted "to rape [P.A.B.'s] ass"; P.A.B. told him to stop and to get off of her.
At some point, O'Connor put lotion on both of their genitals and penetrated P.A.B.'s vagina with his penis, but he did not ejaculate. When he could not maintain an erection, O'Connor became angry, and he threw P.A.B. face down on the bed and attempted to penetrate her anus with his penis, but he was unable to do so.
O'Connor eventually stopped and drove P.A.B. to her friend's apartment. P.A.B. knocked on the door, and when her friend answered, she immediately told him that she had been raped. P.A.B. then reported the sexual assault, and she subsequently met with the police. P.A.B. described O'Connor, his white truck, and the last three numbers of the truck's license plate.
P.A.B. underwent a SART examination, during which she reported feeling pain and soreness in her vaginal area. The forensic examination revealed an abrasion on P.A.B.'s chest, a bruise on P.A.B.'s hymen, and a laceration to P.A.B.'s perineal area, in addition to bruises on P.A.B.'s shoulder, back, clavicle, and right inner thigh.
The police subsequently located O'Connor and talked with him at the construction yard. O'Connor told the police that he had picked up P.A.B. and dropped her off at her friend's residence without incident; he denied having any sexual contact with P.A.B.
When the police interviewed O'Connor a second time, O'Connor again initially denied having any sexual contact with P.A.B. But when the police presented O'Connor with evidence that undermined his account, O'Connor admitted that he had lied when he denied having sexual contact with P.A.B. O'Connor now claimed that P.A.B. voluntarily removed her pants after he drove her back to his trailer, and she started performing oral sex on him. O'Connor stated that he rubbed his hands and penis against P.A.B.'s vagina, but he continued to deny that he had penetrated her with his penis, claiming that he was physically unable to do so.
Later during the same interview, O'Connor again changed his story and said that he had managed to penetrate P.A.B.'s vagina with his penis for a few seconds.
*230A grand jury indicted O'Connor on three counts of first-degree sexual assault.2 At O'Connor's first trial in February 2014, the jury acquitted O'Connor of two counts (digital-vaginal penetration and fellatio), but the jury could not reach a verdict on the remaining count (penile-vaginal penetration).
The State retried the remaining count. At the second trial, the previous evidence was adduced, and O'Connor testified. O'Connor stated that the entire encounter with P.A.B. was consensual. The jury convicted O'Connor of first-degree sexual assault for engaging in penile penetration with P.A.B. without her consent.
Why we uphold the trial court's decision to preclude the admission of evidence of O'Connor's prior acquittals
Prior to O'Connor's second trial, O'Connor's attorney filed a motion to introduce evidence of the acquittals from the first trial. In this motion, O'Connor's attorney announced that he intended to rely on a defense of consent, and he anticipated that the State would seek to introduce the acts for which O'Connor was acquitted as propensity evidence under Alaska Evidence Rule 404(b)(3).3 O'Connor's attorney argued that, if the State introduced evidence of the conduct for which O'Connor was acquitted, then he should be permitted to introduce evidence of the acquittals.
After the prosecutor told the court that she anticipated that P.A.B. would discuss this other conduct as part of her description of the events, the judge preliminarily ruled that evidence of the prior acquittals would be admissible under the Alaska Supreme Court's decision in Hess v. State and this Court's unpublished decision in Espinal v. State .4
But during P.A.B.'s direct examination, when the prosecutor asked P.A.B. to describe what had happened with O'Connor, P.A.B. testified about O'Connor's attempts at penile penetration, but she did not testify to any acts of fellatio or digital penetration. As a result, following P.A.B.'s direct examination, the prosecutor asked the judge to reconsider his ruling about O'Connor's prior acquittals.
O'Connor's attorney did not present a new legal argument in favor of admitting the acquittals. He indicated, however, that he might seek to impeach P.A.B. with her prior statements from the first trial about these acts. He argued that, if he did so, at a minimum there would need to be a limiting instruction advising the jurors to use the evidence of this other conduct solely for purposes of assessing P.A.B.'s credibility, and not for O'Connor's propensity to commit sexual assault. (O'Connor's attorney specifically noted that he was not contending that the prosecutor had instructed P.A.B. not to discuss the conduct for which O'Connor had been acquitted.)
The trial court agreed with the prosecutor that the acquittals were not relevant to P.A.B.'s testimony at that point, since the State had not introduced evidence of the conduct underlying those acquittals to rebut O'Connor's consent defense. And the court ruled that if the defense attorney introduced evidence of these acts during cross-examination for impeachment purposes (and the prosecutor did not suggest that this evidence was otherwise relevant for propensity purposes), then the acquittals would remain inadmissible.
During his cross-examination of P.A.B., O'Connor's attorney impeached P.A.B. with her prior statements that O'Connor had forced her to perform fellatio and had forcibly penetrated her vagina with his fingers. The attorney did not mention the prior trial or the prior charges; he simply confronted *231P.A.B. with the fact that she had made these prior allegations, both before the grand jury and again later "under oath." When pressed as to why she had omitted discussion of these other acts during her testimony on direct examination, P.A.B. said that she "forgot" about those aspects of the assault.
The prosecutor did not address this issue on redirect, and the prosecutor did not argue in closing that these other acts made it more likely that O'Connor engaged in the conduct for which he was on trial. As a result, the jury was never informed of the prior acquittals.
Consistent with O'Connor's request for a limiting instruction, the trial court instructed the jury to restrict its consideration of the other-acts evidence solely to assessing P.A.B.'s credibility. In particular, the court instructed the jury that O'Connor was "on trial solely for one count of penile-vaginal penetration," and that the jury should not consider evidence of other types of sexual penetration "when determining whether the State has satisfied its burden of establishing, beyond a reasonable doubt, that the defendant engaged in sexual penetration, penis to genitals, with P.A.B., without the consent of P.A.B." Rather, the court told the jury that it could consider this evidence only in "assessing [the] witnesses' overall credibility regarding the events on the night in question." In his closing argument, O'Connor's attorney argued - consistent with this instruction - that P.A.B.'s failure to remember such significant aspects of the assault undermined P.A.B.'s credibility.
On appeal, O'Connor argues that the trial court erred in precluding him from introducing evidence of his prior acquittals. In support of his argument, O'Connor relies on the supreme court's decision in Hess .
As a general matter, "a defendant's acquittal of one charge is ... not relevant to prove [the defendant's] factual innocence of the facts underlying that charge."5 Rather, the acquittal "proves only that the state did not prove every element of the crime beyond a reasonable doubt."6
Nevertheless, the Alaska Supreme Court held in Hess that when the State introduces evidence that the defendant has committed another crime, the defense may introduce evidence that the defendant was acquitted of that other crime.7 The court concluded that, in those instances, the acquittal helps the jury weigh the evidence of the prior act. That is, "[e]ven though the defendant's acquittal does not prove that he was innocent of the prior act, a jury may reasonably infer a greater probability of innocence from the fact of the acquittal."8
O'Connor's case is distinguishable from Hess . Here, the other conduct was not admitted as substantive evidence tending to establish O'Connor's guilt. That is, the other conduct was not admitted for propensity purposes, nor was it admitted to contextualize and explain P.A.B.'s account and chronology of the events.
Indeed, this evidence was not introduced by the State at all. Rather, the evidence of these other acts was introduced by O'Connor's attorney, solely to impeach P.A.B. And the parties' use of this evidence was limited throughout the trial to that purpose, consistent with the trial judge's instruction to the jury.
Thus, unlike in Hess , the acquittals in this case were not relevant to rebut an argument by the prosecutor that O'Connor's other acts of sexual assault tended to prove his propensity to commit the charged sexual assault.
Seemingly, the only purpose of the acquittals would have been to suggest that the prior jury had doubts about P.A.B.'s credibility. But under Hess , a prior acquittal is not admissible for this purpose. That is, an acquittal is not admissible to establish the historical facts underlying the prior jury's verdict or the reasons why the prior jury had a reasonable doubt.9 And in any event, O'Connor's attorney never argued that the *232prior jury acquitted him because of doubts about P.A.B.'s credibility, and O'Connor does not argue this on appeal. (In fact, he has not requested or supplied a transcript of the first trial for our review.)
On appeal, O'Connor does not cite any authority that would support admission of the acquittals under these circumstances. Given this record, we uphold the trial court's decision precluding admission of the acquittals.10
Why we uphold the trial court's denial of O'Connor's motion for a judgment of acquittal
After the State rested at O'Connor's second trial, O'Connor's attorney moved for a judgment of acquittal. The trial court denied this motion.
On appeal, O'Connor renews his claim that there was insufficient evidence to support his conviction. O'Connor argues that P.A.B. was not a credible witness and that the physical evidence was consistent with consensual sexual activity.
But when this Court reviews the sufficiency of the evidence to support a conviction, we do not weigh the credibility of witnesses, as witness credibility is exclusively a question for the jury.11 Rather, we view the evidence, and all reasonable inferences from that evidence, in the light most favorable to upholding the jury's verdict.12 Viewing the evidence in that light, we then ask whether a reasonable juror could have concluded that the State proved its case beyond a reasonable doubt.13
Here, the jury could reasonably credit P.A.B.'s testimony about the events and her reports to her friend and to the police that she was sexually assaulted, together with her injuries, to conclude that O'Connor engaged in sexual penetration with P.A.B. without her consent within the meaning of AS 11.41.410(a)(1). The jury could also reasonably consider O'Connor's own shifting statements to the police regarding these events.
Viewing the evidence in the light most favorable to the jury's verdict, we conclude that the State presented sufficient evidence to support O'Connor's first-degree sexual assault conviction.
Why we remand O'Connor's case to the trial court for reconsideration of his request for referral to the three-judge sentencing panel
As a first felony offender convicted of first-degree sexual assault, O'Connor was subject to a presumptive sentencing range of 20 to 30 years.14 Under former AS 12.55.125(o)(1), the court was also required to impose a minimum period of suspended imprisonment of 5 years. O'Connor was fifty-one years old at the time he sexually assaulted P.A.B. He had little criminal history, having been convicted of negligent driving in 1979 and taking a sub-legal Dall sheep in 2003.
Prior to sentencing, O'Connor's attorney filed a motion seeking referral of O'Connor's case to the statewide three-judge sentencing panel based on the non-statutory mitigator of extraordinary potential for rehabilitation.15 In support of this motion, O'Connor's attorney noted O'Connor's minor criminal history, his history of community service, including his service as a member of the Public Safety Advisory Commission and a regular volunteer for the Community Safety Patrol, and his strong support system of friends and family (as evidenced by the letters submitted to the court on O'Connor's behalf). At the *233sentencing hearing itself, O'Connor's attorney further noted that O'Connor had participated in the Static-2002R, an actuarial tool designed to assess the risk of recidivism for sex offenders, and that this test classified him as having a low risk of re-offending.
The State opposed O'Connor's three-judge panel request. The prosecutor argued that O'Connor had to prove "five elements" to establish that he had extraordinary potential for rehabilitation, and that he had failed to do so. (The prosecutor also variously referred to these considerations as "factors" or "steps.")
According to the prosecutor, these five elements were: (1) that the defendant is a first felony offender; (2) that the crime is out of character for the defendant; (3) that the court understands the problems that led the defendant to engage in criminal conduct; (4) that the criminal conduct is unlikely to be repeated; and (5) that there is reason to believe that the defendant can be rehabilitated in a period shorter than the presumptive sentencing range.
The trial court denied O'Connor's request to refer his case to the three-judge panel. Significantly, the court found that O'Connor was unlikely to ever repeat his criminal conduct. However, the court expressed uncertainty as to why O'Connor had committed the sexual assault, and the court therefore concluded that it could not determine whether he could be adequately rehabilitated in a period shorter than the presumptive range. The court ultimately imposed a sentence of 25 years with 5 years suspended (20 years to serve) - the lowest permissible sentence within the presumptive range.
On appeal, O'Connor underscores the trial court's finding that he was unlikely to re-offend, and he argues that this finding was essentially a confirmation of his exceptionally good prospects for rehabilitation, justifying referral of his case to the three-judge panel. He contends that when a court makes a finding that a defendant is unlikely to recidivate, then it becomes less important for the court to identify the precise reasons why the defendant committed the offense.
We agree with O'Connor that the trial court's finding that O'Connor's conduct was "unlikely [to] ever be repeated" was significant and that it lessened the need to understand the precise reasons for O'Connor's criminal conduct. Given that, we are concerned that the trial court adopted the five-factor test proposed by the prosecutor and, as a result, appeared to heavily weigh one particular factor - the reason for O'Connor's conduct - to the exclusion of the totality of the circumstances.
When evaluating whether a defendant has extraordinary potential for rehabilitation, the sentencing court must apply a "totality of the circumstances" test.16 Although the considerations proposed by the prosecutor were relevant to the court's analysis, there is no set "five-factor" test.
Moreover, as O'Connor notes, the court made two findings that are potentially at odds with each other: the judge affirmatively found that O'Connor was unlikely to ever commit sexual assault again, but the judge declined to refer O'Connor's case to the three-judge panel because the judge could not identify the precise reason(s) why O'Connor committed the sexual assault in this case.
The judge's two findings accentuate an ambiguity or potential difficulty in the ways that this Court has described the non-statutory mitigating factor of extraordinary potential for rehabilitation.
In Kirby v. State , this Court declared that a sentencing court is justified in concluding that a defendant has unusually good potential for rehabilitation when "the court is satisfied, after reviewing the totality of the circumstances, that [the defendant] can adequately be treated in the community and need not be incarcerated for the full presumptive term in order to prevent future criminal activity."17
Three years later, in Lepley v. State , this Court attempted to clarify the Kirby test: we declared that a sentencing court should not make "a prediction of successful treatment *234and non-recidivism" unless the court "is reasonably satisfied both that it knows why a particular crime was committed and that the conditions leading to the criminal act will not recur - either because the factors that led the defendant to commit the crime are readily correctable or because the defendant's criminal conduct resulted from unusual environmental stresses unlikely ever to recur."18
In O'Connor's case, the sentencing judge found that O'Connor's criminal conduct was "unlikely [to] ever be repeated," both because it was out of character for him and because O'Connor had already been strongly deterred by the criminal justice process. Thus, even though the judge could not identify the precise reasons for O'Connor's criminal conduct, the judge affirmatively found that O'Connor was unlikely to re-offend.
However, the judge also concluded that, as a legal matter, his inability to identify the precise reasons for O'Connor's sexual assault precluded him from referring O'Connor's case to the three-judge sentencing panel. The judge reasoned that, because he could not identify the precise causes of O'Connor's criminal conduct, he could not predict with any assurance that O'Connor would be rehabilitated in a shorter time frame than the 20-year minimum sentence required by the applicable presumptive sentencing range.
Obviously, there is a tension between the judge's finding that O'Connor will not re-offend and the judge's statement that he could not reasonably predict whether O'Connor could be rehabilitated in less than 20 years.
The tension arises from the fact that, in this context, a finding that the defendant will not re-offend is equivalent to a finding of "rehabilitation." As used in the criminal law, the term "rehabilitation" (or its synonym, "reformation") means that the defendant need not be confined in order to prevent future criminal activity, and that the defendant can be expected to be a law-abiding citizen.
(See Black's Law Dictionary , which defines "rehabilitation" for criminal law purposes as "[t]he process of seeking to improve a criminal's character and outlook so that he or she can function in society without committing other crimes."19 Or, as this Court suggested in Kirby , "rehabilitation" is the converse of recidivism or continued dangerousness.20 )
Although our decision in Lepley could be interpreted as strictly prohibiting a sentencing judge from making a finding of extraordinary potential for rehabilitation unless the judge is able to identify the precise causes of the defendant's criminal behavior, such an interpretation would be unwarranted. While identifying the causes of a defendant's criminal behavior may better assist a judge in predicting rehabilitative potential, sometimes, even after a comprehensive analysis, the ultimate causes of a person's actions remain murky, and a judge may be unable to identify the precise reasons why the person engaged in particular antisocial behavior. But even in these situations, there may still be articulable reasons to conclude that the behavior will not recur.
We interpret Lepley as saying that a sentencing judge may not rely on hunches about the defendant, or personal assessments of the defendant's character, when the judge makes a finding of extraordinary potential for rehabilitation. Rather, the judge must have articulable reasons, based on the evidentiary record, for concluding that the defendant can be rehabilitated earlier than the minimum term of imprisonment prescribed by the applicable presumptive sentencing range - even though these articulable reasons may not precisely identify the ultimate causes of the defendant's criminal behavior.
Here, although the court could not pinpoint the precise reason for O'Connor's conduct, it posited several rationales. Ultimately, the court concluded that O'Connor's conduct *235was unlikely to "ever be repeated" - that, regardless of the reasons for O'Connor's conduct, he had been strongly deterred by the entire criminal process. The court also concluded that O'Connor had prospects for rehabilitation that were "well above average."
We acknowledge that there were factors weighing against a finding of extraordinary potential for rehabilitation: the court found that O'Connor had not expressed remorse for his conduct, and the court found that O'Connor's own evasiveness with the police, and his shifting narrative, suggested that he had exhibited some "level of deception" that contradicted his lifetime of community service. But we conclude that it is appropriate to remand O'Connor's case so that the trial court can, in the first instance, apply a totality of the circumstances test to evaluate O'Connor's request to refer his case to the three-judge panel in light of the guidance we have provided here.
Conclusion
We AFFIRM O'Connor's conviction. We REMAND O'Connor's case for reconsideration of his request for referral to the three-judge panel based on the nonstatutory mitigator of extraordinary potential for rehabilitation, and we retain jurisdiction over this matter. The superior court shall report to us within 90 days of the issuance of this opinion. This deadline may be extended for good cause.