OPINION
PAGE, Justice.This case has a long and complicated procedural history. Respondent, Mohammed Gazizamil Al-Naseer, was initially charged with two counts of criminal vehicular homicide in violation of Minn.Stat. § 609.21, subd. 1(1) (2006) (gross negligence), and Minn.Stat. § 609.21, subd. 1(7) (2006) (leaving the scene), as a result of a traffic accident that caused the death of Kane Thomson. A Clay County jury found Al-Naseer guilty on both counts and the district court subsequently convicted him of both counts, sentencing him to 48 months in prison. On appeal, the court of appeals affirmed the gross-negligence con-vietion but reversed the leaving-the-scene conviction and remanded for a new trial, concluding that the district court incorrectly instructed the jury on the mens rea requirement. State v. Al-Naseer (Al-Naseer I), 678 N.W.2d 679, 694-97 (Minn. App.2004). After granting Al-Naseer’s petition for review, we affirmed the court of appeals’ leaving-the-scene decision but reversed the gross-negligence conviction because the district court failed to instruct the jury on the lesser-included offense of careless driving. State v. Al-Naseer (Al-Naseer II), 690 N.W.2d 744, 750, 753 (Minn.2005).
On remand, after a bench trial at which he waived his right to a jury trial and his right to testify, Al-Naseer was found guilty and convicted of both criminal vehicular homicide (leaving the scene) and careless driving, but found not guilty of criminal vehicular homicide (gross negligence). Al-Naseer again appealed to the court of appeals, which reversed the criminal vehicular homicide (leaving the scene) conviction and remanded for additional findings with respect to Al-Naseer’s mens rea. State v. Al-Naseer (Al-Naseer III), 721 N.W.2d 623, 626 (Minn.App.2006). Both the State and Al-Naseer petitioned for review. We granted the State’s petition but denied Al-Naseer’s petition.
Ultimately, we held that in order for a defendant to be found guilty of criminal vehicular homicide (leaving the scene), the State is required to prove that the defendant knew he had been involved in an accident with a person or another vehicle. On that basis, we
affirm[ed] the reversal of Al-Naseer’s conviction but remand[ed] to the district court to reconsider its verdict based on the present record and to make amended findings in light of the mens rea standard that requires proof that Al-Naseer knew that his vehicle was in*472volved in an accident with a person or another vehicle.
State v. Al-Naseer (Al-Naseer TV), 734 N.W.2d 679, 688-89 (Minn.2007).
Before reconsidering its verdict on remand, the district court allowed the parties to submit briefs. In his brief, Al-Naseer argued that, in light of the new standard we articulated in Al-Naseer TV, he was entitled to a new trial because his constitutional rights to notice of the charges against him and a fair opportunity to defend against those charges would be violated if the court made amended factual findings based on the existing record. The district court refused to consider these arguments, concluding that it did not have the authority to alter our mandate on remand. Al-Naseer also argued that there was insufficient evidence to support his conviction. Specifically finding that Al-Naseer knew he had been involved in an accident with a person or another vehicle, the court again found Al-Naseer guilty of the leaving-the-scene and careless-driving offenses.
Al-Naseer again appealed his leaving-the-scene conviction, and while the court of appeals found that there was sufficient evidence to uphold the verdict, it also found that Al-Naseer was entitled to a new trial because his constitutional right to due process was violated by the district court’s failure to conduct a new trial. State v. Al-Naseer (Al-Naseer V), No. 07-2275, 2009 WL 304738 (Minn.App.2009). The State petitioned for review, challenging the court of appeals’ grant of a new trial based on the denial of due process, and Al-Naseer cross-petitioned, challenging the sufficiency of the evidence. We granted both petitions, requiring us to determine: (1) whether Al-Naseer’s conviction was supported by sufficient evidence; and (2) whether Al-Naseer’s constitutional challenges were properly heard and correctly decided by the court of appeals. Without reaching Al-Naseer’s constitutional challenges, we reverse.
The facts of this case are largely undisputed and are set forth in detail in Al-Naseer II and Al-Naseer TV, and will not be. repeated here except as necessary for resolution of the issues presented in this appeal. Al-Naseer was driving west on Highway 10 in Clay County when his vehicle gradually crossed the fog line and hit Kane Thomson, who was on the right shoulder of the road changing a flat tire. Thomson was crouched down while his friend, Dustin Leingang, stood a foot away holding a flashlight. Thomson had removed the flat tire and placed it between him and the fog line. As Thomson was putting on the spare tire, Leingang felt something brush past his hand with enough force to spin him around, heard a loud thud, looked toward Thomson, and saw Thomson rolling on the shoulder of the road in front of his vehicle. About 150 feet ahead Leingang saw Al-Naseer’s vehicle straddling the fog line neither braking nor accelerating. The vehicle gradually made its way back onto the highway and did not stop. Thomson died from his injuries.
The accident caused significant damage to Al-Naseer’s vehicle: the right headlight was damaged and stopped working, the right front corner of the vehicle was crumpled, and pieces of debris were scattered along the highway for nearly 125 feet from the accident. According to an accident-reconstruction expert, the impact of the accident and the dragging of the flat tire likely jolted the vehicle and caused a loud noise.
A Dilworth police officer spotted Al-Naseer’s vehicle traveling approximately six miles from the accident scene, at about 45 miles an hour, with its hazard lights flashing, both headlights off, and its right *473front tire flat. When stopped by the officer, Al-Naseer got out of his vehicle and immediately went to the vehicle’s damaged right front area.1 Al-Naseer had difficulty communicating in English, but when asked if he had been in an accident, he responded that he had hit something but did not know what it was.
Al-Naseer argues that, the evidence at trial was insufficient to support his leaving-the-scene conviction. In particular, Al-Naseer contends reasonable doubt exists because the evidence admitted to prove the mens rea element of the offense was circumstantial and consistent with rational hypotheses other than guilt. In assessing the sufficiency of the evidence, we review the evidence to determine “whether the facts in the record and the legitimate inferences drawn from them would permit the jury to reasonably conclude that the defendant was guilty beyond a reasonable doubt of the offense of which he was convicted.” State v. Moore, 481 N.W.2d 355, 360 (Minn.1992). The jury’s verdict will be upheld if, “giving due regard to the presumption of innocence and to the state’s burden of proof beyond a reasonable doubt, [the jury] could reasonably have found the defendant guilty.” State v. Pierson, 530 N.W.2d 784, 787 (Minn.1995).
A conviction based on circumstantial evidence, however, warrants heightened scrutiny. State v. Bolstad, 686 N.W.2d 531, 539 (Minn.2004). This heightened scrutiny requires us to consider “whether the reasonable inferences that can be drawn from the circumstances proved support a rational hypothesis other than guilt.” State v. Taylor, 650 N.W.2d 190, 206 (Minn.2002). In other words, the circumstances proved must “be consistent with the hypothesis that the accused is guilty and inconsistent with any other rational hypothesis except that of guilt.” State v. Bias, 419 N.W.2d 480, 484 (1988). “Circumstantial evidence must form a complete chain that, in view of the evidence as a whole, leads so directly to the guilt of the defendant as to exclude beyond a reasonable doubt any reasonable inference other than guilt.” Taylor, 650 N.W.2d at 206.
We recognize that the trier of fact is in the best position to determine credibility and weigh the evidence. Moore, 481 N.W.2d at 360; see also State v. Hough, 585 N.W.2d 393, 396 (Minn.1998) (“We review criminal bench trials the same as jury trials when determining whether the evidence is sufficient to sustain convictions.”). We “will not overturn a conviction based on circumstantial evidence on the basis of mere conjecture.” State v. Lahue, 585 N.W.2d 785, 789 (Minn.1998). The State does not have the burden of removing all doubt, but of removing all reasonable doubt. State v. Hughes, 749 N.W.2d 307, 313 (Minn.2008).
“When reviewing the sufficiency of circumstantial evidence, ‘our first task is to identify the circumstances proved.’ ” State v. Andersen, 784 N.W.2d 320 (Minn. 2010) (quoting State v. Stein, 776 N.W.2d 709, 718 (Minn.2010) (plurality opinion)). In identifying the circumstances proved, we “ ‘defer, consistent with our standard of review, to the jury’s acceptance of the proof of these circumstances and rejection of evidence in the record that conflicted with the circumstances proved by the State.’ ” Id. (quoting Stein, 776 N.W.2d at 718). Then we “ ‘examine independently *474the reasonableness of all inferences that might be drawn from the circumstances proved,’ ” including inferences consistent with rational hypotheses other than guilt. Id. (quoting Stein, 776 N.W.2d at 716). “ ‘We give no deference to the fact finder’s choice between reasonable inferences.’ ” Id. (quoting Stein, 776 N.W.2d at 716). “ ‘In assessing the inferences drawn from the circumstances proved, the inquiry is not simply whether the inferences leading to guilt are reasonable. Although that must be true in order to convict, it must also be true that there are no other reasonable, rational inferences that are inconsistent with guilt.’” Id. (quoting Stein, 776 N.W.2d at 716). This is because “ ‘if any one or more circumstances found proved are inconsistent with guilt, or consistent with innocence, then a reasonable doubt as to guilt arises.’ ” Id. at 338 (Meyer, J., concurring) (quoting State v. Johnson, 173 Minn. 543, 545-46, 217 N.W. 683, 683-84 (1928)).
The court of appeals in Al-Naseer V declined to apply the circumstantial evidence standard of review. Relying on State v. Jones, 516 N.W.2d 545, 549 (Minn. 1994), the court explained that heightened scrutiny “controls only when every element required for conviction has been proved entirely by circumstantial evidence.” Al-Naseer V, 2009 WL 304738, at *4. We have never held that our test for reviewing the sufficiency of the evidence in cases involving circumstantial evidence applies only to cases in which all of the evidence is circumstantial. While we did say in Jones that “a conviction based entirely on circumstantial evidence merits stricter scrutiny,” 516 N.W.2d at 549, there is nothing in Jones that suggests that we intended to limit the application of the test to only convictions based “entirely” on circumstantial evidence. Similarly, we applied heightened scrutiny in State v. Rhodes, holding that “[w]here, as here, the jury verdict is based on circumstantial evidence alone, the conviction warrants a higher standard of review.” 657 N.W.2d 823, 840 (Minn.2003). Like Jones, our holding in Rhodes merely reflects the fact that all of the evidence supporting the conviction was circumstantial, a situation not presented by this case. Neither Jones nor Rhodes should be read as precluding the use of the test in cases in which both direct and circumstantial evidence support the conviction.
That the test for reviewing the sufficiency of the evidence in cases involving circumstantial evidence applies more broadly than suggested by the court of appeals is most evident in first-degree murder cases requiring proof of premeditation. Like the knowledge requirement for a leaving-the-scene conviction, “[p]re~ meditation is a state of mind and, thus, generally proven through circumstantial evidence.” State v. Leake, 699 N.W.2d 312, 319 (Minn.2005); see also State v. Hughes, 749 N.W.2d 307, 312 (Minn.2008) (“Because it ‘is a state of mind,’ premeditation is ‘generally proven through circumstantial evidence.’ ” (quoting Leake, 699 N.W.2d at 319, 321)); see also Al-Naseer TV, 734 N.W.2d at 687 (“The proof of knowledge may be by circumstantial evidence.”).
In reviewing the sufficiency of the evidence in State v. Leake, we applied heightened scrutiny to the element of premeditation, despite the fact that direct evidence established other elements of the offense. We held that the evidence was sufficient to support the verdict because “[although the evidence is circumstantial, it is ‘consistent with the hypothesis that the accused is guilty’ — i.e., that he procured a knife and brought it to Fisher’s bedroom where he stabbed her to death — and is ‘inconsistent with any rational hypothesis except *475that of guilt.’ ” Leake, 699 N.W.2d at 320 (quoting State v. Chomnarith, 654 N.W.2d 660, 664 (Minn.2003)); see also State v. Wallace, 558 N.W.2d 469, 473 (Minn.1997) (stating that, “[c]onsidering the evidence in a light most favorable to the verdict and the totality of the circumstances, the jury could reasonably conclude that the only plausible inference to be drawn was that appellant intended to commit a sex crime against Lott”); State v. Ewing, 250 Minn. 436, 443, 84 N.W.2d 904, 910 (1957) (stating that the circumstances proved “were all consistent with the hypothesis of guilt and inconsistent with the theory of the defendant that the victim may have been the driver of the car”). Consistently, in State v. Kolander, 236 Minn. 209, 216-17, 52 N.W.2d 458, 462-63 (1952), we explained that the district court
properly instructed the jury that “to warrant a finding of guilty on any of the material elements of the crime charged based on circumstantial evidence the facts proven by circumstantial evidence must be consistent with each other, they must be consistent with the theory of guilt and exclude the theory of innocence, and must exclude every other reasonable conclusion except that of the guilt of the defendant.”
Id. (emphasis added). Thus, we conclude that the court of appeals erred when it failed to apply the circumstantial-evidence test to Al-Naseer’s sufficieney-of-the-evi-dence claim.
Applying that test to Al-Na-seer’s leaving-the-scene conviction, we conclude that the evidence presented at trial is insufficient to support the conviction. A person is guilty of criminal vehicular homicide (leaving the scene) if that person, as a result of operating a motor vehicle, causes injury to or the death of another and leaves the scene of the accident in violation of his duties under Minn.Stat. § 169.09, subds. 1 and 6 (2008). Among other things, a conviction for criminal vehicular homicide (leaving the scene) requires the State to prove that the defendant “must have known that there was an accident that involved a person or a vehicle.” Al-Naseer TV, 734 N.W.2d at 687. Thus, for Al-Naseer’s leaving-the-scene conviction to stand, the State had to prove beyond a reasonable doubt that, at the time of the accident, Al-Naseer had actual knowledge of facts imposing a duty on him to stop; that is, the State had to prove Al-Naseer knew he had been in an accident with a person or vehicle. Al-Naseer TV, 734 N.W.2d at 688. If the circumstances proved are consistent with any rational hypothesis other than guilt — meaning a hypothesis in which Al-Naseer did not know he had hit a person or a vehicle— then reasonable doubt exists and Al-Na-seer’s leaving-the-scene conviction cannot stand.
The evidence2 admitted at trial established that the accident occurred on a warm, clear, and dry summer night. Around 11:30 p.m., Thomson and Dustin Leingang were traveling westbound on Highway 10 in Thomson’s vehicle. Approximately half a mile from the Highway 9 intersection, Thomson noticed one of his vehicle’s tires was flat. He pulled onto the *476shoulder of the road and stopped the vehicle three feet, four inches from the white fog line, leaving room for him to change the flat left rear tire without his body being in the lane of traffic. The nearest lighting on the highway was a half-mile back, at the Highway 9 intersection. Lein-gang, who was the only eyewitness to the accident, testified that when Thomson pulled the vehicle onto the shoulder of the road, he activated the vehicle’s hazard lights, left the headlights on, and opened the lighted trunk to retrieve the spare tire. Thus, the rear of the vehicle was illuminated with three lights: a white light on the trunk lid and a flashing hazard light at each corner.
According to Leingang, he was standing about a foot away from the rear of the vehicle holding a flashlight for Thomson, who was crouched down changing the left rear tire. Leingang testified that neither he nor Thomson were over the fog line. Although an occasional vehicle drove by, causing a strong wind, most vehicles moved to the left lane as they passed. Thomson quickly removed the flat tire, placed it on the shoulder of the road between his body and the fog line, and began putting on the spare tire.
As Thomson was putting on the spare tire, Leingang felt something brush past his left hand with enough force to spin him around almost 180 degrees such that he ended up facing the opposite direction. He testified that it was not a sharp hit and could have been a strong wind from a vehicle passing closely. Leingang then heard a loud thud, looked back toward Thomson’s vehicle, and saw Thomson’s body rolling on the shoulder of the road in front of the vehicle. He also saw a vehicle that was halfway over the fog line on the shoulder traveling west on Highway 10. According to Leingang, that vehicle’s brake lights were never activated nor did the vehicle appear to accelerate, but when it was about 150 feet away, it gradually moved back onto the road into the right lane of traffic. Leingang could not see the driver and did not know if the driver checked his mirrors or looked back toward the accident scene.
Other evidence introduced at trial, including testimony from an accident-reconstruction expert and the police officers who investigated the accident, established that the entire incident lasted less than five seconds. After crossing the fog line and brushing past Leingang, Al-Naseer’s vehicle hit Thomson and the flat tire lying on the shoulder of the road next to him. As the vehicle crossed the fog line, it was at an angle sharp enough to miss Leingang and hit Thomson but slight enough so that it did not hit Thomson’s vehicle. Because Al-Naseer’s vehicle did not hit Thomson’s vehicle, it could not have drifted more than three feet, four inches over the fog line. As a result of the impact, Thomson was thrown into the side of his vehicle, causing a dent in the driver’s side door. Thomson was hit with such force that his shoes were knocked from his feet and his body came to rest 25 feet in front of his vehicle. Thomson died from his injuries. The impact also resulted in significant damage to Al-Naseer’s vehicle. It dented the hood, broke the right front headlight assembly, and damaged the right front fender, side marker light, and fog light.
When Al-Naseer’s vehicle passed over the nearly five-inch-thick flat tire that was between Thomson and the fog line, the vehicle’s shock absorbers likely bottomed out, jolting the vehicle. The force and speed of Al-Naseer’s vehicle hitting the tire was sufficient to gouge and scrape the asphalt roadbed. Once out from under Al-Naseer’s vehicle, the tire popped up and rolled onto Highway 10, where it remained until a bystander moved it. The right *477front tire on Al-Naseer’s vehicle went flat, likely due to the impact with the flat tire from Thomson’s vehicle. Other debris from Al-Naseer’s vehicle was scattered up to 125 feet away. According to the testimony of several police officers, even if a person were expecting it, the impact with Thomson and the tire — the jolt when the vehicle bottomed out and the sounds of metal crumpling and glass breaking— would have been loud and startling to anyone in Al-Naseer’s vehicle.
Notwithstanding this evidence, the circumstances proved include evidence that Al-Naseer’s vehicle did not react to either the impending or the actual impact. Lein-gang’s testimony established that Al-Na-seer’s vehicle did not brake or accelerate, even while returning to the highway. There was also testimony that skid marks at the accident scene would have indicated braking, swerving, or accelerating on the part of Al-Naseer’s vehicle, but there were no skid marks either leading up to or leaving the scene of the impact. Despite the noise and feel of the impact and the illumination of Thomson’s vehicle, other than gradually returning to the roadway after 150 feet, there was no reaction by Al-Naseer’s vehicle. Two police officers, one of whom was the accident reconstruc-tionist, testified that the evidence was consistent with Al-Naseer being asleep at the time of the impact. One of the officers also testified that there could have been various reasons for Al-Naseer’s failure to react to the impact of the accident, one of which was that Al-Naseer could have been asleep or otherwise unconscious at the time of the accident.
Finally, the evidence at trial established that shortly after 11:30 p.m. that evening, a Dilworth police officer spotted Al-Na-seer’s vehicle travelling west on Highway 10, about six miles from the accident scene, with a flat tire, no headlights, and its hazard lights flashing. When the officer activated his lights, Al-Naseer pulled his vehicle over onto the right shoulder of the road and immediately got out and started walking to the front of the vehicle. The officer testified that when he approached Al-Naseer, Al-Naseer was trying to put the broken headlight assembly into its proper position. When questioned, Al-Naseer indicated that he knew that he had hit something, but did not know what he had hit.
Based on the evidence presented at trial, the district court found that Al-Naseer must have known he hit Thomson and Thomson’s vehicle, and therefore was guilty of criminal vehicular homicide (leaving the scene). In reviewing these findings, we must first identify the circumstances proved, and then we independently consider the inferences that might be drawn from those circumstances, to determine if there are any rational inferences consistent with a hypothesis other than guilt — namely, whether the circumstances proved support a rational inference that Al-Naseer did not know he was in an accident that involved a person or vehicle at the time he failed to stop. See Andersen, 784 N.W.2d at 329-30 (citing Stein, 776 N.W.2d at 716 (plurality opinion)).
The circumstances proved are that Thomson’s vehicle was stopped on the shoulder of the road with the headlights, trunk light, and both rear hazard lights illuminated. Al-Naseer’s vehicle drifted from the road at a slight angle toward Thomson and hit Thomson, but did not hit Leingang or come into direct contact with Thomson’s vehicle. Al-Naseer’s vehicle also hit the tire that was behind Thomson. Hitting the tire likely caused Al-Naseer’s vehicle’s shock absorbers to bottom out, creating a loud noise and a jolt. The impact with Thomson and the tire caused significant damage to the front right side *478of Al-Naseer’s vehicle. After the impact, Thomson came to rest on the shoulder of the road about 25 feet in front of his own vehicle and the tire ended up on the highway in a lane of traffic. Al-Naseer’s vehicle continued traveling forward, drifting further onto the shoulder of the road toward the ditch. About 150 feet in front of the impact site, Al-Naseer’s vehicle made a gradual shift back toward the highway and returned to the right traffic lane, without the brake lights being activated or appearing to accelerate. Al-Naseer’s vehicle left no skid marks either approaching or leaving the accident scene. If Al-Na-seer looked into his rearview mirror after passing Thomson’s vehicle, the headlights of Thomson’s vehicle would likely have blinded him. After traveling several miles further along Highway 10, Al-Naseer was pulled over by a law enforcement officer. He immediately exited his vehicle and examined the damaged front-right portion of his vehicle. He admitted to the police officer that he knew he had hit something, but did not know what he had hit.3
That Al-Naseer must have known that he hit Thomson or Thomson’s vehicle is a reasonable inference, consistent with guilt, to be drawn from the circumstances proved. But it is not the only reasonable inference to be drawn from those circumstances. The circumstances proved include evidence from which it could be reasonably inferred that Al-Naseer was asleep or otherwise unconscious when his vehicle hit Thomson. Thus, on the element of scienter, the circumstances proved are consistent with a rational hypothesis other than guilt, that is, Al-Naseer was asleep or otherwise unconscious at the time of the accident and therefore would not have known that he had hit a person or a vehicle.
The State contends that an inference that Al-Naseer was asleep or unconscious at the time of the accident is inconsistent with the circumstances proved because the noise and jolt of the accident would have alerted even a sleeping driver. The State further contends that, due to the noise and jolt of hitting Thomson and the tire, Al-Naseer must have known he had been in an “impact.” Finally, the State argues that even if he was asleep at the time of the impact, Al-Naseer was conscious shortly after the impact and would have seen Thomson in his rearview mirror when he awoke, thereby realizing that he had hit Thomson or Thomson’s vehicle.
While not unreasonable, the inferences drawn by the State are not the only rea*479sonable inferences to be drawn from the circumstances proved. After the accident, Al-Naseer’s vehicle did not react to either the noise or jolt of the impact; it did not swerve, brake, or accelerate, but rather drifted past Thomson’s vehicle along the shoulder of the road for another 150 feet, and then gradually returned to the highway. From these circumstances, it can be reasonably inferred that Al-Naseer was asleep or otherwise unconscious at the time of the impact and at the earliest awoke or regained consciousness shortly before his vehicle began to gradually return to the highway. Before awakening or regaining consciousness, Al-Naseer would not have known that he was in an accident. Nor would he necessarily have been aware of the noise and jolt of the impact.
Upon regaining consciousness, Al-Na-seer likely would have recognized at some point that his vehicle’s headlights were not working properly and that the vehicle’s handling had changed dramatically because of the flat tire but, short of stopping and checking, he would not have known that a flat tire was the cause of the handling problem. It can also be inferred from the circumstances proved that, upon waking up or regaining consciousness, Al-Naseer would not have seen Thomson and, even if immediately aware that he had been in an accident, he would not have known the nature of the accident. After the impact, Thomson’s body would have been moving slower than Al-Naseer’s vehicle, and thus Thomson’s body was first to the side of and then behind Al-Naseer. Further, Al-Naseer would not have seen Thomson in his rearview mirror because, as one investigating officer’s testimony indicates, the glare from Thomson’s vehicle’s headlights would have obscured Al-Na-seer’s vision. Thus, it can be inferred that Al-Naseer would have seen nothing more than the bright lights of a vehicle sitting on the side of the road behind him and would not have been able to see what had taken place behind him. Even if these circumstances, as proved, establish that Al-Naseer may have known that there had been an accident, they do not establish that he necessarily would have known that he had been in an accident with a person or a vehicle — as opposed, for example, to hitting a deer, a road sign, or a pothole— an element of the charged offense that the State was required to prove beyond a reasonable doubt.4
*480The State attempts to discredit the inference that AI-Naseer did not know what he hit as unreasonable by arguing that it is based on speculation. Citing State v. Tscheu, 758 N.W.2d 849, 858 (Minn.2008), the State notes that a challenge to a verdict based on circumstantial evidence “may not rely on mere conjecture” or speculation. It is true that we do not set aside verdicts based on speculation. However, we made clear in Tscheu that a defendant is not relying on conjecture or speculation when the defendant, as Al-Naseer does here, “point[s] to evidence in the record that is consistent with a rational theory other than guilt.” Id. at 858. Ultimately, the problem with the State’s argument is that there are reasonable inferences to be drawn from the circumstances proved that do not exclude every rational hypothesis except that of guilt. There is evidence in the record, as discussed above, to support the inference that Al-Naseer did not know what he had hit. Further, given the evidence in the record suggesting that Al-Naseer was asleep or otherwise unconscious when his vehicle hit Thomson, the inference that he did not know what he had hit is reasonable. Because the inference that Al-Naseer did not know what he had hit is consistent with the circumstances proved, our decision is not based on speculation or conjecture.
The dissent states that “the circumstances proved do not reflect the presence of any other object” that would have provided Al-Naseer a reasonable basis to believe the object his vehicle struck was not Thomson’s vehicle. On the record presented in this case, additional evidence of some other object that Al-Naseer’s vehicle could have struck would not inform the answer to the question of Al-Naseer’s knowledge any more than the evidence existing in the current record because, to the extent that Al-Naseer was asleep or otherwise unconscious at the time of the accident, he would not have known what he hit. Even if evidence of damage to something other than a person or vehicle existed, it would not establish Al-Naseer’s subjective knowledge at the time he left the scene of the accident. Although such evidence might weigh on the reasonableness of Al-Naseer’s actual knowledge, it does not affect the proper focus of our review— whether the circumstances proved in this case are consistent with a reasonable inference of innocence.
We made clear in Al-Naseer IV that “a person who does not know what his vehicle hit is not on notice that he has a duty to stop.” 734 N.W.2d at 687 n. 3. In doing so, we expressly rejected the argument that a defendant was guilty of leaving the scene if he/she had “reason to know” that the accident involved a vehicle or person. Al-Naseer IV, 734 N.W.2d at 688. The leaving-the-seene statute, Minn. Stat. § 609.21, subd. 1(7), only imposes a duty to stop if the person knows — i.e., has actual, subjective knowledge — he hit a person or vehicle. Al-Naseer IV, 734 N.W.2d at 688-89. To now suggest that a driver has a duty to stop and investigate even though he does not know what he hit, much less that a failure to stop and investigate renders a driver’s lack of knowledge about what he hit unreasonable as a matter of law, is a dramatic change from established law. We affirm our holding in Al-Naseer IV and decline to impose a new duty on a driver to stop and investigate if he is not aware that he hit a person or vehicle.
To convict Al-Naseer of criminal vehicular homicide “leaving the scene,” the burden was on the State, and not Al-Naseer, to prove beyond a reasonable doubt that “Al-Naseer knew that his vehicle was involved in an accident with a person or another vehicle.” Al-Naseer IV, 734 *481N.W.2d at 688-89. Here, the circumstances proved permit a number of reasonable inferences, including inferences that support the conclusion that Al-Naseer knew that he had hit a person or a vehicle and inferences that support the opposite conclusions, i.e., that Al-Naseer did not know that he had hit a person or a vehicle. The latter inferences are inconsistent with guilt and therefore create reasonable doubt as to Al-Naseer’s guilt. See Andersen, 784 N.W.2d at 337 (Meyer, J., concurring) (quoting Johnson, 173 Minn, at 545-46, 217 N.W. at 683-84). Because of that reasonable doubt, we conclude that there is insufficient evidence to support Al-Na-seer’s conviction of criminal vehicular homicide-leaving the scene. We therefore reverse the court of appeals.5
Reversed.
STRAS, J., not having been a member of this court at the time of the argument and submission, took no part in the consideration or decision of this case.. The record does not indicate whether the left front area of the vehicle was damaged, what caused the left headlight to be off, or whether it was functioning before the collision with Thomson.
. The dissent, suggesting that we have improperly treated conflicting evidence, notes that when "identifying the circumstances proved, we construe conflicting evidence in a light most favorable to the verdict.” While the dissent is correct that we construe conflicting evidence in a light most favorable to the verdict, in this case the evidence is not in conflict. Nor is there any dispute in the circumstances proved. Instead, it is the inferences that can be made from the evidence and the circumstances proved that stand in conflict, and " ‘[w]e give no deference to the fact finder's choice between reasonable inferences.' " Andersen, 784 N.W.2d at 329-30 (quoting Stein, 776 N.W.2d at 716).
. The dissent asserts that Al-Naseer’s statement to the police officer who pulled him over, six miles from the accident scene and perhaps ten minutes later, proves that Al-Naseer left the scene knowing he hit "something.” That assertion is incorrect. Al-Na-seer’s statement to the officer — that he had hit "something” — does not prove he left the scene knowing he hit something. There is no evidence in the record from which it can be inferred when Al-Naseer realized he had hit something. At most, it can be inferred from the circumstances proved that sometime before he got out of his vehicle and went to the damaged right fender, Al-Naseer realized he had hit something.
Similarly, the dissent asserts that the loss of his headlights and "the noticeable change in the operation of his vehicle” also prove that Al-Naseer left the scene knowing he hit "something.” That assertion is not supported by the circumstances proved. The question is not when the headlights on Al-Naseer’s car stopped working or when the vehicle’s operation changed, but when Al-Naseer realized those things had happened, and when Al-Naseer realized that those things necessarily meant he had hit "something.” There is no evidence in the record on this point.
Thus, the fact that Al-Naseer left the scene knowing he had hit something is not a circumstance proven by the lack of operational headlights, the change in the operation of Al-Naseer’s vehicle, or Al-Naseer’s statements to police.
. The dissent, adopting the State’s position, challenges the reasonableness of our conclusion that Al-Naseer did not necessarily know he was in an accident with Thomson. According to the dissent, our conclusion is that “the circumstances proved support a rational hypothesis that before, during, and after the accident, Al-Naseer saw and heard nothing.” This is an inaccurate characterization of our conclusion. Our conclusion that the circumstances proved support a rational hypothesis other than guilt does not rest on the fact that "before, during, and after the accident Al-Naseer saw and heard nothing.” Rather, our conclusion rests on the fact that the evidence in the record establishing the circumstances proved before us does not preclude inferences that Al-Naseer did not know what he had hit. While the record does indicate that at the time Al-Naseer was stopped by the Dilworth police officer Al-Naseer knew that he had hit something, the record does not indicate when he first knew that he had hit something. The only evidence in the record as to what Al-Naseer knew about what he had hit was his statement to the police officer that he did not know what he had hit. Given the silence in the record with respect to what Al-Naseer saw, heard, and knew and when he saw, heard, and knew it, and the evidence that Al-Naseer could have been asleep or otherwise unconscious at the time of the accident, that Al-Naseer knew that he had hit a person or a vehicle is not the only reasonable inference to be drawn from the circumstances proved. Put another way, the circumstances proved do not preclude the reasonable inference that Al-Naseer did not know that he had hit Thompson or his vehicle at the time he left the scene of the accident.
. Because we reverse Al-Naseer’s conviction based on the lack of sufficient evidence, we have no occasion to, and therefore do not, reach the due process issues raised by Al-Naseer.