State of Minnesota v. Nicholas James Firkus

STATE OF MINNESOTA

IN SUPREME COURT

A23-0973

Ramsey County Moore, III, J.
Concurring, Thissen, J.
Concurring, Procaccini, McKeig, Hennesy, JJ.
Concurring in part, dissenting in part, Hudson, C.J.

State of Minnesota,

Respondent,

vs. Filed: February 25, 2026
Office of Appellate Courts
Nicholas James Firkus,

Appellant.

________________________

Keith Ellison, Attorney General, Saint Paul, Minnesota; and

John J. Choi, Ramsey County Attorney, Alexandra Meyer, Assistant County Attorney,
Saint Paul, Minnesota, for respondent.

Robert D. Richman, Saint Louis Park, Minnesota, for appellant.

Keith Ellison, Attorney General, Thomas R. Ragatz, Assistant Attorney General, Saint
Paul, Minnesota for amicus curiae Minnesota Attorney General.

Robert M. Small, Executive Director, Minnesota County Attorneys Association, Saint
Paul, Minnesota;

Travis J. Smith, Murray County Attorney, Slayton, Minnesota; and

Brittany D. Lawonn, Assistant Hennepin County Attorney, Minneapolis, Minnesota for
amicus curiae Minnesota County Attorneys Association.

1
Cathryn Middlebrook, Chief Appellate Public Defender, William Ward, Minnesota State
Public Defender, Saint Paul, Minnesota for amicus curiae Minnesota Board of Public
Defense.

Anders J. Erickson, Johnson Erickson Criminal Defense, Minneapolis, Minnesota, for
amicus curiae Minnesota Association of Criminal Defense Lawyers.

________________________

S Y L L A B U S

1. The circumstantial evidence is sufficient to support appellant’s conviction of

first-degree premeditated murder.

2. A district court should apply the direct-evidence standard in considering a

motion for a judgment of acquittal that is made and decided before the verdict is returned,

even when the State’s proof of an element of the offense is entirely circumstantial.

Affirmed.

O P I N I O N

MOORE, III, Justice.

A Ramsey County grand jury indicted appellant Nicholas James Firkus for

first-degree premeditated murder in connection with the shooting death of his wife Heidi 1

Firkus. 2 Firkus pleaded not guilty and demanded a jury trial. A unanimous jury found that

the State proved beyond a reasonable doubt that Firkus intentionally killed Heidi with

1
Because appellant and the victim share the same last name, this opinion refers to
appellant as “Firkus” and to his wife as “Heidi.”
2
The grand jury also indicted Firkus for the lesser-included offense of second-degree
intentional murder. Because we conclude that the State’s circumstantial evidence was
sufficient to support Firkus’s conviction of first-degree premeditated murder, no further
discussion of the lesser-included offense is required here.

2
premeditation. Based on the jury’s guilty verdict, the district court entered a judgment of

conviction and sentenced Firkus to life without the possibility of release.

On direct appeal to our court, Firkus makes two arguments. First, he argues that the

circumstantial evidence is insufficient to support his conviction of first-degree

premeditated murder. Although Firkus does not deny that he was holding the shotgun that

fatally shot Heidi, he argues that the circumstantial evidence presented at trial is insufficient

because the circumstances proved support a reasonable inference that he shot her without

premeditation or intent during a struggle with an intruder. Second, he argues that the

district court erred when it applied the direct-evidence standard in considering his motions

for judgment of acquittal, which were made and decided before the verdict was returned.

Applying our long-standing two-step circumstantial-evidence test, we conclude that the

circumstantial evidence is sufficient. We also conclude that the district court applied the

correct standard in considering the motions for judgment of acquittal that were made and

decided before the verdict was returned. We therefore affirm.

FACTS

Following a police investigation, a Ramsey County grand jury indicted Firkus for

first-degree premeditated murder, Minn. Stat. § 609.185(a)(1), in connection with the

shooting death of his 25-year-old wife Heidi. Firkus pleaded not guilty and demanded a

jury trial. The following evidence was presented at trial.

A 911 employee testified that the call center received a 911 call from Heidi, which

was made from her house on West Minnehaha Avenue in St. Paul at around 6:30 a.m. on

April 25, 2010. A recording and transcript of the call were admitted into evidence. During

3
the 911 call, Heidi reported that “someone’s trying to break into my house.” After 30 to 40

seconds of Heidi speaking, a loud noise consistent with a gunshot is heard, and the call

ends abruptly. The recording of the 911 call does not contain any sounds of a struggle or

any intruder voice before that noise. About a minute later, Firkus called 911 and told the

dispatcher that “somebody just broke into our house and shot me and my wife.”

First responders testified as follows. When they arrived at the Firkuses’ house, they

found Heidi lying on the floor of the kitchen doorway with a fatal gunshot wound to her

back. Firkus was alive with a gunshot wound to his leg and was transported to a hospital

for medical treatment.

The responding police officers testified as follows. When they arrived at the

Firkuses’ house several minutes after the 911 call, everything appeared calm. The officers

saw no suspicious vehicles or people in the area. The front door was open about an inch.

Photos of the front door of the Firkuses’ house revealed no signs of forced entry or torn

paint or wood. Based on his experience, a police sergeant told the jury that forcing open

the front door with a screwdriver or other tool would have torn away “the paint and the

wood” on the front door frame, and that “the light impressions” on the front door frame

appeared to be “old” and painted over. On cross-examination, defense counsel asked if it

was possible “to use a flat head screwdriver to open a knob lock without tearing away wood

and paint.” The sergeant replied, “In my experience and observations, the possibility that

you’re stating is not probable, but it is possible.” A locksmith testified that if the front door

deadbolt of the Firkuses’ house was not engaged, “the knob lock could be defeated in

15 seconds.”

4
 The district court admitted into evidence the following photograph, which

documented the appearance of the foyer when the responding officers entered the house.

The responding officers observed undisturbed items on the foyer table, which was located

extremely close to the spot where Firkus later told officers he was standing when he fired

the shotgun. One of the officers described the table as “wobbly” and when one of the

officers touched the table, some items on the table fell off. Law enforcement took the

photograph above to show, in part, that the house where the Firkuses lived had a half-moon

window in the front door and from the upstairs landing, a person could see if anyone was

standing outside the front door.

5
 The State also introduced Firkus’s recorded police interviews in which Firkus made

the following statements. The Firkuses’ house was equipped with a security system which

the Firkuses had not activated, but they were “pretty religious” about locking the door

deadbolts. On the night before the shooting, Firkus and Heidi had burgers delivered,

opened a bottle of wine, and watched a movie. When they went to bed, Firkus placed his

two cellphones on his bedside table. On the morning in question, he woke up around

6:00 a.m. to get a drink of water and then heard someone “fiddling with our front door.”

He went to the bedroom closet and loaded his short 20-gauge double-barrel shotgun with

two shells. When loaded, the shotgun’s safety automatically engages and must be pushed

forward and disengaged before the trigger can be pulled. Firkus explained that he had

moved the shotgun from the basement to the bedroom closet about a month earlier. Firkus

did not call 911 on either of his cellphones, which were sitting on his bedside table. Instead,

Firkus woke up Heidi, who was sleeping “like a rock,” told her there was somebody

fiddling with the front doorknob, and directed her to call 911 as Heidi and Firkus headed

to their garage. Firkus explained that in the past, the couple had formulated a plan to escape

out the back door to the garage if anyone ever tried to break into their house. Firkus did

not grab either of his cellphones when they left the bedroom.

Firkus also told the officers that after he and Heidi left their bedroom, they walked

down the stairs, toward the front door of the house. When one of the police officers asked

why Firkus did not wait on the landing, where he would have had an advantage over an

intruder entering the front door, Firkus replied that “the compassionate person in me is not

pulling the trigger.” Firkus further explained that as he reached the front door, a “really

6
big” intruder, who was around 6’1” and 220 pounds, burst through the door, grabbed the

end of the shotgun, pushed it up against Firkus’s chest, and Firkus’s finger slipped onto the

trigger, and he shot Heidi in the back. Firkus watched as “she went straight down” falling

on her face, at which point the intruder rotated the shotgun toward Firkus’s left leg, the

second shotgun shell was fired into his left leg, and then the intruder fled out the front

door. 3 Firkus, who is 5’10”, told the officers that during the struggle, he and the intruder

“never really moved. We didn’t shove back and forth against the wall. We were just in

one spot just kinda trying to figure it out.”

Firkus’s father-in-law testified that on April 26, 2010, the day after the murder,

Firkus said the police would never find the alleged intruder. When he suggested the police

might be able to pull fingerprints off the barrel of the shotgun, Firkus “immediately”

interjected that the intruder had worn gloves. Firkus’s brother-in-law testified that Firkus

appeared “oddly almost giddy or carefree” following Heidi’s death.

A police sergeant testified that within a couple of days after Heidi’s murder, the

police conducted a “sound test” on the Firkuses’ house around 6:15 a.m. Two officers were

upstairs with a video recorder, while one officer downstairs acted as someone entering the

house by opening the screen porch door and then turning the doorknob to the front door

back and forth. In the video recording of this “sound test,” the doorknob jiggling is

inaudible from the bedroom and the bathroom; it is faintly audible from the top of the stairs

3
Medical personnel testified that the second shot was fired in a manner consistent
with a self-inflicted wound. The injury was also consistent with a tangential wound, which
lessens the risk of nerve injury because it affects only the skin and fatty tissue immediately
beneath the skin.

7
between the bedroom and the bathroom. Sounds from the screen porch door were

completely inaudible. Although the sergeant explained that the sound on the video recorder

reflected what he could hear at the time police conducted the test, he agreed that a video

recorder’s sound sensitivity is not necessarily the same as a human ear.

A specialist from the Federal Bureau of Investigation (“FBI”) testified as follows.

In 2020, the FBI created a digital model of the crime scene that showed the firing angles

and ranges for a person Firkus’s height, along with Heidi’s location when she was shot.

The digital model indicated that Firkus was 4 to 9 feet behind Heidi when his finger pulled

the trigger of the shotgun, which was within the Minnesota Bureau of Criminal

Apprehension’s (“BCA”) ballistics testing range. 4 The FBI specialist who helped create

the digital model explained that the model did not include an intruder because the presence

of one or more intruders would not change any of the trajectories calculated in this case.

On cross-examination, the FBI specialist admitted that the trajectory calculations, by

themselves, did not rule out the possibility that an intruder had grabbed the shotgun. On

redirect, the prosecutor asked a question that focused on possible but improbable scenarios.

In response to those questions, the specialist agreed that “anything is possible.”

The State also presented evidence that in 2008, Firkus began to experience financial

problems, which prevented him from consistently making the Firkuses’ mortgage

payments. Ultimately, the Firkuses’ house was in foreclosure and scheduled for eviction

on April 26, 2010, the day after the shooting. Although the Firkuses were members of a

4
The BCA’s ballistic testing indicated that Heidi was between 3 and 10 feet from the
shotgun.

8
close group of couples who discussed the difficulties in their lives (including financial

difficulties), neither Heidi nor Firkus had told friends or family about the foreclosure.

Moreover, when the officers responded to the shooting, they observed that each room was

fully furnished, the kitchen was well-stocked, books remained on the shelves, and the

bedroom closet was filled with clothes on hangers and shelves. Other than a few empty

boxes in the dining area and a few collapsed boxes in the basement, there was no sign the

house was about to be vacated a day later on April 26.

Documentary evidence described the details surrounding the foreclosure as follows.

In April 2009, about a year before Heidi’s death, the bank personally served Firkus with

foreclosure documents and notice of a sheriff’s sale. Heidi was not served with these

documents. Two months later, in June 2009, the bank bought the Firkuses’ house at the

sheriff’s sale. After the Firkuses failed to redeem the house, an eviction hearing was

scheduled for March 8, 2010. The hearing was held a short walk away from where Heidi

worked. At the hearing, Firkus appeared by himself and signed paperwork agreeing to

vacate the house by March 22. The Firkuses did not vacate the house on March 22. Firkus

subsequently negotiated an extension to the day on which they would be locked out of the

house, claiming his grandmother was dying and he needed more time to move. On April 9,

Firkus received a letter informing him that on April 26, the day after the shooting, the

sheriff would physically remove the Firkuses and their property from the house. On

April 18, Heidi invited friends to play badminton at the Firkuses’ house on the weekend of

April 30, four days after the sheriff was scheduled to physically remove the Firkuses from

their house. On April 23, Firkus emailed Heidi about scheduling a meeting with a realtor

9
and contacting U.S. Bank about fraud on their account. By that point, the Firkuses did not

have any active accounts at U.S. Bank, and law enforcement could not find evidence of the

alleged meeting with the realtor. Firkus also emailed Heidi telling her that he needed to

have her all to himself on the night of April 24.

After the State rested its case in chief, Firkus made a motion for a judgment of

acquittal. The district court denied the motion based on its application of the direct-

evidence standard. 5

Firkus then called several witnesses, including B.O., the boyfriend of one of his

former next-door neighbors, who was staying overnight at the neighbor’s house to cat sit

when Heidi was killed. B.O. testified as follows. On the morning in question, B.O. heard

a couple of bangs, looked out the window, and saw no one leaving the Firkuses’ house.

After the shots, he heard a single muffled voice. Although B.O. could “not clearly” make

out what the muffled voice said, he “thought” he heard a male say, “You shot me” or “You

shot her.” B.O. made similar statements to a newspaper reporter later that same day. B.O.

conceded that at first thought the muffled voice was a man, but later he thought the muffled

voice was a woman. Before trial, B.O. listened for the first time to the recording of the 911

call that Firkus made about a minute after Heidi was shot. When he heard Firkus’s voice

in that 911 call, it sounded like the muffled voice he heard on April 25. He added that the

5
Under the direct-evidence standard, a court views the evidence and all resulting
inferences in favor of the State and asks whether the evidence is sufficient to present a fact
question for the jury’s determination, and, if so, whether a reasonable jury could find the
defendant guilty beyond a reasonable doubt. State v. Slaughter, 691 N.W.2d 70, 74–75
(Minn. 2005).

10
statements he made to the newspaper reporter, including his statement that he heard the

muffled voice say, “You shot me” or “You shot her,” were consistent with that 911 call.

The State admitted a recording and transcript of the 911 call that Firkus made about a

minute after Heidi was shot. The statements “You shot me” and “You shot her” do not

appear in the recording or transcript.

At the conclusion of all the evidence but before closing arguments, Firkus made a

second motion for a judgment of acquittal. Again applying the direct-evidence standard,

the district court denied the motion.

During closing argument, defense counsel argued that the State presented

insufficient evidence to prove beyond a reasonable doubt that Firkus intentionally killed

Heidi with premeditation. Acknowledging the inconsistencies in B.O.’s prior statements

and trial testimony, defense counsel argued that the jury should resolve the inconsistencies

in favor of his prior statement that the muffled voice said, “You shot me” or “You shot her,”

because it reflected B.O.’s “best memory of what he heard.” Defense counsel also

emphasized to the jury the testimony of the State’s witnesses that (1) it was possible to

open a knob lock with a prying motion without chipping the paint, and (2) that the FBI

trajectory calculations, standing alone, did not rule out the possibility that an intruder had

grabbed the shotgun.

The jury found Firkus guilty of intentionally killing Heidi with premeditation.

Based on the jury’s guilty verdict, the district court entered a judgment of conviction and

sentenced Firkus to life in prison without the possibility of release.

11
 Firkus filed a direct appeal to our court, arguing that the circumstantial evidence is

insufficient to support his conviction and that the district court erred when it applied the

direct-evidence standard in ruling on his pre-verdict motions for judgment of acquittal.

After we first heard oral arguments, we requested supplemental briefing and reheard oral

argument on six legal questions, which focused on how to identify the circumstances

proved in a circumstantial-evidence analysis, including whether our decisions in State v.

Culver, 941 N.W.2d 134, 144 (Minn. 2020) and State v. Al-Naseer, 788 N.W.2d 469, 476

(Minn. 2010) were consistent with our long-standing circumstantial-evidence test, and also

inquired as to the standard to be applied in considering a motion for a judgment of acquittal

that is made and decided before a verdict is returned.

ANALYSIS

We begin our analysis by considering whether the circumstantial evidence is

sufficient to support Firkus’s conviction of first-degree premeditated murder. 6 Then, we

address the standard that should apply in considering a motion for a judgment of acquittal

that is made and decided before the verdict is returned, when the State’s proof of an element

of the offense is entirely circumstantial.

I.

When the State presents only circumstantial evidence to prove the elements of

premeditation or intent to kill, we apply a two-step test to assess the sufficiency of the

6
Because the parties agree that the State presented no direct evidence that Firkus
acted with premeditation and intent to kill, we consider whether the circumstantial evidence
is sufficient to establish those elements.

12
evidence on those elements. State v. Ulrich, 3 N.W.3d 1, 11 (Minn. 2024). The first step

requires us to “winnow down the evidence presented at trial by resolving all questions of

fact in favor of the jury’s verdict,” which results in “a subset of facts that constitute the

circumstances proved.” State v. Harris, 895 N.W.2d 592, 600 (Minn. 2017) (citation

omitted) (internal quotation marks omitted). At the second step, “we consider whether the

reasonable inferences that can be drawn from the circumstances proved, when viewed as a

whole and not as discrete, isolated facts, ‘are consistent with the hypothesis that the accused

is guilty and inconsistent with any rational hypothesis other than guilt.’ ” State v. Smith,

9 N.W.3d 543, 565 (Minn. 2024) (quoting State v. Hassan, 977 N.W.2d 633, 640

(Minn. 2022)).

To prove that Firkus committed first-degree premeditated murder, the State needed

to establish beyond a reasonable doubt that Firkus intentionally killed Heidi with

premeditation. See State v. Jenkins, 782 N.W.2d 211, 225 (Minn. 2010) (describing the

elements of first-degree premeditated murder); Minn. Stat. § 609.185(a)(1) (same). Firkus

does not deny pulling the trigger of the shotgun that he had just loaded and firing the shot

that killed Heidi. Instead, he argues that the circumstantial evidence is insufficient because

it supports a reasonable inference that he shot Heidi without premeditation and intent

during a struggle with an intruder.

We conclude that when the circumstances proved are identified and viewed as a

whole and not as discrete, isolated facts, they support only one reasonable inference—

Firkus caused Heidi’s death with premeditation and intent to effect her death. In reaching

that conclusion, we apply our long-standing process of identifying the circumstances

13
proved during the first step of our circumstantial-evidence analysis and reject Firkus’s

argument that our prior use of the word “inconsistent” in some of our cases has altered that

test. Next, we consider whether our decisions in Culver, 941 N.W.2d 134, and Al-Naseer,

788 N.W.2d 469, are consistent with our long-standing circumstantial-evidence test. We

then review our existing jurisprudence regarding the second step of the

circumstantial-evidence test and how it interacts with the first step, emphasizing that during

the second step, we view the circumstances proved as a whole and not as discrete, isolated

facts. Finally, we conduct a circumstantial-evidence analysis of the specific facts here and

explain why we disagree with the arguments made by Firkus and our colleagues.

A.

The first step of our circumstantial-evidence test—identifying the circumstances

proved—protects the well-established legal principle that “the jury is in a unique position

to determine the credibility of the witnesses and weigh the evidence before it.” Harris,

895 N.W.2d at 592. In fact, we have repeatedly said the jury is in “the best position” to

evaluate the credibility of evidence, even in cases based on circumstantial evidence. See,

e.g., State v. Moore, 846 N.W.2d 83, 88 (Minn. 2014); State v. Andersen, 784 N.W.2d 320,

329 (Minn. 2010).

In its unique position, a jury “is free to accept part and reject part of a witness’s

testimony.” Harris, 895 N.W.2d at 600 (quoting State v. Landa, 642 N.W.2d 720, 725

(Minn. 2002)); accord State v. Poganski, 257 N.W.2d 578, 581 (Minn. 1977). In evaluating

a witness’s credibility, a jury may consider the witness’s “interest and past inconsistent

statements.” State v. Moore, 438 N.W.2d at 101, 108 (Minn. 1989); see also State v. Asfeld,

14
662 N.W.2d 534, 546 (Minn. 2003) (explaining that in deference to the jury’s “role as

evaluator of witness credibility,” the court was required to assume “the jury disbelieved

Asfeld’s claim that he did not murder [the victim]”). This principle applies with equal force

to the testimony of expert witnesses. State v. Triplett, 435 N.W.2d 38, 44 (Minn. 1989)

(explaining that “[w]eighing the credibility of witnesses, including expert witnesses, is the

exclusive function of the jury”).

Consequently, our precedent does not permit us to re-weigh the evidence and “sit,

in essence, as a 13th juror.” State v. Reek, 942 N.W.2d 148, 166 (Minn. 2020) (declining

to reweigh the testimony of the State’s DNA witness and the defendant’s DNA witness

because the jury heard the witnesses’ testimony, evaluated any differences between them,

and weighed their relative credibility). This prohibition is especially important in contexts

where “appellate review of a cold transcript” cannot substitute for direct observation of the

person’s demeanor. See State v. Munt, 831 N.W.2d 569, 576 (Minn. 2013) (explaining that

a person’s “demeanor plays a fundamental role” in determining “credibility”).

At times, our descriptions of the winnowing-down process have not reflected the

level of precision articulated above. For example, we have sometimes said we must

disregard evidence that is “inconsistent with the jury’s verdict.” See, e.g., State v. Lehman,

3 N.W.3d 875, 878 (Minn. 2024); State v. Cruz, 997 N.W.2d 537, 551 (Minn. 2023)

(quoting Harris, 895 N.W. at 601); State v. Colgrove, 996 N.W.2d 145, 150 (Minn. 2023)

(same). Firkus argues that our use of the phrase “inconsistent with the jury’s verdict”

means any fact that is hypothetically consistent with the verdict must be included in the

circumstances proved regardless of the jury’s unique position to determine the credibility

15
of the witnesses and weigh the evidence before it. We disagree. Our use of this imprecise

phrase was not intended to shift the focus of the winnowing-down process away from

protecting the jury’s unique position to determine the credibility of the witnesses and weigh

the evidence before it, as shown by our contemporaneous references to the jury’s role as

the sole judge of credibility in each case. 7 Lehman, 3 N.W.3d at 878; Cruz, 997 N.W.2d at

551; Colgrove, 996 N.W.2d at 150; 8 Harris, 895 N.W.2d at 600; see also State v. Isaac,

9 N.W.3d 812, 815 (Minn. 2024) (acknowledging “the [fact-finder] is in a unique position

to determine the credibility of the witnesses and weigh the evidence before it” (alteration

in original) (quoting State v. Harris, 895 N.W.2d at 600). We reiterate that, as the

fact-finder, the jury is uniquely positioned to assess witness credibility and weigh the

evidence, and that jurors may accept some parts of a witness’s testimony while rejecting

7
In focusing on whether a piece of evidence is actually or theoretically inconsistent
with the guilty verdict, the separate writings of Justice Thissen and the Chief Justice not
only lose sight of the purpose of the winnowing-down process (protecting the unique
position of the jury to determine the credibility of the witnesses and weigh the evidence
before it), but they also propose an approach that infringes on the jury’s role.
8
We disagree with the concurrence’s view of our decision in Colgrove, 996 N.W.2d
at 150. In Colgrove, the State essentially argued that the evidence of the defendant’s
paranoia and delusions should be excluded from the circumstances proved because, as a
matter of law, “a person experiencing this type of behavior can never form an intent to kill
another person.” Id. at 151. We rejected the State’s legal argument, explaining that we had
consistently held that “[t]he mere fact of a person’s [using intoxicants] does not create a
presumption of intoxication, and the possibility of intoxication does not create the
presumption that a person is rendered incapable of intending to do a certain act.” Id. at
151–52 (citation omitted) (internal quotation marks omitted). Notably, the State did not
argue the defendant’s paranoia and delusions should be excluded from the circumstances
proved to protect the unique position of the jury to determine the credibility of the witnesses
and weigh the evidence before it.

16
others, and that the first step of our circumstantial-evidence test protects this well-

established legal principle. 9

B.

Having reviewed and reiterated our existing jurisprudence regarding the first step

of the circumstantial-evidence test, we consider the following question about how we

determine in the first step whether certain evidence is included in the subset of

circumstances proved, on which we requested and received supplemental briefing:

To determine whether a piece of evidence is a circumstance proved, how
should we reconcile the statement in State v. Culver, 941 N.W.2d 134, 144
(Minn. 2020), that a jury is not compelled to believe statements merely
because they are uncontradicted, with State v. Al-Naseer, 788 N.W.2d 469,
476 (Minn. 2010), where we included as circumstances proved
uncontradicted statements by an eyewitness that were not consistent with the
verdict?

After considering both cases, we conclude that both Culver and Al-Naseer are consistent

with our existing law on whether a piece of evidence is a circumstance proved for purposes

of our circumstantial-evidence test.

In Culver, the defendant argued there was insufficient evidence to support her

conviction of felony deprivation of parenting rights because the circumstances proved

9
Our decision in State v. Allwine, 963 N.W.2d 178, 187 (Minn. 2021), does not alter
our analysis here. In Allwine, we said the circumstances proved included the fact that a
forensic expert found software “needed to access the dark web” on the MacBook Pro laptop
computer and the fact that the defendant “told law enforcement that he had never used the
dark web before.” Id. at 187. Viewed in isolation, this statement could be read to indicate
that the jury had to believe the truth of the defendant’s statement that he never used the
dark web. But when both parts of our statement are viewed together, they reflect a
presumption that the jury found the defendant lied to law enforcement when he said he
never used the dark web—a finding consistent with the jury’s guilty verdict.

17
supported a reasonable inference that the deprivation of parenting time was not substantial.

941 N.W.2d at 143. As part of her argument, the defendant asserted that the content of

online messages she sent to the child’s father stating she was willing to reschedule the

parenting time were part of the circumstances proved in her case. 10 Id. at 144. Consistent

with the purpose of the first step of the circumstantial-evidence test, we protected the legal

principle that a jury is not compelled to believe the testimony of a witness just because it

is uncontradicted by concluding that the circumstances proved did not include the content

of the online messages the defendant sent to the child’s father stating she was willing to

reschedule parenting time. Id. (citing Costello v. Johnson, 121 N.W.2d 70, 76

(Minn. 1963)). Put differently, whether the defendant spoke the truth when she messaged

a willingness to reschedule parenting time was a question of fact for the jury to resolve.

And the absence of any evidence directly contradicting Culver’s statement that she was

willing to reschedule parenting time did not prevent the jury from weighing it against other

evidence, including evidence that the defendant had previously lied about being out of town

at a funeral, and finding that the statement was not true. Id.

Having reviewed the supplemental briefing submitted by the parties and amici, we

conclude that our identification of the circumstances proved in Culver reflects a sound

10
The amended parenting-time order in Culver required that “[a]ll communications
between the Parties . . . occur through Family Wizard,” which is an online application for
co-parenting communication. 941 N.W.2d at 137 n.1.

18
application of the first step of the circumstantial-evidence test, which is designed to protect

the jury’s credibility findings and its weighing of the evidence. 11

In Al-Naseer, the State charged the defendant with criminal vehicular homicide,

which required the State to prove in part that the defendant knew he had been in an accident

with a person or vehicle. 788 N.W.2d 469, 475 (Minn. 2010). The defendant pleaded not

guilty and waived his right to a jury and submitted his case to the district court judge. Id.

at 471. At the bench trial, the only eyewitness testified that the brake lights of the vehicle

driven by the defendant “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.” Id. at 476. The district court found the defendant guilty of criminal vehicular

homicide (leaving the scene). Id. As part of its written verdict, the district court, sitting as

the trier of fact, made the following factual findings: “Immediately after the accident,

Defendant corrected the direction of travel of his vehicle by steering the vehicle in a west-

southwesterly direction,” and “[t]he Defendant did not activate his brakes before or after

the accident.” State v. Al-Naseer, No. K1-02-1163, General Verdict of the Court at 5 (Clay

Cnty. Dist. Ct. filed Feb. 23, 2005). In reaching its guilty verdict, the district court inferred

11
The Minnesota Board of Public Defense and Minnesota Association of Criminal
Defense Lawyers agree that the content of the text messages in Culver was properly
excluded from the circumstances proved because other facts and circumstances in the
evidence furnished a reasonable ground for doubting the content of the text messages,
including the evidence that the defendant lied about being out of town at a funeral on the
days in question. Br. of Minn. Bd. of Pub. Def. & Minn. Ass’n of Crim. Def. Laws. as
Amici Curiae 12 (“Amici Br.”).

19
from these factual findings that “the corrective steering action taken by the Defendant

immediately after the accident establishes that he was aware of his surroundings.” Id.

On appeal, we included the eyewitness testimony about the corrective steering and

the lack of braking in the circumstances proved. Al-Naseer, 788 N.W.2d at 478. But unlike

the district court, we concluded that the circumstances proved, including the corrective

steering and the lack of braking, supported a reasonable inference that the defendant was

asleep or otherwise unconscious when his vehicle hit the victim, and thus the State failed

to prove beyond a reasonable doubt one of the elements of the offense, specifically that the

defendant knew he had been in an accident with a person or vehicle. Id. at 479, 481.

We included the eyewitness testimony in the circumstances proved not because the

testimony was uncontradicted, but because the district court, acting as the trier of fact,

believed the eyewitness testimony and answered relevant questions of fact by expressly

finding that the defendant corrected the vehicle’s direction of travel and never activated the

brakes. And, we never used the word “uncontradicted” in our analysis in Al-Naseer. Thus,

consistent with the purpose of the first step of circumstantial-evidence test, we included

the eyewitness testimony in the circumstances proved to protect the district court’s unique

position to determine the credibility of the witnesses and weigh the evidence before it. 12

12
Firkus asserts that in Al-Naseer, we held that the defendant’s version of events was
one of the circumstances proved. Our reasoning in Al-Naseer does not support Firkus’s
assertion. We only referenced Al-Naseer’s statement to police—he knew he had hit
something but did not know what he had hit—in responding to the dissent’s contention that
Al-Naseer’s statement proved he knew that he hit something when he left the scene of the
accident. Al-Naseer, 788 N.W.2d at 478 n.3. In our analysis of whether the circumstances
proved supported an inference that Al-Naseer knew he hit something when he left the scene

20
 In sum, we conclude that our decisions in Culver and Al-Naseer do not contradict

our existing law. Instead, they both reflect a sound application of the legal principle that

the trier of fact is in a unique position to determine the credibility of the witnesses and

weigh the evidence before it and that the first step of our circumstantial-evidence test

protects that legal principle. 13

C.

Next, we review our existing jurisprudence regarding the second step of the

circumstantial-evidence test. Under the second step of the circumstantial-evidence test, we

consider whether the reasonable inferences that can be drawn from the circumstances

of the accident, we did not mention Al-Naseer’s statement to police. See id. at 478.
Moreover, since our decision in Al-Naseer, we have explicitly held that a defendant’s
alternative version of events is not a circumstance proved. See, e.g., Lehman, 3 N.W.3d at
879.
13
Our request for supplemental briefing did not ask the parties or amici to discuss
whether our application of the first step of the circumstantial-evidence test in Isaac,
9 N.W.3d at 816–17, failed to protect the unique position of the trier of fact to determine
the credibility of the witnesses and weigh the evidence before it. Even so, we address this
issue because it is raised in the concurrence. In Isaac, the defendant was indicted with
first-degree premeditated murder under an aiding and abetting theory of liability. Id. at
814. The shooter shot the victim at approximately 8:04 p.m. Id. We included in the
circumstances proved cell phone records that revealed the shooter called the defendant “at
8:11 p.m., at which point [the defendant] turned his car around and began heading back
northwest toward the Fargo-Moorhead area on Interstate 94.” Id. at 816. Even though the
phone records were inconsistent with the guilty verdict, we properly included them in the
circumstances proved. Because nothing in the record contradicted, impeached, or rendered
improbable the content of the cell phone records or the testimony of the police officer who
reviewed the cell phone records, the finder of fact in Isaac lacked any record-based
reasonable ground for doubting the truth or credibility of this evidence, and thus it could
not disregard the cell phone records. Consequently, our inclusion of the cell phone records
in the circumstances proved did not infringe on the unique position of the trier of fact to
determine the credibility of the witnesses and weigh the evidence before it.

21
proved, when viewed as a whole and not as discrete, isolated facts, are consistent with the

hypothesis that the accused is guilty and inconsistent with any rational hypothesis other

than guilt. Smith, 9 N.W.3d at 565. During the second step, “we do not defer to the

factfinder but examine the reasonableness of the inferences ourselves.” Isaac, 9 N.W.3d

at 818 (citation omitted) (internal quotation marks omitted). “If the circumstances proved

when viewed as a whole, support a reasonable inference ‘that is inconsistent with guilt, the

evidence is not sufficient to support the conviction and we must reverse.’ ” Id. (quoting

State v. Noor, 964 N.W.2d 424, 438 (Minn. 2021)). But we will not reverse a conviction

“based on mere conjecture.” State v. Tscheu, 758 N.W.2d 849, 861 (Minn. 2008). The

second step of the test “does not encroach on the jury’s credibility determinations because

the act of inferring involves the drawing of permissible deductions, not actual fact finding

by the jury.” Harris, 895 N.W.2d at 600–01.

Having reviewed and reiterated our existing jurisprudence regarding the second step

of the circumstantial-evidence test, we consider other questions on which we requested and

received supplemental briefing relating to the interaction between the first step and second

step analysis:

When a factual determination is not based on direct evidence, but rather an
inference drawn from direct evidence, do we exclude from the subset of
circumstances proved an inference that reasonably may be drawn from direct
evidence if the inference is inconsistent with the guilty verdict or do we
include all reasonable inferences in the set of circumstances proved because
we do not defer to the fact-finder’s choice between reasonable inferences?
Does the answer to the prior question change depending on whether the
factual determination concerns an ultimate question of fact (e.g., a factual
determination about mens rea) or does it apply to all factual determinations
that are not based on direct evidence?

22
 The Minnesota Board of Public Defense and Minnesota Association of Criminal

Defense Lawyers argue in their amici brief that a court should never include in “the

circumstances proved an inference—any inference—because an inference is not a

circumstance proved.” Amici Br. 17. This argument aligns with State v. Meany,

115 N.W.2d 247, 255 (Minn. 1962), in which we explained that an “inference” is a

permissible deduction that can be drawn from the proven or admitted facts. Following our

careful consideration of the supplemental briefing submitted by the parties and amici, as

well as our existing case law, we agree with the Minnesota Board of Public Defense and

Minnesota Association of Criminal Defense Lawyers that because an inference is not a

circumstance proved, an inference should never be included in the first step of the

circumstantial-evidence test.

In turn, under the second step of a circumstantial-evidence analysis, we view the

circumstances proved as a whole, not as discrete, isolated facts, in determining whether a

reasonable inference of guilt can be drawn and no reasonable inference inconsistent with

guilt can be drawn. At this step, we do not defer to the fact-finder; instead, we

independently assess the reasonableness of the inferences of guilt and not guilt. This

review does not intrude on the jury’s credibility determinations because drawing inferences

involves evaluating permissible deductions from the facts found, not making new factual

findings. In no case will we reverse a conviction “based on mere conjecture.” Tscheu,

758 N.W.2d at 861.

23
 D.

Having reiterated the law on our circumstantial-evidence test, we now conduct a

circumstantial-evidence analysis of the specific facts here. We begin by winnowing down

the evidence presented at trial by resolving all questions of fact in favor of the jury’s

verdict, which results in a subset of facts that constitute the circumstances proved. See

Harris, 895 N.W.2d at 600. The circumstances proved in this case are as follows.

In 2008, Firkus began to experience financial problems, which prevented him from

consistently making the Firkuses’ mortgage payments. In April 2009, the bank personally

served Firkus with foreclosure documents and notice of a sheriff’s sale regarding the

Firkuses’ house. Heidi was not served with these documents. In June 2009, the bank

bought the Firkuses’ house at the sheriff’s sale. After the Firkuses failed to redeem the

house, an eviction hearing was scheduled for March 8, 2010.

Although the courthouse where the March 8 eviction hearing was held was only a

short walk from where Heidi worked, she did not appear at the hearing. Instead, Firkus

appeared by himself and signed paperwork agreeing to vacate the house by March 22. The

Firkuses did not, however, vacate the house on March 22. Shortly thereafter, Firkus moved

his short 20-gauge double-barrel shotgun and two shotgun shells from their basement to

their bedroom closet.

On April 9, Firkus received a letter informing him that on April 26, the sheriff would

physically remove the Firkuses and their property from the house. On April 18, Heidi

invited friends to play badminton at the Firkuses’ house on the weekend of April 30, four

days after the sheriff was scheduled to physically remove the Firkuses from their house.

24
 On April 23, Firkus emailed Heidi about scheduling a meeting with a realtor and

contacting U.S. Bank about fraud on their account. Law enforcement investigated and

determined that the Firkuses had no active accounts with U.S. Bank at this time, and law

enforcement could find no evidence of the alleged meeting with the realtor. Firkus also

emailed Heidi telling her that he needed to have her all to himself on the night of April 24.

On the night of April 24, Firkus and Heidi had burgers delivered, opened a bottle of

wine, and watched a movie. Even though the sheriff would be physically removing them

from their house in less than 48 hours, not one room in the house was packed for moving.

When they went to bed, Firkus placed his two cellphones on his bedside table.

The next morning, April 25 (the day before the sheriff was scheduled to physically

remove the Firkuses and their property from the house), Firkus woke up before Heidi. He

went to the bedroom closet and loaded his short 20-gauge double-barrel shotgun with two

shells. When loaded, the shotgun’s safety automatically engages. It must be pushed

forward and disengaged before the trigger can be pulled. Firkus did not call 911 on either

of his cellphones, which were sitting on his bedside table. Instead, Firkus woke up Heidi,

who was sleeping “like a rock,” told her there was somebody fiddling with the front

doorknob, and directed her to call 911 as they headed to their garage. Firkus did not grab

either of his cellphones when they left the bedroom.

The house where the Firkuses lived had a half-moon window in the front door. From

the upstairs landing, a person could see if anyone was standing outside the front door. The

house was equipped with a security system which the Firkuses had not activated, but they

were “pretty religious” about locking the door deadbolts. If the front door deadbolt was

25
not engaged, “the knob lock could be defeated in 15 seconds,” according to the locksmith

who testified at trial.

Firkus had his unarmed wife go down the stairs first, straight toward the front door

of the house. As Heidi walked down the stairs, she was talking to the 911 operator on her

cellphone. Firkus followed behind her with the loaded shotgun. Heidi told the 911

operator, “someone’s trying to break into my house.” She did not describe seeing the

intruder or provide any other information explaining why she believed someone was trying

to break into her house. After she provided her street address, there was a period of silence

followed by a gunshot, a scream, and then the 911 call ended. Heidi’s 911 call lasted about

35 seconds (20 seconds longer than the locksmith testified it would have taken an intruder

to defeat the knob lock).

Shotgun fire struck Heidi in her back. Firkus watched as “she went straight down”

falling on her face. A second shot struck Firkus in the leg. The second shot was fired in a

manner consistent with a self-inflicted, tangential wound.

Several minutes later, police officers arrived at the Firkuses’ house. Everything

appeared calm, the officer saw no suspicious vehicles or people in the area. The front door

was open about an inch. Photos of the front door of the Firkuses’ house revealed no signs

of forced entry or torn paint or wood. A police sergeant testified that forcing open the front

door with a screwdriver or other tool would have torn away “the paint and the wood” on

the front door frame, and that “the light impressions” on the front door frame appeared to

be “old” and painted over.

26
 Items on the “wobbly” foyer table—located immediately beside the front door and

near where Firkus stood when he fired the shotgun into Heidi’s back—were undisturbed.

The undisturbed items on the foyer table are pictured above in a photograph one of the

officers took to document the status of the foyer table, shoes, rug, and shotgun when the

officers arrived at the crime scene.

Each room in the house was fully furnished, the kitchen was well-stocked, books

remained on the shelves, and the bedroom closet was filled with clothes on hangers and

shelves. Other than a few empty boxes in the dining area and a few collapsed boxes in the

basement, there was no sign the house was about to be vacated a day later, on April 26.

On April 26, 2010, the day after the murder, Firkus told his father-in-law that he

believed the police would never find the intruder. When his father-in-law suggested the

police might pull fingerprints off the barrel of the shotgun, Firkus “immediately”

interjected that the intruder had worn gloves. Firkus’s brother-in-law later testified that

Firkus appeared “oddly almost giddy or carefree” following Heidi’s death.

Two days after Heidi’s death, the police conducted a “sound test” on the Firkuses’

house around 6:15 a.m. Two officers were upstairs with a video recorder while one officer

downstairs fiddled with the inner doorknob and then the screen porch door. In the video

recording, the sound of the doorknob is inaudible from the bedroom and the bathroom; it

is faintly audible from the top of the stairs between the bedroom and the bathroom. Sounds

from the screen porch door were completely inaudible. The sergeant explained that the

sound on the video recorder reflected what he could hear at the time police conducted the

test.

27
 Having identified the circumstances proved, we now consider whether the

reasonable inferences that can be drawn from the circumstances proved, when viewed as a

whole and not as discrete, isolated facts, are consistent with the hypothesis that the accused

is guilty and inconsistent with any rational hypothesis other than guilt. When so viewed,

we conclude that they support only one reasonable inference—Firkus caused Heidi’s death

with premeditation and intent to effect her death. The alternative hypothesis Firkus

advances—that there was an intruder—is not reasonable considering the lack of fresh

marks on the front door, the undisturbed items on the “wobbly” table next to where Firkus

claimed the struggle with the intruder occurred, and the sound test that revealed that efforts

to defeat the front door’s knob lock were inaudible from the bedroom and the bathroom.

Serving as a backdrop to these circumstances was the Firkuses’ dire financial situation, of

which Heidi appeared entirely unaware. A forced eviction was scheduled for the day after

Heidi’s death. Yet not a single box was packed for moving, and Heidi planned for friends

to visit the house the following weekend. And, on the morning of Heidi’s death, Firkus

sent Heidi downstairs ahead of him, without bringing his cell phones along, following

behind with the shotgun that he had recently moved up to the bedroom and loaded that

morning.

Absent any intruder, Firkus’s undisputed actions, which included loading the

shotgun, falsely telling Heidi there was an intruder and directing her to call 911 on her

cellphone, directing Heidi to walk in front of him toward the alleged danger of an intruder

without taking his cell phones with him, following behind her as they descended the steps,

firing a shotgun shell into her back when she reached the kitchen door and then firing a

28
self-inflicted wound into his own leg, support only one reasonable inference—Firkus

caused the death of Heidi with premeditation and intent to effect her death. Thus, the State

presented sufficient evidence to support Firkus’s conviction of first-degree premediated

murder.

E.

In our circumstantial-evidence analysis, we excluded four discrete facts that Firkus

argues should be included in the circumstances proved: (1) there were compression marks

on the front door frame consistent with someone attempting to pry open the knob lock,

(2) B.O.’s statement that he heard a muffled voice say, “You shot her,” or “You shot me,”

(3) the truth of Heidi’s statement, during the 911 call that Firkus directed her to make, in

which she reported that an intruder was trying to break into the house, and (4) the FBI’s

trajectory calculations reflected in the FBI’s digital model of the crime scene which Firkus

asserts were “completely consistent” with the forensic evidence.

For the following reasons, we conclude that the door frame marks, B.O.’s statement

that the muffled voice said, “You shot her” or “You shot me,” and Heidi’s statement that

there was an intruder are not part of the circumstances proved. The jury was required to

evaluate the credibility of the witnesses regarding this evidence and weigh this testimony

against other evidence. More specifically, during deliberations, the jurors faced these

questions of fact relating to each of those factual issues: (1) did the sergeant accurately

testify that the door frame marks were old, (2) did B.O. hear Firkus say, “You shot me,” or

“You shot her,” (3) was the content of the statement made by Heidi during her 911 call

regarding an intruder true, and (4) whether the FBI’s digital model of the crime scene was

29
probative of whether there was an intruder. The record contained evidence that called into

question each of these statements, including the sergeant’s admission that it was “possible”

that the marks were new; B.O’s ever-changing account of what he heard; and other

evidence, discussed in more detail below, that makes Firkus’s account of an intruder highly

improbable. Under our precedent, we must assume the jury determined the sergeant’s

testimony that the marks were old was credible; assume that the jury determined that B.O.’s

testimony provided no weight as to whether there was an intruder given the lack of detail

and his inconsistent statements; assume the jury determined that the content of the

statement made by Heidi was not true; and assume the jury determined that the FBI’s digital

model was not probative of whether there was an intruder.

The sergeant testified that the marks on the door were old, but conceded on cross-

examination that it was possible, although not probable, that the marks were new. The

marks in question could not be both old and new. The jury, in its unique position to

determine the credibility of the sergeant and weigh the evidence before it, was free to

conclude that the sergeant’s opinion that the marks were old, which was based on his

training and experience, was not called into question by the theoretical possibility adduced

on cross-examination that they were new. The circumstance proved based on the sergeant’s

testimony and other evidence consistent with that testimony is therefore that the marks

were old.

Similarly, although B.O. could “not clearly” make out what the muffled voice said,

he “thought” he heard a man say, “You shot her” or “You shot me.” B.O. made similar

statements to a newspaper reporter later that same day. B.O. conceded that he first thought

30
the muffled voice was a man, but later he thought the muffled voice was a woman. Before

trial, B.O. listened for the first time to the recording of the 911 call that Firkus made about

a minute after Heidi was shot. When he heard Firkus’s voice in that 911 call, it sounded

like the muffled voice he heard on April 25. B.O. added that the statement he made to the

newspaper reporter, including his statement that he heard the muffled voice say, “You shot

her” or “You shot me,” were consistent with that 911 call. But the statements “You shot

her” and “You shot me” do not appear in the recording or transcript of the 911 call that

Firkus placed about a minute after Heidi was shot. So the jury needed to grapple with the

inconsistency in B.O.’s testimony, and in its unique position to determine the credibility of

B.O’s testimony and weigh the evidence before it, the jury was free to conclude that the

B.O.’s testimony, including his statement that he heard the muffled voice say, “You shot

her” or “You shot me,” was not credible. See Moore, 438 N.W.2d at 108 (explaining that

in evaluating the credibility of a witness, a jury may consider the “past inconsistent

statements” of the witness). The circumstances proved therefore do not include B.O.’s

statement that the muffled voice said, “You shot her” or “You shot me.”

As for the contents of Heidi’s 911 call, the record contains other evidence that

allowed the jury to disregard the call’s content on the grounds that the truth of the contents

was improbable. Cf. Culver, 941 N.W.2d at 144. This is especially true because there is a

period of silence, not the sounds of a struggle, before Firkus shoots Heidi in the back. 14

14
To the extent that Firkus argues the absence of any sounds of a struggle can be
explained away by an argument that Heidi’s cellphone was “likely” not on speaker mode,
his argument fails because there is no evidence in the record about whether her phone was

31
Other evidence that the jury weighed supports a factual determination, consistent with the

verdict, that the contents of Heidi’s 911 call merely reflect her belief that an intruder was

present rather than the actual presence of an intruder. This includes evidence that (1) Firkus

decided to wake Heidi, who was sleeping like a rock, and instruct her to call 911 rather

than report the presence of an intruder by calling 911 from either of his own cell phones on

the bedside table, (2) Firkus allowed his unarmed wife to go down the stairs first, straight

toward the front door of the house, (3) substantially more time elapsed between the time

Firkus claimed he heard someone fiddling with the doorknob and when he passed by the

front door than the 15 seconds required to defeat the knob lock, and (4) police found the

front door open by only about an inch upon their arrival. The circumstance proved based

on Heidi’s 911 call is that Firkus falsely told her there was an intruder and directed her to

call 911.

As for Firkus’s reliance on the FBI’s trajectory calculations, the FBI specialist

testified that the presence or absence of an intruder had no effect on the calculations, and

therefore the trajectory calculations are neutral. The specialist testified that the digital

model created in 2020 incorporated data from the ballistics report that revealed that Firkus

was 4 to 9 feet behind his wife when his finger pulled the trigger of the shotgun. The

specialist further testified that the digital model did not include an intruder because the

presence of one or more intruders would not change any of the trajectories calculated

or was not on speaker mode. See State v. Colgrove, 996 N.W.2d at 150 n.5 (explaining that
where an alternative inference has no support in the record, it is speculative; and holding
that an inference that a knife was pre-bent was unreasonable because there was nothing in
the record from which the inference could be drawn).

32
here. 15 The digital model thus provided no probative evidence on whether there was an

intruder.

In sum, we exclude the four discrete facts Firkus argues should be included in the

circumstances proved because the jurors were required to evaluate the credibility of the

witnesses and weigh the evidence, which includes evidence contradicting these alleged

facts.

F.

The separate writings authored by Justice Thissen (“the concurrence”) and the Chief

Justice (“the concurrence/dissent”) adopt approaches that focus on whether a piece of

evidence is actually or theoretically inconsistent with the guilty verdict. Both lose sight of

the purpose of the winnowing-down process in step one of a circumstantial-evidence

analysis—protecting a jury’s unique position to determine the credibility of the witnesses

and weigh the evidence before it. The concurrence’s approach undermines the jury’s

important role by, in effect, allowing an appellate court to sit as a 13th juror, reassessing a

witness’s credibility and reweighing the evidence. By contrast, the concurrence/dissent’s

15
In a typical case involving trajectory calculations, there is a dispute about the
position of the victim’s body when the shot was fired or the location from which the shot
was fired. See, e.g., State v. Hill, 801 N.W.2d 646, 650, 656 (Minn. 2011) (concluding that
the bullet trajectory and wounds contradicted the defendant’s claim that the victim was
standing and facing him at the time of the shooting); State v. Pearson, 775 N.W.2d 155,
162 (Minn. 2009) (concluding that bullet trajectory and wounds were consistent with the
eyewitness testimony that the defendant was outside the van when he fired the first shot
and inconsistent with the defendant’s testimony that he was inside the van when he fired
the first shot in self-defense). Here, however, there is no dispute about the location or
position of Heidi’s body when she was shot in the back or Firkus’s location or stance when
his finger pulled the trigger of the shotgun.

33
approach deprives the jury of its unique position to determine the credibility of the

witnesses and weigh the evidence before it.

For example, the concurrence contends the circumstances proved include Firkus’s

statement that he was in the bedroom when he heard the doorknob jiggle and his statement

that he was at the top of the stairs when he heard the doorknob jiggle, even though “the

content of Firkus’s statements conflicts—meaning that both cannot be true.” (Emphasis

removed.) Then, the concurrence resolves the question of fact about the jiggling of the

doorknob, finding that “Firkus’s claim that he heard an intruder jiggling the door is dubious

given his contradictory statements on the subject.” Because resolving factual disputes

arising from contradictory statements lies squarely within the jury’s unique role of

assessing witness credibility and weighing the evidence, see Moore, 438 N.W.2d at 108

(explaining that a jury may consider a witness’s “interest and past inconsistent statements”

when evaluating credibility), the concurrence’s proposed approach—which includes such

disputed facts among the circumstances proved at step one of the circumstantial-evidence

test and then finds them “dubious” (that is, not credible) during the appellate court’s de

novo review at step two—fails to preserve the jury’s role and permits the appellate court to

improperly reweigh the evidence on appeal.

The concurrence/dissent responds to this concern about apparent appellate-court

fact-finding in the concurrence, observing that “[t]he concurrence conducts an

individualized credibility assessment of every statement Firkus gave to police, particularly

34
the suspect’s description.” 16 Like the concurrence, the concurrence/dissent contends the

circumstances proved include Firkus’s statement that he was in the bedroom when he heard

the doorknob jiggle and his statement that he was at the top of the stairs when he heard the

doorknob jiggle. 17 But unlike the concurrence, the concurrence/dissent does not resolve

16
In discussing the inconsistent statements Firkus made during his police interview,
the concurrence/dissent relies on State v. Webb, 440 N.W.2d 426 (Minn. 1989), and State
v. Race, 383 N.W.2d 656 (Minn. 1986). In Webb, we viewed the inconsistent statements
related to the ownership of a bedspread in a light most favorable to the conviction and
concluded that Webb owned the bedspread found near the victim’s body. 440 N.W.2d at
430–31, 431 n.2. In Race, we discussed the inconsistent statements about the number of
life rafts and then wrote, “[o]n conflicting evidence, the jury, whose task it was, concluded
there was only one life raft on board.” 383 N.W.2d at 659, 662. Consistent with the legal
principle that the jury is in a unique position to determine the credibility of the witnesses
and weigh the evidence before it, our analysis of the inconsistent statements in Webb and
Race protected the unique position of the jury to determine the credibility of the witnesses
and weigh the evidence before it.
In discussing the motive evidence, the concurrence/dissent relies on Bernhardt v.
State, 684 N.W.2d 465 (Minn. 2004), and State v. Hughes, 749 N.W.2d 307 (Minn.
2008). In Bernhardt, we said the dissent placed “undue emphasis on evidence of motive”
and that without something more, the motive evidence was insufficient. 684 N.W.2d at
479. Here, our analysis does not place undue emphasis on the motive evidence and instead
relies on several types of other evidence, including physical evidence and inconsistent
statement evidence. In Hughes, we concluded that the evidence that Hughes’s marriage
had deteriorated to the point of divorce, that Hughes did not believe in divorce, and that he
feared losing custody of his children “provide[d] a basis for the jury to infer [Hughes’s]
motive for killing his wife was to prevent her from leaving him and taking the children.”
749 N.W.2d at 314. Because our decision in Hughes did not set a minimum threshold for
motive evidence and because the State need not prove motive, see State v. Andersen, 784
N.W.2d 320, 331 (Minn. 2010), the concurrence/dissent’s assertion that the “motive
evidence is thinner” in Firkus’s case than it was in Hughes’s case does not alter our analysis.
17
In fact, the concurrence/dissent essentially includes all the evidence presented at
trial (including Firkus’s self-serving statements that there was an intruder) as the
circumstances proved. Ultimately, the concurrence/dissent is concerned that the court is
turning a “blind eye” to any rational inference other than guilt because we fail to evaluate
“the evidence as a whole.” But under the first step of a circumstantial evidence analysis,
we must “winnow down the evidence presented at trial by resolving all questions of fact in

35
this factual dispute during the second step. Instead, it observes that “[i]f a reasonable

fact-finder were to infer, for example, that Firkus was at the top of the stairs, then it was

also reasonable that he could hear someone fiddling with the doorknob.” (Emphasis

added.) In effect, the analysis of the concurrence/dissent leaves the factual dispute about

whether Firkus was in the bedroom or at the top of the stairs unresolved (thereby depriving

the jury of its unique position to determine the credibility of the witnesses and weigh the

evidence before it). In doing so, the concurrence/dissent fails to resolve all questions of

fact in favor of the jury’s verdict during the first step. The concurrence/dissent then

identifies and relies on the competing inferences that can be drawn depending on which of

the contradictory statements is true. In doing so, the concurrence/dissent neglects to view

the reasonable inferences that can be drawn from the circumstances proved as a whole,

which the second step requires, and instead draws inferences from discrete, isolated facts.

We decline to adopt the approaches employed by the concurrence and the

concurrence/dissent because they erroneously focus on whether a piece of evidence is

actually or theoretically inconsistent with the guilty verdict. Instead of protecting the jury’s

unique position to determine the credibility of witnesses and weigh the evidence before it,

those approaches deprive the jury of that position.

favor of the jury’s verdict,” which results in “a subset of facts that constitute ‘the
circumstances proved.’ ” Harris, 895 N.W.2d at 600 (quoting State v. Hawes, 801 N.W.2d
659, 670 (Minn. 2011)). This step protects the jury’s important and unique position to
determine the credibility of the witnesses and weigh the evidence before it. Our careful
application of this well-established law protects against wrongful convictions while
respecting the jury’s unique role in assessing witness credibility and weighing the evidence.

36
 * * *

In sum, we have applied our long-standing two-step test to determine whether the

circumstantial evidence proves that Firkus caused Heidi’s death with premeditation and

intent. In the first step, we winnowed down the evidence by resolving all questions of fact

in favor of the jury’s verdict to determine the circumstances proved. This protected the

jury’s unique position to assess the credibility of witnesses, weigh the evidence, and accept

some but not all parts of a witness’s testimony. We rejected an argument that our decisions

in Culver and Al-Naseer altered this principle. In the second step, we considered whether

the reasonable inferences drawn from the circumstances proved, when viewed as a whole

and not as discrete, isolated facts, were consistent with a hypothesis of guilt and

inconsistent with any rational hypothesis other than guilt. Ultimately, we concluded that

the only reasonable inference that can be drawn from the circumstances proved, when

viewed as a whole and not as discrete, isolated facts, is that Firkus caused Heidi’s death

with premeditation and intent.

II.

This case also raises the question of what standard the district court should apply to

motions for judgment of acquittal that are made and decided before the verdict is returned.

After the State rested its case, and again after all the evidence was submitted, Firkus moved

for a judgment of acquittal, which the district court denied before the verdict was returned.

In denying both motions, the district court applied the direct-evidence standard. Citing the

court of appeals’ decision in State v. Sam, 859 N.W.2d 825, 830 (Minn. App. 2015), Firkus

argues the district court should have applied the circumstantial-evidence standard that

37
appellate courts use when reviewing the sufficiency of circumstantial evidence after a jury

has returned a guilty verdict.

In Sam, the court of appeals held that district courts must apply the

circumstantial-evidence test to motions for judgment of acquittal when the State’s case

rests on circumstantial evidence. 859 N.W.2d at 831. But unlike Firkus’s motions for

judgment of acquittal, which were made and decided before the verdict was returned, the

motion in Sam was made after the jury returned a guilty verdict. Id. at 821. For the reasons

below, we conclude that this temporal distinction is material and that a district court should

apply the direct-evidence standard in considering all motions for a judgment of acquittal

that are made and decided before the verdict is returned.

Before the adoption of the Minnesota Rules of Criminal Procedure in 1975, the

mechanism to seek dismissal of a case during trial before submission to the jury was a

motion for a directed verdict, mirroring the existing Federal Rules of Criminal Procedure.

At that time, we described the standard for deciding a defendant’s motion for a directed

verdict as whether “there was evidence from which the jury could reasonably conclude that

the defendant participated in the crime.” State v. Klotter, 142 N.W.2d 568, 572

(Minn. 1966). This language also mirrored the federal standard for directed verdict

motions.

Upon the adoption of the Minnesota Rules of Criminal Procedure, motions for a

judgment of acquittal replaced motions for a directed verdict, although substantively they

are the same. Minn. R. Crim. P. 26.03, subd. 17(1) (1976) (“Motions for directed verdict

are abolished and motions for judgment of acquittal shall be used in their place.”). The

38
new criminal rule again mirrored the language of the federal rule, both in timing and in

substance. Cf. Fed. R. Crim. P. 29. Like the federal rule, a motion for a judgment of

acquittal may be made at different times. Minn. R. Crim. P. 26.03, subd. 18. First, it may

be made at the close of the prosecution’s case, in which case a district court must rule on

the motion and is prohibited from reserving its ruling. Id., subds. 18(1)(a), 18(2); State v.

Thomas, 891 N.W.2d 612, 616 (Minn. 2017). Second, the motion may also be made at the

close of the defendant’s case, in which case the district court “may reserve ruling on the

motion, submit the case to the jury, and rule before or after verdict.” Minn. R. Crim. P.

26.03, subd. 18(2). Finally, a motion for a judgment of acquittal may be made in a

post-verdict motion, filed within 15 days after the jury is discharged. Id., subd. 18(3)(a).

If the court grants a motion for a judgment of acquittal after a guilty verdict, the court must

make written findings stating the reasons. Id., subds. 18(2), 18(3)(c).

Under the rule, a motion for a judgment of acquittal must be granted “if the evidence

is insufficient to sustain a conviction.” Id., subd. 18(1)(a). Although the text of the rule

and our case law use the word “insufficient,” we have not previously instructed district

courts to apply the appellate standards for reviewing the sufficiency of the evidence after

a verdict has been returned in considering motions for a judgment of acquittal during trial.

Instead, we have more generally described that “[t]he test for granting a motion for a

directed verdict is whether the evidence is sufficient to present a fact question for the jury’s

determination, after viewing the evidence and all resulting inferences in favor of the state.”

State v. Slaughter, 691 N.W.2d 70, 74–75 (Minn. 2005). Just last year, we repeated this

standard in a case involving circumstantial evidence. Allwine, 994 N.W.2d at 537. We

39
stated that a “district court may only deny the motion when the evidence is sufficient to

present a factual question for the jury.” Id. (emphasis added). The corollary of this

directive is that a motion must be granted when no reasonable jury could find a defendant

guilty, even when viewing the evidence in the light most favorable to the verdict. We did

not say that the district court must use the two-step circumstantial-evidence test when

deciding a motion for a judgment of acquittal in a circumstantial-evidence case.

Having weighed the arguments of the parties and the amici, we conclude that in the

context of a motion for judgment of acquittal that is made and decided before a verdict is

returned, the standard proposed by Firkus, and adopted by the court of appeals in Sam, is

unworkable for two reasons: first, the logical impossibility of applying the test prior to a

verdict; and second, the mandatory time restraints in the district court.

As discussed previously, the first step of the circumstantial-evidence review test

requires a court to winnow down the evidence presented at trial by resolving all questions

of fact in favor of the jury’s verdict. We agree with amici Minnesota Board of Public

Defense and Minnesota Association of Criminal Defense Lawyers that “[u]ntil such a

verdict is returned, it is difficult if not impossible to measure whether the evidence would

support it.” Amici Br. 19. Firkus does not explain how a district court is expected to

winnow down the evidence by resolving all questions of fact in favor of the verdict before

there is a verdict.

Further, the proposed standard is particularly unworkable for motions made at the

close of the prosecution’s case when the district court must rule on the motion and may not

40
reserve its ruling. 18 The type of review we now conduct in these cases on appeal is simply

not possible without the benefit of (often thousands of pages of) trial transcript and

extensive briefing by the parties. Requiring a district court to conduct this analysis in the

midst of trial is impractical. Our recent decisions illustrate that reasonable jurists disagree

about what constitutes the circumstances proved and whether the evidence is enough to

sustain the conviction. 19 This case highlights the impracticality of expecting a district court

to undertake such an analysis during trial or before deliberations begin. 20 Instead, we

conclude that when asked to rule on a motion for a judgment of acquittal before a verdict

has been returned, a district court should view the evidence in the light most favorable to

the State and determine whether a jury could reasonably conclude that the defendant was

guilty—the general sufficiency-of-the-evidence standard, 21 not the heightened

18
When a defendant moves for a judgment of acquittal at the close of the State’s
evidence, the district court must hold the State to its burden. The district court must rule
and may not reserve its ruling until after the jury deliberates. Minn. R. Crim. P. 26.03
subd. 18(2). That is because, “[g]iven the presumption of innocence and the state’s burden
to prove the offense, a defendant has no obligation to present any evidence and should not
be put at risk of providing evidence that fills gaps in the state’s case.” Slaughter,
691 N.W.2d at 75. A defendant cannot be compelled to put on a case when the State has
not met its burden.
19
On appeal, the parties at times even disagree about what standard of review applies
in a challenge to the sufficiency of the evidence. E.g., State v. Horst, 880 N.W.2d 24, 39
(Minn. 2016); State v. Silvernail, 831 N.W.2d 594, 598 (Minn. 2013).
20
Although applying the standard to a post-verdict motion after briefing may be
reasonable, that issue is not before the court and need not be decided here, and attempting
to apply it earlier would be infeasible.
21
Likewise, the federal standard is “whether, after viewing the evidence in the light
most favorable to the prosecution, any rational trier of fact could have found the essential

41
circumstantial-evidence test—regardless of the type of evidence produced at trial. If, at

the close of the State’s case, no fact questions remain for the jury to decide—that is to say,

the defendant could not be found guilty even when viewing the evidence in the light most

favorable to the verdict, the motion must be granted.

This standard addresses the central concern of a motion for a judgment of acquittal,

which is that a judge has the power to prevent a case from reaching the jury when the

evidence is clearly insufficient, while not infringing on the jury’s critical role as fact-finder.

Ruling on a motion for a judgment of acquittal requires such a balance: precluding

deliberation on cases that do not merit consideration by a jury because of a clear lack of

evidence while allowing the jury to serve its role as fact-finder, a vital function in our

criminal justice system that provides transparency and legitimacy to a verdict, whether

guilty or not guilty. We believe the decision we announce today is the most workable

solution for this issue that is also consistent with the law. Given that the motions for

judgment of acquittal were made by Firkus and decided by the district court before a verdict

was returned, we conclude that the district court did not err in applying the sufficiency-of-

the-evidence test for direct evidence when it denied the motions.

CONCLUSION

For the foregoing reasons, we affirm the decision of the district court.

Affirmed.

elements of the crime beyond a reasonable doubt.” Jackson v. Virginia, 443 U.S. 307, 319
(1979) (emphasis added).

42
 C O N C U R R E N C E

THISSEN, Justice (concurring).

I concur with the court’s decision. I agree with the court’s analysis and conclusion

in Part II of the opinion. I write separately because I disagree with the approach to

resolving circumstantial-evidence cases that the court adopts. In my view, the court’s

approach is insufficient to ensure consistent application of the circumstantial-evidence test

to different defendants. In addition, the court’s approach collapses the circumstantial-

evidence test into the direct-evidence test in all but name. Accordingly, under the precedent

the court sets in this case, continuing to require litigants and courts to go through additional

hoops of applying the circumstantial-evidence test seems hard to justify. See State v.

Colgrove, 996 N.W.2d 145, 158 n.4 (Minn. 2023) (Thissen, J., dissenting) (explaining that

better jury instructions would eliminate the need for the circumstantial-evidence test).

Based on my review of our precedent, which the court acknowledges has produced

inconsistent results, and to preserve the purpose of the circumstantial-evidence test as a

“heightened standard” intended to protect Minnesotans from facing loss of liberty on

anything less than proof beyond a reasonable doubt, I conclude that a court applying the

circumstantial-evidence test should follow several clear steps. These steps are

straightforward to apply and find support in the language we have used in our cases, the

outcomes in previous circumstantial-evidence cases, and the reason we adopted the

circumstantial-evidence test as a framework for appellate review.

First, the court should determine whether there is direct evidence for each element

of the crime charged. If there is, the court should apply the direct-evidence test.

C-1
 If, however, there is no direct evidence for one or more elements of the crime

charged, the court should apply the circumstantial-evidence test. In applying that test, the

court should first determine the evidence it may consider in assessing an insufficient

evidence argument. This evidence—which we call “the circumstances proved”—is

necessarily different from the evidence the court considers in a direct evidence case.

Circumstances proved include the following: (1) evidence that is consistent with the jury

verdict; (2) uncontradicted evidence from which a jury could have drawn both an inference

consistent with the guilty verdict (or a fact necessary for a determination of guilt) and an

inference consistent with a not guilty verdict; and (3) evidence that supports an inference

inconsistent with a guilty verdict except where (a) the evidence is specifically contradicted

by direct evidence that supports an inference consistent with the guilty verdict, or (b) the

court has confidence the jury was in a position to make a credibility assessment of the

conflicting evidence.

Once the court identifies the circumstances proved, it should consider whether the

circumstances proved, when viewed as a whole, support a reasonable inference that the

defendant is guilty and are inconsistent with any other reasonable inference except that of

guilt. This inquiry is known as the second step of the circumstantial-evidence test, and it

is the critical step for holding the State to its obligation to prove a defendant’s guilt beyond

a reasonable doubt.

Applying the circumstantial-evidence test in this case, I conclude that Firkus’s

conviction must stand. The circumstances proved as a whole support no reasonable

inference except that of guilt.

C-2
 A.

To convict a defendant of first-degree premeditated murder, the State must prove

that the defendant “cause[d] the death of a human being with premeditation and with intent

to effect the death of the person or of another.” Minn. Stat. § 609.185(a)(1). To convict a

defendant of second-degree intentional murder, the State must prove that the defendant

“cause[d] the death of a human being with intent to effect the death of that person or

another, but without premeditation.” Minn. Stat. § 609.19, subd. 1(1). The State must

prove each element of an offense beyond a reasonable doubt. State v. Peterson,

673 N.W.2d 482, 486 (Minn. 2004). The question before us is whether the State produced

sufficient evidence to prove beyond a reasonable doubt that Firkus intended to effect

Heidi’s death and did so with premeditation. In assessing the sufficiency of the evidence—

the existence of facts necessary for conviction—to prove the elements of a crime we apply

either the direct-evidence test or the circumstantial-evidence test to each element,

depending on the nature of the evidence the State produced to prove that element. State v.

Peterson, 910 N.W.2d 1, 6–7 (Minn. 2018).

B.

In cases where the State offered direct evidence—evidence based on personal

knowledge or observation—of the facts necessary for a finding of guilt (i.e., an element of

the crime), we view all the evidence in a light most favorable to the guilty verdict and

assume the fact-finder disbelieved any evidence conflicting with that verdict. State v.

Jones, 4 N.W.3d 495, 500 (Minn. 2024); State v. Ortega, 813 N.W.2d 86, 100 (Minn.

2012). “We will not overturn the verdict if the jury, acting with regard for the presumption

C-3
of innocence and the State’s burden of proof beyond a reasonable doubt, could [have]

reasonably conclude[d] that the defendant was guilty.” Jones, 4 N.W.3d at 502; see also

State v. Huss, 506 N.W.2d 290, 292 (Minn. 1993) (stating that our sufficiency of the

evidence inquiry where the State offered direct evidence of the crime “is limited to whether,

given the evidence as it was presented in court, a jury could have reasonably concluded

that appellant was guilty beyond a reasonable doubt,” and reversing the defendant’s

conviction because the State did not satisfy that standard).

Direct evidence provides a concrete connection to the facts necessary for a

determination of guilt that explains the fact-finder’s verdict. Where a witness provides

direct evidence, their personal knowledge or observations relate directly to the facts

necessary to determine guilt. Accordingly, in those cases we can be confident in assuming

that a fact-finder that found a defendant guilty credited and based its determination on the

direct evidence of the facts necessary to determine guilt and rejected contrary evidence.

In contrast, where the State offered no direct evidence of facts necessary for a

determination of guilt, the fact-finder must have relied on circumstantial evidence. That

is, the jury must have made a leap from facts shown by direct evidence and drawn

inferences indirectly from those facts to determine whether a fact necessary for guilt

existed. In short, the connection between any particular piece of evidence and the facts

necessary for a determination of guilt is more attenuated in circumstantial-evidence cases.

In such cases, it is not always clear which facts the fact-finder credited and which it did

not. The fact-finder necessarily credited some combination of evidence presented to arrive

C-4
at its verdict, but we are less certain about which specific evidence it credited, and so do

not automatically reject all facts inconsistent with the guilty verdict.

This is important because our historical concern in circumstantial evidence cases,

where determining a fact necessary for guilt requires an inferential leap, is that the gap

between direct evidence and the extrapolations juries make in finding a fact necessary for

guilt creates greater space for juries to neglect due deliberation and instead base their

decisions on prejudice, partiality, or hastiness. State v. Tscheu, 758 N.W.2d 849, 869–70

(Minn. 2008) (Meyer, J., concurring) (outlining the history of the circumstantial-evidence

test both as a required jury instruction and as a standard of appellate review); see State v.

Harris, 895 N.W.2d 592, 599 n.4 (Minn. 2017) (stating that the reason for heightened

review in circumstantial-evidence cases is concern over the inferences the jury makes from

facts proved); Colgrove, 996 N.W.2d at 158 (Thissen, J., dissenting).

Consequently, when the State relied solely on circumstantial evidence to establish

facts necessary to find a defendant guilty, we apply a “heightened” two-step test to assess

whether the State proved the defendant guilty beyond a reasonable doubt. Jones, 4 N.W.3d

at 500; State v. Isaac, 9 N.W.3d 812, 815 (Minn. 2024). In doing so, we do not defer to

inferences the fact-finder drew but examine the reasonableness of inferences supporting

the verdict ourselves. Isaac, 9 N.W.3d at 818. Because we cannot be sure which evidence

the jury believed or disbelieved in cases built on circumstantial evidence, the first step of

the circumstantial-evidence test requires us to identify the circumstances proved. Id. at

815. We do so by considering all the evidence presented and winnowing it down,

C-5
excluding evidence we are sufficiently confident the jury did not believe when it

determined that the State proved facts necessary for conviction beyond a reasonable doubt. 1

The State argues that in applying the first step of the circumstantial-evidence test

we should assume the jury found the State’s evidence credible and disregard any evidence

that could support a conclusion other than guilt. The court adopts this approach. By this

method, we include in the circumstances proved only evidence that indisputably supports

1
In Harris, we faced a frontal challenge to the propriety of the circumstantial-
evidence test on the ground that it undermined our traditional deference to jury
determinations. 895 N.W. at 597. In response, we confusingly pledged to reincorporate
appellate respect for jury credibility determinations, id. at 600, into a test born out of
incredulity about jury determinations, where the only evidence of a fact necessary to guilt
is a guess. I use the word confusingly because, by inserting guardrails to protect jury
credibility determinations in step one of the circumstantial-evidence test, we essentially
render meaningless the lack of deference we purportedly give to jury inferences in step two
of the test. In doing so, we distanced ourselves, obscured, and today ultimately undermine
the original concern that motivated us to adopt the circumstantial-evidence test—our
concern that juries may be less deliberate in applying the beyond-a-reasonable-doubt
standard in circumstantial-evidence cases. Our struggles with, and contortions in applying,
the circumstantial-evidence test in the intervening years are the result of that underlying
tension.
I leave for another day the question of whether our decision to adopt a
circumstantial-evidence test a few decades ago was correct. I do, however, make two
observations. First, the idea that appellate courts act as so-called 13th jurors, see infra
note 5, is implicit in the circumstantial-evidence test, and, in light of that foundational
decision, moral panic about that reality seems overstated. Second, our concern that juries
may not rigorously apply the beyond-a-reasonable-doubt burden of proof in circumstantial-
evidence cases has a long history. Minnesota courts historically instructed juries that “to
authorize a conviction, the circumstances should not only be consistent with the prisoner’s
guilt, but they must be inconsistent with any other rational conclusion.” State v. Johnson, 35 N.W. 373, 376 (Minn. 1887). In State v. Turnipseed, 297 N.W.2d 308, 313 (Minn.
1980), we held that this jury instruction language is not constitutionally required, but we
have never said this reminder to the jury is improper. Our jurisprudence of differing
appellate tests for direct-evidence and circumstantial-evidence cases—and our affirmative
choice to step into the role of 13th juror in a limited way—is in large part a response to our
decision to let this historical jury instruction fade away.

C-6
a guilty verdict and exclude any evidence that possibly supports a determination of not

guilty. I reject such a constricted reading of the first step of the test because it collapses

the two-step analysis and makes the circumstantial-evidence test indistinguishable from

the direct-evidence test. With this interpretation, the circumstantial-evidence test no longer

serves any real purpose.

We have not always been precise in the language we use to describe the process of

winnowing down the evidence to the circumstances proved. But based on the outcomes of

several of our decisions discussed below and our reasons for adopting the circumstantial

evidence test—and, importantly, to preserve the possibility of independently assessing

inferences in the second critical step of the circumstantial-evidence test—we cannot

assume that the fact-finder rejected all testimony conflicting with the verdict when we

winnow the evidence in the first step. Rather, we must consider uncontradicted evidence

in the circumstances proved when a fact-finder could have drawn both an inference

consistent with the guilty verdict (or a fact necessary for a determination of guilt) and an

inference consistent with a not guilty verdict from the same piece of evidence. Again,

when a determination of guilt was based on circumstantial evidence, we do not defer to the

fact-finder’s inferences. Id. at 818. Indeed, we have recently said that evidence supporting

an inference that a defendant is not guilty is not inconsistent with a guilty verdict if the jury

could both have found the evidence to be true and still determined that a fact necessary for

the determination of guilt existed. See Colgrove, 996 N.W.2d at 152. In such cases, the

evidence should not be excluded from the circumstances proved.

C-7
 In addition to the situation just described—where a single piece of evidence could

have supported alternative inferences—cases often involve multiple conflicting pieces of

evidence. In that context, a fact-finder may have drawn an inference consistent with a fact

necessary for a determination of guilt from some evidence and also drawn an inference

inconsistent with a fact necessary for a determination of guilt from competing evidence.

In such circumstances, we should exclude from the circumstances proved evidence that

cannot be true in light of the verdict only when one of two conditions exist: (1) the

evidence is specifically contradicted by direct evidence that supports an inference

consistent with the guilty verdict, or (2) we have confidence the jury was in a position to

make a credibility assessment about the conflicting evidence.

First, we generally exclude circumstantial evidence that supports an inference

inconsistent with a guilty verdict when there is other specifically contradictory direct

evidence that supports an inference consistent with the guilty verdict. State v. Stein,

776 N.W.2d 709 (Minn. 2010) (plurality opinion), illustrates this principle. In State v.

Stein, there was testimony and video evidence that the defendant was wearing a black

shirt—the same color as the perpetrator’s shirt—on the night of the crimes. Id. at 718.

DNA taken from the perpetrator’s black shirt, recovered from a crime scene, matched the

defendant. Id. at 712, 718. The defendant, however, testified that he was wearing a white

shirt that night. Id. at 718. In viewing the conflicting evidence in the light most favorable

to the State, we concluded that it was a circumstance proved that the defendant was wearing

a black shirt and not a circumstance proved that the defendant was wearing a white shirt.

Id. We accepted the fact that the defendant wore a black shirt because there was testimony

C-8
and other evidence that specifically contradicted the defendant’s testimony that he wore a

white shirt. Id.

This principle is analogous to the assumption we make in applying the direct-

evidence test—assuming the fact-finder disbelieved any evidence conflicting with the

guilty verdict. But in a circumstantial-evidence case, we should apply the assumption more

narrowly. We should exclude from the circumstances proved only evidence that

contradicts specific direct evidence supporting an inference of guilt. We should not

exclude all evidence supporting inferences inconsistent with a finding of guilt. 2

The reason for this approach is that, where there is direct evidence of a fact

necessary for a determination of guilt, that fact and guilt are coextensive. When a fact-

finder’s determination that the State proved a fact necessary for a determination of guilt

2
By way of example, consider a hypothetical prosecution for first-degree aggravated
armed robbery with a dangerous weapon. See Minn. Stat. § 609.245, subd. 1. One element
of the crime is that the defendant had a dangerous weapon. Id. The State offers the
testimony of Witness A, who states that “I did not see a gun, but I heard gunshots coming
from where the defendant was standing and saw the defendant positioned in a way that
looked like he was firing the shots.” The defendant counters with a police report that says
there was no weapon found on the defendant after his arrest on the scene, there was no
evidence of gunshot residue on his hands, and no gunshots were heard. The jury returns a
guilty verdict.
In an appeal challenging the sufficiency of the evidence, the circumstantial-evidence
test applies because there is no direct evidence that the defendant had a dangerous weapon.
Witness A’s testimony—which is consistent with the verdict—would be a circumstance
proved. The portion of the police report that stated no gunshots were heard would not be
included in the circumstances proved because Witness A’s testimony directly contradicts
that evidence. But the other parts of the police report—which do not support the verdict—
should also be included as a circumstance proved. Those portions of the report are not
necessarily inconsistent with Witness A’s testimony because Witness A provided no direct
evidence that the police found a weapon on the defendant after his arrest on the scene or
gunshot residue on the defendant’s hands.

C-9
was based on an inference the fact-finder drew from the evidence—i.e., circumstantial

evidence—that evidence is not necessarily or solely coextensive with guilt. While we

know the jury believed some combination of the evidence to support its guilty verdict, we

are less certain about which specific evidence the jury believed. If we were to exclude all

evidence from which a jury could have drawn inferences inconsistent with guilt simply

because there was one piece of direct evidence from which it could have drawn an inference

consistent with guilt, we would collapse the first and second steps of the circumstantial-

evidence test into a single inquiry. It would render the circumstantial-evidence test

indistinct from the test we apply to cases proved by direct evidence, undermining our long-

standing commitment to “heightened” review in cases where convictions are based solely

on circumstantial evidence. That is precisely the position the court adopts today.

Even when applying the circumstantial-evidence test, we remain concerned with

respecting the fact-finder’s “unique position to determine the credibility of the witnesses,”

including the freedom “to accept part and reject part of a witness’s testimony.” Harris,

895 N.W.2d at 600 (citations omitted) (internal quotation marks omitted) (discussing the

need to preserve a jury’s credibility assessments of witness testimony and citing

circumstantial-evidence cases for this principle); see also Colgrove, 996 N.W.2d at 158 n.3

(Thissen, J., dissenting) (discussing the application of deference to credibility

determinations in circumstantial-evidence cases). When we have confidence a jury was in

C-10
a position to make credibility determinations, we have more certainty that the jury did not

credit evidence inconsistent with the verdict. 3

In a few circumstantial-evidence cases, we have suggested that our deference to jury

credibility determinations means that all evidence inconsistent with the verdict should be

excluded from the circumstances proved. For instance, in State v. Culver, 941 N.W.2d 134

(Minn. 2020), which the State relies on to support excluding certain non-testimonial

statements the defendant made from circumstances proved, we determined that “the jury

was not compelled to believe the statements that [the defendant] made merely because they

were uncontradicted.” Id. at 144. In doing so, we cited a civil case, Costello v. Johnson,

121 N.W.2d 70, 76 (Minn. 1963), in which we observed that a jury does not have to accept

“at face value” testimony that “was not controverted by . . . testimony on the part of [the

other party].” Id. (holding that the jury could disregard uncontroverted expert testimony).

We reasoned:

[The] weight [of the evidence] and the credibility of the witnesses is usually
for the trier of fact to determine, and it is not compelled to believe any witness
merely because his testimony is uncontradicted. This is particularly true

3
The court understands the concept of “credibility determination” in our
circumstantial-evidence test case law to mean that every piece of evidence that does not
support the guilty verdict must be excluded, regardless of whether it is contradicted or
whether it supports an alternative inference to guilt. As discussed in the body of this
concurrence, however, we have not invariably excluded all evidence that is inconsistent
with a jury verdict. Indeed, if that is what we mean by deference to credibility
determinations—excluding unrefuted evidence even when a jury was not in a position to
assess the credibility of the witness providing the testimony—then we are collapsing the
circumstantial-evidence test into the direct-evidence test and we should have adopted the
dissent’s position in State v. Harris nine years ago for a unitary standard. See Harris,
895 N.W.2d at 605–11 (Lillehaug, J., dissenting).

C-11
 where there are other facts in evidence which refute or modify the
uncontroverted testimony or where testimony is obviously untrue.

Id. (emphasis added).

I believe that, when applying the first step of the circumstantial-evidence test, we

should not use the Costello principle to exclude evidence like the uncontradicted non-

testimonial statements in Culver from circumstances proved. 4 Such an application of the

Costello rule is too blunt an instrument in the first step of the circumstantial-evidence test.

As discussed above, the question in this step focuses on determining whether we are

confident that the jury must have disbelieved a particular fact when it reached its guilty

verdict. When a fact necessary for a determination of guilt requires a jury to draw an

inference from among several pieces of evidence, we do not know which evidence the jury

believed and relied upon in making its determination. Specifically, we do not know if the

jury believed a witness’s testimony about some facts but found evidence about other facts

in the case more compelling. Thus, whether the jury found a witness credible or not is

upstream of the Costello rule that a jury has the choice to disbelieve evidence even if

uncontradicted.

Accordingly, in addition to excluding from the circumstances proved any evidence

that cannot be true if specifically contradictory direct evidence consistent with guilt is true,

we should also exclude evidence that supports an inference that the defendant is not guilty

when other evidence specifically contradicts that evidence (even if the other evidence is

4
I discuss our decision in Culver in more detail in Part D.

C-12
not direct evidence) or we have confidence that the fact-finder was in a position to make a

credibility assessment of the conflicting evidence. 5

The State’s position that we should assume the jury found the State’s evidence

credible and exclude from the circumstances proved all evidence supporting a conclusion

other than guilt is also inconsistent with our case law. For instance, had we applied this

rule in our recent decision in State v. Isaac, 9 N.W.3d 812, it is likely that we would have

reached a different outcome. In Isaac, an aiding-and-abetting first-degree murder case, the

defendant started the evening with the shooter in the Fargo-Moorhead area, then left the

5
The court asserts that, in adopting this approach, I would allow appellate courts to
act as a “13th juror” and reweigh the evidence. Frankly, I can see how the court reaches
that conclusion. The court starts from the perspective that, in step one, we may not consider
any facts inconsistent with the jury verdict—even facts the State entirely failed to refute
that are relevant to proving an element or elements for which no direct evidence exists.
From that viewpoint, any recognition of a fact inconsistent with the verdict is an appellate
court stepping into the jurors’ shoes. Indeed, that is how we have traditionally applied the
direct-evidence test, and the court today is making clear that, practically speaking, the
circumstantial-evidence test is the direct-evidence test in everything but name.
From my perspective—articulated in this concurrence and with support in both our
case law and the very reason for adopting the circumstantial-evidence test—I am not
authorizing appellate courts to reweigh evidence in step one. I am merely saying that when
a court undertakes the step-two analysis considering the reasonableness of both the
inferences the jury drew and alternative inferences, it should be allowed to consider
evidence that the State left unrefuted and evidence that the jury could have accepted as true
and still reached a guilty verdict. I would only exclude that evidence from the step-two
analysis if the record suggests that the jury considered the evidence and rejected it—either
due to directly contradictory evidence or some other showing that the jury was in a position
to reject the evidence because it found the evidence was not credible.
In some sense, by making assumptions about the jury’s decisions when it did not
hear competing direct evidence on a fact necessary for a guilty verdict, the court itself is
stepping into the jury’s shoes. It is also worth noting, on the subject of the “13th juror,”
that in step two of the circumstantial-evidence test, as both the court and I understand it,
we take an explicitly independent and non-deferential look at the inferences that a jury
necessarily made in reaching its verdict.

C-13
shooter and drove toward St. Cloud. Id. at 816. After driving toward St. Cloud for about

an hour, Isaac received a call from the shooter, who had just murdered the victim in

Moorhead. Id. Isaac turned around to meet up with the shooter. Id. We focused heavily

on this fact in reversing Isaac’s conviction, noting that if he had known about the shooter’s

plan beforehand it would not have made sense for Isaac to drive so far in the opposite

direction only to drive back to meet up with the shooter again. See id. at 818. We

concluded that this fact, combined with the other circumstances proved, supported a

reasonable inference that Isaac did not know about the shooter’s plan to murder the victim

until the shooter called him after the murder. Id.

Under the State’s theory, because the facts that Isaac drove toward St. Cloud and

turned around after receiving the shooter’s call would not tend to support his guilt we would

not have considered them as circumstances proved and thus not used them to evaluate

whether there was a rational hypothesis that Isaac was not guilty at step two. Without

considering these as circumstances proved, we likely would have upheld Isaac’s conviction

even with the scant evidence supporting conviction.

Likewise, in State v. Al-Naseer, 788 N.W.2d 469 (Minn. 2010), we overturned

Al-Naseer’s conviction for criminal vehicular homicide based on leaving the scene of a

collision under the circumstantial-evidence test. We considered all the evidence in that

case as circumstances proved. Id. at 475 n.2 (“[I]n this case the evidence is not in

conflict. . . . Instead, it is the inferences that can be made from the evidence and the

circumstances proved that stand in conflict . . . .” (emphasis added)). But under the State’s

version of the circumstantial-evidence test, we would have excluded many pieces of

C-14
evidence we considered, including the sole witness’s testimony that he never saw

Al-Naseer’s vehicle slow down after the collision, police testimony that the evidence was

consistent with Al-Naseer having been asleep at the time of the collision, and Al-Naseer’s

statement to police that he knew he hit something but did not know what. See id. at 477–78.

Despite none of this evidence being in direct conflict with the verdict or other evidence

necessary to establish guilt, the State’s position would have required us to disregard it all

in analyzing whether there was a reasonable inference that Al-Naseer did not know he had

hit a person or vehicle. Like Isaac, this would almost certainly have changed our decision. 6

Once we have winnowed all the evidence presented at trial down to the evidence

that constitutes the circumstances proved, we turn to the second step of the circumstantial-

evidence test: We assess whether the circumstances proved, when viewed as a whole,

support a reasonable inference that the defendant is guilty and are inconsistent with any

other reasonable inference except that of guilt. Harris, 895 N.W.2d at 600–01. Including

evidence in the circumstances proved by itself means nothing when assessing the probative

value of any particular piece of evidence at this second step. We will reverse if the

6
I recognize that, in rare cases, we have reversed convictions under the circumstantial
evidence test where we excluded all evidence inconsistent with a guilty verdict and only
considered evidence consistent with the guilty verdict. See, e.g., State v. Segura, 2 N.W.3d
142 (Minn. 2024). Critical to our decision today, we would have reached the same decision
in one such case, State v. Segura, under the direct-evidence test precisely because, based
solely on evidence consistent with the jury verdict (the only evidence on which we can rely
in direct-evidence cases), we concluded that State did not prove Segura’s guilt beyond a
reasonable doubt. See Huss, 506 N.W.2d at 292.

C-15
circumstances proved do not support a reasonable inference of guilt 7 or support a

reasonable inference that is inconsistent with guilt. Id. at 600. I discuss the second step of

the circumstantial-evidence test as applied to the circumstances proved in this case in

Part E.

It may be true that under the court’s decision today the circumstantial evidence test

gives appellate courts the cover of a flexible off-ramp when we are deeply disturbed by a

jury verdict. But I struggle to understand how such a test, which would seem to flex based

on our level of disturbance, provides consistent results for criminal defendants or is a fair

system that supports court legitimacy.

C.

With that background in place, I return to the question before us: Did the State

produce sufficient evidence to prove beyond a reasonable doubt that Firkus, with

premeditation, intended to effect Heidi’s death?

I start by describing the circumstances proved in this case and then, in Part D, I

explain why I include certain evidence in the circumstance proved that the State urges us

to exclude.

Background

Nicholas and Heidi Firkus met through their church’s high school youth group.

Friends admired Heidi and Firkus for their devotion to Christianity and their marriage.

They were close with other young couples with whom they shared religious values. The

7
There is no dispute in this case that the circumstances proved support a reasonable
inference consistent with the guilty verdict.

C-16
men admired how Firkus treated Heidi. Firkus was seen as a “spiritual leader” within the

group.

In 2007, the Firkuses purchased a home in St. Paul. Soon thereafter, the housing

bubble burst and recession followed. Almost immediately, the Firkuses began missing

payments on their credit card, mortgages, and other debts. For the next two-and-a-half

years, the Firkuses fell deeper into debt, failing to pay off their bills, overdrawing their

bank accounts, and missing mortgage payments. The couple made no payments on their

primary mortgage after September 22, 2008. The bank foreclosed on the Firkuses’ house

and sold it at a sheriff’s sale on June 4, 2009. The bank offered the Firkuses cash payouts

if they agreed to vacate their home before eviction (commonly known as “cash-for-keys”),

but they did not respond to these offers. At a March 8, 2010, eviction hearing, Firkus

agreed to vacate the home by March 22, 2010; Heidi was not present. This date was

subsequently pushed back to April 9, 2010, and then again to April 26, 2010.

At trial, Firkus asserted that Heidi was fully aware of the foreclosure and financial

issues, despite significant evidence to the contrary. Although Heidi worked only a few

blocks away at the time, Firkus attended the March 2010 eviction hearing without her. The

cash-for-keys offers to which the Firkuses never responded would have required Heidi’s

signature. At the time of Heidi’s murder—the day before the Firkuses were to vacate the

home—none of their belongings were packed in boxes or ready to be moved out. Rather,

on that day, Heidi and her friends had plans to get pedicures. The day before her murder,

Heidi and a friend went shopping at the Mall of America and Heidi spent over $100.

Although Heidi had requested time off from work to help her parents move in June 2010,

C-17
she had not requested time off on April 26—the day the Firkuses were to vacate their own

home. The Firkuses had not made any concrete plans for where they would go after moving

out.

Text messages between Firkus and Heidi show that Heidi deferred to Firkus on

financial matters. Heidi told several people that she and Firkus were having issues with

identity fraud, but no activity on their accounts indicated such fraud. Firkus repeatedly

told Heidi that he was corresponding with their bank about identity fraud issues, but there

was no record of communications with the bank on that matter. In 2009, Heidi spoke to

her mother about the couple’s finances and said that “Nick told her they were fine.” None

of Heidi’s friends or family testified that she told them about the foreclosure or eviction.

Heidi did not cash a refund check for the couple’s home insurance policy sent after the

home was sold. Even though the Firkuses were in deep financial trouble, Heidi declined

financial help her mother offered in the summer of 2009. When Heidi’s father called Heidi

about missed payments on her car, for which he had cosigned, Heidi told him she was

unaware of missed payments. Although Firkus did not testify at his trial, he described the

foreclosure and financial issues to police as a “private struggle” for the couple.

Heidi did not feel completely comfortable in her home. In 2009, police shot a

burglar in the neighborhood. A friend’s car had been vandalized while visiting the

Firkuses. One night, the St. Paul police tried to serve a warrant at the Firkus house in the

middle of the night before realizing they were at the wrong house. Heidi told others that

she was afraid to be home alone. In a statement to police, Firkus stated that he normally

set the deadbolt lock on the front door.

C-18
 Firkus owned a shotgun that he used for hunting. He bought this shotgun from a

friend several years before Heidi’s murder. Firkus kept the shotgun in the basement of the

house until about a month before Heidi’s murder, when he moved the shotgun from the

basement to the couple’s upstairs bedroom. The basement was damp and unfinished, and

Firkus said he was trying to prevent the gun from rusting. Although there was no evidence

of rust on the gun, there was pitting, which is consistent with the gun having been recently

cleaned.

Day of the Murder

At 6:32 a.m. on April 25, 2010, Heidi called 911 from her cell phone. The transcript

of the 911 call with Heidi is as follows:

911 Operator 1: State Patrol, 911.

Heidi: Someone’s trying to break into my house.

911 Operator 1: What city are you in, St. Paul?

Heidi: I’m in St. Paul.

911 Operator 1: Hang on just a second. (dialing) What address are you at?

Heidi: 1794

911 Operator 1: 1794 what?

911 Operator 2: 911, where is your emergency?

Heidi: Minne-

911 Operator 1: Stay with-

Heidi: 17, 1794 Minnehaha Avenue.

C-19
 911 Operator 1: Minnehaha, someone’s trying, west?

Heidi: (screams)

The scream occurred about 35 seconds into the call, and at that point Heidi’s phone

disconnected.

Thirty-eight seconds after Heidi hung up with 911, Firkus called two recently dialed

numbers on Heidi’s phone. Sixty-five seconds after Heidi hung up, Firkus called 911. He

told the operator that someone had just broken into his house and shot him and Heidi.

Firkus remained on the phone with 911 until police arrived at 6:38 a.m.

Because there had been a shooting, police treated their response as a high-risk entry;

they kicked open the Firkuses’ front door and four officers rushed into the house. A rug

was wedged against the base of the front door and a pair of shoes in the entryway was

knocked out of place. Police described a table in the entryway as “wobbly.” Items on the

table were undisturbed when police arrived at the scene, but fell off when they touched the

table. A double-barreled shotgun and a pair of blue jeans were on the entryway floor.

Officers found Heidi lying on her back in the kitchen with Firkus kneeling beside her.

Heidi was pronounced dead at the scene.

Firkus had a shotgun wound to his left thigh. He was very emotional. St. Paul

Police officers interviewed Firkus at the hospital while he was receiving treatment for his

wound. Firkus told the officers that he had gotten up to get water from the bathroom and

heard somebody fiddling with the front door. He then told Heidi to call police while he

grabbed his shotgun. Firkus stated that, at first, he and Heidi “called the police from the

closet in case we needed to hide in the closet” but then they decided to head to the garage

C-20
to get out of the house. Although Firkus had two cellphones on his nightstand, he grabbed

only his shotgun and a pair of blue jeans before leaving the room. Firkus later said he

grabbed his jeans because his wallet was in one of the pockets. He said Heidi also grabbed

her wallet before heading out of the bedroom. Firkus followed Heidi down the stairs in an

attempt to get to the garage and out of the house. He reported that when they reached the

entryway, an intruder opened the door and a struggle ensued. The shotgun went off twice

during the struggle, the first shot striking Heidi in the back as she ran through the kitchen

toward the garage, and the second striking Firkus in his left upper thigh. The intruder then

fled.

Firkus participated in a much longer, recorded, interview later that morning at the

police station. This was the last interview Firkus gave to police before he retained counsel.

There were some discrepancies between this statement and Firkus’s initial account at the

hospital. In his second statement, Firkus told police that he got up around 6 a.m. to get

water from the bathroom, went back to sleep for 10 or 15 minutes, and then woke up

because he heard the screen door open. He said he was not concerned at first because the

screen door would sometimes pop open if it was windy, but then he started hearing fiddling

with the front doorknob. Firkus said he grabbed the shotgun from the closet before waking

Heidi. He stated he told Heidi that they had to get to the garage immediately and did not

mention calling the police from the closet.

Firkus described the struggle more extensively during the second interview. He told

police that the shotgun was initially in the three o’clock position across his chest, pointing

toward the kitchen. Firkus stated that after the first shot the intruder pulled down on the

C-21
barrel with his right hand so the gun moved clockwise until the barrel was in the eight

o’clock position, pointing downward toward the front door. According to Firkus, the

shotgun then went off a second time, hitting Firkus in the thigh. Firkus told officers he

was unsure whose finger was on the trigger when he was shot. Firkus recounted that he

dropped the shotgun and the intruder fled. Firkus said he then crawled to Heidi in the

kitchen and rolled her over to try to help her. He found her phone and called 911.

Firkus gave three different descriptions of the intruder. In his 911 call, Firkus told

the operator that the intruder was wearing a hooded sweatshirt and Firkus did not know his

race. When initially asked, Firkus said he could not be certain whether there had been one

or multiple intruders. In his first police interview, at the hospital, Firkus told the officers

that he thought the intruder “might have been black but [he wasn’t] 100% certain.” He

said the intruder was wearing a dark, hooded sweatshirt; gloves; and possibly sunglasses.

He could not recall if the intruder had facial hair. Firkus also said the intruder was a lot

bigger than he was, over six feet tall. In his second interview, at the station, Firkus repeated

this description and said that the hood had been drawn up pretty tight around the intruder’s

face. He also said the intruder was “black, with lighter skin;” around 6’1” or 6’2”; and 215

to 220 pounds.

Immediately after securing the scene on April 25, 2010, police canvassed the

neighborhood for possible witnesses. Within minutes of the shooting, they spoke to B.O.,

who was cat-sitting at his girlfriend’s house immediately next door to the Firkuses. B.O.

stated that around the time of the incident he was in a bedroom with an open window facing

the Firkus household and heard two shots. At the time of trial, a decade later, B.O.’s

C-22
memory was not completely clear on what he heard besides those two shots. On the day

of the murder, B.O. told St. Paul police and a reporter from the Pioneer Press that he had

heard two shots, followed by a male voice saying, “you shot me” or “you shot her.” He

also told the reporter that he heard the male voice say “stop, please.” About a week-and-

a-half to two weeks later, B.O. repeated most of the same story to a defense investigator,

except he told the investigator that he heard at least one voice between shots. Years later,

B.O. stated that he might have heard a woman’s voice. He also testified that, after listening

to Firkus’s 911 call, he believed what he heard could have been consistent with what Firkus

said on the call.

B.O. told police that he had looked out of his window after he heard the two shots

and did not see anyone leaving the Firkus home. At trial, he admitted that if someone had

run right out of the house and turned right into the alley he would not have been able to see

them. None of the Firkuses’ other neighbors testified that they saw anyone or anything

unusual on the street that morning before police arrived.

The day after Heidi’s murder, Firkus told Heidi’s parents that law enforcement

would probably never catch the intruder. He also revealed the foreclosure and financial

troubles during this conversation, and worried that police would view them as a motive for

him to murder Heidi. Firkus later signed the proceeds of Heidi’s life insurance policy over

to her parents.

Police Investigation

Officers did not observe many common signs of forced entry or struggle at the

Firkus house. They noted marks on the door frame by the knob lock. Sergeant Shay

C-23
Shackle testified that these marks were likely old and painted over, though he conceded it

was “possible but not probable” that the marks could be fresh compression marks

consistent with someone using a flathead screwdriver to open the door. Officers searched

the Firkuses’ house, garage, and vehicles for screwdrivers that could have made the marks.

They collected and tested three screwdrivers; Minnesota Bureau of Criminal Apprehension

(BCA) analysts determined that two could not have made the marks on the door frame,

while tests on the third were inconclusive.

Officers preserved portions of the door and door frame as evidence, and a locksmith

examined them and testified at trial more than 10 years later. According to the locksmith,

the deadbolt and doorknob locks were in perfect working order, there was no damage to

the doorknob, and there were no marks near the deadbolt. The locksmith testified that if

the deadbolt had been engaged and an intruder had entered through the door the deadbolt

would have shown obvious signs of damage. If the deadbolt was not engaged, the knob

lock could have been defeated in as little as 15 seconds. The locksmith testified that the

marks by the knob lock were consistent with damage from other burglary scenes.

BCA analysts examined the pattern of the shotgun blast that killed Heidi and

determined that the shotgun had been between three and ten feet away from her at the time

of the shot. Forensic examiners concluded that it was possible for someone of Firkus’s

approximate size to shoot himself in the thigh with that shotgun. The Federal Bureau of

Investigation (FBI) created a physical scale model of the crime scene. Evaluating the

ballistics evidence, Heidi’s autopsy report, and evidence from the crime scene

investigation, the FBI determined that the State’s theory—that Firkus shot Heidi and

C-24
himself—was consistent with the forensic evidence. Although no one conducted a similar

assessment of Firkus’s theory that Heidi was killed during a struggle between Firkus and

an intruder, the FBI expert testified that she could not rule out the possibility that an

intruder shot Firkus and Heidi. There were no identifiable fingerprints or DNA at the scene

from anyone other than Firkus and Heidi.

On April 27, 2010, two days after Heidi’s murder, St. Paul Police conducted a

“sound test” on the Firkuses’ home around 6:15 a.m. Two officers were stationed upstairs

with a video recorder while an officer downstairs fiddled with the inner front doorknob and

then the screen porch door. In the video recording, the doorknob jiggling is inaudible from

the bedroom and the bathroom; it is faintly audible from the top of the stairs between the

bedroom and the bathroom. Sounds from the screen porch door were completely inaudible.

At trial, Sergeant Shackle, who participated in the sound test, stated that the sound on the

video recording reflected what he could hear at the time police conducted the test. He

agreed, however, that a video recorder’s sound sensitivity is not necessarily the same as a

human ear.

Police testified—and photographs confirm—that a person of average height

standing upright outside the front door would be clearly visible to someone coming down

the stairs because of the window in the door. Firkus told police that he was not looking out

of that window when he and Heidi came down the stairs.

An FBI visual information specialist recreated the crime scene in both physical and

digital models. She testified that the goal of the models was to accurately represent to the

jury “what a location looked like, to have a sense of space, and a visual picture of a crime

C-25
scene.” The models used a ballistics report form to depict how far the shooter would have

been from Heidi when she was shot. The model was consistent with Firkus having shot

Heidi. When asked whether the model was also consistent with Firkus’s version of

events—that an intruder shot Heidi—the FBI witness answered “yes, absolutely.” On

redirect examination, the State sought to clarify that answer:

Q. Okay. And so when the defense asks “is what the defendant said
possible?” In theory, anything is possible, right?

...

A. That’s correct. Anything is possible, and I believe our diagrams and
exhibits show th[ose] possibilities.

Police showed Firkus a lineup that included a potential suspect on April 26. Firkus

did not identify anyone from the lineup. In May 2010, Firkus provided a sketch of the

potential intruder to police. Police identified a suspect based on this sketch in 2014. That

suspect was quickly eliminated because he had been in custody at the time of the murder.

These are the facts I would use, in step two of the circumstantial-evidence test, to

determine whether there is support for a reasonable inference that an intruder shot Heidi at

the Firkus home on April 25, 2010, and, hence, a reasonable inference that Firkus did not

kill Heidi with premeditation and intent.

D.

The State encourages us to exclude much of this evidence from the circumstances

proved. For the following reasons, I disagree with the State’s position.

First, the State’s evidence that Firkus intended to shoot and kill Heidi is

circumstantial. See State v. Ulrich, 3 N.W.3d 1, 10 (Minn. 2024) (stating that intent is

C-26
generally proved with circumstantial evidence). For instance, Firkus did not make

statements that he intended to shoot and kill Heidi. Cf. Jones, 4 N.W.3d at 501 (noting

defendant’s statements that he wanted the victim dead were direct evidence of intent).

Moreover, in this case, determining whether Firkus intended to shoot and kill Heidi turns

on Firkus’s defense at trial that the shotgun fired accidentally during his struggle with an

intruder. Importantly, Firkus concedes that if there was no intruder in the house when

Heidi was killed, it is impossible to draw a reasonable inference that he did not intend to

effect Heidi’s death—meaning there would be no reasonable inference inconsistent with

the guilty verdict.

The State offered no direct evidence that there was no intruder in the Firkuses’ home

(for instance, a video recording of the incident or a witness’s direct observation).

Accordingly, the jury’s determination finding Firkus guilty—that is, finding that he, rather

than any intruder, shot Heidi—necessarily required the jury to make an inferential leap

from other evidence. Therefore, in this case, we apply the circumstantial-evidence test to

the questions of intent and whether there was an intruder. As explained in Part B, because

there is no direct evidence in the record that there was no intruder in the house, we should

not exclude all evidence that could possibly support (through inference or otherwise) a

determination that there was no intruder in the house simply because the jury must have

inferred this fact in reaching its guilty verdict. Rather, we should only exclude from the

circumstances proved evidence that could not be true if there is sufficient evidence to

support a guilty verdict; that is, where direct evidence contradicts a specific piece of

circumstantial evidence supporting an inference that there was no intruder or where other

C-27
circumstantial evidence contradicts the evidence that there was no intruder such that we

know the evidence was susceptible to the jury’s credibility assessment.

I emphasize that neither the inference that there was nor that there was not an

intruder is a circumstance proved. The State asserts that it is a circumstance proved that

there was no intruder because the jury read Firkus’s statements to police, and heard a

recording of at least one statement, claiming that there was an intruder, but still returned a

guilty verdict—necessarily rejecting Firkus’s assertion that there was an intruder. As

described above, that is not how step one of the circumstantial-evidence test works. There

is no direct evidence that there was no intruder and, because the jury did not observe Firkus

making his statements, the statements were not susceptible to credibility assessment for

purposes of step one of the circumstantial-evidence test. 8 Thus, we accept the statements

8
Firkus’s statements are different than the defendant’s statement in State v. Webb,
440 N.W.2d 426 (Minn. 1989). In Webb, the jury heard conflicting statements regarding
whether the defendant owned a bedspread found near the victim. Id. at 430–31. The
defendant admitted to owning the bedspread and one of the defendant’s previous
roommates testified that he had seen the defendant with the bedspread. Id. at 430. There
was substantial evidence to the contrary, including uncertainty the roommate expressed in
his testimony, aggressive police interrogation tactics, lack of physical evidence that the
defendant owned the bedspread, and evidence that a nearby hotel used precisely the same
bedspread in its rooms. Id. at 429–30. But because there was direct evidence from the
defendant’s roommate establishing that the defendant owned the bedspread, in addition to
the defendant’s own similar statement, we accepted it as a circumstance proved despite the
conflicting evidence. Id. at 431 n.2.
We emphasize that in Firkus’s case, there is no evidence establishing the presence
of an intruder one way or another, and it is not certain which of Firkus’s statements is more
consistent with the jury’s verdict or which statement the jury must have accepted in finding
Firkus guilty. Thus, it is unnecessary for us to choose between Firkus’s statements in
winnowing down the facts to the circumstances proved. Indeed, as discussed below, the
conflicting nature of Firkus’s statements supports the State’s position that there was no
intruder.

C-28
Firkus made to the police as circumstances proved; we then assess whether they support a

reasonable inference that there was no intruder in step two when we consider all

circumstances proved—including many pieces of evidence suggesting Firkus’s statements

were not true. 9

As I observed in Part B, our prior cases have not always been clear on this point.

For instance, in Culver, 941 N.W.2d at 143–44, we considered a case of alleged felony

deprivation of parental rights in violation of a court order. We excluded from the

circumstances proved email messages between Culver and the father of her child

expressing Culver’s willingness to reschedule parenting time. Id. Culver argued the

written messages showed she did not subjectively intend to “substantially” deprive the

father of parenting time as required to find her guilty. Id. After brief analysis, we excluded

Webb is not a decision that turned on the improbability that the bedspread found
near the victim belonged to the defendant. To the contrary, we held that it was a
circumstance proved that the defendant owned the bedspread despite the conflicting
evidence described above. Id. at 431 & n.2. Our decision in Webb turned on the lack of
other evidence in the case pointing to the defendant’s guilt, such as the lack of testimony
and physical evidence connecting the defendant to the victim and absence of any credible
motive. Id. at 431.
9
The fact that Firkus’s unsworn statements may be evidence that there was no
intruder is not dispositive. Firkus’s statements were inconsistent. Further, there was
substantial other circumstantial evidence before the jury that there was no intruder and the
jury found that evidence convincing. In circumstantial-evidence cases, however, our job
is to make sure that there is no reasonable inference that the defendant was not guilty after
considering the totality of the evidence presented. One aspect of this analysis is
considering whether, in light of all the evidence, the jury properly inferred there was no
intruder even though Firkus made a statement that there was. To foreshadow my
conclusion, I determine that, in light of all the evidence in the case, Firkus’s self-serving
statements (and other evidence supporting a potential inference that there was an intruder)
are not enough to support a reasonable inference that Heidi was killed during a struggle
between Firkus and an intruder.

C-29
those statements from the circumstances proved because “the jury was not compelled to

believe the statements that Culver made merely because they were uncontradicted.” Id. at

144.

As discussed earlier, Culver’s articulation of the law is too blunt in a circumstantial-

evidence case. If a reviewing court excludes every fact—even uncontradicted facts—that

could support an inference that a defendant is not guilty from the pool of circumstances

proved just because the jury may not have believed it, it is applying the direct-evidence

test. Except to the extent we can reverse a conviction today under the direct-evidence test,

proceeding in this way means excluding every fact that could support an inference that the

defendant was not guilty before even reaching the second step of the circumstantial-

evidence test. The second step is rendered meaningless. The “heightened” test for

circumstantial-evidence cases is eliminated.

Rather, in Culver, the emails should not have been excluded from the circumstances

proved. While (as here) there was other circumstantial evidence contradicting Culver’s

messages that she would be willing to reschedule parenting time—notably, the fact that she

never did—there was no direct evidence that she was unwilling to reschedule parenting

time. Furthermore, the written messages themselves were not susceptible to the jury’s

credibility assessment. Consequently, Culver’s written messages should have been

included among the circumstances proved. 10

10
Even if the Culver court had included the messages among the circumstances
proved, the outcome of the case would not have changed. In Culver, the court of appeals
interpreted the deprivation of parental rights statute as requiring the State to prove the

C-30
 We have never held that a court must accept the inverse of a defendant’s statements

claiming innocence—i.e., that a defendant who lied when stating they were innocent is

defendant had a subjective intent to deprive a parent of rights to parenting time.
941 N.W.2d at 138. We held that the relevant statutory phrase “ ‘where the action manifests
an intent substantially to deprive that parent of rights to parenting time’ unambiguously
refers to a condition in which the defendant’s action shows or reveals an objective intent to
substantially deprive a parent of parenting time.” Id. at 141 (emphasis omitted). In
addition, we held that determining whether an accused substantially deprived a parent of
parenting time required considering both the quantitative (the amount of time) and
qualitative (the value of the time) effects of the deprivation. Id. at 143.
Under that legal standard, even if the Culver court had properly included among the
circumstances proved the messages in which Culver claimed that she was willing to
reschedule lost parenting time in the future, there was no reasonable inference to draw, in
view of all the circumstances proved, that she did not engage in actions that showed or
revealed an objective intent to substantially deprive her child’s father of parenting time.
The other circumstances proved included the following:

The child was 3 years old at the time of the offense. In response to Culver’s
request to take the child to a wedding, the court ordered that any changes
should be agreed upon or the parenting time would be as ordered. Culver
and [the child’s father] did not reach an agreement on any modifications to
parenting time to accommodate the wedding. [The child’s father] was
deprived of an overnight visit, which is an important part of the parent-child
relationship. Culver’s actions completely denied [the child’s father] his
court-ordered parenting time, which involved seven visits during the 15-day
period immediately before Culver’s arrest. Culver did not provide any
alternative parenting time to [the child’s father] during the charged period,
even though there were days she was not at a wedding or a funeral.

Id. at 144.
The messages Culver sent may have supported a reasonable inference that she did
not have a subjective intent to deprive the father of parenting time, but they did not support
a reasonable inference that her conduct did not show an objective intent to deprive the
father of parenting time. This conclusion is particularly compelling when one considers
not only the quantitative impact of the deprivation (that she could make up for lost time
later), but the qualitative impact in light of the length of the deprivation and the age of the
child. See id. at 143.

C-31
therefore guilty—as a circumstance proved. 11 Here, the conclusion that Firkus lied about

the presence of an intruder to hide his guilt is an inference. We do not consider, nor give

deference to, a jury’s inferences at step one of the circumstantial-evidence test. Instead,

we determine the circumstances proved and then—at step two—determine whether there

is any rational hypothesis from those circumstances that supports a reasonable inference of

not guilty without giving deference to the verdict. For that reason, it is not a circumstance

proved that there was no intruder.

Conversely, Heidi’s statement to the 911 operator that someone was breaking into

the house does not mean it is a circumstance proved that there was an intruder. The jury

heard Heidi’s 911 call and still rejected Firkus’s theory that there was an intruder. But we

accept the fact that Heidi made the 911 call and her statements in the 911 call as

circumstances proved. First, Heidi’s statements could be true even if there is sufficient

support for the verdict. Heidi never stated that she saw an intruder. As important, no direct

evidence contradicts the fact that Heidi made the statements, and it is not susceptible to a

11
The State points to State v. Gatson, 801 N.W.2d 134, 144 (Minn. 2011), in which
we found it was a circumstance proved that the defendant fabricated an alibi, to support its
argument that “reject[ing] a defendant’s testimony positively establishes the opposite
conclusion.” But in that case the defendant initially provided an alibi, then later admitted
that he drove his co-defendant to and from the crime scene. Id. at 140. Gatson is an
example of a case where the direct evidence contradicted the defendant’s alibi evidence;
the alibi evidence would have been excluded under the test I articulate in this concurrence.
By contrast, Firkus has never stated that there was no intruder. Some details of his story
may conflict—for example, the statements to police regarding whether he was headed to
the bathroom or in the bedroom when he heard noise from the front door—but he has
consistently claimed that there was an intruder.

C-32
credibility assessment. It is a circumstance proved that Heidi told the 911 operator that

someone was breaking into the house.

Evidence of marks on the front door frame by the lock is a circumstance proved.

No one disputes that the marks existed. But the State’s position that it is also a

circumstance proved that the marks were not made by an intruder on the morning Heidi

died is incorrect. It is true that the State offered a police sergeant’s testimony that the marks

were likely old and painted over. But the same witness also testified that it was “possible,”

though “not probable,” that the marks were made more recently by a screwdriver without

stripping away the paint. 12 There is no direct evidence that the marks were not made on

the morning Heidi died. Rather, the entirety of the sergeant’s testimony is included among

the circumstances proved. The same logic applies to the FBI crime scene model of the

Firkuses’ home. The entire testimony of the visual information specialist is a circumstance

proved.

I reach a similar conclusion about Firkus’s conflicting statements to police regarding

when he heard the intruder and how he and Heidi got downstairs. In Firkus’s first statement

to officers at the hospital, he said that he was either in, headed to, or coming from the

bathroom when he heard noises at the front door. According to Firkus, he then woke Heidi

and they either called or debated calling the police from the closet in their bedroom. During

Firkus’s second statement, at the police station, however, he placed himself in the bedroom

when he heard the noises from the front door. Firkus stated that he then grabbed the

12
A locksmith witness also testified that the marks by the knob lock were consistent
with damage from other burglary scenes.

C-33
shotgun before waking Heidi and, when he woke her, told her they were going to get out

of the house; he did not mention potentially calling the police from the closet.

These differences are important in light of the sound test that St. Paul police

conducted after Heidi was killed. During that sound test, officers upstairs used a video

recorder to record another officer fiddling with the doorknob on the front door. In the video

recording, the doorknob jiggling is inaudible from the bedroom and the bathroom, but is

faintly audible from the top of the stairs between the bedroom and the bathroom. Thus,

whether Firkus was in the bathroom, the bedroom, or at the top of the stairs between the

two when he heard jiggling could make a difference in whether Firkus’s statements about

the intruder are true.

Based on all these facts, it is not a circumstance proved that Firkus was in the

bedroom when he claimed to hear someone jiggling the doorknob. 13 There is no direct

evidence that this was the case. Rather, there is evidence, based on one of Firkus’s

statements, that he was in the bedroom at the time he claimed to hear the doorknob jiggle

and there is evidence, based on another of his statements, that he was at the top of the stairs.

Further, while the State introduced these statements at trial, Firkus did not testify. The jury

did not have the opportunity to observe Firkus testify to a particular version of events on

the stand or be cross-examined about the inconsistencies. Although the content of Firkus’s

13
If it were a circumstance proved that Firkus was in the bedroom, then Firkus’s
statement that he heard noise at the door would not be a circumstance proved because direct
evidence from the sound test that noise from the doorknob was inaudible from the bedroom
contradicts his statement that he heard the noise.

C-34
statements conflicts—meaning that both cannot be true—the existence of both statements

is a circumstance proved under our test.

In the same vein, we consider the statements B.O., the next-door neighbor, has given

over the years as circumstances proved. Except for the single consistent element of B.O.’s

account that he heard two gunshots next door around the time of Heidi’s murder, his story

has changed many times. B.O. has, at different times, stated that he heard a male voice

saying, “you shot me” or “you shot her,” that the voice he heard might have been female,

and that he might have heard at least one voice between shots. His testimony did not

establish what occurred on that morning. B.O.’s various statements and testimony at trial

as to what he remembered are circumstances proved, and we examine the inferences from

the testimony and statements in step two.

E.

Having determined the circumstances proved as described above, I turn to step two

of the circumstantial-evidence test—determining whether the circumstances proved, when

viewed as a whole, support a reasonable inference inconsistent with guilt. In reviewing a

conviction based on circumstantial evidence, we will not reverse “on the basis of mere

conjecture.” State v. Cox, 884 N.W.2d 400, 412 (Minn. 2016). Stated differently, a

defendant cannot rely on an inference that has no factual support in the record. See Tscheu,

758 N.W.2d 849, 860–61 (Minn. 2008); State v. Johnson, 995 N.W.2d 155, 163 (Minn.

2023). Rather, to “successfully challenge a conviction based upon circumstantial evidence,

a defendant must point to evidence in the record that is consistent with a rational theory

other than guilt.” State v. Taylor, 650 N.W.2d 190, 206 (Minn. 2002). On the other hand,

C-35
the fact that the inference a jury draws may be more reasonable or more likely than the

alternate inference the defense proposed is not enough to render the alternate inference

unreasonable. If the defense can put forth a single reasonable theory consistent with

finding that the defendant was not guilty based on all the facts proved, reversal is required.

See State v. Andersen, 784 N.W.2d 320, 329 (Minn. 2010). 14

As noted, Firkus concedes that, if there was no intruder, the evidence is sufficient

beyond a reasonable doubt to support the jury’s conclusion that he killed Heidi and acted

with both intent to kill and premeditation. Firkus also agrees that the circumstances proved

support a reasonable inference that there was no intruder. Thus, under the specific facts of

this case, we must determine whether there is a reasonable alternative inference that an

intruder murdered Heidi during a home invasion. After carefully reviewing all the

circumstances proved in their totality, I conclude that Firkus’s narrative about an intruder

is not a reasonable inference that the jury could have drawn.

14
I disagree that reversal is necessary under the circumstantial-evidence test unless
the State’s case is “infallible.” That sets the bar too high. The State’s burden in every
criminal case is to prove the defendant guilty beyond a reasonable doubt. State v. Lehman,
3 N.W.3d 875, 880 (Minn. 2024) (stating that “[t]he State does not have the burden of
removing all doubt, but of removing all reasonable doubt” (quoting Al-Naseer, 788 N.W.2d
at 473)); see State v. Smith, 674 N.W.2d 398, 403 (Minn. 2004) (stating that proof beyond
reasonable doubt “does not require the case to be proved beyond all possibility of doubt,”
but is based on reason and common sense, and approving description of reasonable doubt
as the amount of proof that ordinary men and women rely upon in making their own most
important decisions). Our review of the sufficiency of the evidence proceeds against this
standard of proof. For this reason, our circumstantial-evidence test requires that the
defendant establish that there is a reasonable inference to be drawn from all the
circumstances proved that is inconsistent with guilt.

C-36
 First, it is significant to my analysis that the evidence strongly supports that Firkus

had a motive to kill Heidi. State v. Berndt, 392 N.W.2d 876, 879 (Minn. 1986) (stating

that “if the state can establish a credible motive, credibility is lent to the state’s contention

that the accused committed the crime”). Heidi was shot one day before the Firkuses were

to be evicted from their home. Their home had been sold almost a year earlier. They were

well past the point of foreclosure, not having made a mortgage payment since 2008.

There is substantial evidence that Firkus had hidden the foreclosure from Heidi.

Heidi and Firkus’s text messages indicate that Heidi deferred to Firkus on financial matters.

Heidi had not reached out to anyone about the financial crisis the Firkuses faced; instead,

in 2009, she told her mother that “Nick told her they were fine.” 15 The evidence also shows

that Firkus misled Heidi into believing they were victims of identity fraud, and witnesses

testified that Heidi told them this had happened. Firkus attended the March 2010 eviction

hearing by himself even though Heidi worked nearby. Though Heidi had requested time

off from work to help her parents move, she had not requested time off to move out of her

own home. The day before the Firkuses were to vacate their home, nothing was packed or

ready for the move, and there were no plans for where they would go. Heidi and her

friends, in fact, had plans to get pedicures that day. Firkus had hidden the couple’s serious

financial problems from family, friends, and the couple’s church community. Neither

15
Firkus argues that Heidi must have seen financial correspondence around the house
that would have notified her about the Firkuses’ financial situation. But Firkus offers no
evidence of that. It is not a circumstance proved.

C-37
Firkus nor Heidi had told a single witness about the foreclosure and eviction or their dire

financial situation.

There is no evidence that the Firkuses were unhappy in their marriage or that Firkus

had made any threats against Heidi. But the outward appearance of happiness does not

cancel out the mountain of evidence that Firkus was hiding critically important information

from Heidi about the couple’s finances—information that could have shattered the

relationship.

Second, the State introduced unrefuted evidence that Firkus could have committed

the murder. After creating and analyzing a model of the crime scene, FBI experts

determined that the forensic evidence was consistent with a conclusion that Firkus shot

Heidi and himself.

The circumstances proved also show that the FBI expert could not rule out the

possibility that an intruder shot Firkus in the leg during a struggle over the gun by angling

the gun down after shooting Heidi. But the FBI only tested the State’s theory of the case—

that Firkus could have shot Heidi in the back and then shot himself in the leg. Neither the

FBI nor anyone else tested Firkus’s alternative theory that an intruder shot Firkus in the

leg and Heidi in the back. The FBI witness thus testified that Firkus’s theory was possible,

but it is not true that she testified that the FBI’s crime scene model was equally consistent

with an intruder theory as with the State’s version of events.

Much of the remaining evidence in the case focuses on Firkus’s intruder theory. In

assessing this theory, I start with the fact that no one aside from Firkus saw an intruder

entering or leaving the house. B.O.’s testimony establishes that two shots were fired in the

C-38
house and that someone (male or female) may have said something along the lines of “you

shot her” or “you shot me.” Because of his inconsistent statements, we do not know when

B.O. heard the voices next door (after or in between the two shots). B.O. acknowledged

that he may have heard Firkus as he was talking to the 911 operator after he and Heidi were

shot. B.O.’s testimony does not conclusively establish what he heard.

The details of Firkus’s own story about what happened on the morning Heidi was

shot are based on Firkus’s statements to police, not trial testimony. It is important to my

analysis of Firkus’s intruder theory that his narrative about the events on the morning of

April 25 materially changed over the course of his several statements to police. When

initially asked, Firkus said he could not be certain whether there was one intruder or more,

could not say if the intruder had facial hair, and was unsure of the intruder’s race. At the

police station, he ultimately gave a generic description of the intruder, stating that he was

black with lighter skin, 6’1” or 6’2” and 215 to 220 pounds, and had been wearing a hooded

sweatshirt, sunglasses, and gloves. 16 Just one day after Heidi’s murder, Firkus expressed

to Heidi’s parents that he believed the intruder would never be found.

16
The concurrence/dissent highlights one potential suspect who might have matched
Firkus’s description of the intruder and whose sister lived near the Firkuses. When shown
a lineup that included this suspect, Firkus failed to identify anyone that he recognized. The
suspect was also five inches shorter and 50 pounds lighter than the intruder Firkus
described to police. Firkus told police that the intruder was “much, much taller” and “a lot
bigger” than himself. But this suspect, standing at 5 feet 9 inches, would have been just
one inch taller and about the same weight as Firkus. It is highly doubtful—and certainly
not reasonable—that this person, who was far off from the physical description of the
intruder and who Firkus immediately ruled out when shown a lineup, could have been the
intruder.

C-39
 Further, Firkus’s narrative works only if he heard the intruder jiggling the front

door; absent that starting point, the entire series of events that Firkus claims followed make

no sense. But the reasonableness of Firkus’s claim that he heard an intruder jiggling the

door is dubious given his contradictory statements on the subject. Firkus’s statements place

him either in the bathroom, on the way to the bathroom, or in the bedroom when he first

heard the intruder at the door. The sound test investigators conducted and testimony from

officers who conducted the test indicate that it is very unlikely that someone could have

heard the front doorknob from the bathroom (where Firkus initially told police he was when

he heard the noise) or the bedroom. The video recorder evidence is not conclusive because

a video recorder does not necessarily have the same sound sensitivity as a human ear, 17 but

the sergeant’s testimony that he could not hear the front doorknob jiggling from most of

the upstairs area supports finding that Firkus did not hear jiggling at the front door. 18

Moreover, even if Firkus was on the landing and could have heard the front

doorknob jiggling, the rest of his story turns on a confluence of events taking place in a

17
In a portion of the officer’s testimony, he agreed that the human ear may be more
sensitive to sound than a video recorder. Notably, the sergeant also testified that a human
ear may be less sensitive than a video recorder. We have no way of knowing, and no
testimony establishing, exactly how sensitive the video recorder was compared to the
human ear. What we do know is that the sergeant also testified that the sound the video
recorder documented matches what he heard on the day of the sound test.
18
Because Firkus said in one of his statements that he may have been on the way to
or from the bathroom when he first heard the intruder trying to break in, and because the
sound test evidence demonstrated that the doorknob jiggling was perceptible from the
landing, based on the sound test evidence alone we cannot rule out or find unreasonable
the inference that Heidi was shot while trying to escape an intruder.

C-40
precise and narrow timeframe. 19 According to Firkus’s statements, once he heard the

doorknob jiggling he woke Heidi and told her there was an intruder and she needed to call

911. Firkus loaded two rounds into his shotgun, which happened to be in the bedroom.

Heidi grabbed her wallet and headed down the stairs in front of Firkus, who followed her

with the shotgun. In accordance with Firkus’s directive, Heidi called 911 and reported that

someone was breaking into her home. Heidi did not tell the 911 operator that she had

actually observed an intruder. Thus, while it is plausible that Heidi’s 911 statement

reflected a personal observation, it is equally possible that Heidi was reporting what Firkus

admitted he told her upstairs.

The officers who arrived on the scene observed few common signs of forced entry

or struggle. While there is evidence that a rug was wedged in the door and a pair of shoes

was out of place, there is also evidence that items on a wobbly table in the entryway were

undisturbed when the police arrived, though they fell off easily when police touched the

table.

In addition, a locksmith testified that a break-in was only possible if Firkus

neglected to set the deadbolt the night before. Firkus stated he normally set it. Even if the

deadbolt was not set, there is little evidence of the door being forced open. An investigator

testified about tool marks on the door frame and stated that the most likely explanation was

that the marks were old and had been painted over. As Firkus notes, the investigator also

testified that it “possible but not probable” that the marks were fresh compression imprints

19
The State’s theory of the case is much simpler, requiring far fewer specific events
and less precise timing.

C-41
from a flathead screwdriver. Investigators did not find a screwdriver matching the

impressions at the Firkus house, meaning that any intruder would have had to remember

and taken the screwdriver with them after the struggle and shooting.

In addition, the 911 call did not record sounds of a break-in or struggle. 20 There

were no identifiable fingerprints or DNA at the scene from anyone other than Firkus and

Heidi. 21

Finally, even according to Firkus’s timeline, an intruder would have had to open the

front door at the exact moment Firkus was passing by (and after Heidi had gone by). Given

the timing of the 911 call and the location of Heidi’s body, the shotgun would have had to

go off almost immediately after the intruder entered the house and started struggling with

Firkus, accidentally hitting Heidi squarely in the back and then Firkus in the leg.

As Firkus points out, there are individual circumstances proved that support his

intruder theory. This is true in nearly every case; there will almost always be evidence

that, when viewed in isolation, raises doubt about a defendant’s guilt. In the second step

of the circumstantial-evidence test, however, our job is not to parse through each piece of

20
Firkus does not dispute that no sounds of struggle are audible on the 911 recording,
but insists that is to be expected because the phone was not in speaker phone mode. There
is no evidence in the record to support this argument; it is speculation. See Johnson,
995 N.W.2d at 163 (stating that a party cannot rely on an inference that has no factual
support in the record).
21
In his initial statement to police, Firkus reported that the intruder was wearing
gloves—a fact he now claims explains the lack of fingerprints and DNA. Of course, the
lack of identifiable fingerprints and DNA in no way supports an inference that there was
an intruder; it merely weakens (but does not eliminate) the strength of the lack of
fingerprints and DNA as support for an inference there was no intruder—an inference that
Firkus admits is reasonable.

C-42
evidence in isolation and decide whether it supports an inference inconsistent with guilt.

Our job is to look at the circumstances proved as a whole and determine if there is a

reasonable inference that the defendant is not guilty. Isaac, 9 N.W.3d at 817–18. After

carefully considering the totality of the circumstances proved in this case, including the

strong and nearly unrefuted motive evidence, the fact that the State affirmatively

established that Firkus could have shot Heidi and himself, and the material inconsistencies

in Firkus’s varying statements to police, I am convinced that the only reasonable inference

is that there was no intruder in the home when Heidi was shot.

* * *

In summary, I would apply the circumstantial-evidence test in a manner consistent

with our case law that fulfills the purposes of the test when we first adopted it. Applying

that test here, I would affirm Firkus’s conviction.

C-43
 C O N C U R R E N C E

PROCACCINI, Justice (concurring).

I join in the opinion of the court. I write separately to note that the divergent views

expressed by the members of the court in this case show that the concerns raised by the

dissent in State v. Harris have not diminished. In that dissent, Justice Lillehaug observed

that Minnesota is an outlier in maintaining a dichotomy between direct and circumstantial

evidence in appellate review. State v. Harris, 895 N.W.2d 592, 604 (Minn. 2017)

(Lillehaug, J., dissenting) (noting that “the appellate courts of the United States, 41 other

states, and the District of Columbia” all follow a unified standard of review). He also

asserted that “the distinction between direct and circumstantial evidence is arbitrary,” “the

notion that direct evidence is necessarily more reliable than circumstantial evidence is

outdated,” and “the differing standards of review are confusing and difficult to apply.” Id.

at 606. In my view, the first two points remain valid, and the third is on display

here—where we disagree about the description of the circumstantial-evidence standard, its

application, and the ultimate disposition of this case. Although the parties did not invite us

to reexamine the distinction between direct and circumstantial evidence on appellate

review, and we therefore do not have an opportunity to do so here, I would be open to that

invitation in a future case.

McKEIG, Justice (concurring).

I join in the concurrence of Justice Procaccini.

HENNESY, Justice (concurring).

I join in the concurrence of Justice Procaccini.

C-1
 C O N C U R R E N C E & D I S S E N T

HUDSON, Chief Justice (concurring in part, dissenting in part).

The purpose of the circumstantial-evidence test is to guard against wrongful

convictions. For almost a century, this court has acknowledged the unique posture of

circumstantial evidence cases by applying a separate standard of appellate review. See

State v. Johnson, 217 N.W. 683, 684 (Minn. 1928). And although the State has repeatedly

asked us to overturn this precedent, we have declined to do so because of the basic principle

for which it stands: “our duty to ensure that defendants not be convicted based on

insufficient evidence.” State v. Harris, 895 N.W.2d 592, 599 (Minn. 2017). In fidelity to

that duty, I disagree with the court in its articulation of the circumstantial-evidence test and

its conclusion that the only reasonable inference that can be drawn from the evidence is

that Firkus caused the death of Heidi with premeditation and intent to effect her death.

Thus, I would reverse Firkus’s conviction. I therefore respectfully dissent from Part I of

the court’s opinion. 1

I agree, in part, with Justice Thissen’s concurrence (“the concurrence”).

Specifically, I agree with the concurrence’s articulation of the circumstantial-evidence test,

but I disagree with the concurrence’s conclusion that no reasonable inference may be drawn

from the circumstances proved except that of guilt.

1
I agree with the court’s holding in Part II that the district court did not err when it
applied the direct-evidence standard in considering Firkus’s motion for judgment of
acquittal made prior to the verdict.

C/D-1
 A.

“[A] circumstantial evidence case is the only case in the judicial system in which

every witness can tell the truth and an innocent person can be convicted.” C. Rauch Wise,

Circumstantial Case—Circumspect Conviction, Nat’l Ass’n of Crim. Def. Lawyers 2

(2022). Circumstantial evidence is unique because it “always requires an inferential step

to prove a fact that is not required with direct evidence.” Id. It is at this inferential step

where we serve as a bulwark between the jury and the defendant’s presumption of

innocence. Firkus’s case provides a clear example of the power of inference. The evidence

presented at trial provided two equally reasonable scenarios: that Firkus intentionally shot

Heidi, or that a struggle with an intruder caused Firkus to shoot her. Jurors are unlikely to

envision a scenario of innocence on their own in light of narrative evidence of guilt.

Eugenée M. Heeter, Comment, Chance of Rain: Rethinking Circumstantial Evidence Jury

Instructions, 64 Hastings L.J. 527, 538 (2013). Thus, it is our job as the reviewing court

to ensure that the State has disproven any reasonable scenario of innocence. In my view,

the State did not do so here, and the court errs by holding otherwise.

The court gets it wrong in large part because its articulation of the circumstantial-

evidence test is flawed. In winnowing down the evidence to determine the circumstances

proved, the court practically discards all evidence that is inconsistent with the guilty

verdict. 2 It is no surprise then that at the second step of the test, the court concludes that

2
The court makes a distinction between “resolving all questions of fact in favor of
the jury’s verdict” and disregarding evidence that is inconsistent with the jury’s verdict, but
this is a distinction without significance. The practical effect is the same—excluding any
evidence that might support a rational hypothesis other than guilt.

C/D-2
the circumstances proved when viewed as a whole are consistent with the hypothesis that

Firkus is guilty and inconsistent with any rational hypothesis other than guilt. 3 The

circumstantial-evidence test, as articulated by the court today, does nothing to guard

against wrongful convictions. If anything, it upholds wrongful convictions by design.

The court’s main critique of the separate writings is that they propose an approach

that infringes on the jury’s role as the sole judge of credibility. But the court is just as

guilty of this charge. An approach that resolves all fact questions in favor of the jury’s

verdict infringes on the jury’s role because a jury “is free to accept part and reject part of a

witness’s testimony.” See Harris, 895 N.W.2d at 600. While criticizing us for doing so,

the court ironically usurps the jury’s role by retroactively resolving all fact questions in

favor of the verdict regardless of the jury’s actual credibility determinations. 4

For these reasons, I disagree with Part I of the court’s opinion. Instead, I join Justice

Thissen’s concurrence in its articulation of the circumstantial-evidence test. The court, the

concurrence, and I agree that the factual scenario in which there was no intruder and Firkus

killed Heidi with a shotgun blast to the back is a reasonable inference consistent with guilt.

3
As used throughout the concurrence/dissent, the terms “hypothesis” and “theory”
are interchangeable.
4
Under the court’s articulation of the circumstantial-evidence test, our long-standing
commitment to heightened review in circumstantial evidence cases is an illusory one. If
our commitment to a heightened standard means anything, it cannot be the case that we
disregard all evidence that is inconsistent with the verdict under both the direct-evidence
test and the circumstantial-evidence test. Otherwise, the two tests collapse into one. As
Justice Thissen observes in his concurrence, implicit in our precedent adopting a
heightened standard in circumstantial evidence cases is an acknowledgment that appellate
courts have to act as so-called 13th jurors to ensure that defendants are not convicted based
on insufficient evidence.

C/D-3
But it is here that we part ways because that is not the only reasonable inference to be

derived from the circumstantial evidence with which we are presented. “[T]o succeed on

his challenge to the verdict, [Firkus] may not rely on mere conjecture. He must instead

point to evidence in the record that is consistent with a rational theory other than guilt.”

See State v. Tscheu, 758 N.W.2d 849, 858 (Minn. 2008) (citation omitted). I believe Firkus

has done so here. I therefore disagree with the court’s and concurrence’s conclusion at the

second step of the circumstantial-evidence test that the circumstances, as a whole, support

only one rational inference—that Firkus intentionally shot Heidi. See Harris, 895 N.W.2d

at 600–01 (articulating the second step of the circumstantial-evidence test). I believe there

are two rational inferences to be drawn in this case.

To sustain a conviction of first-degree murder based purely on circumstantial

evidence, “[a] detailed review of the evidence in [the] case, made in a light most favorable

to the verdict, [must] exclude, beyond a reasonable doubt, all reasonable inferences other

than that of appellant’s guilt.” See State v. Scharmer, 501 N.W.2d 620, 622 (Minn. 1993);

Minn. Stat. § 609.185(a)(1) (2024) (defining the elements of first-degree premeditated

murder). At the second step of the circumstantial-evidence test, “we do not defer to the

fact[-]finder but examine the reasonableness of the inferences ourselves.” State v. McInnis,

962 N.W.2d 874, 890 (Minn. 2021).

The concurrence correctly identifies the circumstances proved. A non-exclusive list

of the circumstances proved consistent with guilt are as follows: the FBI crime scene

model supported a non-intruder theory; the distance from the shotgun to Heidi’s wound is

consistent with Firkus shooting her from the base of the stairs; there was no tipped-over

C/D-4
furniture or other disruption of the items in the hall; the tool marks outside the front door

lock could be old and painted over, rather than freshly created; there were no DNA or

fingerprints that conclusively matched a person who is not Firkus or Heidi; Firkus’s thigh

wound could be consistent with a self-inflicted gunshot; Firkus and Heidi were facing

serious financial difficulties that could provide Firkus with a motive to kill her; the police’s

video recorder sound test indicated that Firkus could not have heard the doorknob when he

was in the bedroom or bathroom, as he told police; B.O. and the other neighbors did not

see anyone leaving the Firkus home; Firkus had gunshot residue on his hands; the

locksmith testified that a break-in was only possible if Firkus neglected to set the deadbolt

the night before, although Firkus usually set the lock every night; the details in Firkus’s

statements to police changed in material ways over the course of the investigation; Firkus

could not identify a suspect from a police lineup; and there were no sounds of a break-in

or struggle on Heidi’s 911 call.

But these circumstances proved have an inverse analogue that is not consistent with

guilt. Specifically, the concurrence correctly noted the following were also circumstances

proved: Firkus made several statements to police, including his repeated assertions that

there was an intruder; Heidi’s 911 call stated that there was an intruder; the locksmith

witness testified that the marks by the front door lock were consistent with damage from

other burglary scenes; the FBI crime scene model was consistent with Firkus’s intruder

theory; the FBI model and BCA ballistics analysis indicated that Firkus could have suffered

a gunshot wound to the thigh during a struggle; Firkus provided a suspect description to

the police and to a police sketch artist; Firkus’s statements indicated that he could have

C/D-5
been in either the bedroom or the bathroom when he heard the doorknob jiggle; the video

recorder may not have been as sensitive to sound as the human ear; on the day of the

shooting, B.O. told a reporter that he heard two gunshots from next door and soon afterward

heard a male voice saying “you shot me” or “you shot her”; the entryway rug and shoes

were disturbed; the front door was propped open one inch when the police arrived; the

intruder would not have been visible to the neighbors if he had ran to the right of the Firkus

home when fleeing; the door knob lock could have been defeated in as little as 15 seconds

if the deadbolt was not engaged; Firkus stated that the intruder was wearing gloves, which

would make him less likely to leave DNA or fingerprints at the scene; there was a previous

burglary in the same neighborhood and Heidi felt uncomfortable in the home; the Firkuses

outwardly had a loving relationship; one fingerprint on the front screen door did not match

Firkus or Heidi, and another fingerprint from the interior of the door was inconclusive;

there were multiple DNA samples that were insufficient for testing; and Firkus did not

receive any of Heidi’s life insurance money. These are over a dozen circumstances proved

that are consistent with a rational theory other than guilt and yet the court excludes the vast

majority of them. 5

But there is more. Firkus offers additional support for his intruder theory. To

explain the lack of disturbance of the wobbly console table in the entryway, he points out

5
The court claims that I essentially include “all the evidence” presented at trial in the
circumstances proved. But I only include what I must. It is well-established precedent that
the positive testimony of an unimpeached witness “cannot be arbitrarily disregarded by
either court or jury, for reasons resting wholly in their own minds, and not based upon
anything appearing on the trial.” See O’Leary v. Wangensteen, 221 N.W. 430, 431
(Minn. 1928).

C/D-6
that four police officers entered the house after his 911 call without disturbing any items

on the table. He notes that the FBI’s model “illustrate[s] that [Firkus] could have struggled

with an intruder in the area of the stairs, and not knocked into any furniture in the

entryway.” And he asserts that under the State’s theory, in order to shoot Heidi in the

kitchen and then himself in the leg, he would have had to rotate his body 180 degrees

between shots. He says, “[t]he State offered no explanation for why, if [Firkus] staged this

encounter to look like a struggle with an intruder, he would have pirouetted between shots.”

Finally, to rebut the State’s claim that he was suspiciously calm during his police

interviews, Firkus notes that he was sedated at the hospital for his injury and had previously

been hysterical.

In addition, Firkus presented other plausible evidence that undercut the State’s

theory. For example, B.O. testified at trial that he told a Pioneer Press reporter, on the day

of the shooting, that after hearing gunshots from next door, he heard a male voice cry out

“you shot her” or “you shot me.” In support of its narrative that there was no intruder, the

State posited at trial that the male voice B.O. heard was actually Firkus talking to 911—

rather than Firkus talking to an intruder. In response, Firkus notes: (1) the 911 call

transcript does not include the words “you shot her” or “you shot me,” which indicates that

he was directing the words at the intruder when he was shot, rather than relaying it on a

phone call; and (2) the use of the pronoun “you” is circumstantial evidence that another

person was in the room to whom Firkus was speaking. It is also noteworthy that, at trial,

B.O. did not give a precise timeline, but he heard the voice yell out soon after hearing the

gunshots. Presumably the exclamation and the shots were close in time, but it took Firkus

C/D-7
almost 65 seconds after Heidi was shot to dial 911. It is therefore a reasonable inference

that the male voice B.O. heard was reacting to the gunshots in the heat of the moment, not

talking to police on the phone over one minute later. 6 Furthermore, B.O.’s comments

shortly after the murder to the reporter are circumstances proved, and he told the reporter

that it sounded like the male voice was “scared” and that the voice had said “stop, please.”

These circumstances proved support Firkus’s intruder theory.

Firkus also points to the tight timeline between the end of Heidi’s 911 call and the

start of Firkus’s 911 call as additional proof that Firkus could not have staged the crime as

the State argued. 7 Firkus initially struggled to call 911 from Heidi’s phone. Thirty-eight

seconds after Heidi was shot, Firkus made his first attempt to contact 911, but accidentally

dialed two other numbers that were in Heidi’s recent calls list. The State’s theory would

give Firkus 38 seconds to shoot Heidi, shoot himself in the leg, disturb the rug and shoes

at the entryway, prop the front door open one inch, and crawl or walk to Heidi’s body to

call 911 from her phone. Firkus asserts that the timeline is “impossible.”

6
It is another possible inference that B.O. lingered at the window for over one minute
after hearing the gunshots before going downstairs to try to contact police on his computer,
and therefore overheard both the gunshots and the 911 call. But it is also a circumstance
proved that B.O.’s testimony at trial 10 years after the murder differed from what he
initially told the reporter and police in 2010.
7
Conversely, the concurrence argues that Firkus’s intruder theory requires too tight
of a timeline between the intruder breaking open the door, the struggle, and the first shots
ringing out. I mention Firkus’s argument merely to show that he has pointed to evidence
in the record in support of his theory. Cf. Tscheu, 758 N.W.2d at 858.

C/D-8
 B.

We need not find that the State’s version of events is impossible in order to conclude

that the State did not meet its burden in this case. I acknowledge that an appellant casting

doubt on the plausibility of the State’s theory is not in itself sufficient to challenge a

conviction. See State v. Race, 383 N.W.2d 656, 662 (Minn. 1986). “Our cases do not

require that the evidence be consistent only with a specific theory advanced by the

prosecution.” Id. Rather, “[o]ur decisions make clear . . . [a defendant] must show that his

claim is consistent with a rational hypothesis other than guilt.” Id.

We have repeatedly said that to support a conviction, the circumstantial evidence

“must point unerringly to the accused’s guilt.” State v. McArthur, 730 N.W.2d 44, 49

(Minn. 2007) (citation omitted) (internal quotation marks omitted). “Unerring” means

“[i]ncapable of error” or “infallible.” Unerring, Black’s Law Dictionary (12th ed. 2024);

see also Merriam-Webster’s Collegiate Dictionary 1366 (11th ed. 2020) (defining

“unerring” as “committing no error: faultless, unfailing”). I cannot say that the State has

“infallibl[y]” proven beyond a reasonable doubt that there are no reasonable hypotheses

inconsistent with guilt. 8 There are a litany of circumstances proved that are consistent with

8
The concurrence asserts that I set the bar too high by requiring the State to put on
an “infallible” or “unerring” case. I agree with the concurrence that the State’s burden—
as articulated in our case law—is to disprove beyond a reasonable doubt any reasonable
hypothesis inconsistent with guilt. See State v. Andersen, 784 N.W.2d 320, 330
(Minn. 2010). But I disagree with the concurrence’s contention that the burden I have
articulated is too high. Circumstantial evidence cases are subject to enhanced scrutiny by
a reviewing court, hence the reason why the court does not apply the direct evidence test.
See Bernhardt v. State, 684 N.W.2d 465, 477 (Minn. 2004). I am merely using a plain
language analysis to reiterate our requirement that “the circumstantial evidence must do

C/D-9
an intruder theory. And no expert in the record conclusively foreclosed an intruder theory.

To the contrary, all of them testified at trial that the intruder scenario was at least possible,

even if not highly likely. Cf. Harris, 895 N.W.2d at 602–03 (reversing a circumstantial

evidence ineligible-person-in-possession-of-a-firearm conviction in part because the

expert testified at trial that approximately 25 percent of the general population could not

be excluded as a source of DNA found on the firearm); State v. Al-Naseer, 788 N.W.2d

469, 477–78, 481 (Minn. 2010) (reversing a circumstantial evidence vehicular homicide

conviction in part because police officers testified that there could be various reasons why

the defendant did not react to the car accident, including reasons that were inconsistent

with guilt). “[I]f 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 474 (quoting

Andersen, 784 N.W.2d at 338 (Meyer, J., concurring)). Because there are multiple

circumstances proved consistent with a rational hypothesis other than guilt, the State has

not “unerringly” proven its case beyond a reasonable doubt. McArthur, 730 N.W.2d at 49.

Here, the circumstances proved do not rule out Firkus’s intruder theory; in fact, the

evidence supports it. The entire testimony of the FBI visual investigation specialist is a

circumstance proved. And the FBI witness testified that their model of the crime scene

was equally consistent with an intruder theory as with the State’s version of events. 9

more than give rise to suspicion of guilt; it must point unerringly to the accused’s guilt.”
Id. (quoting Scharmer, 501 N.W.2d at 622).
9
The FBI visual investigation specialist testified that they did not specifically build
a model to disprove Firkus’s intruder theory. But when asked whether Firkus’s theory was

C/D-10
Similarly, the police sergeant testified it was possible that the marks were freshly made by

an intruder. In addition, there was a crumpled rug and an askew pair of shoes in the hallway

that were consistent with a potential struggle, and the front door was propped open one

inch. Unlike in Tscheu, it is not a distant theoretical possibility that a third party entered

the Firkuses home, struggled with Firkus over the gun, and caused him to kill Heidi in the

process. In Tscheu, we pointed to the presence of DNA consistent with Tscheu’s profile

found under the victim’s fingernails and in her rectum as support for the only reasonable

hypothesis that Tscheu forcibly restrained her, penetrated her, and then drowned her within

a short period of time. Id. at 860. The physical evidence made it implausible to us “that

he engaged in consensual vaginal sex with [the victim] and that [the victim] was

subsequently attacked and murdered by a different person” who left no physical evidence

of their presence. Id. at 861. Here, however, the physical evidence of a third party can be

readily inferred from the crime scene.

The concurrence points to the lack of identifiable fingerprints or DNA at the scene

as support for its inference that there was no intruder. But one crime scene investigator

testified at trial that there was a latent fingerprint on the exterior screen door that did not

match either Firkus or Heidi, and there was another inconclusive print found on the interior

frame of the main door. The two unknown fingerprints are circumstances proved. And

although the DNA analyst at the scene was able to retrieve multiple DNA samples, at least

two of them were insufficient to perform DNA profiling. This, too, is a circumstance

possible, the specialist testified that “[a]nything is possible, and I believe our diagrams and
exhibit show that [sic] possibilities.”

C/D-11
proved that is not inconsistent with an intruder theory. See, e.g., Andersen, 784 N.W.2d at

332 (finding that “the lack of footprints around the crime scene is [either] not inconsistent

with Andersen’s guilt, or consistent with his innocence”).

Nevertheless, the court and concurrence find that Firkus’s intruder theory is not a

reasonable inference that the jury could draw. This threatens to collapse the heightened

two-step analysis we have sought to apply to circumstantial evidence cases into one step

that is deferential to the guilty verdict. Cf. State v. Jones, 4 N.W.3d 495, 500 (Minn. 2024)

(defining the one-step standard of review for direct evidence as “view[ing] the evidence in

a light most favorable to the verdict and assum[ing] the fact-finder disbelieved any

testimony conflicting with that verdict when considering whether sufficient evidence

supports the jury’s guilty verdict” (citation omitted) (internal quotation marks omitted)).

Relying on inconsistencies in Firkus’s interviews with police, the concurrence

concludes that Firkus’s intruder theory was unreasonable for the jury to believe. The

concurrence conducts an individualized credibility assessment of every statement Firkus

gave to police, particularly the suspect’s description. The concurrence asserts that Firkus’s

varying descriptions render it “highly doubtful” and “certainly not reasonable” to believe

Firkus’s recounting of the intruder was accurate. But the job of the reviewing court is not

to determine whether every statement the defendant provides to police matches with a

specific theory of the case. As the concurrence itself states, “the fact that the inference

drawn by the jury may be more reasonable or more likely than the alternative inference

proposed by the defense is not enough to render the alternative inference unreasonable.” It

is possible that Firkus’s inconsistencies mean that he is lying about what happened that

C/D-12
night; but it is also possible that because of the panic of the moment, his memories of the

intrusion were jumbled and confusing.

Firkus does not have an obligation to explain every discrepancy in the circumstances

proved in order to make his theory reasonable. Rather, Firkus must “point to evidence in

the record that is consistent with a rational theory other than guilt.” Tscheu, 758 N.W.2d

at 858. As I explained above, he has identified evidence in the record to support his intruder

theory. It is the State—not Firkus—that retains the affirmative burden to remove all

reasonable doubt. Andersen, 784 N.W.2d at 330 (“The State does not have the burden of

removing all doubt, but of removing all reasonable doubt.”). And to the extent the

concurrence argues that the simplicity of the State’s case renders it more believable, this

court has never held that the defendant must present a simpler theory of the case in order

to identify a rational theory inconsistent with guilt. See, e.g., Al-Naseer, 788 N.W.2d at

477 (overturning the defendant’s conviction in part because of expert testimony that there

could have been “various reasons,” some of which were more likely than others, that the

defendant was not aware that he had hit a person while driving).

There are two cases that are particularly helpful in guiding my analysis of whether

Firkus’s theory is reasonable given some inconsistencies in the facts: State v. Webb,

440 N.W.2d 426 (Minn. 1989), and Race, 383 N.W.2d at 656. In both cases, the State’s

theory of guilt—and thus, the conviction—rested almost exclusively on disputed facts and

inconsistent defendant testimony.

In Webb, a first-degree premeditated murder case, the State presented no direct

evidence connecting the defendant to the murder victim. See 440 N.W.2d at 429–30. Nor

C/D-13
did the police find any physical evidence that the murder occurred in the defendant’s

apartment as the State alleged, despite taking hundreds of forensic samples. Id. at 430.

There was conflicting evidence that defendant owned the bedspread found near the victim’s

body. Id. The defendant told police that he owned the bedspread, but his testimony was

contradicted by one of his former roommates, who stated that he did not have such a

bedspread. Id. The bedspread was also common to many other homes in the community

and a nearby hotel. Id. We ultimately considered it a circumstance proved that the

bedspread belonged to the defendant, see id. at 431 n.2, but we reversed his conviction. Id.

at 431. Although we “view[ed] the evidence in a light most favorable to the conviction,

we [could] not say that the circumstantial evidence in [the] case [was] consistent only with

the appellant’s guilt, and inconsistent with any other rational hypothesis.” Id. at 430. That

is because “[o]ther circumstances in [the] case undercut the state’s hypothesis of the

appellant’s guilt.” Id. at 431. Notably, we did not require Webb to disprove the

circumstances proved aligned with his guilt, and we did not require him to resolve any

inconsistencies in his personal statements or to provide an explanation as to why they were

inconsistent. 10

In Race, another first-degree murder case, we noted a lack of direct evidence and a

defendant whose statements to authorities contained numerous “[s]ignificant

inconsistencies.” 383 N.W.2d at 662. Race owned a boat that he frequently took out on

10
We noted “[a]ppellant’s admission that he owned the bedspread may have been the
product of an overly assertive interrogation by the police,” id. at 431, given that he
“suffered some sort of mental deficiencies.” Id. at 428.

C/D-14
Lake Superior. Id. at 657–58. One evening, he took his wife on the boat; she disappeared,

and her body was later found on the lakeshore. Id. at 660. In statements to police, Race

gave conflicting stories. He initially alleged that his boat engine flooded, that he helped

his wife into a life raft, and while he was towing her to shore, she floated away and

drowned. Id. at 659. In the first version of his story to police, there were two life rafts on

board. The first one would not hold air, so he tossed the raft overboard, inflated a second

raft, and helped his wife into it. Id. Police, however, later discovered that there was one

raft still aboard the boat, which had slash marks in the underside consistent with a knife.

Id. at 660. Race changed his story to assert that he had kept the first raft in the boat, rather

than throwing it overboard. Id.

The trial “focused on the existence or nonexistence of an alleged second raft” and

“these inconsistencies [in Race’s story] were a major factor in resolving the issue.” Id. In

weighing the sufficiency of the circumstantial evidence supporting Race’s conviction, we

determined that “the jury, whose task it was, concluded there was only one life raft on

board.” 11 Id. at 662. Therefore, “the only rational inference to be drawn from the

circumstantial evidence is that appellant murdered his wife.” Id. We noted that there was

no physical, forensic, or testimonial evidence indicating that the second raft existed. Id.

But primarily, we found that the “[s]ignificant inconsistencies in appellant’s statements to

authorities substantially diminished the credibility of his assertion of the existence of two

11
The existence of a second raft was crucial to the question of guilt because, if it
existed, Race’s account of an accident remained plausible but if it did not, Race’s account
was a fabrication designed to explain incriminating physical evidence.

C/D-15
rafts.” Id. The lack of evidence in the record, as well as Race’s diminished credibility,

underlaid our decision to sustain Race’s conviction. 12 Id.

In Firkus’s case, the concurrence points to “material” inconsistencies in Firkus’s

statements to police. When first asked, Firkus could not say whether there was one intruder

or multiple intruders. Nor could he identify specific physical characteristics of the intruder.

In a subsequent police interview, Firkus changed his story to say that the man was wearing

a hooded sweatshirt, sunglasses, and gloves, and provided a physical description. 13

Similarly, Firkus at first claimed to police that he was walking to the bathroom to

get a drink of water when he heard the front doorknob jiggling, and then later said that he

was in the bedroom when he heard it. Relying primarily on the police’s video recorder

sound test of the doorknob jiggling, the concurrence says that “it is very unlikely that

someone could have heard the front doorknob from the bathroom . . . or the bedroom.” But

the concurrence then immediately says that “[t]he video recorder evidence is not conclusive

because a video recorder does not necessarily have the same sound sensitivity as a human

ear.” The concurrence agrees that both Firkus’s and the State’s versions of Firkus’s

location when he heard the doorknob are circumstances proved. I question, then, how

12
Race had also taken out several life insurance policies on his wife several months
before her death, which he later used to pay off a home mortgage and personal loans. Id.
at 661.
13
Testimony by an investigator at trial indicated that the police found a suspect who
largely matched Firkus’s description—although he was shorter and weighed less—whose
sister lived nearby to the Firkus home. It is a circumstance proved that the police found a
suspect, but it is also a circumstance proved that Firkus could not pick a suspect out of a
later police lineup.

C/D-16
Firkus’s version is less likely to have occurred. The police did not use a decibel meter to

pick up the sound of the doorknob. The sound test occurred in 2010, when recording

technology was significantly less advanced than it is today. And it is not unreasonable to

believe that someone who lives in a home is more attuned to slight sounds than someone—

like a police officer—who does not live there. Unlike in Race, where the physical evidence

of the slashed life raft indicated that the defendant’s initial story that he threw the raft

overboard was untrue, there is no concrete evidence rendering one of Firkus’s statements

untrue. As the concurrence said, nothing about where Firkus was standing when he first

heard the intruder is “conclusive.”

Firkus’s situation is more akin to that in Webb, described above. Our reversal of

Webb’s conviction was despite conflicting evidence on whether he owned the bedspread

found near the body because we focused on other circumstances proved that undercut the

state’s hypothesis of Webb’s guilt. Webb, 440 N.W.2d at 430. As I have enumerated

above, there is a long list of conflicting inferences in this case, ranging from the ballistics

analysis to the unknown fingerprint found at the scene to B.O.’s witness testimony. The

concurrence itself finds the video recorder result to be “not conclusive,” and agreed that

the intruder theory cannot be ruled out as unreasonable “based on the sound test evidence

alone.” There are many conflicting inferences in this case, which makes what we explained

in Webb equally true here: the “circumstances [proved] may cast a suspicion of guilt on

the appellant, but they in no way exclude other rational inferences which can also be drawn

from these circumstances.” Id. at 431.

C/D-17
 Finally, we turn to the motive evidence. “In a first-degree murder prosecution, the

state has no burden of establishing a motive for the crime. Nonetheless, if the state can

establish a credible motive, credibility is lent to the state’s contention that the accused

committed the crime.” State v. Berndt, 392 N.W.2d 876, 879 (Minn. 1986). In

circumstantial evidence cases, “[m]otive evidence includes threats by the defendant,

conduct by the victim known to have angered the defendant, and plans or desires of the

defendant that would be furthered by the victim’s death.” McArthur, 730 N.W.2d at 49.

But here, the only motive evidence is of the last type—that Firkus was embarrassed by

their financial situation and needed to kill Heidi to prevent her from finding out about the

upcoming foreclosure.

In State v. Hughes, we weighed motive evidence the State presented at trial that the

defendant’s “marriage had deteriorated to the point of divorce,” that he was “distraught

about their separation,” that he was fearful “divorce would lead to his loss of custody of

the children,” and that his wife “planned to discuss child custody arrangements” on the day

the defendant murdered her. 749 N.W.2d 307, 314 (Minn. 2008). This, we felt, was

sufficient to “provide[] a basis for the jury to infer that appellant’s motive for killing his

wife was to prevent her from leaving him and taking the children.” Id.; see also Andersen,

784 N.W.2d at 331 (noting motive evidence that the victim had recently ended his business

relationship with the defendant and that their personal relationship was deteriorating).

Here, however, the State’s motive evidence is thinner. There is circumstantial

evidence that Heidi did not know about the impending foreclosure, namely that she had not

talked to her parents about it, had not taken off work the day of the foreclosure, and had

C/D-18
not begun packing up the house. But there was additional testimony that suggested she did

know about the foreclosure and was merely embarrassed to discuss it publicly, given that

all of the foreclosure documents were addressed to both her and Firkus and that financial

documents were scattered throughout the house where she could easily see them. 14 There

was also significant testimony at trial about the strength and love of what was presumably

a happy relationship. See Berndt, 392 N.W.2d at 881 (rejecting the State’s narrative that

the appellant wanted to kill his children and noting “[t]he uncontradicted evidence was that

appellant loved [his children] and enjoyed their company”).

Although Heidi’s knowledge of their financial situation is unclear, it does not rise

to the level of a clear breakdown in the marriage as we have seen in cases like Hughes,

where the defendant and his victim were actively separated and fighting over child

custody. 15 See 749 N.W.2d at 314. And unlike in Race, where the defendant took out life

insurance policies in his wife’s name several months before her murder and used the money

to pay off his personal debts, 383 N.W.2d at 661, Firkus signed over Heidi’s life insurance

money—a policy that she was enrolled in automatically at her job—to her parents, despite

14
The concurrence argues that it is not a circumstance proved that Heidi had seen the
foreclosure documents scattered around the house. I agree that the evidence introduced at
trial is insufficient to determine whether she actually saw the documents, just as it is
insufficient to know whether she actually saw an intruder when she called 911 on that
fateful morning. But there is ample testimony from the investigating officers at trial that
there were “dozens and dozens of foreclosure documents” easily accessible in their home
office.
15
In Hughes, there was also substantial other direct and circumstantial evidence
indicating the defendant’s involvement in the killing, including neighbors’ observations
that the defendant fled the house within minutes of hearing gunshots rather than rendering
aid to his wife. 749 N.W.2d at 315.

C/D-19
his indisputably dire financial situation. The concurrence says that the “outward

appearance of happiness” does not cancel out the fact that Firkus was hiding information

from Heidi that could shatter their marriage. But thousands, if not millions, of American

couples have experienced dire financial circumstances, up to and including home

foreclosure. The vast majority of those people do not commit murder in order to avoid

their financial obligations. And in Berndt, we explicitly rejected the proposed theory that

the defendant had a financial motive to kill his wife because he was not a named beneficiary

of his deceased wife’s life insurance policy and stood to gain no other financial benefits.

See 392 N.W.2d at 879. Here, the State could not identify any benefit that Firkus would

receive if his wife was murdered.

As with the other evidence in this case, the State’s motive evidence provides two

conflicting but reasonable inferences arising from the circumstances proved: one where

Firkus hid the foreclosure from Heidi and felt it necessary to kill her to prevent her from

finding out about his financial mismanagement, and one where Firkus loved Heidi and told

her about the situation, which both of them were too embarrassed to discuss publicly. We

have explicitly rejected “plac[ing] undue emphasis on evidence of motive,” particularly in

lieu of looking at the other evidence in the record. Bernhardt, 684 N.W.2d at 479. The

circumstances proved in this case can give rise to reasonable scenarios both consistent and

inconsistent with guilt. Without more substantial evidence establishing that Firkus had a

strong desire to kill Heidi, that killing her would help him avoid foreclosure, or at the very

least that their marriage was breaking down, I cannot countenance the concurrence’s

C/D-20
reliance on the State’s thin motive evidence to assert that the only reasonable inference is

that Firkus killed Heidi.

In sum, the standard in circumstantial evidence cases is that “[t]he evidence [must]

form a complete chain leading so directly to appellant’s guilt as to exclude beyond a

reasonable doubt any rational hypothesis except that of his guilt.” See Scharmer,

501 N.W.2d at 622 (emphasis added). The State established one complete chain of

inferences: that there was no intruder and Firkus shot Heidi in the back. But that chain is

easily broken by inferring a different conclusion from the same set of circumstances

proved. If a reasonable fact-finder were to infer, for example, that Firkus was at the top of

the stairs, then it was also reasonable that he could hear someone fiddling with the

doorknob, and it was further reasonable for him to assume that someone was trying to break

into his home. Or if the reasonable fact-finder infers that the fingerprint on the main door

that did not match Firkus or Heidi belonged to another unknown person, then it is also

reasonable to conclude that the intruder left that fingerprint there. Or if the reasonable

fact-finder considers Firkus’s thigh wound, the FBI ballistics model, and the testimony of

medical personnel at trial, then they may equally reasonably infer that the wound was

self-inflicted as that it was caused by a struggle.

When faced with a binary set of choices, choosing one or the other of two

circumstances proved is sufficient to break the State’s chain of guilt. Instead of evaluating

the evidence as a whole to determine whether there is another rational theory inconsistent

with guilt, the court “turns a blind eye to any rational inference other than guilt.”

Bernhardt, 684 N.W.2d at 479. That is not the law, and the court abdicates the

C/D-21
responsibility it has accepted for nearly a century to be a check on juries in circumstantial

evidence cases. See Johnson, 217 N.W. at 684. As this court succinctly stated in

Bernhardt, “[i]f our standard on circumstantial evidence means anything, it means that

appellant cannot be convicted on this record that does not exclude other rational

hypotheses.” 684 N.W.2d at 479.

Because I disagree with the court’s articulation of the circumstantial-evidence test,

and conclude that the State did not prove beyond a reasonable doubt that there was no other

reasonable hypothesis inconsistent with guilt, I respectfully dissent as to Part I of the

court’s opinion.

C/D-22