*281David Newbern, Justice.

This child abuse case resulted in the appellant’s conviction of first degree murder. The sole issue on appeal is whether the state’s evidence was sufficient to sustain the conviction. We hold there was no evidence of the “. . . premeditated and deliberated purpose of causing the death of another person . . .” required for conviction of first degree murder by Ark. Stat. Ann. § 41-1502(l)(b) (Repl. 1977). However, we find the evidence was sufficient to sustain a conviction of second degree murder, described in Ark. Stat. Ann. § 41-1503(l)(c) (Repl. 1977), as the appellant was shown to have caused his son’s death by delivering a blow to his abdomen or chest “. . . with the purpose of causing serious physical injury. . . .” The conviction is thus modified from one of first degree murder to one of second degree murder and affirmed.

The facts of this case are as heart-rending as any we are likely to see. The appellant is six feet two inches tall and weighs 300 pounds. His son, Ronnie Midgett, Jr., was eight years old and weighed between thirty-eight and forty-five pounds. The evidence showed that Ronnie Jr. had been abused by brutal beating over a substantial period of time. Typically, as in other child abuse cases, the bruises had been noticed by school personnel, and a school counselor as well as a SCAN worker had gone to the Midgett home to inquire. Ronnie Jr. would not say how he had obtained the bruises or why he was so lethargic at school except to blame it all, vaguely, on a rough playing little brother. He did not even complain to his siblings about the treatment he was receiving from the appellant. His mother, the wife of the appellant, was not living in the home. The other children apparently were not being physically abused by the appellant.

Ronnie Jr.’s sister, Sherry, aged ten, testified that on the Saturday preceding the Wednesday of Ronnie Jr.’s death their father, the appellant, was drinking whiskey (two to three quarts that day) and beating on Ronnie Jr. She testified that the appellant would “bundle up his fist” and hit Ronnie Jr. in the stomach and in the back. On direct examination she said that she had not previously seen the appellant beat Ronnie Jr., but she had seen the appellant choke him for no particular reason on Sunday nights after she and Ronnie Jr. returned from church. On cross-examination, Sherry testified that Ronnie Jr. had lied and her father was, on that Saturday, trying to get him to tell the truth. *282She said the bruises on Ronnie Jr.’s body noticed over the preceding six months had been caused by the appellant. She said the beating administered on the Saturday in question consisted of four blows, two to the stomach and two to the back.

On the Wednesday Ronnie Jr. died, the appellant appeared at a hospital carrying the body. He told hospital personnel something was wrong with the child. An autopsy was performed, and it showed Ronnie Jr. was a very poorly nourished and underdeveloped eight-year-old. There were recently caused bruises on the lips, center of the chest plate, and forehead as well as on the back part of the lateral chest wall, the soft tissue near the spine, and the buttocks. There was discoloration of the abdominal wall and prominent bruising on the palms of the hands. Older bruises were found on the right temple, under the chin, and on the left mandible. Recent as well as older, healed, rib fractures were found.

The conclusion of the medical examiner who performed the autopsy was that Ronnie Jr. died as the result of intra-abdominal hemorrhage caused by a blunt force trauma consistent with having been delivered by a human fist. The appellant argues that in spite of all this evidence of child abuse, there is no evidence that he killed Ronnie Jr. having premeditated and deliberated causing his death. We must agree.

It is true that premeditation and deliberation may be found on the basis of circumstantial evidence. That was the holding in House v. State, 230 Ark. 622, 324 S.W.2d 112 (1959), where the evidence showed a twenty-four-year-old man killed a nineteen-year-old woman with whom he was attempting to have sexual intercourse. The evidence showed a protracted fight after which the appellant dumped the body in a water-filled ditch not knowing, according to House’s testimony, whether she was dead or alive. Although it is not spelled out, presumably the rationale of the opinion was that Hoiise had time to premeditate during the fight and there was substantial evidence he intended the death of the victim when he left her in the water. Our only citation of authority on the point of showing premeditation and deliberation by circumstantial evidence in that case was Weldon v. State, 168 Ark. 534, 270 S.W. 968 (1925), where we said:

The very manner in which the deadly weapons were used *283was sufficient to justify the jury in finding that whoever killed Jones used the weapons with a deliberate purpose to kill. Jones’ body was perforated three times through the center with bullets from a pistol or rifle, and was also horribly mutilated with a knife. The manner, therefore, in which these deadly weapons were used tended to show that the death of Jones was the result of premeditation and deliberation.

While a fist may be a deadly weapon, the evidence of the use of the fist in this case is not comparable to the evidence in House v. State, supra, and Weldon v. State, supra, where there was some substantial evidence consisting of other circumstances that the appellant who dumped the apparently immobile body in the water and walked away and the appellant who wielded the deadly weapons intended and premeditated that death occur. Nor do we have in this case evidence of any remark made or other demonstration that the appellant was abusing his son in the hope that he eventually would die.

The annotation at 89 A.L.R. 2d 396 (1963) deals with the subject of crimes resulting from excessive punishment of children. While some of the cases cited are ones in which a parent or step-parent flew into a one-time rage and killed the child, others are plain child abuse syndrome cases like the one before us now. None of them, with one exception, resulted in affirmance of a first degree murder conviction. Several were decisions in which first degree murder convictions were set aside for lack of evidence of premeditation and deliberation. See, e.g., People v. Ingraham, 232 N.Y. 245, 133 N.E. 575 (1921); Pannill v. Commonwealth, 185 Va. 244, 38 S.W.2d 457 (1946). The case cited in the annotation in which a first degree murder conviction was affirmed is Morris v. State, 270 Ind. 245, 384 N.E.2d 1022 (1979). There the appellant was left alone for about fifteen minutes with his five-month-old baby. When the child’s mother returned to their home she found the baby had been burned severely on one side. About a month later, the appellant and his wife were engaged in an argument when the baby began to whine. The appellant laid the baby on the floor, began hitting the baby in the face and then hit the baby’s head on the floor, causing the baby’s death. At the time of the offense, the Indiana law required malice, purpose, and premeditation to convict of first degree murder. In discussing the *284premeditation requirement, the court said only:

Premeditation which also may be inferred from the facts and circumstances surrounding the killing, need not long be deliberated upon, but may occur merely an instant before the act. [Citation omitted.] It is clear from the facts adduced at trial regarding the burning and beating of the child that the jury could well have inferred that his killing was perpetrated purposely and with premeditated malice. [384 N.E.2d at 1024]

No explanation is given for the quantum leap from “the facts,” horrible as they were, to the inference of premeditation. We made the same error in Burnett v. State, 287 Ark. 158, 697 S.W.2d 95 (1985), another child abuse case in which the facts were particularly repugnant, where we said:

Premeditation, deliberation and intent may be inferred from the circumstances of the case, such as the weapon used and the nature, extent and location of the wounds inflicted . . . . [T]he weapon used was a fist which struck the abdomen with such force as to rupture the colon. The child sustained fingernail scratches, four broken ribs, and other internal damage, as well as numerous bruises due to blows with a fist all over his body. The required mental state for first degree murder can be inferred from the evidence of abuse, which is substantial. [287 Ark. at 162-163, 697 S.W.2d at 98]

The problem with these cases is that they give no reason, like the reasons found in House v. State, supra, and Weldon v. State, supra, to make the inference of premeditation and deliberation.

In Simmons v. State, 227 Ark. 1109, 305 S.W.2d 119 (1957), the appellant was antagonized more than once by his victims. After the first time he went home and got his shotgun to use, he said, for hunting squirrels. We modified the conviction from first degree murder to second degree murder, noting that the appellant had opportunities to kill the victims after he had obtained his weapon but before he shot them. His having let those opportunities pass negated premeditation and deliberation. We said:

There is no testimony of any witness, aside from the *285testimony of appellant in open court and his written confession, from which the jury could have found the existence of premeditation and deliberation. Neither do we find any circumstance which amounts to substantial evidence upon which a finding of premeditation and deliberation could be based. Consequently we are led to conclude that the jury must have resorted to speculation rather than substantial evidence in arriving at a verdict of murder in the first degree. [227 Ark. at 110-111, 305 S.W.2d at 120]

The appellant argues, and we must agree, that in a case of child abuse of long duration the jury could well infer that the perpetrator comes not to expect death of the child from his action, but rather that the child will live so that the abuse may be administered again and again. Had the appellant planned his son’s death, he could have accomplished it in a previous beating.

In this case the evidence might possibly support the inference that the blows which proved fatal to Ronnie Jr. could have been struck with the intent to cause his death developed in a drunken, misguided, and overheated attempt at disciplining him for not having told the truth. Even if we were to conclude there was substantial evidence from which the jury could fairly have found the appellant intended to cause Ronnie Jr.’s death in a drunken disciplinary beating on that Saturday, there would still be no evidence whatever of a premeditated and deliberated killing.

In Ford v. State, 276 Ark. 98, 633 S.W.2d 3, cert. den. 459 U.S. 1022 (1980), we held that to show the appellant acted with a premeditated and deliberated purpose, the state must prove that he (1) had the conscious object to cause death, (2) formed that intention before acting, and (3) weighed in his mind the consequences of a course of conduct, as distinguished from acting upon sudden impulse without the exercise of reasoning power. Viewing the evidence most favorable to the appellee, the circumstances of this case are not substantial evidence the appellant did (2) and (3), as opposed to acting on impulse or with no conscious object of causing death. The jury was thus forced to resort to speculation on these important elements.

A clear exposition of the premeditation and deliberation requirement which separates first degree from second degree *286murder is found in 2 W. LaFave and A. Scott, Jr., Substantive Criminal Law § 7.7 (1986):

Almost all American jurisdictions which divide murder into degrees include the following two murder situations in the category of first degree murder: (1) intent-to-kill murder where there exists (in addition to the intent to kill) the elements of premeditation and deliberation, and (2) felony murder where the felony in question is one of five or six listed felonies, generally including rape, robbery, kidnapping, arson and burglary. Some states instead or in addition have other kinds of first degree murder.

(a) Premeditated, Deliberate, Intentional Killing. To be guilty of this form of first degree murder the defendant must not only intend to kill but in addition he must premeditate the killing and deliberate about it. It is not easy to give a meaningful definition of the words “premeditate” and “deliberate” as they are used in connection with first degree murder. Perhaps the best that can be said of “deliberation” is that it requires a cool mind that is capable of reflection, and of “premeditation” that it requires that the one with the cool mind did in fact reflect, at least for a short period of time before his act of killing.

It is often said that premeditation and deliberation require only a “brief moment of thought” or a “matter of seconds,” and convictions for first degree murder have frequently been affirmed where such short periods of time were involved. The better view, however, is that to “speak of premeditation and deliberation which are instantaneous, or which take no appreciable time,. . . destroys the statutory distinction between first and second degree murder,” and (in much the same fashion that the felony-murder rule is being increasingly limited) this view is growing in popularity. This is not to say, however, that premeditation and deliberation cannot exist when the act of killing follows immediately after the formation of the intent. The intention may be finally formed only as a conclusion of prior premeditation and deliberation, while in other cases the intention may be formed without prior thought so that premeditation and deliberation occurs only *287with the passage of additional time for “further thought, and a turning over in the mind.” [Footnotes omitted.]

The evidence in this case supports only the conclusion that the appellant intended not to kill his son but to further abuse him or that his intent, if it was to kill the child, was developed in a drunken, heated, rage while disciplining the child. Neither of those supports a finding of premeditation or deliberation.

Perhaps because they wish to punish more severely child abusers who kill their children, other states’ legislatures have created laws permitting them to go beyond second degree murder. For example, Illinois has made aggravated battery one of the felonies qualifying for “felony murder,” and a child abuser can be convicted of murder if the child dies as a result of aggravated battery. See People v. Ray, 399 N.E.2d 977 (Ill. App. 1979). Georgia makes “cruelty to children” a felony, and homicide in the course of cruelty to children is “felony murder.” See Bethea v. State, 304 S.E. 2d 713 (Ga. 1983). Idaho has made murder by torture a first degree offense, regardless of intent of the perpetrator to kill the victim, and the offense is punishable by the death penalty. See State v. Stuart, 715 P.2d 833 (Idaho 1985). California has also adopted a murder by torture statute making the offense murder in the first degree without regard to the intent to kill. See People v. Demond, 59 Cal. App. 3d 574, 130 Cal.Rptr. 590 (1976). Cf People v. Steger, 128 Cal. Rptr. 161, 546 P.2d 665 (1976), in which the California Supreme Court held that the person accused of torture murder in the first degree must be shown to have had a premeditated intent to inflict extreme and prolonged pain in order to be convicted.

All of this goes to show that there remains a difference between first and second degree murder, not only under our statute, but generally. Unless our law is changed to permit conviction of first degree murder for something like child abuse or torture resulting in death, our duty is to give those accused of first degree murder the benefit of the requirement that they be shown by substantial evidence to have premeditated and deliberated the killing, no matter how heinous the facts may otherwise be. We understand and appreciate the state’s citation of Burnett v. State, supra, but, to the extent it is inconsistent with this opinion, we must overrule it.

*288The dissenting opinion begins by stating the majority concludes that one who starves and beats a child to death cannot be convicted of murder. That is not so, as we are affirming the conviction of murder; we are, however, reducing it to second degree murder. The dissenting opinion’s conclusion that the appellant starved Ronnie Jr., must be based solely on the child’s underdeveloped condition which could, presumably, have been caused by any number of physical malfunctions. There is no evidence the appellant starved the child. The dissenting opinion says it is for the jury to determine the degree of murder of which the appellant is guilty. That is true so long as there is substantial evidence to support the jury’s choice. The point of this opinion is to note that there was no evidence of premeditation or deliberation which are required elements of the crime of first degree murder. The dissenting opinion cites two child abuse cases in which first degree murder convictions have been affirmed. One is Morris v. State, supra, with which we dealt earlier in this opinion. The other, is Lindsey v. State, 501 S.W.2d 647 (Tex. Crim. App. 1973), in which the opinion does not say the conviction was for first degree murder. In fact, the issue there was whether the killing occurred with “intent and malice” which are obviously not the same as premeditation and deliberation.

In this case we have no difficulty with reducing the sentence to the maximum for second degree murder. Dixon v. State, 260 Ark. 857, 545 S.W.2d 606 (1977). The jury gave the appellant a sentence of forty years imprisonment which was the maximum for first degree murder, and we reduce that to twenty years which is the maximum imprisonment for second degree murder. Just as walking away from the victim in the water-filled ditch in House v. State, supra, after a protracted fight, and the “overkill” and mutilation of the body in Weldon v. State, supra, were circumstances creating substantial evidence of premeditation and deliberation, the obvious effect the beatings were having on Ronnie Jr. and his emaciated condition when the final beating occurred are circumstances constituting substantial evidence that the appellant’s purpose was to cause serious physical injury, and that he caused his death in the process. That is second degree murder, § 41-1503(l)(c). Therefore, we reduce the appellant’s sentence to imprisonment for twenty years.

Affirmed as modified.

*289Hickman, Hays, and Glaze, JJ., dissent.

Darrell Hickman, Justice,

dissenting. Simply put, if a parent deliberately starves and beats a child to death, he cannot be convicted of the child’s murder. In reaching this decision, the majority overrules a previous unanimous decision and substitutes its judgment for that of the jury. The majority has decided it cannot come to grips with the question of the battered child who dies as a result of deliberate, methodical, intentional and severe abuse. A death caused by such acts is murder by any legal standard, and that fact cannot be changed—not even by the majority. The degree of murder committed is for the jury to decide—not us.

Convictions for murder resulting from child abuse have become more common in our courts. That is probably because such cases are being reported more often and prosecutors are more apt to seek retribution.

The decision of what charge to file in a homicide case rests with the prosecuting attorney. He has the duty to prove the charge. The decision of whether the state has proved the crime rests with the jury. Our role is only to determine if substantial evidence exists to support the verdict.

Sometimes the facts may warrant a charge of second degree murder. We have affirmed convictions for second degree murder in two such cases. Boone v. State, 282 Ark. 274, 668 S.W.2d 17 (1984); Limber v. State, 264 Ark. 479, 572 S.W.2d 402 (1978).

Whether the particular acts of child abuse amount to first degree murder depend on the particular facts and circumstances in each case. Just as in any other murder case, the state must prove each element of the crime. For a first degree murder conviction, the state must prove premeditation and deliberation.

We have never held motive relevant to murder, nor do we even try to look into the warped minds that commit murder to make their acts rational. Parker v. State, 290 Ark. 158, 717 S.W.2d 800 (1986). Consequently, circumstantial evidence usually plays a strong part in determining intent in any murder case.

In this case the majority, with clairvoyance, decides that this parent did not intend to kill his child, but rather to keep him alive *290for further abuse. This is not a child neglect case. The state proved Midgett starved the boy, choked him, and struck him several times in the stomach and back. The jury could easily conclude that such repeated treatment was intended to kill the child.

In Burnett v. State, supra, the state chose to seek a first degree murder conviction. The child was killed in an extremely horrible way. He was malnourished and dehydrated, bruises on his face and upper and lower extremities, four broken ribs, a ruptured colon, and abrasions. His life was made intolerable and insufferable until at last a blow killed him. The parents, who could not have been unaware or innocent, were found guilty of killing him, which they did. We unanimously upheld that jury verdict. It was no “quantum leap” on our part (whatever that means), just a decision based on the facts and the law. The majority unanimously joined in the Burnett decision.

The facts in this case are substantial to support a first degree murder conviction. The defendant was in charge of three small children. The victim was eight years old and had been starved; he weighed only 38 pounds at the time of his death. He had multiple bruises and abrasions. The cause of death was an internal hemorrhage due to blunt force trauma. His body was black and blue from repeated blows. The victim’s sister testified she saw the defendant, a 30 year old man, 6’2” tall, weighing 300 pounds, repeatedly strike the victim in the stomach and back with his fist. One time he choked the child.

The majority is saying that as a matter of law a parent cannot be guilty of intentionally killing a child by such deliberate acts. Why not? Is it because it is inconceivable to rational people that a parent would intend to kill his own child? Evidently, this is the majority’s conclusion, because they hold the intention of Midgett was to keep him alive for further abuse, not kill him. How does the majority know that? How do we ever know the actual or subliminal intent of a defendant? “If the act appellant intended was criminal, then the law holds him accountable, even though such result was not intended.” Hankins v. State, 206 Ark. 881, 178 S.W.2d 56 (1944); see also Black v. State, 215 Ark. 618, 222 S.W.2d 816 (1949). There is no difference so far as the law is concerned in this case than in any other murder case. It is simply a question of proof. This parent killed his own child, and the *291majority cannot accept the fact that he intended to do just that.

Undoubtedly, the majority could accept it if the child were murdered with a bullet or a knife; but they cannot accept the fact, and it is a fact, that this defendant beat and starved his own child to death. His course of conduct could not have been negligent or unintentional.

Other states have not hesitated to uphold a conviction for first degree murder in such cases. Morris v. State, 384 N.E.2d 1022 (Ind. 1979); Lindsey v. State, 501 S.W.2d 647 (Tex. 1973). The fact that some states (California and Idaho) have passed a murder by torture statute is irrelevant. Those statutes may make it easier to prosecute child murderers, but they do not replace or intend to replace the law of murder. Whether murder exists is a question of the facts—not the method. The majority spends a good deal of effort laboring over the words “premeditation and deliberation,” ignoring what the defendant did. Oliver Wendell Holmes said: “We must think things not words . . .” Holmes, “Law in Science and Science in Law,” Collected Legal Papers, p. 238 (1921). If what Midgett did was deliberate and intentional, and that is not disputed, and he killed the child, a jury can find first degree murder.

I cannot fathom how this father could have done what he did; but it is not my place to sit in judgment of his mental state, nor allow my human feelings to color my judgment of his accountability to the law. The law has an objective standard of accountability for all who take human life. If one does certain acts and the result is murder, one must pay. The jury found Midgett guilty and, according to the law, there is substantial evidence to support that verdict. That should end the matter for us. He is guilty of first degree murder in the eyes of the law. His moral crime as a father is another matter, and it is not for us to speculate why he did it.

I would affirm the judgment.

Hays and Glaze, JJ., join in the dissent.

*291-ASupplemental Opinion on Denial of Rehearing

June 29, 1987

731 S.W.2d 774

1. Appeal & error — no basis for granting rehearing. — Where appellant only reargues a question raised on appeal he presents no basis for granting rehearing. [Ark. Sup. Ct. R. 20(g).]

2. Criminal law — first degree murder. — Ark. Stat. Ann. § 41-1502(3) (Supp. 1985) provides that first degree murder is a class Y felony, and a person convicted of a class Y felony may be sentenced to a term of not less than ten years and not more than forty years, or life.

3. Appeal & error — insufficient proof to support jury’s VERDICT OF A HIGHER OFFENSE — APPELLATE COURT MAY REDUCE sentence. — If the evidence proves insufficient to support a jury’s verdict of a higher offense, the trial court may sentence the defendant for a lesser included offense where the evidence clearly shows the commission of the latter and the appellate court, in its discretion, may reduce the sentence to that prescribed for the lesser offense.

Petition for Rehearing; denied.

David Newbern, Justice.

The appellant only reargues the question whether there was evidence of premeditation and deliberation and thus presents no basis for granting rehearing. Arkansas Supreme Court and Court of Appeals Rule 20(g). However, both parties have pointed out an error of law in our opinion. We stated that the maximum sentence for first degree murder is forty years imprisonment. We should have said the maximum term of years is forty. The statute provides first degree murder is a class Y felony, Ark. Stat. Ann. § 41-1502(3) (Supp. 1985), and a person convicted of a class Y felony may be sentenced to a term of not less than ten years and not more than forty years, or life. The jury thus sentenced the appellant to the maximum determinate sentence for first degree murder, although it was not the ultimate maximum sentence, i.e., life imprisonment.

The appellant argues, in his response to the appellee’s request for rehearing, that we should grant a new trial so that a jury may set the sentence for second degree murder. We decline to do so, as we regard the maximum sentence to be supported by the evidence in the record, and as we said in Collins v. State, 261

*291-BArk. 195, 548 S.W.2d 106 (1977):

If the evidence proves insufficient to support a jury’s verdict of a higher offense, the trial court may sentence the defendant for a lesser included offense where the evidence clearly shows the commission of the latter (and this court, in its discretion, may reduce the sentence to that prescribed for the lesser offense). Caton v. State, 252 Ark. 420, 479 S.W.2d 537 [1972]. This rule applies in murder cases as well as for other felonies. Simpson v. State, 56 Ark. 8, 19 S.W. 99 [1892]. [261 Ark. at 209, 548 S.W.2d at 114.]

The sentence has been reduced to one within the range prescribed for second degree murder, and we find that sentence to be justified by the evidence in the record.

Rehearing denied.

Hickman, Hays, and Glaze, JJ., would grant.

MEYER, Justice.

Defendant was convicted of the first-degree murder of his father, Clyde Forrest. The State having stipulated before trial to the absence of any statutory aggravating factors under N.C.G.S. § 15A-2000, the case was tried as a noncapital case, and defendant was sentenced accordingly to life imprisonment. In his appeal to this Court, defendant brings forward three assignments of error relative to the guilt-innocence phase of his trial. Having considered the entire record and each of these assignments in turn, we find no error in defendant’s trial. We therefore leave undisturbed defendant’s conviction and life sentence.

The facts of this case are essentially uncontested, and the evidence presented at trial tended to show the following series of events. On 22 December 1985, defendant John Forrest admitted his critically ill father, Clyde Forrest, Sr., to Moore Memorial Hospital. Defendant’s father, who had previously been hospitalized, was suffering from numerous serious ailments, including severe heart disease, hypertension, a thoracic aneurysm, numerous pulmonary emboli, and a peptic ulcer. By the morning of 23 December 1985, his medical condition was determined to be unbeatable and terminal. Accordingly, he was classified as “No Code,” meaning that no extraordinary measures would be used to save his life, and he was moved to a more comfortable room.

On 24 December 1985, defendant went to the hospital to visit his ailing father. No other family members were present in his father’s room when he arrived. While one of the nurse’s assistants was tending to his father, defendant told her, “There is no need in doing that. He’s dying.” She responded, “Well, I think he’s better.” The nurse’s assistant noticed that defendant was sniffing as though crying and that he kept his hand in his pocket during their conversation. She subsequently went to get the nurse.

When the nurse’s assistant returned with the nurse, defendant once again stated his belief that his father was dying. The *189nurse tried to comfort defendant, telling him, “I don’t think your father is as sick as you think he is.” Defendant, very upset, responded, “Go to hell. I’ve been taking care of him for years. I’ll take care of him.” Defendant was then left alone in the room with his father.

Alone at his father’s bedside, defendant began to cry and to tell his father how much he loved him. His father began to cough, emitting a gurgling and rattling noise. Extremely upset, defendant pulled a small pistol from his pants pocket, put it to his father’s temple, and fired. He subsequently fired three more times and walked out into the hospital corridor, dropping the gun to the floor just outside his father’s room.

Following the shooting, defendant, who was crying and upset, neither ran nor threatened anyone. Moreover, he never denied shooting his father and talked openly with law enforcement officials. Specifically, defendant made the following oral statements: “You can’t do anything to him now. He’s out of his suffering.” “I killed my daddy.” “He won’t have to suffer anymore.” “I know they can burn me for it, but my dad will not have to suffer anymore.” “I know the doctors couldn’t do it, but I could.” “I promised my dad I wouldn’t let him suffer.”

Defendant’s father was found in his hospital bed, with several raised spots and blood on the right side of his head. Blood and brain tissue were found on the bed, the floor, and the wall. Though defendant’s father had been near death as a result of his medical condition, the exact cause of the deceased’s death was determined to be the four point-blank bullet wounds to his head. Defendant’s pistol was a single-action .22-calibre five-shot revolver. The weapon, which had to be cocked each time it was fired, contained four empty shells and one live round.

At the close of the evidence, defendant’s case was submitted to the jury for one of four possible verdicts: first-degree murder, second-degree murder, voluntary manslaughter, or not guilty. After a lengthy deliberation, the jury found defendant guilty of first-degree murder. Judge Cornelius accordingly sentenced defendant to the mandatory life term.

Defendant assigns three specific errors relative to his conviction at trial: first, that the trial court committed reversible error *190in its instruction to the jury concerning the issue of malice; second, that the trial court committed reversible error in its submission of the first-degree murder charge to the jury because there was insufficient evidence of premeditation and deliberation; third and finally, that the trial court committed reversible error when, during jury deliberation, it inquired into the jury’s numerical division and subsequently instructed the jury about deliberating toward a verdict. We deal with each assignment of error in turn.

I.

[1] In his first assignment of error, defendant asserts that the trial court committed reversible error in its instruction to the jury concerning the issue of malice. Defendant makes three specific arguments in support of his position on this assignment of error. First, states defendant, the instruction permitting an inference of malice from the use of a deadly weapon on these particular facts constituted an impermissible shift of the burden of persuasion on the issue of malice to defendant. Second, continues defendant, the trial court erred in giving incomplete instructions on the element of malice and in thereby improperly suggesting that the mitigating evidence presented at trial neither negated malice nor showed heat of passion. Third, concludes defendant, the trial court erred more generally in giving instructions on malice which were simply erroneous and misleading. We find each of defendant’s arguments unpersuasive, and we therefore overrule this assignment of error.

On the issue of malice, the trial court consistently instructed the jury as follows:

Malice means not only hatred, ill-will or spite, as it is ordinarily understood; to be sure that’s malice. But it also means that condition of the mind that prompts a person to take the life of another intentionally, or to intentionally inflict serious bodily harm which proximately results in his death without just cause, excuse or justification.

If the State proves beyond a reasonable doubt that the defendant killed the victim with a deadly weapon, or intentionally inflicted a wound upon the victim with a deadly weapon that proximatley [sic] caused the victim’s death you may infer, first, that the killing was unlawful. Second, that it *191was done with malice. But you are not compelled to do so. You may consider this, along with all other facts and circumstances in determining whether the killing was unlawful and whether it was done with malice.

I charge that it is not a legal defense to the offense of murder if the defendant, John Forrest, at the time of the shooting believed his father, Clyde Forrest, to be terminally ill or in danger of immediate death. But you may consider such belief in determining whether the killing was done with malice.

It is this instruction to which defendant now assigns error.

Defendant first argues that, on the particular facts of this case, the trial court’s instruction permitting an inference of malice from the use of a deadly weapon improperly shifted the burden of persuasion on the issue of malice to defendant. Here, claims defendant, where the facts presented tended to show a distraught son who wanted merely to end his father’s suffering, the evidence in fact negated the element of malice. According to defendant, there was no rational connection here between the fact proved (intentional use of a dangerous weapon) and the fact inferred (malice). Therefore, concludes defendant, use of an inference under these circumstances was tantamount to shifting the burden of persuasion to defendant, because first, the jury was encouraged to draw the inference regardless of any other evidence presented, and second, it was told, in effect, that the inference could not be overcome — that the direct evidence was not a “legal defense.” We cannot agree.

The instruction employed by the trial court is in accord with the North Carolina Pattern Jury Instructions and with extensive North Carolina case law. See State v. Reynolds, 307 N.C. 184, 297 S.E. 2d 532 (1983); State v. Patterson, 297 N.C. 247, 254 S.E. 2d 604 (1979). Significantly, the trial court did not instruct the jury that malice should be presumed. On the contrary, the trial court instructed the jury that it “may infer” that the killing was unlawful and committed with malice, but that it was not compelled to do so. The trial court properly instructed the jury that it should consider this permissive inference along with all the other facts and circumstances, including defendant’s belief that his father was terminally ill or in danger of immediate death, in *192deciding whether the State had proven malice beyond a reasonable doubt. Defendant’s first argument therefore lacks merit.

[2] Defendant argues second that the trial court erred in giving incomplete instructions on the issue of malice, thereby improperly suggesting that any mitigating evidence presented did not negate malice or show heat of passion. While conceding that the instruction here was technically correct, defendant claims that it was nevertheless inadequate and misleading in that it failed to define what was meant by the phrase “just cause, excuse or justification.” According to defendant, there is abundant evidence in the record that, upon seeing his father at the hospital, he was overwhelmed by the futile, horrible suffering before him and that, in a highly emotional state, he killed to bring relief to the man he deeply loved. The jury instruction employed by the trial court, concludes defendant, because it did not instruct on heat of passion, for all intents and purposes precluded the jury from considering these critical facts in mitigation of the offense. We do not agree with defendant, and we hold that a heat of passion jury instruction on facts such as those of the case at bar is improper.

In essence, defendant asks this Court to hold that his extreme distress over his father’s suffering was adequate provocation, as in the “heat of passion” doctrine, to negate the malice element required for a murder conviction. Our Court has held on numerous occasions that, under certain circumstances, one who kills another human being in the “heat of passion,” produced by adequate provocation sufficient to negate malice, is guilty of manslaughter rather than murder. State v. Robbins, 309 N.C. 771, 309 S.E. 2d 188 (1983); State v. Jones, 299 N.C. 103, 261 S.E. 2d 1 (1979). A killing in the “heat of passion” on sudden and adequate provocation means a killing without premeditation under the influence of a sudden passion which renders the mind incapable of cool reflection. State v. Jones, 299 N.C. 103, 261 S.E. 2d 1; State v. Jennings, 276 N.C. 157, 171 S.E. 2d 447 (1970).

Significantly, our Court has narrowly construed the requirement under the “heat of passion” doctrine that provocation be adequate and reasonable. We have held, for example, that mere words or insulting language, no matter how abusive, can never be adequate provocation and can never reduce murder to manslaughter under the “heat of passion” doctrine. State v. McCray, 312 *193N.C. 519, 324 S.E. 2d 606 (1985); State v. Montague, 298 N.C. 752, 259 S.E. 2d 899 (1979). We have held as adequate provocation an assault or threatened assault, State v. Montague, 298 N.C. 752, 259 S.E. 2d 899; State v. Williams, 296 N.C. 693, 252 S.E. 2d 739 (1979), and the discovery of the deceased spouse and a paramour in the act of intercourse, State v. Ward, 286 N.C. 304, 210 S.E. 2d 407, vacated in part 428 U.S. 903, 49 L.Ed. 2d 1207 (1974).

We are unwilling to hold that, as in the case at bar, where defendant kills a loved one in order to end the deceased’s suffering, adequate provocation to negate malice is necessarily present. The “heat of passion” doctrine is meant to reduce murder to manslaughter when defendant kills without premeditation and deliberation and without malice, but rather under the influence of the heat of passion suddenly aroused which renders the mind temporarily incapable of cool reflection. State v. Jones, 299 N.C. 103, 261 S.E. 2d 1. Here, irrefutable proof of premeditation and deliberation is clearly present. This defendant, though clearly upset by his father’s condition, indicated by his actions and his statements that his crime was premeditated and deliberate.

The instruction employed by the trial court was correct, and we reject this second of defendant’s arguments that the jury instructions constitute reversible error.

[3] Defendant argues third that the trial court committed reversible error in giving instructions on the issue of malice which were erroneous and generally misleading. Defendant’s objection here is essentially a grammatical one and is directed at that portion of the jury instruction which reads as follows:

[Malice] also means that condition of the mind that prompts a person to take the life of another intentionally, . . . without just cause, excuse or justification.

The trial court, argues defendant, failed to explicitly and specifically qualify the particular definition of malice as “that condition of the mind that prompts a person to take the life of another intentionally” with the important phrase “without just cause, excuse or justification.” This, claims defendant, almost certainly led the jury to conclude that the intentional shooting alone required them to find malice, despite any evidence to the contrary. The *194trial court, adds defendant, should have defined malice in its instruction as follows:

That condition of the mind which prompts a person, without just cause, excuse or justification to take the life of another intentionally

or

to intentionally inflict serious bodily harm which proximately results in his death.

We do not agree, and we therefore decline defendant’s invitation to adopt a new jury instruction concerning the issue of malice. The instruction employed by the trial court is consistent with the North Carolina Pattern Jury Instructions and is the very instruction we have previously expressly approved on numerous occasions. State v. Reynolds, 307 N.C. 184, 297 S.E. 2d 532; State v. Patterson, 297 N.C. 247, 254 S.E. 2d 604. Moreover, the instruction used at trial is, on its face, in essence the same as that for which defendant argues. Defendant’s third argument in support of this assignment of error is without merit, and the assignment as a whole is hereby overruled.

II.

[4] In his second assignment of error, defendant asserts that the trial court committed reversible error in denying his motion for directed verdict as to the first-degree murder charge. Specifically, defendant argues that the trial court’s submission of the first-degree murder charge was improper because there was insufficient evidence of premeditation and deliberation presented at trial. We do not agree, and we therefore overrule defendant’s assignment of error.

We recently addressed this very issue in the case of State v. Jackson, 317 N.C. 1, 343 S.E. 2d 814 (1986). Our analysis of the relevant law in that case is instructive in the case at bar:

Before the issue of a defendant’s guilt may be submitted to the jury, the trial court must be satisfied that substantial evidence has been introduced tending to prove each essential element of the offense charged and that the defendant was the perpetrator. State v. Hamlet, 312 N.C. 162, 321 S.E. 2d 837 (1984); State v. Powell, 299 N.C. 95, 261 S.E. 2d 114 (1980). *195Substantial evidence must be existing and real, but need not exclude every reasonable hypothesis of innocence. State v. Williams, 308 N.C. 47, 301 S.E. 2d 335, cert. denied, 464 U.S. 865, 78 L.Ed. 2d 177, reh’g denied, 464 U.S. 1004, 78 L.Ed. 2d 704 (1983). In considering a motion to dismiss, the trial court must examine the evidence in the light most favorable to the State, and the State is entitled to every reasonable intendment and inference to be drawn therefrom. State v. Hamlet, 312 N.C. 162, 321 S.E. 2d 837; State v. Bright, 301 N.C. 243, 271 S.E. 2d 368 (1980). Contradictions and discrepancies in the evidence are for the jury to resolve and do not warrant dismissal. State v. Brown, 315 N.C. 40, 337 S.E. 2d 808 (1985); State v. Powell, 299 N.C. 95, 261 S.E. 2d 114.

First-degree murder is the intentional and unlawful killing of a human being with malice and with premeditation and deliberation. State v. Fleming, 296 N.C. 559, 251 S.E. 2d 430 (1979); N.C.G.S. § 14-17 (1981 and Cum. Supp. 1985). Premeditation means that the act was thought out beforehand for some length of time, however short, but no particular amount of time is necessary for the mental process of premeditation. State v. Brown, 315 N.C. 40, 337 S.E. 2d 808; State v. Myers, 299 N.C. 671, 263 S.E. 2d 768 (1980). Deliberation means an intent to kill, carried out in a cool state of blood, in furtherance of a fixed design for revenge or to accomplish an unlawful purpose and not under the influence of a violent passion, suddenly aroused by lawful or just cause or legal provocation. State v. Hamlet, 312 N.C. 162, 321 S.E. 2d 837; State v. Bush, 307 N.C. 152, 297 S.E. 2d 563 (1982). The phrase “cool state of blood” means that the defendant’s anger or emotion must not have been such as to overcome his reason. State v. Myers, 299 N.C. 671, 263 S.E. 2d 768.

Premeditation and deliberation relate to mental processes and ordinarily are not readily susceptible to proof by direct evidence. Instead, they usually must be proved by circumstantial evidence. State v. Buchanan, 287 N.C. 408, 215 S.E. 2d 80 (1975). Among other circumstances to be considered in determining whether a killing was with premeditation and deliberation are: (1) want of provocation on the part of the deceased; (2) the conduct and statements of the defendant before and after the killing; (3) threats and declarations of the *196defendant before and during the course of the occurrence giving rise to the death of the deceased; (4) ill-will or previous difficulty between the parties; (5) the dealing of lethal blows after the deceased has been felled and rendered helpless; and (6) evidence that the killing was done in a brutal manner. State v. Brown, 315 N.C. 40, 337 S.E. 2d 808; State v. Williams, 308 N.C. 47, 301 S.E. 2d 335, cert. denied, 464 U.S. 865, 78 L.Ed. 2d 117, reh’g denied, 464 U.S. 1004, 78 L.Ed. 2d 704. We have also held that the nature and number of the victim’s wounds is a circumstance from which premeditation and deliberation can be inferred. State v. Bullard, 312 N.C. 129, 322 S.E. 2d 370 (1984); State v. Brown, 306 N.C. 151, 293 S.E. 2d 569, cert. denied, 459 U.S. 1080, 74 L.Ed. 2d 642 (1982).

Jackson, 317 N.C. at 22-23, 343 S.E. 2d at 827.

As in Jackson, we hold in the present case that there was substantial evidence that the killing was premeditated and deliberate and that the trial court did not err in submitting to the jury the question of defendant’s guilt of first-degree murder based upon premeditation and deliberation. Here, many of the circumstances that we have held to establish a factual basis for a finding of premeditation and deliberation are present. It is clear, for example, that the seriously ill deceased did nothing to provoke defendant’s action. Moreover, the deceased was lying helpless in a hospital bed when defendant shot him four separate times. In addition, defendant’s revolver was a five-shot single-action gun which had to be cocked each time before it could be fired. Interestingly, although defendant testified that he always carried the gun in his job as a truck driver, he was not working on the day in question but carried the gun to the hospital nonetheless.

Most persuasive of all on the issue of premeditation and deliberation, however, are defendant’s own statements following the incident. Among other things, defendant stated that he had thought about putting his father out of his misery because he knew he was suffering. He stated further that he had promised his father that he would not let him suffer and that, though he did not think he could do it, he just could not stand to see his father suffer any more. These statements, together with the other circumstances mentioned above, make it clear that the trial court did not err in submitting to the jury the issue of first-*197degree murder based upon premeditation and deliberation. Accordingly, defendant’s second assignment of error is overruled.

III.

[5] In his third assignment of error, defendant asserts that the trial court committed reversible error when it inquired into the numerical division of the deliberating jury and when it subsequently instructed the jury about deliberating toward a verdict. Defendant claims that the trial court’s actions taken in context were sufficiently coercive of the jury as to deny him a fair trial. We have recently addressed this very issue in a similar case, and we simply do not agree.

During its deliberation at trial, the jury returned to the courtroom on several occasions with a specific question. On one such occasion, the exchange between the trial court and the jury proceeded as follows:

[COURT]: Mrs. Kelly, as Foreperson of the jury, you have submitted a question to the Court. You have indicated that you are unable at this time to come to a unanimous decision. You would like the Court to advise you. Is that your question?

Foreperson: Yes, sir.

COURT: Listen very carefully to what I ask you. I’m going to ask you the numerical division. I don’t want you to tell me which way; just tell me the division numberwise the way the jury is now constituted.

FOREPERSON: You mean in numbers?

COURT: Yes, ma’m [sic].

FOREPERSON: Eleven to one.

COURT: And has that number remained the same throughout the proceedings, or has it shifted from time to time?

FOREPERSON: No, sir. It has been constant.

COURT: Members of the jury, your Foreperson has indicated that you’ve been unable to reach a verdict at this particular point. The Court wants to emphasize the fact that it is your duty to do whatever you can to reach a verdict *198in this matter. You should reason the matter over together as reasonable men and women and to reconcile your differences if you can without the surrendering of your conscientious convictions. But no juror should surrender his honest conviction as to the weight or effect of the evidence solely because of the opinion of a fellow juror, or for the mere purpose of returning a verdict. The Court will now let. you return to the jury room to continue with your deliberations, and when you’ve reached a unanimous verdict — please see if you can reach a unanimous verdict. If you can please knock on the door.

It was this inquiry and instruction by the trial court to which defendant now assigns error.

The disposition of defendant’s assignment of error is controlled by our recent decision in the very similar case of State v. Bussey, 321 N.C. 92, 361 S.E. 2d 564 (1987). There, as here, defendant challenged the trial court’s inquiry into the numerical division of the deliberating jury and its instruction concerning deliberating further toward a verdict. In Bussey, we reaffirmed our holding in State v. Fowler, 312 N.C. 304, 322 S.E. 2d 389 (1984), that such cases are to be decided by employing a totality of the circumstances test. Bussey, 321 N.C. 92, 361 S.E. 2d 564. Therefore, the proper analysis here is whether, upon consideration of the totality of the circumstances, the inquiry and instruction of the trial court were unduly coercive.

As in Bussey, we hold here that, when viewed in the totality of the circumstances, the challenged inquiry and instruction were not coercive of the jury’s verdict. Our exact analysis in Bussey is appropriate in the case at bar as well:

The record shows that the presiding judge made it perfectly clear from the outset that he did not wish to be told whether the majority favored guilt or innocence. He was at all times respectful of the jury, never impugning its efforts or threatening it with being held for unreasonable periods of time to accomplish a unanimous verdict. The judge was confronted with a report of deadlock .... He properly exercised his discretion to hold the jurors to their duty to deliberate thoroughly together before concluding that they were indeed unable to agree. The judge’s additional instructions in re*199sponse to the first inquiry of the jury hew closely to the language of N.C.G.S. § 15A-1235. They are notable for the balance he achieved between recalling the jurors to their duty to deliberate fully and reminding them that their duty also required them to stand fast for their convictions after full reflection. Nor is there the slightest reference in his remarks to burdens on the administration of justice, to wasted court resources, or to the necessity of empanelling another jury in the event of a mistrial. The trial judge’s instructions and remarks were well within the rules established in State v. Fowler, 312 N.C. 304, 322 S.E. 2d 389, and State v. Alston, 294 N.C. 577, 243 S.E. 2d 354 (1978).

Id. at 97, 361 S.E. 2d at 567.

In oral argument, defendant placed particular emphasis upon the final two sentences of the trial court’s instruction to the jury, claiming that this amounted to pleading by the court that the jury reach a verdict. Specifically, the court concluded its instruction to the jury as follows:

The Court will now let you return to the jury room to continue with your deliberations, and when you’ve reached a unanimous verdict — please see if you can reach a unanimous verdict. If you can please knock on the door.

Assuming, arguendo, as defendant argues, that these two sentences, if taken out of context, might be considered of questionable propriety, we find that, in the context of the court’s total instruction and, in particular, of its admonishment to the jury that no juror should surrender any conscientious convictions, this passage is not coercive and does not constitute error in the court’s instructions.

In conclusion, having reviewed the record and each of defendant’s assignments of error, we find that defendant had a fair trial, free of prejudicial error. Accordingly, we leave undisturbed defendant’s conviction of the first-degree murder of Clyde Forrest and his sentence of life imprisonment.

No error.

*200Chief Justice Exum

dissenting.

Almost all would agree that someone who kills because of a desire to end a loved one’s physical suffering caused by an illness which is both terminal and incurable should not be deemed in law as culpable and deserving of the same punishment as one who kills because of unmitigated spite, hatred or ill will. Yet the Court’s decision in this case essentially says there is no legal distinction between the two kinds of killing. Our law of homicide should not be so roughly hewn as to be incapable of recognizing the difference. I believe there are legal principles which, when properly applied, draw the desirable distinction and that both the trial court and this Court have failed to recognize and apply them.

The difference, legally, between the two kinds of killings hinges on the element of malice, the former being without, and the latter with, malice. The absence of malice, however, does not mean the killing is justified or excused so as not to be unlawful; it means simply that the killing is mitigated so as not to be murder but manslaughter. Our cases have traditionally recognized the distinction between mitigation and excuse in the law of homicide. State v. Hankerson, 288 N.C. 632, 220 S.E. 2d 575 (1975), and cases therein cited and discussed.

The error in the trial court’s instructions stems from the failure to recognize this difference between mitigation and excuse. The trial court instructed that malice was “that condition of mind that prompts a person to take the life of another intentionally .. . without just cause, excuse or justification.” This instruction, correct insofar as it goes, is incomplete. The trial court should have added “and without mitigation.”

Failure to include circumstances in mitigation as capable of rebutting malice, in effect, precluded the jury from considering at all defendant’s reasons for killing his father on the issue of whether he acted with malice. The instructions were that only matters which excused the killing altogether were sufficient to rebut the element of malice! The trial court then told the jury that defendant’s reasons for killing his father would not excuse the killing, saying,

*201I charge that it is not a legal defense to the offense of murder if the defendant, John Forrest, at the time of the shooting believed his father, Clyde Forrest, to be terminally ill or in danger of immediate death.

Although the trial court followed this immediately with, “But you may consider such belief in determining whether the killing was done with malice,” he gave the jury no theory by which the circumstances might in law rebut the inference of malice which arose from the intentional killing with a deadly weapon. In essence this instruction was superfluous because the jury had already been told that only legal defenses, as opposed to circumstances in mitigation, could be considered on the issue of malice. At best the instructions were conflicting on the crucial element in the case. Ordinarily this kind of error calls for a new trial. State v. Parrish, 275 N.C. 69, 165 S.E. 2d 230 (1969).

The jury’s confusion concerning the malice instructions is revealed by their three requests that the trial court repeat them and the trial court’s finally submitting them to the jury in writing.

For this error in the trial court’s instructions, I vote to give defendant a new trial.

*142Opinion

KENNARD, J.

On January 26, 2001, two dogs owned by defendant Marjorie Knoller and her husband, codefendant Robert Noel, attacked and killed Diane Whipple in the hallway of an apartment building in San Francisco. Defendant Knoller was charged with second degree murder (Pen. Code, § 189)1 and involuntary manslaughter (§ 192, subd. (b)); codefendant Noel, who was not present at the time of the attack on Whipple, was charged with involuntary manslaughter but not murder. Both were also charged with owning a mischievous animal that caused the death of a human being, in violation of section 399.

After a change of venue to Los Angeles County, a jury convicted defendants on all counts. Both moved for a new trial. (See § 1181, subd. 6 [a trial court may grant a new trial when “the verdict or finding is contrary to law or evidence”].) The trial court denied Noel’s motion. It granted Knoller’s motion in part, giving her a new trial on the second degree murder charge, but denying her motion for a new trial on the other two crimes of which she was convicted (involuntary manslaughter and possession of a mischievous animal that causes death).

With respect to Knoller, whose conviction of second degree murder was based on a theory of implied malice, the trial court took the position that, to be guilty of that crime, Knoller must have known that her conduct involved a high probability of resulting in the death of another. Finding such awareness lacking, the trial court granted Knoller’s motion for a new trial on the second degree murder conviction.

The trial court sentenced both defendants to four years’ imprisonment, the maximum term for involuntary manslaughter (§ 193, subd. (b)), staying the sentences for the section 399 violations. Defendants appealed from their convictions, and the People appealed from the order granting Knoller a new trial on the murder count. The Court of Appeal consolidated the appeals.

The Court of Appeal reversed the trial court’s order granting Knoller a new trial on the second degree murder charge. It remanded the case to the trial court for reconsideration of the new trial motion in light of the Court of Appeal’s holding that implied malice can be based simply on a defendant’s conscious disregard of the risk of serious bodily injury to another. In all other respects, the Court of Appeal affirmed the convictions of both defendants.

*143Both defendants petitioned this court for review. We granted only Knoller’s petition, limiting review to two questions: “(1) Whether the mental state required for implied malice includes only conscious disregard for human life or can it be satisfied by an awareness that the act is likely to result in great bodily injury,”2 and “(2) Whether the trial court abused its discretion in granting Knoller’s motion for new trial under Penal Code section 1181 [, subdivision 6].”

With respect to the first issue, we reaffirm the test of implied malice we set out in People v. Phillips (1966) 64 Cal.2d 574 [51 Cal.Rptr. 225, 414 P.2d 353] and, as mentioned on page 152, post, reiterated in many later cases: Malice is implied when the killing is proximately caused by “ ‘an act, the natural consequences of which are dangerous to life, which act was deliberately performed by a person who knows that his conduct endangers the life of another and who acts with conscious disregard for life.’ ” (People v. Phillips, supra, at p. 587.) In short, implied malice requires a defendant’s awareness of engaging in conduct that endangers the life of another—no more, and no less.

Measured against that test, it becomes apparent that the Court of Appeal set the bar too low, permitting a conviction of second degree murder, based on a theory of implied malice, if the defendant knew his or her conduct risked causing death or serious bodily injury. But the trial court set the bar too high, ruling that implied malice requires a defendant’s awareness that his or her conduct had a high probability of resulting in death, and that granting defendant Knoller a new trial was justified because the prosecution did not charge codefendant Noel with murder. Because the trial court used an incorrect test of implied malice, and based its decision in part on an impermissible consideration, we conclude that it abused its discretion in granting Knoller a new trial on the second degree murder count. It is uncertain whether the court would have granted the new trial had it used correct legal standards. We therefore remand the matter to the Court of Appeal, and direct it to return the case to the trial court with directions to reconsider defendant Knoller’s new trial motion in light of the views set out in this opinion.

*144I. Facts and Proceedings

In 1998, Pelican Bay State Prison inmates Paul Schneider and Dale Bretches, both members of the Aryan Brotherhood prison gang, sought to engage in a business of buying, raising, and breeding Presa Canario dogs. This breed of dog tends to be very large, weighing over 100 pounds, and reaching over five feet tall when standing on its hind legs. A document found in defendants’ apartment describes the Presa Canario as “a gripping dog . . . [f] . . . always used and bred for combat and guard . . . [and] used extensively for fighting . . . .”

Prisoners Schneider and Bretches relied on outside contacts, including Brenda Storey and Janet Coumbs, to carry out their Presa Canario business. Schneider told Coumbs that she should raise the dogs.

As of May 1990, Coumbs possessed four such dogs, named Bane, Isis, Hera, and Fury. Hera and Fury broke out of their fenced yard and attacked Coumbs’s sheep. Hera killed at least one of the sheep and also a cat belonging to Coumbs’s daughter. Coumbs acknowledged that Bane ate his doghouse and may have joined Fury in killing a sheep.

Defendants Knoller and Noel, who were attorneys representing a prison guard at Pelican Bay State Prison, met inmate Schneider at the prison sometime in 1999. In October 1999, defendants filed a lawsuit on behalf of Brenda Storey against Coumbs over the ownership and custody of the four dogs. Coumbs decided not to contest the lawsuit and to turn the dogs over to defendants. Coumbs warned Knoller that the dogs had killed Coumbs’s sheep, but Knoller did not seem to care.

Defendant Knoller thereafter contacted Dr. Donald Martin, a veterinarian for 49 years, and on March 26, 2000, he examined and vaccinated the dogs. With his bill to Knoller, Dr. Martin included a letter, which said in part: “I would be professionally amiss [sic] if I did not mention the following, so that you can be prepared. These dogs are huge, approximately weighing in the neighborhood of 100 pounds each. They have had no training or discipline of any sort. They were a problem to even get to, let alone to vaccinate. You mentioned having a professional hauler gather them up and taking them. . . . Usually this would be done in crates, but I doubt one could get them into anything short of a livestock trailer, and if let loose they would have a battle, [f] To add to this, these animals would be a liability in any household, reminding me of the recent attack in Tehama County to a boy by large dogs. *145He lost his aim and disfigured his face. The historic romance of the warrior dog, the personal guard dog, the gaming dog, etc. may sound good but hardly fits into life today.” Knoller thanked Dr. Martin for the information and said she would pass it on to her client.

On April 1, 2000, both defendants and a professional dog handler took custody of the dogs from Coumbs. Bane then weighed 150 pounds and Hera 130 pounds. Coumbs told both defendants that she was worried about the dogs, that Hera and Fury should be shot, and that she was also concerned about Bane and Isis.

Hera remained for a short time at a kennel in San Mateo County while Bane was sent to a facility in Los Angeles County. Both defendants soon became concerned for the health of the two dogs. On April 30, 2000, defendants brought Hera to their sixth-floor apartment at 2398 Pacific Avenue in San Francisco. Bane arrived in September 2000. Codefendant Noel purchased dog licenses, registering himself and Knoller as the dogs’ owners.

A later search of defendants’ apartment showed that they frequently exchanged letters with Pelican Bay inmates Schneider and Bretches. Over 100 letters were sent and received between March and December 2000, apparently under the guise of attorney-client correspondence.3 In the letters, defendants discussed a commercial breeding operation, considering various names such as GuerraHund Kennels, Wardog, and finally settling on Dog-O-War. Prisoners Schneider and Bretches’s notes on a Web site for the business described Bane as “Wardog,” and “Bringer of Death: Ruin: Destruction.”

Between the time defendants Noel and Knoller brought the dogs to their sixth-floor apartment in San Francisco and the date of the fatal mauling of Diane Whipple on January 26, 2001, there were about 30 incidents of the two dogs being out of control or threatening humans and other dogs. Neighbors mentioned seeing the two dogs unattended on the sixth floor and running down the hall. Codefendant Noel’s letters to prisoner Schneider confirmed this, mentioning one incident when defendant Knoller had to let go of the two dogs as they broke from her grasp and ran to the end of the hall. Noel *146described how the dogs even pushed past him and “took off side by side down the hall toward the elevator in a celebratory stampede!! 240 lbs. of Presa wall to wall moving at top speed!!!” In a letter to inmate Schneider, defendant Knoller admitted not having the upper body strength to handle Bane and having trouble controlling Hera.

When neighbors complained to defendants Noel and Knoller about the two dogs, defendants responded callously, if at all. In one incident, neighbors Stephen and Aimee West were walking their dog in a nearby park when Hera attacked their dog and “latched on” to the dog’s snout. Noel was unable to separate the dogs, but Aimee threw her keys at Hera, startling Hera and causing Hera to release her grip on the Wests’ dog. On another day, Stephen West was walking his dog when he encountered Noel with Bane. Bane lunged toward West’s dog, but Noel managed to pull Bane back. When Stephen West next saw Noel, West suggested that Noel muzzle the dogs and talk to dog trainer Mario Montepeque about training them; Noel replied there was no need to do so. Defendants Knoller and Noel later encountered Montepeque, who advised defendants to have their dogs trained and to use a choke collar. Defendants disregarded this advice. On still another occasion, when dog walker Lynn Gaines was walking a dog, Gaines told Noel that he should put a muzzle on Bane; Noel called her a “bitch” and said the dog Gaines was walking was the problem.

There were also instances when defendants’ two dogs attacked or threatened people. David Moser, a fellow resident in the apartment building, slipped by defendants Knoller and Noel in the hallway only to have their dog Hera bite him on the “rear end.” When he exclaimed, “Your dog just bit me,” Noel replied, “Um, interesting.” Neither defendant apologized to Moser or reprimanded the dog. Another resident, Jill Cowen Davis, was eight months pregnant when one of the dogs, in the presence of both Knoller and Noel, suddenly growled and lunged toward her stomach with its mouth open and teeth bared. Noel jerked the dog by the leash, but he did not apologize to Davis. Postal carrier John Watanabe testified that both dogs, unleashed, had charged him. He said the dogs were in a “snarling frenzy” and he was “terrified for [his] life.” When he stepped behind his mail cart, the dogs went back to Knoller and Noel. On still another occasion, the two dogs lunged at a six-year-old boy walking to school; they were stopped less than a foot from him.

One time, codefendant Noel himself suffered a severe injury to his finger when Bane bit him during a fight with another dog. The wound required surgery, and Noel had to wear a splint on his arm and have two steel pins placed in his hand for eight to 10 weeks.

*147Mauling victim Diane Whipple and her partner Sharon Smith lived in a sixth-floor apartment across a lobby from defendants. Smith encountered defendants’ two dogs as often as once a week. In early December 2000, Whipple called Smith at work to say, with some panic in her voice, that one of the dogs had bitten her. Whipple had come upon codefendant Noel in the lobby with one of the dogs, which lunged at her and bit her in the hand. Whipple did not seek medical treatment for three deep, red indentations on one hand. Whipple made every effort to avoid defendants’ dogs, checking the hallway before she went out and becoming anxious while waiting for the elevator for fear the dogs would be inside. She and Smith did not complain to apartment management because they wanted nothing to do with defendants Knoller and Noel.

On January 26, 2001, Whipple telephoned Smith to say she was going home early. At 4:00 p.m., Esther Birkmaier, a neighbor who lived across the hall from Whipple, heard dogs barking and a woman’s “panic-stricken” voice calling, “Help me, help me.” Looking through the peephole in her front door, Birkmaier saw Whipple lying facedown on the floor just over the threshold of her apartment with what appeared to be a dog on top of her. Birkmaier saw no one else in the hallway. Afraid to open the door, Birkmaier called 911, the emergency telephone number, and at the same time heard a voice yelling, “No, no, no” and “Get off.” When Birkmaier again approached her door, she could hear barking and growling directly outside and a banging against a door. She heard a voice yell, “Get off, get off, no, no, stop, stop.” She chained her door and again looked through the peephole. Whipple’s body was gone and groceries were strewn about the hallway. Birkmaier called 911 a second time.

At 4:12 p.m., San Francisco Police Officers Sidney Laws and Leslie Forrestal arrived in response to Birkmaier’s telephone calls. They saw Whipple’s body in the hallway; her clothing had been completely ripped off, her entire body was covered with wounds, and she was bleeding profusely. Defendant Knoller and the two dogs were not in sight.

The officers called for an ambulance. Shortly thereafter, defendant Knoller emerged from her apartment. She did not ask about Whipple’s condition but merely told the officers she was looking for her keys, which she found just inside the door to Whipple’s apartment.

An emergency medical technician administered first aid to Whipple, who had a large, profusely bleeding wound to her neck. The wound was too large to halt the bleeding, and Whipple’s pulse and breathing stopped as paramedics arrived. She was revived but died shortly after reaching the hospital.

*148An autopsy revealed over 77 discrete injuries covering Whipple’s body “from head to toe.” The most significant were lacerations damaging her jugular vein and her carotid artery and crushing her larynx, injuries typically inflicted by predatory animals to kill their prey. The medical examiner stated that although earlier medical attention would have increased Whipple’s chances of survival, she might ultimately have died anyway because she had lost one-third or more of her blood at the scene. Plaster molds of the two dogs’ teeth showed that the bite injuries to Whipple’s neck were consistent with Bane’s teeth.

Animal control officer Andrea Runge asked defendant Knoller to sign over custody of the dogs for euthanasia. Knoller, whom Runge described as “oddly calm,” agreed to sign over Bane, but she refused to sign over Hera for euthanasia and she refused to help the animal control officers with the animals, saying she was “unable to handle the dogs.” When tranquilizer darts malfunctioned and failed to quiet Bane, “come-along” poles were used by animal control officers backed up by officers with guns drawn. Hera too was controlled by officers with “come-along” poles.

On February 8, 2001, both defendants appeared on the television show Good Morning America and basically blamed mauling victim Whipple for her own death. Defendant Knoller claimed that Whipple had already opened her apartment door when something about her interested Bane. He broke away, pulled Knoller across the lobby, and jumped up on Whipple, putting his paws on either side of her. Knoller said she pushed Whipple into Whipple’s apartment, fell on top of Whipple, and then tried to shield Whipple with her own body. But Whipple’s struggles must have been misinterpreted by the dog, and when Whipple struck Knoller with her fist, the dog began to bite Whipple. Knoller claimed that Whipple had ample opportunity to just slam the door of her apartment or stay still on the floor.

Codefendant Noel did not testify, but he presented evidence of positive encounters between the two dogs and veterinarians, friends, and neighbors. Defendant Knoller did testify in her own defense. She referred to herself, her husband, and Pelican Bay prisoner Schneider as the “triad,” and she spoke of Schneider as her “son.” The two dogs had become a focal point in the relationship. She denied reading literature in the apartment referring to the vicious nature of the dogs. She thought the dogs had no personality problems requiring a professional trainer. She denied receiving or otherwise discounted any warnings about the two dogs’ behavior and she maintained that virtually all the witnesses testifying to incidents with the dogs were lying. She said she never walked both dogs together. Ordinarily, she would walk Hera and codefendant Noel would walk Bane, because she had insufficient body strength to control Bane. But after Noel was injured while breaking up a fight *149between Bane and another dog, Knoller would sometimes walk Bane, always on a leash. She said she had just returned from walking Bane on the roof of the apartment building, and had opened the door to her apartment while holding Bane’s leash, when Bane dragged her back across the lobby toward Whipple, who had just opened the door to her own apartment. The other dog, Hera, left defendants’ apartment and joined Bane, who attacked Whipple. Knoller said she threw herself on Whipple to save her. She denied that Hera participated in the attack. She acknowledged not calling 911 to get help for Whipple.

Asked whether she denied responsibility for the attack on Whipple, Knoller gave this reply: “I said in an interview that I wasn’t responsible but it wasn’t for the—it wasn’t in regard to what Bane had done, it was in regard to knowing whether he would do that or not. And I had no idea that he would ever do anything like that to anybody. How can you anticipate something like that? It’s a totally bizarre event. I mean how could you anticipate that a dog that you know that is gentle and loving and affectionate would do something so horrible and brutal and disgusting and gruesome to anybody? How could you imagine that happening?”

In rebuttal, the prosecution presented evidence that the minor character of defendant Knoller’s injuries—principally bruising to the hands—indicated that she had not been as involved in trying to protect mauling victim Whipple as she had claimed. Dr. Randall Lockwood, the prosecution’s expert on dog behavior, testified that good behavior by a dog on some occasions does not preclude aggressive and violent behavior on other occasions, and he mentioned the importance of training dogs such as Bane and Hera not to fight.

The jury found Knoller guilty of second degree murder; it also found both Knoller and Noel guilty of involuntary manslaughter and owning a mischievous animal that caused the death of a human being. Both defendants moved for a new trial. The trial court denied Noel’s motion. We quote below the pertinent statements by the trial court in granting Knoller’s motion for a new trial on the second degree murder count.

The trial court observed: “The law requires that there be a subjective understanding on the part of the person that on the day in question—and I do not read that as being January 26th, 2001 because by this time, with all of the information that had come out dealing with the dogs, the defendants were fully on notice that they had a couple of wild, uncontrollable and dangerous dogs that were likely going to do something bad. [f] Is the ‘something bad’ death? That is the ultimate question in the case. There is no question but that the something bad was going to be that somebody was going to be badly hurt. I defy either defendant to stand up and tell me they had no idea that *150those dogs were going to hurt somebody one day. But can they stand up and say that they knew subjectively—not objectively and that’s an important distinction—that these dogs were going to stand up and kill somebody?” (Italics added.)

The trial court continued: “I am guided by a variety of principles. One of them is that public emotion, public outcry, feeling, passion, sympathy do not play a role in the application of the law. The other is that I am required to review all of the evidence and determine independently rather than as a jury what the evidence showed. I have laid out most of the evidence as it harms the defendants in this case. Their conduct from the time that they got the dogs to the time—to the weeks after Diane Whipple’s death was despicable. . . .

“There was one time on the stand, Ms. Knoller, when I truly believed what you said. You broke down in the middle of a totally scripted answer and you actually, instead of crying, you actually got mad and you said you had no idea that this dog could do what he did and pounded the table. I believed you. That was the only time, but I did believe you.” The court then described the definition of second degree murder as requiring that one “subjectively knows, based on everything, that the conduct that he or she is about to engage in has a high probability of death to another human being.” (Italics added.)

The trial court went on: “What we have in this case as it relates to Ms. Knoller is the decision to take the dog outside, into the hallway, up to the roof, go to the bathroom, bring it back down and put it in the apartment. There was no question but that taking the dog out into the hallway by that very act exposed other people in the apartment, whether they are residents there or guests, invitees to what might happen with the dog. When you take everything as a totality, the question is whether or not as a subjective matter and as a matter of law Ms. Knoller knew that there was a high probability that day, or on the day before on the day after,—I reject totally the argument of the defendants that she had to know when she walked out the door—she was going to kill somebody that morning. The Court finds that the evidence does not support it.” (Italics added.)

The trial court concluded it had “no choice, . . . taking the Legislature’s scheme, the evidence that was received, as despicable as it is, but to determine not that [defendant Knoller] is acquitted of second degree murder but to find that on the state of the evidence, I cannot say as a matter of law that she subjectively knew on January 26th that her conduct was such that a human being was likely to die.” (Italics added.)

The trial court mentioned another consideration: “The Court also notes a great troubling feature of this case that Mr. Noel was never charged [with *151murder] as Ms. Knoller was. In the Court’s view, given the evidence, Mr. Noel is more culpable than she. Mr. Noel personally knew that she could not control those dogs. He could not control those dogs. Mr. Noel was substantially haughtier than she was. In brushing off all of the incidents that happened out in the street, Mr. Noel knew as a theological certainty that that dog, which had recently been operated on, was taking medication that had given it diarrhea, was going to go out into the hallway or out into the street possibly, at the hands of Ms. Knoller. He . . . left her there to do that. [f] . . . And yet Mr. Noel was not charged [with murder]. Equality of sentencing and the equal administration of justice is an important feature in any criminal court. That played a role as well.” The trial court then granted defendant Knoller’s motion for a new trial on the second degree murder count.

As noted earlier, both defendants as well as the prosecution appealed. The Court of Appeal reversed the trial court’s order granting Knoller’s motion for a new trial on the second degree murder count. It disagreed with the trial court that a second degree murder conviction, based on a theory of implied malice, required that Knoller recognized “her conduct was such that a human being was likely to die.” The Court of Appeal held that a second degree murder conviction can be based simply on a defendant’s “subjective appreciation and conscious disregard of a likely risk of . . . serious bodily injury.” In all other respects, the Court of Appeal affirmed both defendants’ convictions.

II. The Elements of Implied Malice

Murder is the unlawful killing of a human being, or a fetus, with malice aforethought. (§ 187, subd. (a).) Malice may be express or implied. (§ 188.) At issue here is the definition of “implied malice.”

Defendant Knoller was convicted of second degree murder as a result of the killing of Diane Whipple by defendant’s dog, Bane. Second degree murder is the unlawful killing of a human being with malice aforethought but without the additional elements, such as willfulness, premeditation, and deliberation, that would support a conviction of first degree murder. (See §§ 187, subd. (a), 189.) Section 188 provides: “[Mjalice may be either express or implied. It is express when there is manifested a deliberate intention to take away the life of a fellow creature. It is implied, when no considerable provocation appears, or when the circumstances attending the killing show an abandoned and malignant heart.”

The statutory definition of implied malice, a killing by one with an “abandoned and malignant heart” (§ 188), is far from clear in its meaning. Indeed, an instruction in the statutory language could be misleading, for it “could lead the jury to equate the malignant heart with an evil disposition or *152a despicable character” (People v. Phillips, supra, 64 Cal.2d at p. 587) instead of focusing on a defendant’s awareness of the risk created by his or her behavior. “Two lines of decisions developed, reflecting judicial attempts ‘to translate this amorphous anatomical characterization of implied malice into a tangible standard a jury can apply.’ ” (People v. Nieto Benitez (1992) 4 Cal.4th 91, 103 [13 Cal.Rptr.2d 864, 840 P.2d 969], quoting People v. Protopappas (1988) 201 Cal.App.3d 152, 162-163 [246 Cal.Rptr. 915].) Under both lines of decisions, implied malice requires a defendant’s awareness of the risk of death to another.

The earlier of these two lines of decisions, as this court observed in People v. Nieto Benitez, supra, 4 Cal.4th at pages 103-104, originated in Justice Traynor’s concurring opinion in People v. Thomas (1953) 41 Cal.2d 470, 480 [261 P.2d 1], which stated that malice is implied when “the defendant for a base, antisocial motive and with wanton disregard for human life, does an act that involves a high degree of probability that it will result in death.” (We here refer to this as the Thomas test.) The later line dates from this court’s 1966 decision in People v. Phillips, supra, 64 Cal.2d at page 587: Malice is implied when the killing is proximately caused by “ ‘an act, the natural consequences of which are dangerous to life, which act was deliberately performed by a person who knows that his conduct endangers the life of another and who acts with conscious disregard for life.’ ” (The Phillips test.)

In People v. Watson (1981) 30 Cal.3d 290, 300 [179 Cal.Rptr. 43, 637 P.2d 279], we held that these two definitions of implied malice in essence articulated the same standard. Concerned, however, that juries might have difficulty understanding the Thomas test’s concept of “wanton disregard for human life,” we later emphasized that the “better practice in the future is to charge juries solely in the straightforward language of the ‘conscious disregard for human life’ definition of implied malice,” the definition articulated in the Phillips test. (People v. Dellinger (1989) 49 Cal.3d 1212, 1221 [264 Cal.Rptr. 841, 783 P.2d 200].) The standard jury instructions thereafter did so. (See CALJIC No. 8.11; Judicial Council of Cal. Crim. Jury Instns. (2006) CALCRIM No. 520.) Since 1989, our decisions have articulated the standard we set out in Dellinger and in CALJIC No. 8.11. (See, e.g., People v. Randle (2005) 35 Cal.4th 987, 994 [28 Cal.Rptr.3d 725, 111 P.3d 987]; People v. Taylor (2004) 32 Cal.4th 863, 867-868 [11 Cal.Rptr.3d 510, 86 P.3d 881]; People v. Lasko (2000) 23 Cal.4th 101, 107 [96 Cal.Rptr.2d 441, 999 P.2d 666]; People v. Hansen (1994) 9 Cal.4th 300, 308 [36 Cal.Rptr.2d 609, 885 P.2d 1022]; People v. Whitfield (1994) 7 Cal.4th 437, 450 [27 Cal.Rptr.2d 858, 868 P.2d 272]; People v. Nieto Benitez, supra, 4 Cal.4th at pp. 104, 111.) The trial court here instructed the jury in the language of CALJIC No. 8.11.

*153III. The Court of Appeal’s Test for Implied Malice

As discussed in the preceding part, the great majority of this court’s decisions establish that a killer acts with implied malice only when acting with an awareness of endangering human life. This principle has been well settled for many years, and it is embodied in the standard jury instruction given in murder cases, including this one. The Court of Appeal here, however, held that a second degree murder conviction, based on a theory of implied malice, can be based simply on a defendant’s awareness of the risk of causing serious bodily injury to another.

In support of that view, the Court of Appeal pointed to three decisions of this court: People v. Conley (1966) 64 Cal.2d 310 [49 Cal.Rptr. 815, 411 P.2d 911] (Conley), People v. Poddar (1974) 10 Cal.3d 750 [111 Cal.Rptr. 910, 518 P.2d 342] (Poddar), and People v. Coddington (2000) 23 Cal.4th 529 [97 Cal.Rptr.2d 528, 2 P.3d 1081] (Coddington). We discuss each case below.

In Conley, supra, 64 Cal.2d 310, the defendant, after consuming copious quantities of alcohol, went to the home of his former lover and her husband, where he shot and killed both of them. He was convicted of two counts of first degree murder. The issue on appeal was whether the trial court should have instructed the jury on diminished mental capacity caused by intoxication. This court held that it should have so instructed because “[a]n awareness of the obligation to act within the general body of laws regulating society . . . is included in the statutory definition of implied malice in terms of the abandoned and malignant heart . . . .” (Id. at p. 322.) In explaining that holding, Conley stated that a person who carefully weighs the course of action he is about to take and chooses to kill his victim, after considering the reasons for and against it, “is normally capable also of comprehending the duty society places on all persons to act within the law.” (Ibid.) Conley continued: “If, despite such awareness, he does an act that is likely to cause serious injury or death to another, he exhibits that wanton disregard for human life or antisocial motivation that constitutes malice aforethought.” (Ibid., italics added.)4 It is this sentence from Conley on which the Court of Appear relied. But that language from Conley described the defendant’s act (the objective component of implied malice), not the defendant’s mental state (the subjective component of implied malice); it is therefore irrelevant to the issue here, which concerns the subjective component—whether the defendant must be aware of the risk of death or only a risk of serious bodily injury.

*154Conley, supra, 64 Cal.2d 310, did not discuss whether implied malice could be based merely on a defendant’s awareness of the risk of serious bodily injury to another but not the risk of death resulting from the defendant’s actions. That issue, presented here, did not arise in Conley, because there the defendant, who said he was going to kill the victims and did so, could not claim he was aware only of the risk of causing serious bodily injury.

In cases decided shortly before and after Conley, we reiterated the established definition of implied malice as requiring an awareness of the risk that the defendant’s conduct will result in the death of another. One year before Conley was filed, we stated in People v. Washington (1965) 62 Cal.2d 777, 780, 782 [44 Cal.Rptr. 442, 402 P.2d 130], that implied malice required a “conscious disregard for life.” Conley did not at all suggest that it intended to depart from the view expressed in Washington. And two months after Conley, this court in People v. Phillips, supra, 64 Cal.2d at page 582, endorsed its earlier statement in Washington that implied malice requires a “conscious disregard for life.’'’ (Italics added.)

We now turn to Poddar, supra, 10 Cal.3d 750, the second of the three decisions that the Court of Appeal cited. In that case, the defendant went to the home of a woman he had dated casually, shot her with a pellet gun, and then killed her with a knife. He was convicted of second degree murder. This court held that the trial court’s jury instruction on second degree murder was defective because it did not explain the concept of diminished capacity as set out in Conley, supra, 64 Cal.2d 310. (Poddar, supra, 10 Cal.3d at pp. 757-759.) In its discussion of diminished capacity, Poddar stated that to prove implied malice, “it must be shown that the accused was both aware of his duty to act within the law and acted in a manner likely to cause death or serious bodily injury despite such awareness.” (Id. at p. 758, italics added.) As in Conley, Poddar referred to serious bodily injury in describing the defendant’s act, the objective component of implied malice. Poddar did not say that the defendant’s mental state, the subjective component of implied malice, at issue here, could be satisfied by proof that the defendant acted with an awareness that his conduct could cause serious bodily injury. Indeed, the defendant in Poddar never claimed that he was unaware that his acts could cause death.

Even if the above discussed language from Conley, supra, 64 Cal.2d at page 322, and from Poddar, supra, 10 Cal.3d at page 758, could be viewed as implying that a second degree murder conviction, on a theory of implied malice, could be based simply on a defendant’s awareness of the risk of causing serious bodily injury, rather than death, that language would lack authoritative force. “ ‘It is axiomatic that language in a judicial opinion is to *155be understood in accordance with the facts and issues before the court. An opinion is not authority for propositions not considered.’ ” (Kinsman v. Unocal Corp. (2005) 37 Cal.4th 659, 680 [36 Cal.Rptr.3d 495, 123 P.3d 931], quoting Chevron U.S.A., Inc. v. Workers’ Comp. Appeals Bd. (1999) 19 Cal.4th 1182, 1195 [81 Cal.Rptr.2d 521, 969 P.2d 613].) “An appellate decision is not authority for everything said in the court’s opinion but only ‘for the points actually involved and actually decided.’ ” (Santisas v. Goodin (1998) 17 Cal.4th 599, 620 [71 Cal.Rptr.2d 830, 951 P.2d 399].) Because the facts and issues in Conley, supra, 64 Cal.2d 310, and in Poddar, supra, 10 Cal.3d 750, did not encompass the question whether implied malice could be based on a defendant’s awareness of the risk of serious bodily injury alone, the language the Court of Appeal cited from Conley and Poddar lacks authoritative force.

This brings us to Coddington, supra, 23 Cal.4th 529, the last in the trio of decisions relied on by the Court of Appeal. In that case, the defendant lured teenage girls to his mobilehome by telling them they would star in an antidrug video, and then raped them and committed other sexual offenses. He killed two older women who had accompanied the girls as chaperones. The defendant was convicted of two counts of first degree murder with special circumstances, as well as various other offenses, and he was sentenced to death.

Among the many issues the defendant in Coddington raised on appeal was a claim that the trial court had erred in not instructing the jury on second degree murder based on implied malice. Responding to that claim, the Attorney General argued in Coddington that such an instruction was not needed because there was no evidence that the defendant’s offense was less than first degree murder, and that the defendant’s conduct proved that he “acted with actual or presumptive knowledge that serious bodily injury was likely to occur.” (Coddington, supra, 23 Cal.4th at p. 592, italics added.) This court rejected the Attorney General’s argument, explaining that such a mental state (actual or presumptive knowledge that serious bodily injury is likely to occur) “permits an inference of implied malice . . . and does not support a conclusion that no instruction on second degree murder on a theory of implied malice was necessary.” (Ibid.)

Notwithstanding Coddington’s offhand comment that knowledge of the risk of serious bodily injury permits an inference of implied malice, Coddington reiterated the established rule that a trial court must instruct on second degree murder based on implied malice whenever there is evidence “from which the jury could have inferred that appellant acted without intent to kill even though his conduct posed a high risk of death.” (Coddington, supra, 23 Cal.4th at p. 593, italics added.) Thus, Coddington’s offhand comment cannot be viewed as implicitly overruling the decisions of this court *156discussed earlier (see, ante, at p. 152) declaring that implied malice requires an awareness of the risk of death.

In sum, the three decisions on which the Court of Appeal relied lack persuasive force. Neither Conley, supra, 64 Cal.2d 310, nor Poddar, supra, 10 Cal.3d 750, addressed the issue presented here: whether implied malice can be based on a defendant’s awareness of the risk of great bodily injury but not death resulting from the defendant’s actions. With respect to the comment in Coddington, supra, 23 Cal.4th at page 592, suggesting that knowledge of the likelihood of serious bodily injury permits an inference of implied malice, it is inconsistent not only with the holding in that case but also with the views expressed in other decisions of this court. (See, ante, at p. 152.) We conclude that a conviction for second degree murder, based on a theory of implied malice, requires proof that a defendant acted with conscious disregard of the danger to human life. In holding that a defendant’s conscious disregard of the risk of serious bodily injury suffices to sustain such a conviction, the Court of Appeal erred.

IV. The Trial Court’s Grant of a New Trial on the Second Degree Murder Charge

We now turn to the second issue raised by the petition for review—whether the trial court abused its discretion in granting defendant Knoller a new trial on the second degree murder charge. Such an abuse of discretion arises if the trial court based its decision on impermissible factors (see People v. Carmody (2004) 33 Cal.4th 367, 378 [14 Cal.Rtpr.3d 880, 92 P.3d 369]) or on an incorrect legal standard (see Linder v. Thrifty Oil Co. (2000) 23 Cal.4th 429, 435—436 [97 Cal.Rptr.2d 179, 2 P.3d 27]; In re Carmaleta B. (1978) 21 Cal.3d 482, 496 [146 Cal.Rptr. 623, 579 P.2d 514]).

In granting Knoller a new trial, the trial court properly viewed implied malice as requiring a defendant’s awareness of the danger that his or her conduct will result in another’s death and not merely in serious bodily injury. (See, ante, at pp. 149-151.) But the court’s ruling was legally flawed in other respects. As we explain below, the trial court based its ruling on an inaccurate definition of implied malice, and it inappropriately relied on the prosecutor’s failure to charge codefendant Noel with murder.

As discussed earlier in part II, this court before its decision in People v. Dellinger, supra, 49 Cal.3d 1212, had defined implied malice in two similar but somewhat different ways. Under the Thomas test, malice is implied when “the defendant for a base, antisocial motive and with wanton disregard for human life, does an act that involves a high degree of probability that it will result in death.” (People v. Thomas, supra, 41 Cal.2d at p. 480 (conc. opn. of *157Traynor, J.); see also Poddar, supra, 10 Cal.3d at pp. 756-757.) Under the Phillips test (People v. Phillips, supra, 64 Cal.2d at p. 587), malice is implied when the killing is proximately caused by “ ‘an act, the natural consequences of which are dangerous to life, which act was deliberately performed by a person who knows that his conduct endangers the life of another and who acts with conscious disregard for life.’ ” In People v. Dellinger, supra, 49 Cal.3d 1212, we observed that although these two tests “articulated one and the same standard” (id. at p. 1219), the Thomas test contained “obscure phraseology” and had “become a superfluous charge,” so that the “better practice in the future” would be for trial courts to instruct juries in the “straightforward language” of the Phillips test (Dellinger, at p. 1221).5

Here, the trial court properly instructed the jury in accordance with the Phillips test. But when the court evaluated defendant Knoller’s new trial motion, it relied on language from the Thomas test, and as explained below, its description of that test was inaccurate. The court stated that a killer acts with implied malice when the killer “subjectively knows, based on everything, that the conduct that he or she is about to engage in has a high probability of death to another human being” and thus the issue in this case was “whether or not as a subjective matter and as a matter of law Ms. Knoller knew that there was a high probability” that her conduct would result in someone’s death. (Italics added.) But “high probability of death” is the objective, not the subjective, component of the Thomas test, which asks whether the defendant’s act or conduct “involves a high degree of probability that it will result in death.” (People v. Thomas, supra, 41 Cal.2d at p. 480 (conc. opn. of Traynor, J.).) The subjective component of the Thomas test is whether the defendant acted with “a base, antisocial motive and with wanton disregard for human life.” (Ibid.) Nor does the Phillips test require a defendant’s awareness that his or her conduct has a high probability of causing death. Rather, it requires only that a defendant acted with a “ ‘conscious disregard for human life’ ” (People v. Dellinger, supra, 49 Cal.3d at p. 1221; see People v. Phillips, supra, 64 Cal.2d at p. 587).

As just shown, in treating the objective component of the Thomas test as the subjective component of that test, the trial court applied an erroneous definition of implied malice in granting defendant Knoller a new trial on the second degree murder charge.

In ruling on Knoller’s motion for a new trial, the trial court also commented that, in its view, codefendant Noel was more culpable than defendant Knoller, and that the district attorney’s failure to charge Noel with *158murder was a “troubling feature of this case” that “played a role as well” in the court’s decision to grant Knoller a new trial on the second degree murder charge. Dissimilar charging of codefendants, however, is not among the grounds for a new trial in section 1181. Although section 1181 states that a defendant’s new trial motion may be granted only on the grounds stated in that section, several courts have held that new trials may nonetheless be granted on grounds not enumerated in the statute when necessary to protect a defendant’s constitutional right to a fair trial. (See, e.g., People v. Oliver (1975) 46 Cal.App.3d 747, 751 [120 Cal.Rptr. 368] [judicial misconduct]; People v. Davis (1973) 31 Cal.App.3d 106, 109 [106 Cal.Rptr. 897] [unexpected absence of witness].) No published decision, however, has ever approved granting a new trial based on differential treatment of defendants. (See generally People v. Belmontes (1988) 45 Cal.3d 744, 810-813 [248 Cal.Rptr. 126, 755 P.2d 310] [disposition of codefendant’s case is irrelevant to jury’s determination at penalty phase of capital case].)

We specifically do not address whether a new trial could be granted on such a ground, an issue that would involve significant separation of powers considerations. Even assuming a new trial could be granted on such a ground, it is not justified here. Defendant Knoller and codefendant Noel were not similarly situated with regard to their dog Bane’s fatal mauling of Whipple in the hallway of the apartment building where they all lived. The immediate cause of Whipple’s death was Knoller’s own conscious decision to take the dog Bane unmuzzled through the apartment building, where they were likely to encounter other people, knowing that Bane was aggressive and highly dangerous and that she could not control him. Bringing a more serious charge against the person immediately responsible for the victim’s death was a permissible exercise of prosecutorial discretion, not grounds for a new trial.

V. Conclusion and Disposition

In sum, the trial court abused its discretion in granting defendant Knoller a new trial on the second degree murder charge. That court erroneously concluded both that Knoller could not be guilty of murder, based on a theory of implied malice, unless she appreciated that her conduct created a high probability of someone’s death, and that a new trial was justified because the prosecution did not charge codefendant Noel with murder. It is uncertain whether the trial court would have reached the same result using correct legal standards. Moreover, the Court of Appeal, in reversing the trial court’s order, also erred, mistakenly reasoning that implied malice required only a showing that the defendant appreciated the risk of serious bodily injury. Under these circumstances, we conclude that the matter should be returned to the trial court to reconsider its new trial order in light of the views set out in this opinion.

*159The Court of Appeal’s judgment is reversed and the matter is remanded to that court, with directions to return the case to the trial court for reconsideration of defendant Knoller’s new trial motion in accord with the views expressed in this opinion.

George, C. J., Baxter, J., Werdegar, J., Chin, J., Moreno, J., and Corrigan, J., concurred.

The petition of appellant The People for a rehearing was denied July 18, 2007.

Opinion

MOSK, J.

We consider in this case whether a prosecution for involuntary manslaughter (Pen. Code, § 192, subd. (b)) and felony child endangerment (id., § 273a, subd. (1)) can be maintained against the mother of a child who died of meningitis after receiving treatment by prayer in lieu of medical attention. We conclude that the prosecution is permitted by statute as well as the free exercise and due process clauses of the state and federal Constitutions.

Defendant Laurie Grouard Walker is a member of the Church of Christ, Scientist (hereafter the Church). Her four-year-old daughter, Shauntay, fell *119ill with flu-like symptoms on February 21, 1984, and four days later developed a stiff neck. Consistent with the tenets of her religion, defendant chose to treat the child’s illness with prayer rather than medical care.1 Defendant contacted an accredited Christian Science prayer practitioner who thereafter prayed for Shauntay and visited the child on two occasions. Defendant also engaged a Christian Science nurse who attended Shauntay on February 27 and again on March 6 and 8.2 Shauntay nevertheless lost weight, grew disoriented and irritable during the last week of her illness, and died on March 9 of acute purulent meningitis after a period of heavy and irregular breathing. During the 17 days she lay ill, the child received no medical treatment.

The People charged defendant with involuntary manslaughter and felony child endangerment based on allegations that her criminal negligence proximately caused Shauntay’s death. Defendant moved to dismiss the prosecution (Pen. Code, § 995) on the grounds that (1) her conduct was specifically protected by law, and (2) the statutes under which she had been charged failed to provide fair notice that her conduct was criminal. The court denied her motion.

Defendant petitioned the Court of Appeal for a writ of prohibition and a stay. (Pen. Code, § 999a.) The petition and stay request were summarily denied, and defendant petitioned for review in this court. We granted the petition and transferred the matter to the Court of Appeal with directions to issue an alternative writ of prohibition. After further briefing and oral argument, the Court of Appeal again denied defendant’s petition. She thereafter filed a second petition for review in this court, which we also granted. *120Defendant and amici curiae offer a variety of statutory and constitutional arguments in support of their claim that the prosecution of defendant under Penal Code section 192, subdivision (b) (hereafter section 192(b)), and section 273a, subdivision (1) (hereafter section 273a(l)), is barred as a matter of law. For the reasons set forth below, we reject their contentions and conclude that defendant can be prosecuted as charged.

I. Statutory Contentions

A. Section 270 as a complete defense to prosecution

Defendant first contends that the provisions of Penal Code section 270 (hereafter section 270) provide a complete defense to any prosecution based on her treatment of Shauntay’s illness with prayer rather than medical care. Section 270 enumerates certain necessities that parents must furnish their children and imposes misdemeanor liability for the failure to do so. As enacted in 1872, the statute provided that “Every parent of any child who willfully omits, without lawful excuse, to perform any duty imposed upon him by law, to furnish necessary food, clothing, shelter, or medical attendance for such child, is guilty of a misdemeanor.” (Pen. Code (1st ed. 1872) § 270.) The Legislature amended the provision in 1925 by inserting the phrase “or other remedial care” after “medical attendance.” (Stats. 1925, ch. 325, § 1, p. 544.) The statute was again amended in 1976 to specify that “treatment by spiritual means through prayer alone” constitutes “other remedial care.” (Stats. 1976, ch. 673, § 1, p. 1661.)3

1.

As a threshold consideration we must ascertain whether prayer treatment constitutes an acceptable substitute for medical care under the terms of section 270, as defendant contends. If it does not, then a fortiori the statute provides no defense to prosecutions under separate manslaughter and child endangerment provisions for the use of prayer in lieu of medicine. This determination hinges on whether “other remedial care,” defined in section 270 to include prayer, represents an alternative to “medical attendance” or rather identifies a distinct and additional necessity that parents must provide their children.

*121In People v. Arnold (1967) 66 Cal.2d 438, 452 [58 Cal.Rptr. 115, 426 P.2d 515], we considered the contention that section 270 allows parents to provide children with “an accepted alternative to medical attendance: ‘other remedial care,’ namely enemas, compresses, and prayer.” The case involved the appeal of a mother convicted of misdemeanor-manslaughter after unsuccessfully treating her child’s illness with prayer. Although reversing on unrelated grounds, the Arnold court summarily rejected in dictum the defendant’s interpretation of section 270, reasoning that “The phrase ‘other remedial care’ . . . does not sanction unorthodox substitutes for ‘medical attendance’; it indicates one of the multiple necessities which the parent must provide.” {Ibid.)

While the Arnold decision predates the 1976 amendment specifying that “other remedial care” includes prayer, the court’s reasoning remains fatal to a defense based on treatment by spiritual means: regardless of its content, “other remedial care” constitutes “one of the multiple necessities” under Arnold, thus operating in addition to rather than in lieu of the responsibility to furnish medical attendance. Because the 1976 amendment “did not address the contention [in Arnold] that other remedial care could not act as a substitute to standard medical treatment,” the Court of Appeal in the case at bar concluded that defendant’s provision of prayer did not supplant her separate responsibility to furnish medical care under section 270.

Well-settled principles guide our review of the statutory analysis set forth in Arnold and embraced by the decision below. “ ‘ “The fundamental rule of statutory construction is that the court should ascertain the intent of the Legislature so as to effectuate the purpose of the law. [Citations.]” ’ In determining such intent, the court turns first to the words of the statute. ‘[W]here . . . the language is clear, there can be no room for interpretation.’ ” (Regents of University of California v. Public Employment Relations Bd. (1986) 41 Cal.3d 601, 607 [224 Cal.Rptr. 631, 715 P.2d 590], citations omitted.)

Section 270 requires that parents “furnish necessary clothing, food, shelter or medical attendance, or other remedial care . . . .” In our view, this language is sufficiently clear to reject the dictum in Arnold and conclude that the Legislature intended “other remedial care” to constitute a substitute for “medical attendance.” We begin by noting the repetition of “or” to introduce both “medical attendance” and “other remedial care.” The first use of the word, preceding “medical attendance,” denotes that clothing, food, shelter, and medical attendance represent distinct necessities each of which must be provided a child; it would be superfluous if the succeeding phrase, “or other remedial care,” introduced yet another necessity into the statutory scheme. We have often observed that courts *122should give significance to every word, phrase, and sentence of an act, and that any construction rendering certain words surplusage should be avoided. (Palos Verdes Faculty Assn. v. Palos Verdes Peninsula Unified School Dist. (1978) 21 Cal.3d 650, 659 [147 Cal.Rptr. 359, 580 P.2d 1155].) To give significance to the initial use of the word “or,” its second use must be taken to mean that “other remedial care” operates as an alternative to the immediately preceding phrase rather than as an additional necessity appended to the entire sentence in the manner of an afterthought.

The definition of certain pivotal words in the statute bolsters this interpretation. “Remedial” is defined as “affording a remedy: intended for a remedy or for the removal or abatement of a disease or of an evil.” (Webster’s New Internat. Diet. (3d ed. 1961) p. 1920.) “Remedy,” in turn, is defined as “something that relieves or cures a disease: a medicine or application that serves or helps to terminate disease and restore health.” {Ibid.) Finally, “other” is defined as “not the same: different.” {Id. at p. 1598.) When these definitions are substituted for the words of the statute, the provision penalizes parents who fail to provide “clothing, food, shelter or medical attendance, or [different] care [intended to relieve or cure a disease].” It thus is apparent that the Legislature intended “other remedial care” to represent an alternative to medical attendance under the terms of section 270.

Any doubt regarding this interpretation cannot survive examination of the legislative history of the 1976 amendment defining “other remedial care” to include prayer. When the members of the Assembly considered the amendment, contained in Assembly Bill No. 3843, 1975-1976 Regular Session, they had before them the third reading analysis of the legislation prepared by the Assembly Office of Research. The analysis stated: “Under this bill, the parents may not be liable for failing to provide for the health of the child because they choose treatment by prayer rather than common medical treatment . . . .” (Assem. Office of Research, 3d reading analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.).) Similarly, the members of the Senate received an analysis of the legislation prepared by either the Republican or Democratic Caucus. Both caucus analyses stated that the amendment would shield from liability those parents who provide prayer in lieu of medical care for their children. (Sen. Democratic Caucus, 3d reading analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.); Sen. Republican Caucus, 3d reading analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.).) While these materials are not dispositive evidence of legislative intent (Shippen v. Department of Motor Vehicles (1984) 161 Cal.App.3d 1119, 1126 [208 Cal.Rptr. 13]), they are significant insofar as their contents do not contradict the plain language of the statute. {Commodore Home Systems, Inc. v. Superior Court (1982) 32 Cal.3d 211, 219 [185 Cal.Rptr. *123270, 649 P.2d 912].) We accordingly conclude that section 270 exempts parents who utilize prayer treatment from the statutory requirement to furnish medical care, and overrule People v. Arnold, supra, 66 Cal.2d 438, 452, to the extent it concludes to the contrary.

2.

We next consider whether section 270 bars the prosecution of defendant under the manslaughter and child endangerment statutes. (§§ 192(b), 273a(l).) Again we turn to its plain language for initial guidance. Citing the statutory provision that “treatment by spiritual means through prayer alone . . . shall constitute ‘other remedial care’, as used in this section ” (italics added), the Court of Appeal concluded that section 270 expressly precludes any extension of its religious exemption to other statutes. This analysis confuses the statutory limitation on the definition of “other remedial care” with a limitation on possible defenses implied by that definition.

Following the rule of the last antecedent, the phrase “as used in this section” must modify “other remedial care.” (See People v. Baker (1968) 69 Cal.2d 44, 46 [69 Cal.Rptr. 595, 442 P.2d 675].) The language therefore either (1) qualifies the five earlier references to “other remedial care” in the statute, none of which mentions “treatment by spiritual means through prayer alone,” or (2) distinguishes the definition of the phrase in section 270 from its use in numerous other statutes, most of which make no reference to treatment by spiritual means. (See, e.g., Code Civ. Proc., § 1209.5; Welf. & Inst. Code, §§ 305, subd. (c), 369, subds. (a)-(g), 625, subd. (c), 739, subds. (a)-(g), 11452, subd. (6), 14059; see also Stats. 1987, chs. 1353, 1485.) Neither of these qualifications on the statutory language bears on whether the defense implied by the definition of “other remedial care,” as used in section 270, should apply to the charges against defendant under sections 192(b) an 273a(l).

Defendant conversely asserts that the plain language of section 270 requires the extension of its religious exemption to her prosecution. She focuses on the reference in the statute to the provision of “necessary clothing, food, shelter or medical attendance, or other remedial care . . . .” (Italics added.) Observing that “necessary” is defined, inter alia, as “absolutely required: essential, indispensable” (Webster’s New Internat. Diet. (3d ed. 1961) p. 1511), she contends that there can be no circumstance involving the illness of a child in which the use of prayer in lieu of medicine is unlawful.

It is true that the statute recognizes “other remedial care” as an acceptable substitute for “medical attendance” when care is “necessary”; *124however, this conclusion does not address the question whether compliance with the terms of section 270 absolves defendant of liability under all other provisions of the Penal Code. “When used in a statute words must be construed in context, keeping in mind the nature and obvious purpose of the statute where they appear . . . .” (People v. Black (1982) 32 Cal.3d 1, 5 [184 Cal.Rptr. 454, 648 P.2d 104]; People v. Alday (1973) 10 Cal.3d 392, 395 [110 Cal.Rptr. 617, 515 P.2d 1169].) Conduct that is legal in one statutory context thus may be actionable under separate statutes created for different legislative purposes. It follows that the legality of defendant’s conduct under the terms of section 270 cannot be read to create a parallel exemption from prosecution under sections 192(b) and 273a(l) unless the statutes reflect some shared legislative objective.4 (Milligan v. City of Laguna Beach (1983) 34 Cal.3d 829, 835 [196 Cal.Rptr. 38, 670 P.2d 1121]; see also People v. Caudillo (1978) 21 Cal.3d 562, 585 [146 Cal.Rptr. 859, 580 P.2d 274]; Pen. Code, § 4.) We turn to the purpose of section 270.

“Rather than punishment of the neglectful parents, the principal statutory objectives [of section 270] are to secure support of the child and to protect the public from the burden of supporting a child who has a parent able to support him.” (People v. Sorensen (1968) 68 Cal.2d 280, 287 [66 Cal.Rptr. 7, 437 P.2d 495, 25 A.L.R.3d 1093].) The provision is “designed to supplement civil statutes for effective enforcement of child support obligations.” (Note, Criminal Nonsupport and a Proposal for an Effective Felony-Misdemeanor Distinction (1986) 37 Hastings L.J. 1075, 1079; County of Ventura v. George (1983) 149 Cal.App.3d 1012, 1015 [197 Cal.Rptr. 245].) A parent thus may comply with the requirements of the statute simply by providing financial assistance to another individual with physical custody of the child. (Lyons v. Municipal Court (1977) 75 Cal.App.3d 829, 843 [142 Cal.Rptr. 449].) A parent who is financially incapable of furnishing support, without fault, is likewise excused from compliance. (People v. Sorensen, supra, 68 Cal.2d at p. 287.) Indeed, the dispositional section of the statutory scheme explicitly contemplates protection of the fisc by stipulating that “If the children are receiving public assistance, all fines, penalties or forfeitures *125imposed and all funds collected from the defendant [for violations of section 270 or section 270a] shall be paid to the county department. Money so paid shall be applied first to support . . . and any balance remaining shall be applied to future needs, or be treated as reimbursement for past support furnished from public assistance funds.” (Pen. Code, § 270d; see People v. Sorensen, supra, 68 Cal.2d at p.287.)

Disputing this settled understanding of section 270 as a fiscal support provision, defendant asserts that the objective of the statute is to protect children from serious injury rather than to secure certain routine necessities at parental expense.5 She argues that her interpretation is supported by the notes of the Code Commission of 1870-1872. We will consider the code commissioners’ notes when they do not conflict with other persuasive evidence of legislative intent and particularly “ ‘where the commission’s comment is brief, because in such a situation there is ordinarily strong reason to believe that the legislators’ votes were based in large measure upon the explanation of the commission proposing the bill.’ ” (Keeler v. Superior Court (1970) 2 Cal.3d 619, 630 [87 Cal.Rptr. 481, 470 P.2d 617, 40 A.L.R.3d 420], quoting Van Arsdale v. Hollinger (1968) 68 Cal.2d 245, 249-250 [66 Cal.Rptr. 20, 437 P.2d 508].) The explanatory note for section 270 reads: “Consult the Civil Code for the provisions reported defining the duty of parental support. As to the criminality of a willful omission to perform this duty.—See Reg. vs. Chandler [(1855) 6 Cox Crim. Cas. 519], Reg. vs. Gray [(1857) 7 Cox Crim. Cas. 326], Reg. vs. S_ [(1851) 5 Cox Crim. Cas. 279], Reg. vs. Philpot[t] [(1853) 6 Cox Crim. Cas. 140].” (Code comrs. note, Pen. Code (1st ed. 1872) p. 117.) Because the note says nothing regarding the purpose of the statute, any construction of legislative intent resting on its contents requires the implausible assumption that the legislators sought out and read the four English common law cases cited therein and then conformed their understanding of the statute to the reasoning of those opinions.

Yet even if we were to so assume, the cases fail to substantiate defendant’s interpretation. She emphasizes that each involved allegations of a child suffering physical injury as a result of a parent’s failure to provide basic necessities, and asserts that child endangerment was thus the harm to be *126averted by imposition of criminal liability. To the contrary, the first case cited in the code commission note held that a mother could not be found guilty of neglecting to feed her child when she had lacked the financial means to do so, even though public assistance had been available to otherwise provide for the starving child. “It is admitted in the case that there was no evidence of her having the means; that being admitted, it is no answer to say that she might have procured the means by applying to the relieving officer.” (Reg. v. Chandler, supra, 6 Cox Crim. Cas. at p. 520.) This disposition suggests that failure to provide financial support, rather than injury per se, was the gravamen of the common law crime. Indeed, the continued existence under section 270 of a complete defense for parents who lack without fault the means of support significantly distinguishes the misdemeanor provision from sections 192(b) and 273a(l), which recognize no insolvency exception to felony charges stemming from severe neglect.

We thus reaffirm our determination in Sorensen, supra, 68 Cal.2d 280, that section 270 requires able parents to furnish certain routine necessities for their children so that the public need not unnecessarily assume that obligation. (See also Davis v. Stroud (1942) 52 Cal.App.2d 308, 315 [126 P.2d 409].) While certainly reflecting concern for the general welfare of children, the fiscal objectives of this support provision are so manifestly distinguishable from the specific purposes of the involuntary manslaughter and felony child-endangerment statutes—designed to protect citizens from immediate and grievous bodily harm—that section 270 cannot be read to create express exemptions from prosecution under those separate provisions as a matter of parallel construction. The Legislature has determined that the provision of prayer is sufficient to avert misdemeanor liability for neglecting one’s financial responsibility to furnish routine child support. This hardly compels the conclusion that in so doing the Legislature intended to create an unqualified defense to felony manslaughter and child endangerment charges for those parents who continue to furnish prayer alone in the rare instance when a gravely ill child lies dying for want of medical attention.6

In the absence of support from the plain language and purpose of section 270, defendant points to the legislative history of certain amendments to the *127statute as evidence of an intent to exempt prayer treatment from the reach of sections 192(b) and 273a(l). She specifically contends that the 1925 and 1976 amendments to section 270 were enacted to bar manslaughter prosecutions against Christian Scientists. With respect to the 1925 amendment, which added the phrase “or other remedial care” to the statute, defendant rests her account of its legislative history on certain descriptive passages contained in the 1920 and 1925 annual reports of the Christian Science Committee on Publication for Southern California. She is not well served by these materials. First, the documents are so removed from the legislative process that any construction of the statute based on their contents would be patently unreliable. Second, even if we were to consider the reports, they provide virtually no support for her contention.7

Defendant next contends that the 1976 amendment identifying prayer treatment as a form of “other remedial care” was similarly intended to shield Christian Scientists from manslaughter prosecutions. She observes that the amendment was sponsored by the Church in response to our dictum in Arnold, a case involving a misdemeanor-manslaughter prosecution. While legislative materials demonstrate that the amendment was indeed sponsored by the Church in response to Arnold, there is no evidence that the Legislature intended the modification to affect manslaughter, as opposed to misdemeanor, liability. (See Sen. Com. on Judiciary, Analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.); Sen. Republican Caucus, 3d reading analysis of Assem. Bill No. 3843, supra.) The portion of the Arnold opinion cited in the legislative materials as inspiring the amendment focused exclusively on the underlying misdemeanor liability of the parent; none of the documents mentions the relationship between the discussion of section 270 in Arnold and the manslaughter charge elsewhere involved in the case. (Ibid.) While the ensuing amendment necessarily precluded *128misdemeanor-manslaughter prosecutions based on violations of section 270 by eliminating misdemeanor liability under that statute, the Legislature left untouched the possibility of an involuntary manslaughter prosecution based as here on allegations of criminal negligence in the “commission of a lawful act which might produce death, in an unlawful manner . . . (§ 192 (b).)8

The historical materials documenting the enactment of Assembly Bill No. 3843 demonstrate that the members of the Legislature were well aware the legislation left open the possibility of manslaughter and child endangerment prosecutions, but simply declined to extend their amendatory efforts beyond section 270. A staff analysis prepared for the Assembly Committee on Criminal Justice observed that “The bill appears unclear in two respects. First, Section 273a makes it a wobbler (10 year top) for any person to permit a minor under his care or custody to suffer any physical harm or injury. Thus, though the parents may not be hable for failing to provide for the health of the child because they choose treatment by prayer rather than common medical treatment, they would be liable if the child suffered any physiological harm. Second, no exception is made under the manslaughter statutes for parental liability should the child die. If treatment by prayer is to be recognized in part, the parents should not be liable for the results of using a permitted mode of healing.” (Assem. Com. on Criminal Justice, Analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.).) Despite the opinion offered in the final sentence of the staff analysis, no amendments were made to eliminate potential liability under sections 192(b) and 273a(l). The committee passed the bill on April 29, 1976.

When the full Assembly considered the legislation on May 6, 1976, its members had before them the third reading analysis of the bill prepared by the Assembly Office of Research. This analysis incorporated, nearly verbatim, the observations regarding manslaughter and child endangerment liability contained in the earlier analysis prepared for the Criminal Justice Committee. (Assem. Office of Research, 3d reading analysis of Assem. Bill No. 3843, supra.) Again, no amendments were offered in response to the observations. The bill passed and moved to the Senate.

The Senate Committee on Judiciary received a four-page analysis of the legislation. Under the heading “Comment” and entirely capitalized, unlike *129any other portion of the document, were the following questions: “Do the provisions of this bill conflict with section 273a of the Penal Code which makes it a crime for any person to willfully cause or permit a minor under his care or custody to suffer any physical harm or injury? []j] Might a parent be immune from liability for failure to provide for the health of the child because they [s/e] choose treatment by prayer rather than common medical treatment, but incur liability if the child suffers any harm?” (Sen. Com. on Judiciary, Analysis of Assem. Bill No. 3843, supra, capitalization omitted.) As in the Assembly, the Senate committee members chose to disregard the issues raised so prominently in their staff analysis and passed the measure to the full Senate without addressing manslaughter and child endangerment liability. No other reference to sections 192(b) and 273a(l) appears in the historical materials documenting the enactment of Assembly Bill No. 3843.

The ineluctable conclusion we must draw from these materials is that the members of the Legislature were fully conscious of the potential liability remaining under sections 192(b) and 273a(l) for conduct they had legalized with respect to section 270, but simply chose to leave the matter unaddressed. Needless to say, considered silence is an insufficient basis to infer that the Legislature, by amending a misdemeanor support provision, actually exempted from felony liability all parents who offer prayer alone to a dying child. “The failure of the Legislature to change the law in a particular respect when the subject is generally before it and changes in other respects are made is indicative of an intent to leave the law as it stands in the aspects not amended.” (Cole v. Rush (1955) 45 Cal.2d 345, 355 [289 P.2d 450, 54 A.L.R.2d 1137], disapproved on another point in Vesely v. Sager (1971) 5 Cal.3d 153, 167 [95 Cal.Rptr. 623, 486 P.2d 151].) The plain language, purpose, and legislative history of section 270 thus fail to establish a discernible legislative intent to exempt prayer treatment, as a matter of law, from the reach of the manslaughter and felony child-endangerment statutes. (Accord, Note, California's Prayer Healing Dilemma (1987) 14 Hastings Const.L.Q. 395, 401-404.)

B. Expressions of legislative intent in related statutes

Defendant next contends that an intent to exempt prayer treatment from conduct within the reach of sections 192(b) and 273a(l) is implied by a number of other civil and criminal measures relating to the provision of prayer in lieu of medical care to children. She first cites a plethora of statutes exempting prayer practitioners and their facilities from medical licensure requirements9 or variously accommodating individuals who *130choose to rely on such treatment for their own care.10 These accommodative provisions, however, evince no legislative sanction of prayer for the treatment of children in life-threatening circumstances.

More useful are statutes dealing with the definition of neglected or abused children for purposes of the state’s child welfare services program (Welf. & Inst. Code, § 16500 et seq.), the activities of the Office of Child Abuse Prevention (id. § 18950 et seq.), and a criminal provision requiring certain individuals to report instances of suspected child abuse (Pen. Code, § 11165 et seq.). Utilizing substantially similar language, each of these three statutes provides that children receiving treatment by prayer shall not “for that reason alone” be considered abused or neglected for its purposes. (Welf. & Inst. Code, §§ 16509.1 (hereafter W&I section 16509.1) and 18950.5 (hereafter W&I section 18950.5); Pen. Code, § 11165.2 (hereafter section 11165.2).) Defendant cites these provisions as evidence that the Legislature does not consider prayer treatment to be a threat to the health of children, and thus that the imposition of criminal liability for the results of its use is inconsistent with legislative intent.

The Attorney General urges a different construction of the statutory language. He contends that the phrase “for that reason alone ” (italics added) denotes that a child receiving prayer treatment can still fall within the reach of the statutory definitions if the provision of such treatment, coupled with a grave medical condition, combine to pose a serious threat to the physical well-being of the child.11 While defendant contends the phrase *131merely indicates that a child receiving prayer treatment can come within the purview of the statutes for other reasons, her construction renders the use of the word “alone” surplusage and thus must be rejected under the rule that “ ‘[ejvery word, phrase and provision employed in a statute is intended to have meaning and to perform a useful function. . . (White v. County of Sacramento (1982) 31 Cal.3d 676, 681 [183 Cal.Rptr. 520, 646 P.2d 191], quoting Clements v. T R. Bechtel Co. (1954) 43 Cal.2d 227, 233 [273 P.2d 5].)

The code section immediately preceding W&I section 16509.1 in the child welfare services chapter strongly corroborates the interpretation offered by the Attorney General. That section reads: “Cultural and religious child-rearing practices and beliefs which differ from general community standards shall not in themselves create a need for child welfare services unless the practices present a specific danger to the physical or emotional safety of the child.” (Welf. & Inst. Code, § 16509 (hereafter W&I section 16509).) It is fundamental that “the language of a particular code section must be construed in light of and with reference to the language of other sections accompanying it and related to it with a view to harmonizing the several provisions and giving effect to all of them.” (Johnson v. Superior Court (1984) 159 Cal.App.3d 573, 582 [205 Cal.Rptr. 605].) The provision of prayer treatment in lieu of medical care to a gravely ill child doubtless constitutes a religious child-rearing practice “which differ[s] from general community standards . . . .” Nor is there any question that W&I sections 16509 and 16509.1 are related and intended to express a unified statutory objective. To harmonize their provisions accordingly, the use of the word “alone” in W&I section 16509.1 must be construed to signify that treatment by prayer will not constitute neglect for purposes of the child welfare services chapter except in those instances when such treatment, coupled with a sufficiently grave health condition, presents “a specific danger to the physical .. . safety of the child.”12 *132While section 11165.2 and W&I section 18950.5 lack similar companion provisions, there are persuasive reasons why we should interpret them in the same manner. First, each employs the words “for that reason alone” in a context identical to the use of the phrase in W&I section 16509.1. Identical language appearing in separate provisions dealing with the same subject matter should be accorded the same interpretation. (Ford Dealers Assn. v. Department of Motor Vehicles (1982) 32 Cal.3d 347, 359 [185 Cal.Rptr. 453, 650 P.2d 328].) Because each statute deals with the relationship of prayer treatment to the definition of child neglect or abuse, we are obliged to construe their shared language in a consistent fashion. (Dieckmann v. Superior Court (1985) 175 Cal.App.3d 345, 356 [220 Cal.Rptr. 602].) Second, both section 11165.2 and W&I section 18950.5 explicitly refer to the language of W&I section 16509.1 to define the conduct excepted from their definitions of neglect and abuse.13 This fact further suggests that the Legislature intended the several statutes to share a common meaning.

Finally, the most telling indication that the statutes should be construed together in the manner urged by the Attorney General is their mutual interrelation with the child dependency provisions of Welfare and Institutions Code section 300 (hereafter W&I section 300). Furnishing the state with its most powerful tool to intercede on behalf of children threatened at the hands of their parents, W&I section 300 delineates the circumstances under which a child can be removed from parental custody and declared a dependent of the court. Section 11165.2 and W&I sections 16509.1 and 18950.5 are each components of separate acts connected in some significant fashion to the child dependency proceedings outlined in W&I section 300. In sum, the three acts (1) require that suspected instances of child abuse or neglect be reported to the agency responsible for initiating child dependency proceedings under W&I section 300 (Pen. Code, § 11166, subds. (a), (b), & (g)); (2) provide services to neglected or abused children identified through dependency proceedings (Welf. & Inst. Code, §§ 16506, subd. (a), 16507, 16508, subd. (a)); and (3) fund child-abuse prevention projects in cooperation with local welfare agencies responsible for supervising dependency *133proceedings (Welf. & Inst. Code, § 18964, subd. (f)(3)). This intimate interrelation of statutory objectives, revolving around the identification and enforcement provisions of W&I section 300, counsels us to interpret the language of the related acts with reference to the provisions of the dependency statute to “achieve a uniform and consistent legislative purpose.” (Isobe v. Unemployment Insurance Appeals Bd. (1974) 12 Cal.3d 584, 591 [116 Cal.Rptr. 376, 526 P.2d 528]; People v. Caudillo, supra, 21 Cal.3d 562, 585.)

On September 30, 1987, the Governor signed into law Senate Bill No. 243, 1987-1988 Regular Session, which revised W&I section 300 in its entirety. (Stats. 1987, ch. 1485, § 4 [No. 5 Deering’s Adv. Legis. Service, pp. 5779-5780].) Although the legislation will not take effect until January 1, 1989, its provisions dealing with the relationship of prayer treatment to dependency proceedings are critically significant to our interpretive task insofar as they represent the Legislature’s most recent and detailed articulation of the protection to be assured seriously ill children receiving such care.

Newly amended W&I section 300 provides in pertinent part: “Any minor who comes within any of the following descriptions is within the jurisdiction of the juvenile court which may adjudge that person to be a dependent child of the court. . . . [fl] (b) The minor has suffered, or there is substantial risk that the minor will suffer, serious physical harm or illness, ... by the willful or negligent failure of the parent... to provide the minor with adequate food, clothing, shelter, or medical treatment. . . . Whenever it is alleged that a minor comes within the jurisdiction of the court on the basis of the parent’s . . . willful failure to provide adequate medical treatment or specific decision to provide spiritual treatment through prayer, the court shall give deference to the parent’s . . . medical treatment, nontreatment, or spiritual treatment through prayer alone in accordance with the tenets and practices of a recognized church or religious denomination by an accredited practitioner thereof and shall not assume jurisdiction unless necessary to protect the minor from suffering serious physical harm or illness.'” (Italics added.)

Thus in any circumstance involving the threat of “serious physical harm or illness,” the Legislature has empowered the juvenile court to intercede and assume custody for the express purpose of assuring medical care for a child whose parent is furnishing spiritual treatment by prayer alone. The expression of legislative intent is clear: when a child’s health is seriously jeopardized, the right of a parent to rely exclusively on prayer must yield. This intent is implicit in the enumeration of necessities a parent must furnish to avert a dependency proceeding under W&I section 300; conspicuously absent from the list is any substitute for adequate medical treatment. *134It follows that the only tenable construction of the related provisions defining the relationship of prayer treatment to child neglect or abuse is the analysis offered by the Attorney General.

While dependency proceedings are civil rather than criminal, their relevance to our inquiry is plain. Parents possess a profound interest in the custody of their children. (In re Carmaleta B. (1978) 21 Cal.3d 482, 489 [146 Cal.Rptr. 623, 579 P.2d 514]; Holt v. Superior Court (1960) 186 Cal.App.2d 524, 526-527 [9 Cal.Rptr. 353].) “Custody embraces the sum of parental rights with respect to the rearing of a child, including its care. It includes . . . the right to direct his activities and make decisions regarding his care and control, education, health, and religion.” (Burge v. City & County of San Francisco (1953) 41 Cal.2d 608, 617 [262 P.2d 6].) The United States Supreme Court has termed this constellation of parental interests “essential” (Meyers v. Nebraska (1923) 262 U.S. 390, 399 [67 L.Ed. 1042, 1045, 43 S.Ct. 625, 29 A.L.R. 1446]), among the “basic civil rights of man” (Skinner v. Oklahoma (1942) 316 U.S. 535, 541 [86 L.Ed. 1655, 1660, 62 S.Ct. 1110]), and “[r]ights far more precious . . . than property rights” (May v. Anderson (1953) 345 U.S. 528, 533 [97 L.Ed. 1221, 1226, 73 S.Ct. 840]). Consistent with the gravity of the prerogative at stake, parents involved in W&I section 300 proceedings are assured notice and a due process hearing (In re Robert P. (1976) 61 Cal.App.3d 310, 318 [132 Cal.Rptr. 5]) while those who are indigent receive appointed counsel (In re Christina H. (1986) 182 Cal.App.3d 47, 49 [227 Cal.Rptr. 41]; Cal. Rules of Court, rules 1334(c), 1363(c)). The Legislature’s willingness to intrude on a parental interest of such magnitude to assure that children receiving prayer treatment are spared serious physical harm certainly evinces no contrary intent with respect to the application of the penal laws, which in significant respects constitute a less intrusive method of advancing the state’s paramount interest in the protection of its children.

Defendant’s argument by analogy to civil neglect and dependency provisions therefore corroborates rather than refutes our previous determination that the Legislature has created no exemption under sections 192(b) and 273a(l) for parents who are charged with having killed or endangered the lives of their seriously ill children by providing prayer alone in lieu of medical care. The legislative design appears consistent: prayer treatment will be accommodated as an acceptable means of attending to the needs of a child only insofar as serious physical harm or illness is not at risk. When a child’s life is placed in danger, we discern no intent to shield parents from the chastening prospect of felony liability.

C. Defendant’s conduct and the standard for criminal culpability

Taking a wholly different tack, defendant next contends that she cannot be convicted under either the manslaughter or felony child-endangerment *135statutes regardless of the availability of a religious exemption. She rests this contention on a claim that the People will be unable to prove the degree of culpability necessary to convict her under either provision, both of which require criminal negligence in the commission of an offending act. (People v. Penny (1955) 44 Cal.2d 861, 879 [285 P.2d 926]; People v. Peabody (1975) 46 Cal.App.3d 43, 47 [119 Cal.Rptr. 780].) We have defined criminal negligence as “ ‘aggravated, culpable, gross, or reckless, that is, the conduct of the accused must be such a departure from what would be the conduct of an ordinarily prudent or careful man under the same circumstances as to be incompatible with a proper regard for human life, or, in other words, a disregard of human life or an indifference to consequences. . . . [Such negligence] is ordinarily to be determined pursuant to the general principles of negligence, the fundamental of which is knowledge, actual or imputed, that the act of the slayer tended to endanger life.’” (People v. Penny, supra, 44 Cal.2d at pp. 879-880.) Defendant makes two arguments for the claim that her conduct cannot, as a matter of law, constitute such negligence.

She first contends that the defenses recognized at English common law are available to her under Civil Code section 22.2, which reads: “The common law of England, so far as it is not repugnant to or inconsistent with the Constitution of the United States, or the Constitution or laws of this State, is the rule of decision in all the courts of this State.” She cites two English cases from the 19th century in support of the proposition that the common law recognized treatment by prayer in lieu of medicine as legally insufficient to constitute criminal negligence.14 While we note that common law defenses, with limited exceptions, are unavailable in California (Keeler v. Superior Court, supra, 2 Cal.3d at pp. 631-632), we need look no further than the cases themselves to dispose of defendant’s contention.

The opinion of the court in Regina v. Wagstaffe (Cen.Crim.Ct. 1868) 10 Cox. Crim. Cas. 530, consists of a vaguely worded jury charge. The court instructed the jury that criminal negligence “was a very wide question. . . . At different times people had come to different conclusions as to what might be done with a sick person. . . . [A] man might be convicted of manslaughter because he lived in a place where all the community was of a contrary opinion, and in another he might be acquitted because they were all of his opinion. . . .” (Id. at p. 532.) The court asked rhetorically wheth*136er it was “intended by God Almighty that persons should content themselves by praying for His assistance, without helping themselves, or resorting to such means as were within their reach for that purpose?” (ibid.), and concluded with the observation that the defendants appeared sincere and affectionate. Although the defendants were subsequently acquitted, the fact that the jury itself resolved the question of criminal negligence negates the claim that the court in Wagstqffe recognized a legal defense to the charge. Furthermore, its jury instructions merely restated the principle that criminal negligence is a question of fact to be determined in light of contemporary community standards, which at the time made the particular question a close one.

The second case cited by defendant makes this point quite clearly. In Regina v. Hines (1874) 80 Cent. Crim. Ct. 309, the court dismissed an indictment for manslaughter against a parent who had exclusively prayed for an ill child.15 Although the court ruled that the conduct was not criminally negligent as a matter of law, to state the holding is to refute its application 114 years later: the court considered and rejected the proposition that a parent who treated a child by spiritual care “instead of calling in a doctor to apply blisters, leeches, and calomel,” was guilty of criminal negligence. (Id. at p. 312.) Were blisters, leeches and calomel the medical alternative to prayer today, quite likely defendant’s reliance on Hines would more fully resonate with this court. Medical science has advanced dramatically, however, and we may fairly presume that the community standard for criminal negligence has changed accordingly. Nineteenth-century English common law thus fails to establish a defense, as a matter of law, to charges arising today for criminal negligence in the death of a child treated by prayer alone.16

Defendant next contends that her actions are legally insufficient to constitute criminal negligence under the definition of that conduct established in the decisions of this court. Emphasizing her sincere concern and good faith in treating Shauntay with prayer, she claims that her conduct is incompatible with the required degree of culpability. Defendant does not dispute, however, that criminal negligence must be evaluated objectively. (People v. Watson (1981) 30 Cal.3d 290, 296-297 [17 Cal.Rptr. 43, 637 P.2d 279]; People v. Penny, supra, 44 Cal.2d 861, 880.) The question is whether *137“a reasonable person in defendant’s position would have been aware of the risk involved . . . .” (People v. Watson, supra, 30 Cal.3d at p. 296.) If so, “defendant is presumed to have had such an awareness.” (Ibid.)

The significance of this principle was well illustrated in People v. Burroughs (1984) 35 Cal.3d 824 [201 Cal.Rptr. 319, 678 P.2d 894], a case involving a “self-styled ‘healer’ ” who provided “ ‘deep’ abdominal massages” to a leukemic who thereafter died of a massive abdominal hemorrhage. (Id. at pp. 826, 828.) We observed that “There is no allegation made, nor was there any evidence adduced at trial, that [the defendant] at any time harbored any intent even to harm [the victim] in the slightest fashion.” (Id. at p. 834.) “Indeed, nowhere is it claimed that defendant attempted to perform any action with respect to [the victim] other than to heal him . . . .” (Id. at p. 833.) Nonetheless, we determined that the defendant could be charged with criminally negligent involuntary manslaughter. (Id. at p. 836.) The relevant inquiry, then, turns not on defendant’s subjective intent to heal her daughter but on the objective reasonableness of her course of conduct.17

In view of this standard, we must reject defendant’s assertion that no reasonable jury could characterize her conduct as criminally negligent for purposes of sections 192(b) and 273a(l). As the court in People v. Atkins (1975) 53 Cal.App.3d 348 [125 Cal.Rptr. 855], observed in affirming the involuntary manslaughter and felony child-endangerment conviction of a parent whose child died for want of medical care, criminal negligence “could have been found to have consisted of the [mother’s] failure to seek prompt medical attention for [her son], rather than waiting several days. There is evidence she knew, or should have known, that [her son] was seriously injured. . . . Viewing [the evidence] in the light most favorable to the prosecution, there is substantial evidence here of involuntary manslaughter based on the lack of due caution and circumspection in omitting to take the child to a doctor.” (Id. at p. 360.) When divorced of her subjective intent, the alleged conduct of defendant here is essentially indistinguishable.

Defendant’s arguments to the contrary are not persuasive. She first asserts that the various statutory exemptions enacted for Christian Scientists demonstrate a legislative acceptance of the reasonableness of their spiritual care that is incompatible with a finding of “gross, culpable, or reckless” *138negligence. As discussed at length above, however, California’s statutory scheme reflects not an endorsement of the efficacy or reasonableness of prayer treatment for children battling life-threatening diseases but rather a willingness to accommodate religious practice when children do not face serious physical harm. Indeed, the relevant statute suggest that prayer treatment for gravely ill children is sufficiently unreasonable to justify the state in taking the draconian step of depriving parents of their rights of custody. {Ante, at pp. 132-134.)

The two cases cited by defendant in support of her claim are clearly distinguishable. In People v. Rodriguez (1960) 186 Cal.App.2d 433 [8 Cal.Rptr. 863], the court reversed the involuntary manslaughter conviction of a mother who had left her children alone at home where one died in a fire. The court ruled that the mother’s conduct did not reflect a course of conduct sufficiently reckless to justify a finding of criminal negligence. (Id. at pp. 440-441.) In terms of unreasonableness, however, the failure of defendant to seek medical attention for a child who sickened and died over a 17-day period is plainly more egregious than the decision of Mrs. Rodriguez to leave her children alone at home for an afternoon. In Somers v. Superior Court (1973) 32 Cal.App.3d 961 [108 Cal.Rptr. 630], the court granted a writ of prohibition barring the manslaughter prosecution of a police officer who had shot a fleeing youth whom the officer mistook for a felon. The court observed that the situation was “tense and menacing” because of earlier reports of robberies in the vicinity, that the victim matched the description of a suspect and appeared to be carrying a shotgun, and that the victim continued to flee after the officer had shouted “Stop, police.” {Id. at pp. 965, 968-970.) Again, the objective unreasonableness of defendant’s course of conduct, compared with the officer’s actions in Somers, is of an evidently greater magnitude.

In sum, we reject the proposition that the provision of prayer alone to a seriously ill child cannot constitute criminal negligence as a matter of law. Whether this defendant’s particular conduct was sufficiently culpable to justify conviction of involuntary manslaughter and felony child endangerment remains a question in the exclusive province of the jury.

II. Constitutional Defenses

A. Free exercise under the First Amendment

In the absence of a statutory basis to bar defendant’s prosecution, we necessarily reach her constitutional claims. Defendant and the Church first contend that her conduct is absolutely protected from criminal *139liability by the First Amendment to the United States Constitution and article I, section 4, of the California Constitution. We do not agree.

The First Amendment bars government from “prohibiting the free exercise” of religion. Although the clause absolutely protects religious belief, religiously motivated conduct “remains subject to regulation for the protection of society.” (Cantwell v. Connecticut (1940) 310 U.S. 296, 303-304 [84 L.Ed. 1213, 1218, 60 S.Ct. 900, 128 A.L.R. 1352].) To determine whether governmental regulation of religious conduct is violative of the First Amendment, the gravity of the state’s interest must be balanced against the severity of the religious imposition. (Wisconsin v. Yoder (1972) 406 U.S. 205, 221 [32 L.Ed.2d 15, 28, 92 S.Ct. 1526].) If the regulation is justified in view of the balanced interests at stake, the free exercise clause requires that the policy additionally represent the least restrictive alternative available to adequately advance the state’s objectives. (Thomas v. Review Bd, Ind. Empl. Sec. Div. (1981) 450 U.S. 707, 718 [67 L.Ed.2d 624, 634, 101 S.Ct. 1425].)

Defendant does not dispute the gravity of the governmental interest involved in this case, as well she should not. Imposition of felony liability for endangering or killing an ill child by failing to provide medical care furthers an interest of unparalleled significance: the protection of the very lives of California’s children, upon whose “healthy, well-rounded growth . . . into full maturity as citizens” our “democratic society rests, for its continuance . . . .” (Prince v. Massachusetts (1944) 321 U.S. 158, 168 [88 L.Ed. 645, 653, 64 S.Ct. 438].) Balanced against this interest is a religious infringement of significant dimensions. Defendant unquestionably relied on prayer treatment as an article of genuine faith, the restriction of which would seriously impinge on the practice of her religion. We note, however, that resort to medicine does not constitute “sin” for a Christian Scientist (Schneider, Christian Science and the Law: Room for Compromise?, supra, 1 Colum. J.L. & Soc. Probs. at pp. 87-88), does not subject a church member to stigmatization (Talbot, The Position of the Christian Science Church, supra, 26 N.E. Med. J. at p. 1642), does not result in divine retribution (Schneider, op. cit. supra, at pp. 87-88), and, according to the Church’s amicus curiae brief, is not a matter of church compulsion.

Regardless of the severity of the religious imposition, the governmental interest is plainly adequate to justify its restrictive effect. As the United States Supreme Court stated in Prince v. Massachusetts, supra, 321 U.S. at page 170 [88 L.Ed. at p. 654], “Parents may be free to become martyrs themselves. But it does not follow they are free, in identical circumstances, to make martyrs of their children before they have reached the age of full legal discretion when they can make that choice for themselves.” The court *140in Prince considered a free-exercise claim asserted by parents whose religious beliefs required that their children sell religious tracts in violation of child labor laws. If parents are not at liberty to “martyr” children by taking their labor, it follows a fortiori that they are not at liberty to martyr children by taking their very lives. As the court explained, “The right to practice religion freely does not include liberty to expose the community or child to communicable disease or the latter to ill health or death.” (Id. at pp. 166-167 [88 L.Ed. at p. 653]; accord, Wisconsin v. Yoder, supra, 406 U.S. at pp. 233-234 [32 L.Ed.2d at p. 35].)

In an attempt to avoid this inexorable conclusion, the Church argues at length over the purportedly pivotal distinction between the governmental compulsion of a religiously objectionable act and the governmental prohibition of a religiously motivated act. Accepting arguendo the force of the distinction, we find that it has no relevance in a case involving an interest of this magnitude. As the court in Prince recognized, parents have no right to free exercise of religion at the price of a child’s life, regardless of the prohibitive or compulsive nature of the governmental infringement. Furthermore, the United States Supreme Court has specifically sustained the compulsion of religiously prohibited conduct for interests no more compelling than here implicated. In Jacobson v. Massachusetts (1905) 197 U.S. 11, 39 [49 L.Ed. 643, 655, 25 S.Ct. 358], the court upheld a law compelling the vaccination of children for communicable diseases in the face of parental religious objections. In United States v. Lee (1982) 455 U.S. 252, 261 [71 L.Ed.2d 127, 135, 102 S.Ct. 1051], the court upheld a law requiring that the Amish violate the tenets of their faith by participating in the Social Security system. And in Gillette v. United States (1971) 401 U.S. 437, 462 [28 L. Ed.2d 168, 188, 91 S. Ct. 828], the court upheld the government’s right to compel certain conscientious objectors to make war despite the religious character of their objections. We see no basis in these precedents for the conclusion that parents may constitutionally insulate themselves from state compulsion so long as their life-threatening religious conduct takes the form of an omission rather than an act.

The imposition of felony liability for failure to seek medical care for a seriously ill child is thus justified by a compelling state interest. To survive a First Amendment challenge, however, the policy must also represent the least restrictive alternative available to the state. Defendant and the Church argue that civil dependency proceedings advance the governmental interest in a far less intrusive manner. This is not evident. First, we have already observed the profoundly intrusive nature of such proceedings; it is not clear that parents would prefer to lose custody of their children pursuant to a disruptive and invasive judicial inquiry than to face privately the prospect of criminal liability. Second, child dependency proceedings advance the *141governmental interest only when the state learns of a child’s illness in time to take protective measures, which quite likely will be the exception rather than the rule: “Under ordinary circumstances,. . . the case of a true believer in faith healing will not even come to the attention of the authorities, unless and until someone dies.” (Comment, Religious Beliefs and the Criminal Justice System: Some Problems of the Faith Healer, supra, 8 Loyola L. A. L.Rev. at pp. 403-404.) Finally, the imposition of criminal liability is reserved for the actual loss or endangerment of a child’s life and thus is narrowly tailored to those instances when governmental intrusion is absolutely compelled.

We conclude that an adequately effective and less restrictive alternative is not available to further the state’s compelling interest in assuring the provision of medical care to gravely ill children whose parents refuse such treatment on religious grounds. Accordingly, the First Amendment and its California equivalent do not bar defendant’s criminal prosecution. (Accord, Craig v. State (1959) 220 Md. 590 [155 A.2d 684, 690]; People v. Pierson (1903) 176 N.Y. 201 [68 N.E. 243, 245]; Owens v. State (1911) 6 Okla.Crim. 110 [116 P. 345, 347-348]; Commonwealth v. Barnhart (1985) 345 Pa.Super. 10 [497 A.2d 616, 623-624]; Note, California’s Prayer Healing Dilemma, supra, 14 Hastings Const.L.Q. at pp. 412.)

B. Due process right to fair notice of illegal conduct

Article I, section 7, of the California Constitution and the Fourteenth Amendment to the United States Constitution both assure that no person shall be deprived of “life, liberty, or property without due process of law.” Among the implications of this constitutional command is that the state must give its citizenry fair notice of potentially criminal conduct. This requirement has two components: “due process requires a statute to be definite enough to provide (1) a standard of conduct for those whose activities are proscribed and (2) a standard for police enforcement and for ascertainment of guilt.” (Burg v. Municipal Court (1983) 35 Cal.3d 257, 269 [198 Cal.Rptr. 145, 673 P.2d 732], cert. den. 466 U.S. 967 [80 L.Ed.2d 812, 104 S.Ct. 2337]; see also Kolender v. Lawson (1983) 461 U.S. 352, 357-358 [75 L.Ed.2d 903, 908-909, 103 S.Ct. 1855].) Defendant contends that sections 192(b) and 273a(l), when read together with section 270, violate this constitutional dictate.

We initially observe that these statutes do not invite standardless law enforcement. Unlike typical due process challenges involving an ambiguously worded statute applied in an arbitrary and unforeseeable manner (see, e.g„ Lanzetta v. New Jersey (1939) 306 U.S. 451 [83 L.Ed. 888, 59 S.Ct. 618]), we consider here three separate provisions that clearly identify their *142respective proscriptions. (People v. Harris (1966) 239 Cal.App.2d 393, 395-397 [48 Cal.Rptr. 677] [upholding the validity of § 273a on fair notice grounds]; People v. Wilson (1947) 78 Cal.App.2d 108, 114 [177 P.2d 567] [same, § 192(b)]; People v. Yates (1931) 114 Cal.App.Supp. 782, 789 [298 P. 961] [same, § 270].) Even if we accept arguendo defendant’s contention that the intersection of the statutes creates uncertainty on the part of law enforcement officials regarding the legality of prayer treatment when a child’s life is endangered or lost, the officials are nevertheless required to make only one discretionary judgment: whether or not to prosecute conduct otherwise within the reach of the felony statutes in view of the provisions of section 270. This discretion certainly is not “of such a standardless sweep [that it] allows policemen, prosecutors, and juries to pursue their personal predilections.” (Smith v. Goguen (1974) 415 U.S. 566, 575 [39 L.Ed.2d 605, 613, 94 S.Ct. 1242].)

With respect to the remaining component of the due process analysis, defendant makes two arguments why the statutory scheme fails to provide fair notice. She first contends that sections 192(b) and 273a(1) provide no notice of the point at which lawful prayer treatment becomes unlawful, thus requiring her “at peril of life, liberty or property to speculate as to the meaning of penal statutes.” (Lanzetta v. New Jersey, supra, 306 U.S. at p. 453 [83 L.Ed. at p. 890].) She frames her argument in the form of a rhetorical question: “Is it lawful for a parent to rely solely on treatment by spiritual means through prayer for the care of his/her ill child during the first few days of sickness but not for the fourth or fifth day?” Justice Holmes correctly answers: “[T]he law is full of instances where a man’s fate depends on his estimating rightly, that is, as the jury subsequently estimates it, some matter of degree. . . . ‘An act causing death may be murder, manslaughter, or misadventure, according to the degree of danger attending it’ by common experience in the circumstances known to the actor.” (Nash v. United States (1913) 229 U.S. 373, 377 [57 L.Ed. 1232, 1235, 33 S.Ct. 780]; see also Coates v. City of Cincinnati (1971) 402 U.S. 611, 614 [29 L.Ed.2d 214, 217, 91 S.Ct. 1686].) The “matter of degree” that persons relying on prayer treatment must estimate rightly is the point at which their course of conduct becomes criminally negligent. In terms of notice, due process requires no more. {Burg v. Municipal Court, supra, 35 Cal.3d at p. 270.)

Defendant contends in conclusion that the statutory scheme violates her right to fair notice by allowing punishment under sections 192(b) and 273a(l) for the same conduct that is assertedly accommodated under section 270. She argues in essence that the statutes issue “inexplicably contradictory commands” (Raley v. Ohio (1959) 360 U.S. 423, 438 [3 L.Ed.2d 1344, 1356, 79 S.Ct. 1257]) and thus violate due process by precluding “an ordinary person [from] intelligently choosflng], *143in advance, what course it is lawful for him to pursue.” (Connally v. General Construction Co. (1926) 269 U.S. 385, 393 [70 L.Ed. 322, 329, 46 S.Ct. 126].)18

In considering whether a legislative proscription is sufficiently clear to satisfy the requirements of fair notice, “we look first to the language of the statute, then to its legislative history, and finally, to California decisions construing the statutory language.” (Pryor v. Municipal Court (1979) 25 Cal.3d 238, 246 [158 Cal.Rptr. 330, 599 P.2d 636]; People v. Mirmirani (1981) 30 Cal.3d 375, 383 [178 Cal.Rptr. 792, 636 P.2d 1130].) We thus require citizens to apprise themselves not only of statutory language but also of legislative history, subsequent judicial construction, and underlying legislative purposes (People v. Grubb (1965) 63 Cal.2d 614, 620 [47 Cal.Rptr. 772, 408 P.2d 100]). (See generally Amsterdam, The Void-For-Vagueness Doctrine in the Supreme Court (1960) 109 U. Pa. L.Rev. 67.) These principles express the strong presumption that legislative enactments “must be upheld unless their unconstitutionality clearly, positively, and unmistakably appears. [Citations.] A statute should be sufficiently certain so that a person may know what is prohibited thereby and what may be done without violating its provisions, but it cannot be held void for uncertainty if any reasonable and practical construction can be given to its language.” (Lockheed Aircraft Corp. v. Superior Court (1946) 28 Cal.2d 481, 484 [171 P.2d 21, 166 A.L.R. 701], citations omitted.)

As we have discussed at length above, the purposes of the statutes here at issue are evidently distinguishable: sections 192(b) and 273a(l) *144protect against grievous and immediate physical harm while section 270 assures the routine provision of child support at parental expense. (Ante, at pp. 124-126.) In light of these distinguishable objectives, it cannot be said that the legality of defendant’s conduct under section 270 constitutes an “inexplicably contradictory command” with respect to the separate requirements of sections 192(b) and 273a(l). Indeed, the legislative history of section 270 specifically demonstrates the Legislature’s unwillingness to extend the statute’s religious exemption to the felony provisions. (Ante, at pp. 127-129.) Sections 270, 192(b), and 273a(l) thus provided constitutionally sufficient notice to defendant that the provision of prayer alone to her daughter would be accommodated only insofar as the child was not threatened with serious physical harm or illness.19

III. Disposition

We conclude that the prosecution of defendant for involuntary manslaughter and felony child endangerment violates neither statutory law nor the California or federal Constitution. The judgment of the Court of Appeal is affirmed.

Lucas, C. J., Panelli, J., Arguelles, J., Eagleson, J., and Kaufman, J., concurred.

MOSK, J., Concurring

My opinion prepared for the court holds as a matter of statutory construction that Penal Code section 270 (hereafter section 270) provides no religious defense to charges arising under the manslaughter and felony child-endangerment statutes. Because of this holding, the majority chose not to reach the Attorney General’s separate contention that an extension of section 270’s religious exemption to this felony prosecution would import into the proceeding a defense that offends the establishment clauses of the state and federal Constitutions. The issue, however, has been timely raised and thoroughly briefed, and its importance is manifest. I believe we should address it in this case for the guidance of the Legislature, so that any further legislative efforts to accommodate religious *145practice will comply with this constitutional command. As will appear, in my view the statutory exemption as it now reads plainly violates the establishment clauses.

The California and federal Constitutions admonish the Legislature “to make no law respecting an establishment of religion.” (U.S. Const., 1st Amend.; Cal. Const., art. I, § 4.) Two broad classes of legislation fall under this proscription: laws “affording a uniform benefit to all religions” and laws “that discriminate among religions.” (Larson v. Valente (1982) 456 U.S. 228, 252 [72 L.Ed.2d 33, 52-53, 102 S.Ct. 1673], italics in original.) The constitutionality of the first class of enactments is traditionally measured against three criteria delineated by the United States Supreme Court in Lemon v. Kurtzman (1971) 403 U.S. 602 [29 L.Ed.2d 745, 91 S.Ct. 2105]. Under Lemon, a law must first have a secular legislative purpose; second, its principal or primary effect must neither advance nor inhibit religion; and third, it must not foster an excessive governmental entanglement with religion. (Id. at pp. 612-613 [29 L.Ed.2d at pp. 755-756].)

Laws in the second class strike closer to the heart of the establishment clause prohibition and thus require more demanding scrutiny. (Larson v. Valente, supra, 456 U.S. at p. 252 [72 L.Ed.2d at p. 52].) As the Larson court observed, “The clearest command of the Establishment Clause is that one religious denomination cannot be officially preferred over another.” (Id. at p. 244 [72 L.Ed.2d at p. 47].) The essential attribute of this constitutional dictate is governmental neutrality with respect to matters of faith: “Government in our democracy, state and national, must be neutral in matters of religious theory, doctrine, and practice. It may not be hostile to any religion or to the advocacy of no-religion; and it may not aid, foster, or promote one religion or religious theory against another or even against the militant opposite. The First Amendment mandates governmental neutrality between religion and religion, and between religion and nonreligion.” (Epperson v. Arkansas (1968) 393 U.S. 97, 103-104 [21 L.Ed.2d 228, 234, 89 S.Ct. 266].)1 If a law effects a preference among religions, the governmental policy is presumptively suspect and subject to strict scrutiny. (Larson v. Valente, supra, 456 U.S. at p.246 [72 L.Ed.2d at p. 49].)

The court in Larson considered a provision exempting from the reporting and registration requirements of a charitable solicitations act only those *146religious organizations receiving over 50 percent of their contributions from members or affiliated organizations. (Id. at p. 231 [72 L.Ed.2d at p. 39].) By plainly discriminating among religions in the allocation of the statutory exemption, the provision granted a denominational preference requiring strict scrutiny. (Id. at p. 246.) While acknowledging that the state had “a significant interest in protecting its citizens from abusive practices in the solicitation of funds for charity, and that this interest retains importance when the solicitation is conducted by a religious organization” (id. at p. 248 [72 L.Ed.2d at p. 50]), the court nonetheless determined that the 50 percent rule was not sufficiently tailored to the statutory objective to withstand constitutional scrutiny. (Id. at p. 251 [72 L.Ed.2d at p. 52].)

Section 270 similarly allocates its religious benefit on a selective basis. The statute excludes from criminal liability any parent who provides a minor with “treatment by spiritual means through prayer alone in accordance with the tenets and practices of a recognized church or religious denomination, by a duly-accredited practitioner thereof’’'' (Italics added.) The provision thus affords no protection for parents who otherwise treat their children “by spiritual means through prayer alone.” Specifically denied the exemption are (1) parents not affiliated with a “recognized” church or religious denomination who nonetheless provide prayer treatment on the basis of personal religious beliefs or the teachings of an unrecognized sect, and (2) parents who provide prayer treatment in accordance with the tenets of a recognized denomination that does not “accredit” prayer “practitioners.”

These excluded believers are not the fanciful product of a strained reading of the statutory language. In People v. Arnold (1967) 66 Cal.2d 438 [58 Cal.Rptr. 115, 426 P.2d 515], this court considered a religious exemption claimed under section 270 by a member of “the Church of the First Born,” described as “a religious group believing in faith healing.” (Id. at p. 442, fn. 1.) Whether, and on what basis, a court would determine that the Church of the First Born constitutes a “recognized” religion is a serious question not easily answered.2 Furthermore, while the opinion in Arnold states that members of the group prayed with the defendant for a cure, there is no indication that they were “duly accredited practitioner[s]” of prayer treatment. *147{Ibid.) If not, Mrs. Arnold would have been denied the current statutory exemption even if the Church of the First Born had been “recognized.” Indeed, certain well-known denominations decline to term anyone a “healer”: “The so-called Pentecostal sects have some members who actively seek and encourage ‘divine intervention,’ but they do not ordinarily perform acts that are thought to ‘heal’ a sick person. In this sense, there are no ‘healers’; the cure is thought to come directly from God.” (Comment, Religious Beliefs and the Criminal Justice System: Some Problems of the Faith Healer (1975) 8 Loyola L.A. L.Rev. at pp. 413-414.)

Also denied the statutory exemption are parents whose use of prayer treatment stems from personal religious beliefs rather than the tenets of a recognized church or denomination. In other jurisdictions such parents have repeatedly prevailed on establishment and equal protection grounds against similarly formulated provisions. In Davis v. State (1982) 294 Md. 370 [451 A.2d 107], the court considered a challenge to a compulsory immunization statute exempting children whose parents objected because the procedure “‘conflicts with the tenets and practice of a recognized church or religious denomination of which he is an adherent or member ....’” {Id. at pp. 108-109, italics deleted.) The plaintiff “rested his objection [to the immunization] on his personal religious views rather than the tenets of any recognized church or religious denomination of which he was a member or adherent.” (Id. at p. 109.) In Dalli v. Board of Education (1971) 358 Mass. 753 [267 N.E.2d 219], the plaintiff fell outside a similar exception to a compulsory immunization statute because she objected on the basis of “her personal ‘belief in the Bible, and its teachings.’ ” (Id. at p. 220.) Finally, the court in Brown v. Stone (Miss. 1979) 378 So.2d 218, certiorari denied 449 U.S. 887 [66 L.Ed.2d 112, 101 S.Ct. 242], considered the claim of a parent whose minister submitted the following statement: “ ‘Be it known that the [Cjhurch of Christ as a religious body does not teach against the use of . . . immunizations or vaccinations .... However, [plaintiff] who is a member of the . . . Church of Christ . . . does have strong convictions against the use of any kind of medications and we respect his views.’ ” (Id. at pp. 219-220; see also Maier v. Besser (1972) 73 Mise.2d 241 [341 N.Y.S.2d 411, 412]; cf. Lewis v. Califano (3d Cir. 1980) 616 F.2d 73, 75.) Were these parents charged with failure to furnish medical attendance under section 270, they would face conviction regardless of their alternative provision of “treatment by spiritual means through prayer alone.”

The one group of parents squarely protected by the terms of the statute are Christian Scientists, whose denomination sponsored the 1976 amendment to section 270 enacting its religious exemption. It is thus more than fortuity that the word “practitioner,” used by Christian Scientists to formal*148ly designate their healers, also appears in section 270 to describe the required providers of the exempted treatment. As the analysis of the amendment prepared for the Senate Committee on Judiciary frankly observed, the purpose of the legislation was to “Ensure that no parent who uses Christian Science methods to heal his or her child shall be liable for not providing recognized medical attention for the children.” (Sen. Com. on Judiciary, Analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.), italics added; accord, Sen. Democratic Caucus, 3d reading analysis of Assem. Bill No. 3843 (1975-1976 Reg. Sess.).)

By sparing the favored from criminal liability while condemning others for failure to cloak identical conduct in the mantle of a sanctioned denomination or procedure, the religious exemption of section 270 operates without neutrality “in matters of religious theory, doctrine, and practice,” and thus cannot survive in the absence of a compelling state interest in its discriminatory effect. Unlike the exemption in Larson, however, which advanced an independent secular objective, the only discernible state interest in this exemption is religious accommodation per se. While accommodation has been sustained as a legitimate objective when it “reflects nothing more than the governmental obligation of neutrality in the face of religious differences” (Sherbert v. Verner (1963) 374 U.S. 398, 409 [10 L.Ed.2d 965, 83 S.Ct. 1790]), here the accommodation reflects nothing less than a denominational preference in the face of indistinguishable religious conduct. Manifestly this is not a compelling objective in the constitutional sense.

If the Legislature wishes to exempt from criminal liability those parents who rely on prayer treatment in lieu of medical care, the establishment clause requires at a minimum that the exemption be granted irrespective of denominational affiliation or practice. (Gillette v. United States (1971) 401 U.S. 437, 454 [28 L.Ed.2d 168, 183, 91 S. Ct. 828]; Lewis v. Califano, supra, 616 F.2d at p. 78 [“The establishment clause requires the government to extend the same benefits it currently extends to Christian Scientists ... to all individuals who sincerely believe in faith healing”]; Developments in the Law: Religion and the State (1987) 100 Harv.L.Rev. 1606, 1738 [“The only legislative accommodations that can withstand establishment clause scrutiny are those that accommodate all religious objectors equally”].)3 The conclusion is thus inescapable that the religious exemption found in section 270 violates the establishment clauses of the California and federal Constitutions. (Accord, Dalli v. Board of Education, supra, 267 N.E.2d at p. 223; Davis v. State, supra, 451 A.2d at pp. 113-114; State v. Miskimens (1984) 22 *149Ohio Misc.2d 43 [490 N.E.2d 931, 934]; Maier v. Besser, supra, 341 N.Y.S.2d at p. 414; Kolbeck v. Kramer (1964) 84 N.J.Super. 569 [202 A.2d 889, 892]; Note, California's Prayer Healing Dilemma (1987) 14 Hastings Const.L.Q. 395, 412-414; Comment, Religious Beliefs and the Criminal Justice System: Some Problems of the Faith Healer (1975) 8 Loyola L.A. L.Rev. 396, 429.)4

The exemption in section 270 is also invalid under the criteria set forth in Lemon v. Kurtzman, supra, 403 U.S. at pages 612-613 [29 L.Ed.2d at pages 755-756], even though “the Lemon v. Kurtzman ‘tests’ are intended to apply to laws affording a uniform benefit to all religions, and not to provisions . . . that discriminate among religions.” (Larson v. Valente, supra, 456 U.S. at p. 252 [72 L.Ed.2d at pp. 52-53], italics in original, fn. omitted.) As noted above, Lemon requires that a statute (1) have a secular purpose, (2) neither advance nor inhibit religion, and (3) foster no excessive entanglement between government and religion.

I have already observed that the accommodative purpose of this statutory exemption reflects a nonsecular preference among adherents of prayer treatment rather than a neutral governmental response to genuine religious differences. While one might charitably argue that the exemption has the effect of identifying indicia of sincere religious conduct, thus facilitating administration of the statute, discrimination subject to the strictest scrutiny *150cannot be justified on the basis of administrative convenience alone. (Cf. Frontiero v. Richardson (1973) 411 U.S. 677, 690 [36 L.Ed.2d 583, 594, 93 S.Ct. 1764].) Furthermore, the indicia are underinclusive and more likely to complicate rather than facilitate administration by requiring theological and social judgments that law enforcement officials and courts are not equipped to make. These administrative complications are closely linked to the troubling entanglement of church and state that the provision invites and Lemon forbids.

To apply section 270, law enforcement officials and courts are required to evaluate “the tenets and practices” of various religions, searching for a doctrinal sanction of “spiritual treatment by prayer alone”; they are called upon to consider whether individual healers have been “duly accredited” by a particular denomination; and most disturbing, they are required to ascertain whether a particular religious group is “recognized.” This last inquiry requires prosecutors and law enforcement officials to judge in their discretion whether a particular religious group has reached the critical mass of size and acceptance necessary for statutory protection, and leaves courts with nothing but subjective experience and belief to guide the required determination.

The assistant legal affairs secretary to the Governor clearly anticipated the foregoing troubling scenario in her preenactment analysis of the 1976 amendment to section 270: “The bill requires that the religion or denomination be ‘recognized.’ No further definition is provided. While this would constitute a severe problem in the medical emergency situation, it would not present a problem where there is sufficient opportunity to argue the problem of religion.” It is precisely the entangling prospect of public officials arguing “the problem of religion” as an aspect of their ongoing enforcement of section 270, coupled with the politically divisive implications of their judgments, that the establishment clause seeks to avert. (Larson v. Valente, supra, 456 U.S. at pp. 252-255 [72 L.Ed.2d at pp. 52-55].)

Defendant maintains that even if the statutory exemption violates the establishment clause, courts should nevertheless construe it in an edited fashion to avert the constitutional difficulties raised by its facially preferential language. This cannot be done. It is of course fundamental that “If feasible within bounds set by their words and purposes, statutes should be construed to preserve their constitutionality.” (Conservatorship of Hofferber (1980) 28 Cal.3d 161, 175 [167 Cal.Rptr. 854, 616 P.2d 836].) This is a limited interpretive function, however, and in no way delegates to courts the authority to rewrite the work of the Legislature. As Justice Tobriner observed with respect to the constitutionality of another provision of section 270, “If elimination of objectionable parts of a statute requires a wholesale *151rewriting, a court’s attempt to do so transgresses both the legislative intent and the judicial function.” (In re King (1970) 3 Cal.3d 226, 237 [90 Cal.Rptr. 15, 474 P.2d 983], cert. den. 403 U.S. 931 [29 L.Ed.2d 709, 91 S.Ct. 2249]; sec Metromedia, Inc. v. City of San Diego (1982) 32 Cal.3d 180, 187 [185 Cal.Rptr. 260, 649 P.2d 902].) Here, too, the unconstitutional aspect of the statute so clearly manifests a particular legislative intent that its elimination would amount to amendment by judicial fiat.

The Legislature has repeatedly designed statutory exemptions for parental use of prayer treatment with precisely the language found in section 270. Welfare and Institutions Code section 16509.1 excludes from its definition of neglect the provision of “treatment solely by spiritual means through prayer alone in accordance with the tenets and practices of a recognized church or religious denomination by a duly accredited practitioner thereof . . . .” Section 11165.2 of the Penal Code and Welfare and Institutions Code section 18950.5 incorporate this definition by reference in their separate provisions defining the relationship of prayer treatment to child abuse and neglect. Finally, newly amended Welfare and Institutions Code section 300 adopts the language nearly verbatim in its provision defining the relationship of prayer treatment to child dependency proceedings. (See also Welf. & Inst. Code, § 5006.) Repeated use of the precise language evinces a legislative affinity for an invalid formulation that cannot be ignored.

Had the Legislature confronted the choice of extending its religious exemption to all parents who sincerely rely on prayer treatment, no matter how unorthodox or unconventional their creed may appear, or alternatively to none at all, one cannot presume that it would have chosen the former rather than the latter option. The statutory provision thus must be considered as written. If the Legislature seeks to accommodate the practice of prayer treatment, it must more clearly evince its intent to do so in a nonpreferential manner to avert the fatal constitutional defects afflicting section 270.

Kaufman, J., concurred.

BROUSSARD, J., Concurring and Dissenting

I agree with the majority that a prosecution may be maintained against petitioner for involuntary manslaughter. (Pen. Code, § 192, subd. (b).)1 However, I cannot agree that the child endangerment provisions of section 273a are applicable to cases where the parent has omitted to provide necessary medical attendance. Rather the failure to provide necessary medical attendance is made punishable by section 270, and section 273a is not applicable to omissions to *152provide care but to active conduct endangering the child. Moreover, even if the failure to provide necessary medical attendance were punishable under section 273a, the prayer exemption of section 270 must be read into section 273a or the exemption is pointless. It is overwhelmingly clear that the Legislature sought to preclude child endangerment liability of persons coming within the religious exemption in section 270, particularly where prayer is successful, and to apply section 273a to such persons would defeat the legislative intent rendering the religious exemption meaningless.

Section 270 provides in part: “If a parent of a minor child willfully omits, without lawful excuse, to furnish necessary . . . medical attendance ... he or she is guilty of a misdemeanor . . .”2 Section 273a provides that a person who “willfully causes or permits” child abuse or endangerment of the child’s health or person is guilty of a felony or a misdemeanor. The section is divided into two subdivisions which use identical language to describe the conduct proscribed except that subdivision (1) of the section applies “under *153circumstances or conditions likely to produce great bodily harm or death” and subdivision (2) of the section applies “under circumstances or conditions other than those likely to produce great bodily harm or death.” Subdivision (1) provides that its violation is punished by imprisonment in the county jail not exceeding one year or in the state prison for two, four or six years. Violation of subdivision (2) is a misdemeanor.3

We must interpret the statutes “in accordance with applicable rules of statutory construction, fundamental among which are those which counsel that the aim of such construction should be the ascertainment of legislative intent so that the purpose of the law may be effectuated [citation]; that a statute should be construed with reference to the entire statutory system of which it forms a part in such a way that harmony may be achieved among the parts [citations]; and that courts should give effect to statutes ‘according to the usual, ordinary import of the language employed in framing them.’ [Citations.]” (Merrill v. Department of Motor Vehicles (1969) 71 Cal.2d 907, 918 [80 Cal.Rptr. 89, 458 P.2d 33]; People v. Woodhead (1987) 43 Cal.3d 1002, 1007-1008 [239 Cal.Rptr. 656, 741 P.2d 154]; People ex rel. Younger v. Superior Court (1976) 16 Cal.3d 30, 40 [127 Cal.Rptr. 122, 544 P.2d 1322].)

When we follow the fundamental rules of statutory construction it is clear that sections 270 and 273a are both concerned with the protection of the health and person of children, that section 270 is applicable to a willful failure to provide necessary care and that section 273a is not applicable to a failure to provide medical care but to willful active conduct causing harm or endangering the child’s health or person.

There can be no rational doubt that the Legislature intended that section 270 should be applicable where a parent fails to provide medical care endangering the health or person of a child. The language of the section speaks of the omission to “furnish necessary . . . medical attendance.” Medical attendance is only necessary when the health or person of the child is *154endangered. The plain language of the first sentence of the section shows that the section applies to child endangering conduct. If the parent in the instant case had not come within the prayer exemption, it is clear that she could be prosecuted under section 270. (See People v. Arnold (1967) 66 Cal.2d 438, 451, 452 [58 Cal.Rptr. 115, 426 P.2d 515] [upholding misdemeanor manslaughter charge on the basis of section 270 prior to adoption of the religious exemption].)

Section 270 is not merely an economic regulation requiring reimbursement of those providing medical attendance. While the third sentence of the section provides that a parent is not relieved of criminal liability because another furnishes the medical care, that sentence may not be read as prohibiting criminal liability when no one provides the necessary support. People v. Sorenson (1968) 68 Cal.2d 280, 287 [66 Cal.Rptr. 7, 437 P.2d 495, 25 A.L.R.3d 1093], the principal case relied upon by the majority in concluding that the purpose of section 270 is reimbursement (maj. opn., p. 124), expressly states that “ ‘the principal statutory objectives are to secure support of the child’ ” and to protect the public fisc. If there was any doubt as to the legislative intent that the statute should be applied when the child does not receive necessary medical attendance as well as when the state seeks reimbursement, such doubt is dispelled by the adoption of the religious exemption in the last sentence of the section. As the majority recognize, the legislative history shows that the Legislature sought to “shield from liability those parents who provide prayer in lieu of medical care” (maj. opn., pp. 122-123), and it would be absurd to conclude that by adopting that provision the Legislature intended only to exempt a parent from a duty to pay for medical care which was not furnished.

Accordingly section 270, like section 273a, is applicable to child endangerment, and both sections are applicable whether or not the child is injured (see People v. Peabody (1975) 46 Cal.App.3d 43, 46 [119 Cal.Rptr. 780]; People v. Harris (1966) 239 Cal.App.2d 393, 398 [48 Cal.Rptr. 677]). Both sections are found in the same chapter of the Penal Code. The statutes are in pari materia. (People v. Caudillo (1978) 21 Cal.3d 562, 585 [146 Cal.Rptr. 859, 580 P.2d 274]; see 2A Sutherland, Statutory Construction (Sands, 4th ed. 1984) § 51.03, p.467.) Accordingly, it is our duty to construe them together and harmonize them.

We cannot reject application of the pari materia rule on the grounds that section 270 deals not only with child endangerment but also with reimbursement or that section 273a deals not only with child endangerment but also with child abuse. If the pari materia rule were limited to identical statutes, it would serve no purpose at all and could never be applied. The basis of the pari materia rule is that both statutes share the same purpose or *155object. (Ibid.) The pari materia rule applies although the statutes may have additional dissimilar objectives so long as they also share the same common objective.

When we harmonize the statutes, the result is clear. Section 270 imposes a duty upon parents to provide the identified “necessary . . . medical attendance,” and imposes criminal liability when the parent “willfully omits” to do so, thereby endangering the child. Section 273a imposes criminal liability for willfully causing or permitting child endangerment. To avoid conflict between the sections, section 273a should not be construed to apply when the asserted criminal conduct is the omission to perform the duties imposed by section 270, but only when the basis of the child endangerment is active conduct endangering the child, willfully causing or permitting child endangerment. The only active conduct shown by the evidence is that petitioner prayed. Prayer is not prohibited by section 273a.

Moreover, even if it is concluded that the failure to provide necessary medical attendance is punishable under section 273a in cases where the section 270 prayer exemption is inapplicable, we may not apply section 273a to cases where that exemption applies. The conclusion is unavoidable that the Legislature intended to exempt parents who utilize prayer treatment from the statutory requirement to provide “necessary . . . medical attendance.” As pointed out above, medical attendance is necessary when its absence endangers the health or person. To hold that section 273a applies to parents who utilize prayer treatment in accordance with the exemption in section 270 means that those exempt may always be prosecuted under section 273a for child endangerment and that, since injury is unnecessary for child endangerment, it would be irrelevant whether God answered the prayers.

The legislative intent to provide some exemption from criminal liability is overwhelmingly clear, although the extent of the exemption may not be clear. The exemption is obviously directed at the duty to protect the child by securing medical attendance imposed by section 270. It would be unrealistic and contrary to all of the legislative history we have been furnished to conclude that the exemption is directed at the economic aspect of section 270.4 There is nothing in the legislative history to indicate that the Legisla*156ture sought to eliminate a nonexistent duty to pay for medical services which were never rendered or was concerned primarily with reimbursement for medical services paid for by others.

The religious exemption must be applied to the child endangerment provisions of section 273a or the legislative intent is totally defeated. It must be applied to cases where the failure to provide necessary medical attendance endangers the child’s health but does not result in harm.

I would reverse the judgment of the Court of Appeal with directions to grant the petition for writ of prohibition insofar as it seeks dismissal of the section 273a charge and to deny it insofar as it seeks dismissal of the manslaughter charge.

Petitioner’s application for a rehearing was denied January 9, 1989.

Opinion

FRANSON, Acting P. J.

Introduction

This appeal challenges the California felony-murder rule as it applies to an unintentionally caused death during a high speed automobile chase following the commission of a nonviolent, daylight burglary of an unattended motor vehicle. Solely by force of precedent we hold that the felony-murder rule applies and respondents can be prosecuted for first degree murder.

Statement of the Case and Facts

Respondents were charged by information with murder (Pen. Code, § 187) and several counts of burglary. In response to a Penal Code section 995 motion to set aside the information, the trial court dismissed the murder charge and amended the information to substitute a vehicular manslaughter charge under Penal Code section 192, subdivision 3, paragraph (a). The People have appealed.

The pertinent facts are as follows: On Sunday, February 20, 1977, at about 8:30 a.m., uniformed Cadet Police Officer Guy Ballesteroz was on routine patrol in his vehicle, proceeding southbound on Blackstone Avenue in the City of Fresno. As the officer approached the Fresno Dodge car lot, he saw an older model Plymouth parked in front of the lot. He also saw respondents rolling two tires apiece toward the Plymouth. *622His suspicions aroused, the officer radioed the dispatcher and requested that a police unit be sent.

Officer Ballesteroz kept the respondents under observation as he proceeded past the car lot and stopped at the next intersection. As he reached that point he saw the respondents stop rolling the tires and walk to the Plymouth on the street. Ballesteroz made a U-turn and headed northbound on Blackstone. The respondents got into the Plymouth and drove away “really fast.” Thereafter, a high speed chase ensued which eventually resulted in respondents’ car running a red light at the intersection of Blackstone and Barstow Avenues and striking another automobile which had entered the intersection. The driver of the other automobile was killed. Respondents were arrested at the scene. The chase from the car lot covered some 7 miles and lasted approximately 10 to 12 minutes. During the chase the respondents’ car narrowly missed colliding with several other cars including two police vehicles that were positioned to block their escape.

Later investigation revealed that four locked Dodge vans at the car lot had been forcibly entered and the spare tires removed. Fingerprints from both of the respondents were found on the jack stands in some of the vans.

Respondents May Be Prosecuted for First Degree Felony Murder

Penal Code section 189 provides, in pertinent part: “All murder . . . which is committed in the perpetration of, or attempt to perpetrate, arson, rape, robbeiy, burglary, mayhem, or [lewd acts with a minor], is murder of the first degree; . . .” (Italics added.) This statute imposes strict liability for deaths committed in the course of one of the enumérated felonies whether the killing was caused intentionally, negligently, or merely accidentally. (People v. Cantrell (1973) 8 Cal.3d 672, 688 [105 Cal.Rptr. 792, 504 P.2d 1256]; People v. Coefield (1951) 37 Cal.2d 865, 868 [236 P.2d 570].) Malice is imputed and need not be shown. (People v. Burton (1971) 6 Cal.3d 375, 384-385 [99 Cal.Rptr. 1, 491 P.2d 793]; People v. Ireland (1969) 70 Cal.2d 522, 538 [75 Cal.Rptr. 188, 450 P.2d 580, 40 A.L.R. 3d 1323].) The purpose of the felony-murder rule is to deter felons from killing negligently or accidentally. (People v. Washington (1965) 62 Cal.2d 111, 781 [44 Cal.Rptr. 442, 402 P.2d 130]; see Holmes, The Common Law, pp. 56-57.)

*623Burglary falls expressly within the purview of California’s first degree felony-murder rule. Any burglaiy within Penal Code section 459 is sufficient to invoke the rule. (People v. Talbot (1966) 64 Cal.2d 691, 705 [51 Cal.Rptr. 417, 414 P.2d 633]; People v. Thomas (1975) 44 Cal.App.3d 573, 575 [117 Cal.Rptr. 855]; People v. Earl (1973) 29 Cal.App.3d 894, 900 [105 Cal.Rptr. 831].) Whether or not the particular burglary was dangerous to human life is of no legal import. (Earl, supra.)

The meaning of murder committed “in the perpetration of’ a felony within Penal Code section 189 also is clear. The Supreme Court has stated that this language does not require a strict causal relation between the felony and the killing; it is sufficient if both are “parts of one continuous transaction.” (People v. Welch (1972) 8 Cal.3d 106, 118 [104 Cal.Rptr. 217, 501 P.2d 225]; People v. Mason (1960) 54 Cal.2d 164, 169 [4 Cal.Rptr. 841, 351 P.2d 1025].) Flight following a felony is considered part of the same transaction as long as the felon has not reached a “place of temporary safety.” (People v. Salas (1972) 7 Cal.3d 812, 822 [103 Cal.Rptr. 431, 500 P.2d 7, 58 A.L.R.3d 832]; People v. Boss (1930) 210 Cal. 245, 250 [290 P. 881].) Whether the defendant has reached such a place of safety is a question of fact for the jury. Respondents’ reliance on People v. Ford (1966) 65 Cal.2d 41, 56 [52 Cal.Rptr. 228, 416 P.2d 132] for the proposition that this is a legal question is misplaced. In Ford, the court held that it was a question of law in that particular case because many hours had elapsed between the felony and the killing, and there was no evidence that the defendant was attempting to escape at the time of the killing.

Respondents argue that although California has applied the felony-murder rule to escaping robbers no case has applied the rule to escaping burglars. They cite dicta in People v. Boss, supra, 210 Cal. 245, 251, to support the distinction between those escaping from robberies: “Robbery, unlike burglary is not confined to a fixed locus, but is frequently spread over considerable distance and varying periods of time. The escape of the robbers with the loot, by means of arms, necessarily is as important to the execution to the plan as gaining possession of the property.” (Id., at p. 251.)

This distinction does not withstand analysis. A burglary predicated on theft can be committed with equal or greater violence than a robbery, and leaving the scene with the stolen property is equally important. Moreover, the Boss dicta has not been cited to support such a distinction in any other California case. Furthermore, other states do not draw a distinction *624between burglary and robbery flight. To the contrary, the felony-murder rule has been applied to unintended deaths in the course of burglary flight. (See, e.g., People v. Hickman (1973) 12 Ill.App.3d 412 [297 N.E.2d 582]—flight by armed defendants who had burglarized warehouse in nighttime; Gore v. Leeke (1973) 261 S.C. 308 [199 S.E.2d 755]; Commonwealth v. Carey (1951) 368 Pa. 157 [82 A.2d 240]—armed defendant fled from residence he burglarized at night; State v. Ryan (1937) 192 Wash. 160 [73 P.2d 735]; Lakes v. State (1937) 61 Okla.Crim. 252 [67 P.2d 457];State v. Adams (1936) 339 Mo. 926 [98 S.W.2d 632, 108 A.L.R. 838]—defendants fleeing after burglarizing a filling station in nighttime; Francis v. State (1919) 104 Neb. 5 [175 N.W. 675]—armed defendants fleeing after burglarizing store buildings; Conrad v. State (1906) 75 Ohio St. 52 [78 N.E. 957]—fleeing defendants had burglarized a home; see generally Felony-Murder Rule—“Termination of Felony” (1974) 58 A.L.R.3d 851, 962-975.) Thus, the trial court erred in striking the murder count premised upon the felony-murder rule.

We deem it appropriate, however, to make a few observations concerning the irrationality of applying the felony-murder rule in the present case. In People v. Washington, supra, 62 Cal.2d 777, 783, a case limiting the rule’s application to killings committed by the defendant or his accomplice, our Supreme Court stated: “The felony-murder rule has been criticized on the grounds that in almost all cases in which it is applied it is unnecessary and that it erodes the relation between criminal liability and moral culpability. [Citations.] Although it is the law in this state (Pen. Code, § 189), it should not be extended beyond any rational function that it is designed to serve.” (Italics added.) In People v. Phillips (1966) 64 Cal.2d 574 [51 Cal.Rptr. 225, 414 P.2d 353], the court elaborated: “We have thus recognized that the felony-murder doctrine expresses a highly artificial concept that deserves no extension beyond its required application. Indeed, the rule itself has been abandoned by the courts of England, where it had its inception. It has been subjected to severe and sweeping criticism.” (Id., at pp. 582-583, fns. omitted.) The Phillips court explained, “The felony-murder doctrine has been censured not only because it artificially imposes malice as to one crime because of defendant’s commission of another but because it anachronistically resurrects from a bygone age a ‘barbaric’ concept that has been discarded in the place of its origin. . . .” (Id., at p. 583, fn. 6.)

The Supreme Court has recently reaffirmed its dislike of the felony-murder rule in People v. Henderson (1977) 19 Cal.3d 86, 92-93 [137 Cal.Rptr. 1, 560 P.2d 1180]. The literature is replete with criticism of the *625rule. See, for example, Perkins on Criminal Law (2d ed. 1969) page 44; Packer, The Case for Revision of the Penal Code (1961) 13 Stan. L. Rev. 252, 259.

In People v. Satchell (1971) 6 Cal.3d 28 [98 Cal.Rptr. 33, 489 P.2d 1361, 50 A.L.R.3d 383], our Supreme Court reversed a second degree felony-murder conviction arising out of a killing by an ex-felon in possession of a concealed weapon (a sawed-off shotgun) in violation of Penal Code section 12021. The court concluded that the felony must be viewed in the abstract and not on the basis of the particular facts of the case; that the carrying of a concealed weapon by an ex-felon is not a felony inherently dangerous to human life “. . . because we can conceive of such a vast number of situations wherein it would be grossly illogical to impute malice, . . .” (6 Cal.3d at p. 40.) The court accepted the defendant’s argument that of the many activities that are punishable as felonies only some clearly manifest a propensity for dangerous acts by the perpetrator; hence, it cannot be said theoretically that a felon who is armed with a concealable weapon presents a danger significantly greater than a nonfelon similarly armed. The court, however, pointed out that independent of the felony-murder rule the prosecution was still free to prove any degree of murder or manslaughter that the evidence might substantiate. (Id., atp. 33, fn. 11.)

Satchell also considered the propriety of a felony-murder instruction based upon a violation of Penal Code section 12020. This section provides that any person in possession of certain weapons, including a sawed-off shotgun, is guilty of a felony. The court concluded that this offense abstractly viewed also is not inherently dangerous to human life since it makes no distinction between the innocent gun collector and the hardened criminal. Again the court noted that if such possession was of an extremely reckless nature indicating a “conscious disregard for human life,” malice could be imputed via ordinary murder principles. (6 Cal.3d atp. 42.)

In People v. Lopez (1971) 6 Cal.3d 45 [98 Cal.Rptr. 44, 489 P.2d 1372], the Satchell reasoning was used to reverse a second degree murder conviction based on the underlying felony of escape from a county jail. (Pen. Code, § 4532.) “We cannot conclude that those who commit nonviolent escapes such as those here suggested thereby perpetrate an offense which should logically serve as the basis for the imputation of malice aforethought in a murder prosecution. Because section 4532 draws no relevant distinction between such escapes and the more violent *626variety, it proscribes an offense which, considered in the abstract, is not inherently dangerous to human life and cannot properly support a second degree felony-murder instruction.” (Italics original.) (6 Cal.3d at pp. SI-52, fn. omitted.)

And in People v. Morales (1975) 49 Cal.App.3d 134 [122 Cal.Rptr. 157], it was held that grand theft from the person of another is not an inherently dangerous felony to support a felony-murder charge: “It is apparent that the offense can readily be perpetrated without any significant hazard to human life; . . . Only in the unusual case would a taking from the person involve a substantial danger of death without the thief using force against his victim. If the thief does use force, either to effect the taking or to resist the victim’s efforts to retrieve the property [citation], the crime becomes robbery, and will support application of the felony-murder rule for that reason. Where the thief abstains from the use of force, he thereby removes the chief source of danger to human life; in such case the purpose of the felony-murder rule, ‘to deter felons from killing negligently or accidentally’ [citation], has already been achieved, and thus there would be no rational purpose to be served in extending the doctrine to cover the nonforceful larceny.” (49 Cal.App.3d at p. 143.)

Finally, the grand theft of an automobile in violation of Vehicle Code section 10851 followed by a high speed chase resulting in an unintended death does not constitute the commission of a felony inherently dangerous to human life so as to support the felony-murder doctrine (People v. Williams (1965) 63 Cal.2d 452, 458, fn. 5 [47 Cal.Rptr. 7, 406 P.2d 647], disapproving People v. Pulley (1964) 225 Cal.App.2d 366 [37 Cal.Rptr. 376], on the ground that Pulley erroneously looked to the particular facts of the felony in determining its dangerous character).

Therefore, one may cogently ask: If possession of a concealed weapon by an ex-felon, escape from a county jail, and grand theft of an automobile are not felonies per se dangerous to human life so as to provide a basis for the felony-murder rule, how may the theft of personal property from an unattended vehicle without the use of weapons be deemed inherently dangerous? Furthermore, if such a burglary will not support a second degree felony murder, how can it rationally be used to support a first degree felony murder?

If we were writing on a clean slate, we would hold that respondents should not be prosecuted for felony murder since viewed in the abstract, an automobile burglary is not dangerous to human life. The present case *627demonstrates why this is so. Respondents committed the burglary on vans parked in a dealer’s lot on a Sunday morning. There were no people inside the vans or on the lot at the time. The respondents were not armed and presumably had no expectation of using violence during the burglary.

Furthermore, treating the flight as part of the burglary to bootstrap the entire transaction into one inherently dangerous to human life simply begs the issue; flight from the scene of any crime is inherently dangerous. So, if a merchant in pursuit of a fleeing shoplifter is killed accidentally (by falling and striking his head on the curb or being hit by a passing automobile), the thief would be guilty of first degree felony murder assuming the requisite intent to steal at the time of the entry into the store. (Cf. People v. Earl, supra, 29 Cal.App.3d 894.) Such a harsh result destroys the symmetry of the law by equating an accidental killing resulting from a petty theft with a premeditated murder. In no sense can it be said that such a result furthers the ostensible purpose of the felony-murder rule which is to deter those engaged in felonies from killing negligently or accidentally. (People v. Washington, supra, 62 Cal.2d 111, 781.) On the other hand, if the flight is divorced from the burglary the latter can be objectively evaluated as to its dangerous propensities, and the instant burglary would not be deemed dangerous to human life and would be outside the purview of the felony-murder rule.

As was pointed out in People v. Earl, supra, 29 Cal.App.3d 894, 898, when the felony-murder statute was enacted in 1872, Penal Code section 459 required that the burglary occur in the nighttime and involve the entry of a “house, room, apartment, or tenement, or any tent, vessel, water craft, or railroad car . . . As defined, burglary was per se a crime dangerous to human life based on the probability of human occupancy of the described enclosures. Such danger to life was a common element in all of the felonies specified in section 189 (i.e., arson, rape, robbery, mayhem, or lewd acts upon a child). Contrary to the holding in People v. Talbot, supra, 64 Cal.2d at page 705, it rationally can be argued that the Legislature did not intend to include an automobile burglary within the felony-murder rule. This conclusion is strongly supported by the legislative classification of burglary into degrees. Penal Code section 460 provides that every burglary of an inhabited dwelling house, trailer coach, or building committed in the nighttime, is burglary of the first degree. All other kinds of burglary are of the second degree. It would be reasonable to include only first degree burglary in the Penal Code section 189 definition of felony murder.

*628Nonetheless, as previously explained the force of precedent requires the application of the first degree felony-murder rule to the instant case.

Respondents Also May Be Prosecuted for Second Degree Murder

For the guidance of the trial court, we observe that respondents may also be prosecuted for ordinary second degree murder. Second degree murder is an unlawful killing with malice aforethought but not willful, premeditated or deliberate. (People v. Jeter (1964) 60 Cal.2d 671 [36 Cal.Rptr. 323, 388 P.2d 355]; People v. Brust (1957) 47 Cal.2d 776, 783 [306 P.2d 480].) Malice is implied when the circumstances attending the killing demonstrate “an abandoned and malignant heart.” (Pen. Code, § 188.) This simply means that malice may be implied when the defendant does an act with a high probability that it will result in death and does it with a base antisocial motive and with wanton disregard for human life. (People v. Washington, supra, 62 Cal .2d 111, 782.)

In People v. Pulley, supra, 225 Cal.App.2d 366, the defendants stole an automobile and got involved in a 75- to 80-mile-per-hour chase with the police. They ran through a red light and caused a multi-car collision, killing one of the drivers. The court stated: “By any reasonable standard, stealing and driving a stolen car and endeavoring to escape pursuing officers with the stolen car, entering an intersection against all rules of the road at 70 to 80 miles per hour and crashing with other cars lawfully proceeding therein, are highly dangerous. Violence in evading the police is within the ambit of risk. Death here was not a freak coincidence, but an expectable incident of the felony, part of the risk that is set in motion by the original crime.” {Id., at p. 373.) The court upheld the application of the second degree felony-murder rule based upon the automobile theft.

The Supreme Court subsequently disapproved Pulley on the ground that the court erroneously looked to the particular facts of the case in determining whether it was “inherently dangerous” so as to support a second degree felony murder conviction. (People v. Williams, supra, 63 Cal.2d 452, 458, fn. 5.) However, the Supreme Court thereafter made clear that the Pulley fact pattern demonstrated sufficiently extreme and wanton recklessness to establish malice aforethought and second degree murder. (People v. Satchell, supra, 6 Cal.3d 28, 33-34, fn. 11; see also People v. Phillips, supra, 64 Cal.2d 574, 581; Note (1967) 55 Cal.L.Rev. 329, 340, fn. 58.)

*629Respondents contend that their conduct falls within a specific proscription of Penal Code section 192, subdivision 3, paragraph (a), vehicular manslaughter, the unlawful killing by a grossly negligent use of the automobile. They argue that where general and specific statutes both punish the same conduct, the specific must control. This argument is unavailing. The respondents’ conduct was more than grossly negligent. The conduct clearly presents an issue of fact as to whether or not respondents exhibited a wanton and reckless disregard for human life. Respondents drove at high speeds through main thoroughfares of Fresno in an attempt to elude Officer Ballesteroz. At one point in the chase they drove on the wrong side of Herndon Avenue and caused oncoming cars to swerve off of the road to avoid a head-on collision. They then made a U-turn and sped back to Blackstone Avenue, ran a red light and caused other traffic to stop to avoid a collision. Respondents then drove down Blackstone at speeds estimated between 60 and 75 miles per hour and headed straight at two oncoming police vehicles which were attempting to block their flight. Respondents did not reduce their speed as they approached the officers’ vehicles, and only a last minute maneuver by the officers avoided a possible fatal collision. At the next intersection respondents’ vehicle which “hadn’t slowed down very much” ran the red light and struck and killed the driver of the other car. Under these facts the foreseeability of serious injury or death was apparent to respondents. (Cf. People v. Pulley, supra, 225 Cal.App.2d at p. 373; see Witkin, Cal. Crimes (1978 Supp.) § 327, p. 327.) In light of the Supreme Court’s language in People v. Satchell, supra, 6 Cal.3d 28, that a high speed flight from police in an automobile may support a second degree murder conviction, the respondents’ argument must fail.

The judgment is reversed.

Hopper, J., concurred.

Ginsburg, J.,* concurred only in the reversal of the judgment of dismissal.

Respondents’ petitions for a hearing by the Supreme Court were denied February 8, 1979. Bird, C. J., Tobriner, J., and Mosk, J., were of the opinion that the petitions should be granted.

*1132Opinion

KENNARD, J.

Murder is the unlawful killing of a human being, with malice aforethought. (Pen. Code, § 187, subd. (a).) But under the second degree felony-murder rule, the prosecution can obtain a conviction without showing malice if the killing occurred during the commission of an inherently dangerous felony. Is the crime of driving with a willful or wanton disregard for the safety of persons or property while fleeing from a pursuing police officer (Veh. Code, § 2800.2)1 an inherently dangerous felony for purposes of the second degree felony-murder rule? We conclude it is not.

I

At 12:40 a.m. on May 23, 2002, California Highway Patrol Officer Gary Stephany saw defendant driving a Chevrolet Tahoe (a sport utility vehicle) without a rear license plate, and signaled him to pull over. Defendant stopped on the side of the road. But when Officer Stephany and his partner, Officer Wayne Bernard, got out of their patrol car, defendant restarted the engine and sped to a nearby freeway. The officers gave chase at speeds of up to 90 miles per hour and radioed for assistance. Defendant left the freeway and drove onto a surface street, turning off his car’s headlights. He ran two stop signs and a red light, and he drove on the wrong side of the road. His speed was 15 to 20 miles over the posted speed limit of 50 miles per hour. At some point, he made a sharp turn onto a small dirt road and escaped.

Minutes later, Officer Anthony Arcelus and his partner, Officer Bret Boss, who had been monitoring the pursuit on their car radio, saw the Tahoe with its headlights on again and took up the chase. Officer Arcelus, who was driving, estimated the Tahoe’s speed at more than 80 miles per hour, and he saw it run a stop sign and a traffic light. By then, the car’s headlights were again turned off. Up to that point, the chase had taken place in rural parts of Fresno County. When the Tahoe started heading toward downtown Fresno, Officer Arcelus gave up the pursuit, fearing that the high-speed chase might cause an accident.

About a minute after Officer Arcelus stopped chasing the Tahoe, he saw it run a red light half a mile ahead of him and collide with a car driven by Jeanette Rodriguez. Rodriguez was killed and her husband, a passenger in the car, was seriously injured. It turned out that the Tahoe that defendant was driving had been stolen earlier that day. Defendant, who was also injured in the crash, was arrested and charged with murder (Pen. Code, § 187), with *1133causing serious bodily injury while evading a police officer (§ 2800.3), and with evading a police officer in willful or wanton disregard for the safety of persons or property (§ 2800.2).

At trial, the prosecution called as a witness Laurie Bennett, defendant’s passenger during the chase. She was evasive about the events leading up to the accident. Ultimately, she admitted that she had told the truth when she explained to a police officer that five or six times during the chase she had begged defendant to let her get out of the car, and that defendant had run a red light at the intersection where the fatal accident occurred. An accident reconstruction expert testified that at the time of the accident the Tahoe was traveling over 80 miles per hour, and Rodriguez’s car was traveling close to the posted speed limit of 35 miles per hour. John Mikkelson, a pipeline inspector working near the intersection where the accident occurred, said he looked at the signal immediately after hearing the crash of the two colliding cars and saw that it was green for cars traveling in Rodriguez’s direction (and thus presumably red for defendant).

Forensic toxicologist Roger Peterson, a witness for the defense, testified that defendant had a “high amount” of methamphetamine in his bloodstream at the time of the accident. A person under the influence of methamphetamine, Peterson said, might drive at excessive speeds, might have trouble staying in a single lane, and might not notice traffic lights and signs. Defendant also had marijuana in his bloodstream, but not enough to be under the influence. Victim Rodriguez’s bloodstream contained morphine (a metabolite of heroin) and benzoyleconine (a metabolite of cocaine). Based on this evidence, toxicologist Peterson expressed his opinion that Rodriguez was under the influence of heroin, and possibly cocaine when the accident occurred.

Defendant testified on his own behalf. He admitted stealing the Tahoe and fleeing from the Highway Patrol officers. He did so because his probation officer had told him he would go to prison if he was again caught in a stolen car. He could only remember bits and pieces of the chase. He described himself as a skilled driver; his cousin, a race car driver, had taught him to drive “sprint cars” at a racetrack. He saw the victims’ car before the accident but could not recall hitting it. He could not remember what color the signal light was when he entered the intersection but admitted it was “most likely” red when the car he was driving crashed into the Rodriguez car.

The trial court instructed the jury: “Every person who unlawfully kills a human being during the commission of violation of California Vehicle Code section 2800.2, a felony inherently dangerous to human life, is guilty of the crime of murder in violation of Section 187 of the Penal Code. [][] In order to *1134prove this crime, each of the following elements must be proved: [(j[] 1. A human being was killed; [f] 2. The killing was unlawful; and [(J[] 3. The' killing occurred during the commission of violation of California Vehicle Code section 2800.2, a felony inherently dangerous to human life. A violation of Vehicle Code section 2800.2 is a felony inherently dangerous to human life.” The trial court did not instruct the jury that malice is an element of murder or that the jury could convict defendant if it found that he acted with express or implied malice when he killed victim Rodriguez.

In his closing argument, defense counsel contended that defendant did not violate section 2800.2 because he did not drive with willful and wanton disregard for life or property; that even if defendant violated section 2800.2 while fleeing from the officers he was not doing so when the accident occurred, because by then the officers were no longer chasing him; and that defendant might not have caused the accident because there was a reasonable doubt that he ran a red light at the time of the incident.

During its deliberations, the jury sent the trial court this note: “It appears in the instructions if there is a guilty verdict in [section] 2800.2 then there must be a guilty verdict for [Penal Code section] 187, yes or no?” The court replied that it was “not in a position to say yes or no”; it then reread the instructions on felony murder and causation. The jury convicted defendant of all counts.

The Court of Appeal affirmed. As pertinent here, it rejected defendant’s contention that he could not be convicted under the second degree felony-murder rule because section 2800.2 is not an inherently dangerous felony. And it rejected defendant’s contention that he could not be convicted of murder because his conduct fit within section 2800.3 (causing death or serious bodily injury by willful flight from a pursuing peace officer), which he claimed is a “special statute” that bars his conviction for the more general crime of murder.

We granted defendant’s petition for review on these two issues: “1. Whether the offense of driving in willful or wanton disregard for the safety of persons or property while fleeing from a pursuing police officer (Veh. Code, § 2800.2) is a felony inherently dangerous to human life for purposes of the second degree felony-murder rule, [f] 2. Whether the offense of proximately causing death or serious bodily injury by willful flight from a pursuing police officer (Veh. Code, § 2800.3) is a more specific offense precluding application of the second degree felony-murder rule where death occurs during the offense of driving in willful or wanton disregard for the safety of persons or property while fleeing from a pursuing police officer (Veh. Code, § 2800.2).”

*1135H

Because the second degree felony-murder rule is a court-made rule, it has no statutory definition. This court has described it thusly: “A homicide that is a direct causal result of the commission of a felony inherently dangerous to human life (other than the . . . felonies enumerated in Pen. Code, § 189) constitutes at least second degree murder.” (People v. Ford (1964) 60 Cal.2d 772, 795 [36 Cal.Rptr. 620, 388 P.2d 892], italics added.) The rule “eliminates the need for proof of malice in connection with a charge of murder.” (People v. Robertson (2004) 34 Cal.4th 156, 165 [17 Cal.Rptr.3d 604, 95 P.3d 872].) It is not an evidentiary presumption but a substantive rule of law (see People v. Dillon (1983) 34 Cal.3d 441, 472-476 [194 Cal.Rptr. 390, 668 P.2d 697]; see also People v. Patterson (1989) 49 Cal.3d 615, 626 [262 Cal.Rptr. 195, 778 P.2d 549]), which is based on the theory that “when society has declared certain inherently dangerous conduct to be felonious, a defendant should not be allowed to excuse himself by saying he was unaware of the danger to life because, by declaring the conduct to be felonious, society has warned him of the risk involved.” (People v. Patterson, supra, 49 Cal.3d at p. 626.)

Because the second degree felony-murder rule is “a judge-made doctrine without any express basis in the Penal Code” (People v. Dillon, supra, 34 Cal.3d at p. 472, fn. 19), its constitutionality has been questioned (see People v. Patterson, supra, 49 Cal.3d at p. 641 (conc. opn. of Panelli, J.)). And, as we have noted in the past, legal scholars have criticized the rule for incorporating “an artificial concept of strict criminal liability that ‘erodes the relationship between criminal liability and moral culpability.’ ” (Id. at p. 621.) Therefore, we have repeatedly stressed that the rule “ ‘deserves no extension beyond its required application.’ ” (Id. at p. 622; see also People v. Burroughs (1984) 35 Cal.3d 824, 829 [201 Cal.Rptr. 319, 678 P.2d 894]; People v. Phillips (1966) 64 Cal.2d 574, 582 [51 Cal.Rptr. 225, 414 P.2d 353].)

“In determining whether a felony is inherently dangerous [under the second degree felony-murder rule], the court looks to the elements of the felony in the abstract, ‘not the “particular” facts of the case,’ i.e., not to the defendant’s specific conduct.” (People v. Hansen (1994) 9 Cal.4th 300, 309 [36 Cal.Rptr.2d 609, 885 P.2d 1022].) That is, we determine whether the felony “by its very nature . . . cannot be committed without creating a substantial risk that someone will be killed . . . .” (People v. Burroughs, supra, 35 Cal.3d at p. 833; see also People v. Robertson, supra, 34 Cal.4th at p. 166.)

*1136Felonies that have been held inherently dangerous to life include shooting at an inhabited dwelling (People v. Hansen, supra, 9 Cal.4th at p. 311), poisoning with intent to injure (People v. Mattison (1971) 4 Cal.3d 177 [93 Cal.Rptr. 185, 481 P.2d 193]), arson of a motor vehicle (People v. Nichols (1970) 3 Cal.3d 150, 163 [89 Cal.Rptr. 721, 474 P.2d 673]; but see People v. Henderson (1977) 19 Cal.3d 86, 96 [137 Cal.Rptr. 1, 560 P.2d 1180]), grossly negligent discharge of a firearm (People v. Clem (2000) 78 Cal.App.4th 346, 353-354 [92 Cal.Rptr.2d 727]; see also People v. Robertson, supra, 34 Cal.4th at pp. 168-169 [quoting Clem with approval]), manufacturing methamphetamine (People v. James (1998) 62 Cal.App.4th 244, 271 [74 Cal.Rptr.2d 7]), kidnapping (People v. Greenberger (1997) 58 Cal.App.4th 298, 377 [68 Cal.Rptr.2d 61]; People v. Pearch (1991) 229 Cal.App.3d 1282, 1299 [280 Cal.Rptr. 584]), and reckless or malicious possession of a destructive device (People v. Morse (1992) 2 Cal.App.4th 620, 646 [3 Cal.Rptr.2d 343]).

Felonies that have been held not inherently dangerous to life include practicing medicine without a license under conditions creating a risk of great bodily harm, serious physical or mental illness, or death (People v. Burroughs, supra, 35 Cal.3d at p. 833); false imprisonment by violence, menace, fraud, or deceit (People v. Henderson, supra, 19 Cal.3d at pp. 92-96); possession of a concealable firearm by a convicted felon (People v. Satchell (1971) 6 Cal.3d 28, 35-41 [98 Cal.Rptr. 33, 489 P.2d 1361]); possession of a sawed-off shotgun (id. at pp. 41-43); escape (People v. Lopez (1971) 6 Cal.3d 45, 51-52 [98 Cal.Rptr. 44, 489 P.2d 1372]); grand theft (People v. Phillips, supra, 64 Cal.2d at pp. 580-583); conspiracy to possess methedrine (People v. Williams (1965) 63 Cal.2d 452, 458 [47 Cal.Rptr. 7, 406 P.2d 647]); extortion (People v. Smith (1998) 62 Cal.App.4th 1233, 1236-1238 [72 Cal.Rptr.2d 918]); furnishing phencyclidine (People v. Taylor (1992) 6 Cal.App.4th 1084, 1099 [8 Cal.Rptr.2d 439]); and child endangerment or abuse (People v. Lee (1991) 234 Cal.App.3d 1214, 1229 [286 Cal.Rptr. 117]).

m

In determining whether section 2800.2 is an offense inherently dangerous to life, we begin by reviewing the statutory scheme. Three statutes punish those who flee from police officers: sections 2800.1, 2800.2, and 2800.3.

Section 2800.1 states that any motorist who “with the intent to evade, willfully flees or otherwise attempts to elude” a peace officer pursuing on a motor vehicle or bicycle is, under specified circumstances, guilty of a misdemeanor.

*1137Under section 2800.3, when “willful flight or attempt to elude a pursuing peace officer in violation of Section 2800.1 proximately causes death or serious bodily injury to any person,” the offense is a wobbler (an offense that can be a felony or a misdemeanor, at the trial court’s discretion), punishable by up to five years in prison.

Section 2800.2, which was the basis for defendant’s conviction under the second degree felony-murder rule, provides:

“(a) If a person flees or attempts to elude a pursuing peace officer in violation of Section 2800.1 and the pursued vehicle is driven in a willful or wanton disregard for the safety of persons or property, the person driving the vehicle, upon conviction, shall be punished by imprisonment in the state prison, or by confinement in the county jail .... The court may also impose a fine ... or may impose both that imprisonment or confinement and fine.

“(b) For purposes of this section, a willful or wanton disregard for the safety of persons or property includes, but is not limited to, driving while fleeing or attempting to elude a pursuing peace officer during which time either three or more violations that are assigned a traffic violation point count under Section 12810 occur, or damage to property occurs.”

In concluding that section 2800.2 is an inherently dangerous felony, the Court of Appeal relied heavily on People v. Johnson (1993) 15 Cal.App.4th 169 [18 Cal.Rptr.2d 650]. There the Court of Appeal, construing an earlier version of section 2800.2 that was essentially the same as what is now subdivision (a) of that section, held that driving with “willful or wanton disregard for the safety of persons or property” was inherently dangerous to life. We need not decide, however, whether Johnson was correct, because in 1996, three years after Johnson was decided, the Legislature amended section 2800.2 to add subdivision (b). (Stats. 1996, ch. 420, § 1.) Subdivision (b) very broadly defines the term “willful or wanton disregard for the safety of persons or property,” as used in subdivision (a), to include any flight from an officer during which the motorist commits three traffic violations that are assigned a “point count” under section 12810, or which results in “damage to property.”

Violations that are assigned points under section 12810 and can be committed without endangering human life include driving an unregistered vehicle owned by the driver (§§ 40001, 12810, subds. (e), (g)(1)), driving with a suspended license (§§ 14601, 12810, subd. (i)), driving on a highway at slightly more than 55 miles per hour when a higher speed limit has not been posted (§§ 22349, subd. (a), 12810, subd. (e)), failing to come to a *1138complete stop at a stop sign (§§ 22450, 12810, subd. (e)), and making a right turn without signaling for 100 feet before turning (§§ 22108, 12810, subd. (e)).

The Court of Appeal here rejected defendant’s contention that because of the broad definition of the phrase “willful or wanton disregard for the safety of persons or property” in subdivision (b) of section 2800.2, violations of section 2800.2 are not inherently dangerous to life for the purposes of the second degree felony-murder rule. The Court of Appeal quoted People v. Sewell (2000) 80 Cal.App.4th 690 [95 Cal.Rptr.2d 600], which concluded that subdivision (b) “did not change the elements of the section 2800.2 offense, in the abstract, or its inherently dangerous nature.” (Sewell, at p. 694.) But, as we pointed out in the preceding paragraph, subdivision (b) greatly expanded the meaning of the quoted statutory phrase to include conduct that ordinarily would not be considered particularly dangerous.2

The Attorney General contends that when the Legislature amended section 2800.2 to add subdivision (b), it did not intend to make the second degree felony-murder rule inapplicable to violations of that section. The legislative history of the amendment makes no mention, however, of the second degree felony-murder rule; nor does the legislative history pertaining to the original enactment in 1988 of section 2800.2 (Stats. 1988, ch. 504, § 3, p. 1919). In all likelihood, the Legislature did not consider the effect that either the statute’s original enactment or its amendment would have on murder prosecutions. In the absence of any evidence of legislative intent, we assume that the Legislature contemplated that we would determine the application of the second degree felony-murder rule to violations of section 2800.2 based on our long-established decisions holding that the rule applies only to felonies that are inherently dangerous in the abstract. (People v. Robertson, supra, 34 Cal.4th at p. 166; People v. Hansen, supra, 9 Cal.4th at p. 309; People v. Phillips, supra, 64 Cal.2d at p. 582; People v. Williams, supra, 63 Cal.2d at p. 458, fn. 5.) As we have explained in this opinion, a violation of section 2800.2 is not, in the abstract, inherently dangerous to *1139human life. Therefore, the second degree felony-murder rule does not apply when a killing occurs during a violation of section 2800.2.

Thus, the trial court here erred when it instructed the jury that it should find defendant guilty of second degree murder if it found that, while violating section 2800.2, defendant fatally injured Jeanette Rodriguez when their cars collided.3 The parties have not briefed the question of whether the trial court’s instructional error was prejudicial. That is a matter to be considered by the Court of Appeal on remand.4

Conclusion

Nothing here should be read as saying that a motorist who kills an innocent person in a hazardous, high-speed flight from a police officer should not be convicted of murder. A jury may well find that the motorist has acted with malice by driving with conscious disregard for the lives of others, and thus is guilty of murder. (See generally People v. Watson (1981) 30 Cal.3d 290 [179 Cal.Rptr. 43, 637 P.2d 279].) But, as we have explained, not all violations of section 2800.2 pose a danger to human life. Therefore, the prosecution may not (as it did here) resort to the second degree felony-murder rule to remove from the jury’s consideration the question whether a killing that occurred during a violation of section 2800.2 was done with malice.5

We reverse the judgment of the Court of Appeal, which upheld defendant’s conviction for second degree murder, and remand the matter to that court for further proceedings consistent with this opinion.

George, C. J., Werdegar, J., Chin, J., and Moreno, J., concurred.

*1140BROWN, J., Concurring and Dissenting.

I concur with the majority’s holding that defendant’s conviction for second degree felony murder must be reversed and the case remanded for further proceedings. However, for the reasons set forth in my dissenting opinion in People v. Robertson (2004) 34 Cal.4th 156, 186-192 [17 Cal.Rptr.3d 604, 95 P.3d 872] (dis. opn. of Brown, J.), I cannot countenance the majority’s continued allegiance to this dubious doctrine.

Here, defendant was convicted solely on a second degree felony-murder theory. The majority appears to acknowledge the rule is constitutionally and analytically suspect: “Because the second degree felony-murder rule is ‘a judge-made doctrine without any express basis in the Penal Code’ [citation], its constitutionality has been questioned [citation]. And, as we have noted in the past, legal scholars have criticized the rule for incorporating ‘an artificial concept of strict criminal liability that “erodes the relationship between criminal liability and moral culpability.” ’ [Citation.] Therefore, we have repeatedly stressed that the rule ‘ “deserves no extension beyond its required application.” ’ [Citations.]” (Maj. opn., ante, at p. 1135.) I agree, but I would go farther and abrogate the rule entirely. (People v. Robertson, supra, 34 Cal.4th 156, 191 (dis. opn. of Brown, J.) [“Because the second degree felony-murder rule is suspect I believe it would not be missed if we abandoned it”].) As the facts of this case conclusively demonstrate, the application of the second degree felony-murder rule remains irredeemably arbitrary.

The majority concludes, based on a technical parsing of the provision’s grammar, that a violation of Vehicle Code section 2800.2 is not an inherently dangerous felony for purposes of second degree felony murder. However, a commonsense construction of the statute’s language leads to the opposite conclusion—a conclusion that is considerably less counterintuitive. As one lower court stated in addressing the same issue we review here, “It would seem clear as a matter of logic that any felony whose key element is ‘wanton disregard’ for human life necessarily falls within the scope of ‘inherently dangerous’ felonies. ... [][]... [f] .. . [A]part from the ‘wanton disregard’ element, one must also be engaged in the act of fleeing from a pursuing peace officer whose vehicle is displaying lights and sirens. Any high-speed pursuit is inherently dangerous to the lives of the pursuing police officers. In even the most ethereal of abstractions, it is not possible to imagine that the ‘wanton disregard’ of the person fleeing does not encompass disregard for the safety of the pursuing officers.” (People v. Johnson (1993) 15 Cal.App.4th 169, 173-174 [18 Cal.Rptr.2d 650].) Unlike the majority, I find the Court of Appeal’s statement in Johnson persuasive.

*1141Indeed, I agree with Justice Baxter that if any offense should easily qualify as inherently dangerous, Vehicle Code section 2800.2 certainly would. “Subdivision (a) of section 2800.2 gives clear and specific notice that one who, in order to elude police pursuit, drives with reckless indifference to safety is guilty of a felony. Such reckless driving is, of course, inherently dangerous—by definition, it creates a substantial risk that someone will be killed.” (Dis. opn. of Baxter, J., post, at p 1143.) Although it is possible to imagine slow motion pursuits where neither people nor property are harmed, the facts of this case present the more likely scenario: defendant greatly exceeded the speed limit, ran stop signs and stoplights, drove the wrong way on a street and entered downtown Fresno, where the pursuing police officer broke off his chase because he determined that it was too dangerous to proceed. Unfortunately, although the police officer avoided injury by breaking off his pursuit, defendant still entered an intersection on a red light and collided with another vehicle, killing its driver.

“[R]easonable judges can disagree about the legitimacy of contracting or expanding the statutory definition of a felony in order to conclude that a particular violation should be deemed inherently dangerous.” (People v. Robertson, supra, 34 Cal.4th 156, 186 (dis. opn. of Brown, J.).) Two other Court of Appeal decisions have concluded that a violation of Vehicle Code section 2800.2 is an inherently dangerous felony. (People v. Sewell (2000) 80 Cal.App.4th 690, 693-697 [95 Cal.Rptr.2d 600]; People v. Johnson, supra, 15 Cal.App.4th at pp. 173-174.) In this case, two members of this court and a unanimous Court of Appeal reached the same conclusion. The fact that such variations are not just possible, but actually inevitable, suggests a level of arbitrariness we should make every effort to eliminate from the criminal law. For that reason, as well as other concerns discussed more fully in Robertson, I would abrogate the nonstatutory second degree felony-murder rule and leave it to the Legislature to define precisely what conduct subjects a defendant to strict criminal liability.

BAXTER, J., Dissenting.

I respectfully dissent. In early morning darkness, defendant, driving a stolen vehicle, led police officers on a perilous and extended chase over Fresno County roads. He ran three stop signs and a red light, and even proceeded on the wrong side of a divided highway, while operating the vehicle far in excess of posted speed limits. Finally, as he dashed on city streets toward downtown Fresno at speeds between 80 and 90 miles per hour, he ran a second red light and collided with another vehicle. Both occupants of that car were ejected onto the street. One perished.

*1142As a result of Ms reckless actions, defendant suffered a conviction for violation of VeMcle Code section 2800.2,1 wMch makes it a felony to flee police pursuit by driving with “a willful or wanton disregard for the safety of persons or property.” (Id., subd. (a).) Because Ms violation of this law had fatal consequences, defendant was also convicted of second degree murder on a felony-murder theory—i.e., that he caused death in the commission of an inherently dangerous felony. Instructions told the jury that the “willful or wanton disregard” necessary to both convictions required an intentional act performed with a “conscious disregard” for safety.

The majority concede that defendant (1) violated section 2800.2, the reckless flight statute, by “willful[ly] or wanton[ly] disregard[ing]” human safety during a flight from the police, and (2) thereby produced a human fatality. Nonetheless, though two Court of Appeal decisions have concluded otherwise (People v. Sewell (2000) 80 Cal.App.4th 690, 693-697 [95 Cal.Rptr.2d 600] (Sewell); People v. Johnson (1993) 15 Cal.App.4th 169, 173-174 [18 Cal.Rptr.2d 650] (Johnson)), the majority insist that this statute cannot support a felony-murder conviction.

The majority invoke the premise that second degree felony murder only occurs in the commission of a felony which is inherently dangerous in the abstract—one which, by its very nature, cannot be committed without creating a substantial risk, or a high probability, that someone will be killed. (E.g., People v. Hansen (1994) 9 Cal.4th 300, 309 [36 Cal.Rptr.2d 609, 885 P.2d 1022] (Hansen).) Relying on the peculiar construction of the reckless flight statute, including its recent amendment, the majority posit that even if defendant violated section 2800.2 in a potentially lethal way, it can be violated without creating a lethal danger to persons.

The majority focus upon subdivision (b) of section 2800.2, which was added in 1996. This subdivision states that “[f]or purposes of [section 2800.2],” driving with “a willful or wanton disregard for the safety of persons or property includes . . . driving . . . during which time . . . tiiree or more [traffic] violations that are assigned a traffic violation point count under [s]ection 12810 occur . . . .” The majority reason that, because some statutory “points” violations are not inherently dangerous, one can commit the unitary felony described in both subdivisions of section 2800.2 in a way that does not place human life at risk.

*1143I am not persuaded. Subdivision (a) of section 2800.2 gives clear and specific notice that one who, in order to elude police pursuit, drives with reckless indifference to safety is guilty of a felony. Such reckless driving is, of course, inherently dangerous—by definition, it creates a substantial risk that someone will be killed. Moreover, there is no doubt that defendant committed exactly the reckless endangerment of human life forbidden by the statute. As I explain in further detail below, his conviction for violating section 2800.2, as well as his felony-murder conviction, were unambiguously based on the dangerous recklessness of his flight from the police.

Hence, the principal reason noted by the majority for limiting the second degree felony-murder rule should not bar a felony-murder finding here. The statute’s express words placed defendant on notice that the particular conduct he was committing—recklessly unsafe driving to elude police pursuit—was both felonious and inherently dangerous, and thus a basis for murder liability if death resulted. This is a case where “ ‘society has declared certain inherently dangerous conduct to be felonious,’ ” and “ ‘defendant should [therefore] not be allowed to excuse himself by saying he was unaware of the danger to life because, by declaring the conduct to be felonious, society has warned him of the risk involved.’ ” (Maj. opn., ante, at p. 1135, quoting People v. Patterson (1989) 49 Cal.3d 615, 626 [262 Cal.Rptr. 195, 778 P.2d 549] (Patterson).)

Conversely, the principal reason for applying the felony-murder rule is present. The purpose of the felony-murder doctrine “ ‘is to deter those engaged in felonies from killing negligently or accidentally.’ ” (Hansen, supra, 9 Cal.4th 300, 308, quoting People v. Satchell (1971) 6 Cal.3d 28, 43 [98 Cal.Rptr. 33, 489 P.2d 1361].) Because the doctrine absolves the prosecution from proving malice, it properly applies when “ ‘the killer is engaged in a felony whose inherent danger to human life renders logical an imputation of malice on the part of all who commit it.’ ” (Ibid.)

Those requirements are met here. It is appropriate to deter persons from killing negligently or accidentally—as did defendant—while engaged—as was defendant—in recklessly unsafe driving to elude police pursuit, a specific form of conduct made felonious by section 2800.2, subdivision (a). Moreover, the inherent danger such conduct poses to human life is so clear that it is logical to impute malice to anyone who commits it.

Under such circumstances, it perverts reason to refuse to apply the felony-murder rule simply because subdivision (b) of section 2800.2 may additionally describe a nondangerous felony. Where society has warned, in plain statutory words, that the particular conduct committed by the defendant *1144is both dangerous and felonious, it should not matter that the statute may forbid nondangerous conduct as well.2

It is worth noting that, although the Legislature elected to include subdivision (b) as part of section 2800.2, it could just as easily have added a separate section, establishing a distinct felonious offense of committing three “points” violations while driving to elude a peace officer. This would equally have satisfied the apparent legislative purpose to deter flight from the police by expanding the circumstances under which driving to elude a pursuing police officer would constitute a felony. (See, e.g., Sen. Com. on Crim. Proc., Analysis of Assem. Bill No. 1999 (1995-1996 Reg. Sess.) as amended Apr. 29, 1996, p. 6.)

However, neither the Legislature’s desire to create a felony of flight with “points” violations, nor its choice of methodology to accomplish this result, can deprive the reckless conduct described in subdivision (a) of all force as an inherently dangerous felony. If subdivision (a) described an inherently dangerous felony before the addition of subdivision (b) in 1996 (see Johnson, supra, 15 Cal.App.4th 169, 173-174), the unchanged words of that subdivision equally do so following the 1996 amendment (see Sewell, supra, 80 Cal.App.4th 690, 693-697).

In sum, where the defendant committed inherently dangerous conduct expressly made felonious by subdivision (a) of section 2800.2, and was accused, tried, and convicted solely under that subdivision, I submit that subdivision (a) is the provision we should analyze to determine if it qualifies as a basis for felony murder.

As I have indicated above, this is such a case. The prosecution avoided all reliance on subdivision (b) of section 2800.2, and that subdivision was never an issue in the trial below. When referring to section 2800.2, the information cited only the language of subdivision (a), charging that defendant “drove with a willful or wanton disregard for the safety of persons or property.” In his testimony, one of the pursuing officers, Gary Stephany, ticked off the numerous Vehicle Code violations defendant had committed during the *1145high-speed chase, but the import of Stephany’s testimony was simply to emphasize defendant’s extreme and ongoing recklessness. The prosecutor never asked Stephany, or any other witness, whether defendant committed “points” violations.

In his argument to the jury concerning section 2800.2, the prosecutor emphasized that defendant’s various reckless traffic violations proved he had acted with “willful or wanton disregard,” i.e., with conscious indifference to the consequences for safety. The entire focus of the prosecutor’s argument was on defendant’s culpable state of mind while fleeing the police, as evidenced by his inherently dangerous driving maneuvers. The prosecutor never suggested that technical “points” violations committed by defendant were a basis for finding him guilty of violating section 2800.2, or of murder.

Most crucially, all reference to subdivision (b) was omitted from the jury instruction on section 2800.2. When describing the elements of a violation of that statute, the instruction said only that “ ‘[wjillful or wanton’ means an act intentionally performed with a conscious disregard for the safety of persons or property. It does not necessarily include an intent to injure.” (Italics added.)

Thus, beyond doubt, the jury convicted defendant of murder based solely on proof that he fled police pursuit by driving with reckless indifference to safety, conduct which is both inherently dangerous and expressly felonious under subdivision (a) of section 2800.2. It follows that the felony underlying defendant’s murder conviction was an inherently dangerous one. Accordingly, I submit, a murder conviction could be premised on his commission of this felony.

But even if section 2800.2 were not an inherently dangerous felony that could support a felony-murder conviction, any error in instructing the jury to the contrary was harmless. Indeed, by remanding to the Court of Appeal for a determination on the issue of prejudice, the majority imply that a no-prejudice conclusion is plausible.

I find such a conclusion inescapable, for two reasons. First, a reasonable jury, properly instructed on an implied malice theory of second degree murder, could not have failed to find, on this evidence, the elements of malicious murder. Second, though this jury received no explicit instructions on malice, it necessarily did find, under the instructions which were given, that defendant killed maliciously.

“ ‘Implied malice, for which the second degree felony-murder doctrine acts as a substitute [fn. omitted], has both a physical and a mental component. . . . *1146The mental component is the requirement that the defendant “knows that his conduct endangers the life of another and . . . acts with a conscious disregard for lifer [Citation].’ ” (People v. Robertson (2004) 34 Cal.4th 156, 165 [17 Cal.Rptr.3d 604, 95 P.3d 872], quoting Patterson, supra, 49 Cal.3d 615, 626, italics added.)

In my view, the evidence that defendant acted with actual knowledge and conscious disregard of the lethal probabilities is conclusive. No other inference can be drawn from his decision to maintain his careening flight from the police over highways and streets carrying innocent travelers. During the high-speed chase, he swerved over a median boundary of a divided highway and travelled in lanes reserved for oncoming vehicles, risking a head-on collision. With equal purpose, he ran at least three stop signs and two red lights, forcing cross-traffic to yield and creating the obvious danger of crashes at these controlled intersections. In his frantic efforts to outdistance his pursuers, he “fishtailed” into curves and comers, giving rise to the peril of spin-out accidents that could involve other vehicles. The fatal collision occurred in the midst of this course of conduct, as defendant, still trying to escape, ran a red light at high speed.

As a matter of law, I believe, these intentional acts went beyond mere gross negligence or reckless indifference. Defendant—by his own evaluation a skilled driver—must have acted with full awareness that he was thereby placing human life at risk. He cannot evade this responsibility through his inherently implausible testimony that he remembered only bits and pieces of the chase, and could not recall whether the traffic light was red at the fatal intersection when he entered it.

In any event, this jury was told that the felony described in section 2800.2 required a conscious disregard for safety. Hence, in finding that defendant caused death in the commission of this felony, the jury necessarily found that he understood the danger, and chose to act anyway, thus exhibiting implied malice. Under these circumstances, reliance on a felony-murder theory, even if error, must be deemed harmless. (Cf., e.g., People v. Sedeno (1974) 10 Cal.3d 703, 720-721 [112 Cal.Rptr. 1, 518 P.2d 913].)3

Having concluded that defendant’s violation of section 2800.2 was an “inherently dangerous” felony for purposes of the felony-murder rule, I must *1147address his second argument. Defendant urges that he cannot be prosecuted and punished for felony murder in any event because the Legislature intends section 2800.3 as the only statute under which deaths caused by vehicular flight from the police may be prosecuted and punished without proof of malice.4

Defendant invokes the principle that a special statute defining an offense preempts a more general statute encompassing the same conduct. (See, e.g., In re Williamson (1954) 43 Cal.2d 651, 654 [276 P.2d 593].) But this rule applies only “(1) when each element of the general statute corresponds to an element on the face of the special statute, or (2) when it appears from the statutory context that a violation of the special statute will necessarily or commonly result in a violation of the general statute. [Citation.]” (People v. Watson (1981) 30 Cal.3d 290, 295-296 [179 Cal.Rptr. 43, 637 P.2d 279].)

Neither circumstance is present here. Both statutes permit punishment, without proof of malice, for fatalities caused by vehicular flight from the police, but the two statutes do not govern such killings identically. Section 2800.3—in defendant’s view, the more specific death-related statute— sanctions such a homicide regardless of the fleeing motorist’s degree of driving care. By contrast, to be convicted of felony murder under the more “general” statute, section 2800.2, the fleeing motorist must have driven with “a willful or wanton disregard for safety.” Thus, the facial elements of the two statutes differ, and a fatality punishable under section 2800.3 will not necessarily or commonly constitute a felony murder under section 2800.2.

Through an exhaustive analysis of legislative history, defendant argues that the Legislature’s “carefully crafted” scheme for punishment of pursuit-related fatalities precludes prosecution of such homicides as felony murder. His contention lacks merit. Vehicular flight from the police, with its potential for death, injury, and property damage, has been a growing problem. The Legislature has responded over the years by expanding the punishment for such conduct under various circumstances. Among other things, it has provided in section 2800.3 that one whose flight, however carefully conducted, causes death or injury may receive up to five years in prison, a $10,000 fine, or both. But nothing indicates the Legislature intended a similar limitation on the punishment of persons who commit homicides while engaged in the inherently dangerous felony of flight from police pursuit with *1148willful and wanton disregard for safety. Every consideration of logic and common sense suggests that such homicides may properly be treated under the law of felony murder.

For all these reasons, I would affirm the judgment of the Court of Appeal.

*1178Opinion

CHIN, J.

In this murder case, the trial court instructed the jury on second degree felony murder with shooting at an occupied vehicle under Penal Code section 246, the underlying felony.1 We granted review to consider various issues concerning the validity and scope of the second degree felony-murder rule.

We first discuss the rule’s constitutional basis. Although the rule has long been part of our law, some members of this court have questioned its constitutional validity. We conclude that the rule is based on statute, specifically section 188’s definition of implied malice, and hence is constitutionally valid.

Next we reconsider the contours of the so-called merger doctrine this court adopted in People v. Ireland (1969) 70 Cal.2d 522 [75 Cal.Rptr. 188, 450 P.2d 580] (Ireland). After reviewing recent developments, primarily some of our own decisions, we conclude the current state of the law in this regard is untenable. We will overrule some of our decisions and hold that all assaultive-type crimes, such as a violation of section 246, merge with the charged homicide and cannot be the basis for a second degree felony-murder instruction. Accordingly, the trial court erred in instructing on felony murder in this case. We also conclude, however, that this error, alone, was not prejudicial.

We reverse the judgment of the Court of Appeal, which had found the same error prejudicial. However, the Court of Appeal also found a second error, a finding not before us on review. We remand the matter to the Court of Appeal to decide whether the two errors, in combination, were prejudicial.

I. Facts and Procedural History

We take our facts primarily from the Court of Appeal’s opinion.

Judy Onesavanh and Sophal Ouch were planning a party for their son’s birthday. Around 9:00 p.m. on September 13, 2003, they and a friend, Bounthavy Onethavong, were driving to the store in Stockton in a blue Mitsubishi that Onesavanh’s father owned. Onesavanh’s brother, George, also drives the car. The police consider George to be highly ranked in the Asian Boys street gang (Asian Boys).

*1179That evening Ouch was driving, with Onesavanh in the front passenger seat and Onethavong behind Ouch. While they were stopped in the left turn lane at a traffic light, a blue Honda with tinted windows pulled up beside them. When the light changed, gunfire erupted from the Honda, hitting all three occupants of the Mitsubishi. Onethavong was killed, having received two bullet wounds in the head. Onesavanh was hit in the back and seriously wounded. Ouch was shot in the cheek and suffered a fractured jaw.

Ouch and Onesavanh identified the Honda’s driver as “T-Bird,” known to the police to be Rathana Chan, a member of the Tiny Rascals Gangsters (Tiny Rascals), a criminal street gang. The Tiny Rascals do not get along with the Asian Boys. Chan was never found. The forensic evidence showed that three different guns were used in the shooting, a .22, a .38, and a .44, and at least six bullets were fired. Both the .38 and the .44 struck Onethavong; both shots were lethal. Only the .44 was recovered. It was found at the residence of Sokha and Mao Bun, brothers believed to be members of a gang.

Two months after the shooting, the police stopped a van while investigating another suspected gang shooting. Defendant was a passenger in the van. He was arrested and subsequently made two statements regarding the shooting in this case. He admitted he was in the backseat of the Honda at the time; T-Bird was the driver and there were two other passengers. Later, he also admitted he fired a .38-caliber firearm. He said he did not point the gun at anyone; he just wanted to scare them.

Defendant, who was 16 years old at the time of the shooting, was tried as an adult for his role in the shooting. He was charged with murder, with driveby and gang special circumstances, and with two counts of attempted murder, discharging a firearm from a vehicle, and shooting into an occupied vehicle, all with gang and firearm-use allegations, and with street terrorism. At trial, the prosecution presented evidence that defendant was a member of the Tiny Rascals, and that the shooting was for the benefit of a gang. Defendant testified, denying being a member of the Tiny Rascals or being involved in the shooting.

The prosecution sought a first degree murder conviction. The court also instructed the jury on second degree felony murder based on shooting at an occupied motor vehicle (§ 246) either directly or as an aider and abettor. The jury found defendant guilty of second degree murder. It found the personal-firearm-use allegation not true, but found that a principal intentionally used a firearm and the shooting was committed for the benefit of a criminal street *1180gang. The jury acquitted defendant of both counts of attempted murder, shooting from a motor vehicle, and shooting at an occupied motor vehicle. It convicted defendant of being an active participant in a criminal street gang.

The Court of Appeal, in an opinion authored by Justice Morrison, reversed the murder conviction and otherwise affirmed the judgment. It found two errors in the case. It held the trial court had properly admitted defendant’s first statement that he had been in the car but that the court should have excluded his subsequent statement that he had fired a gun. It concluded that the latter statement was procured by a false promise of leniency. It found this error harmless beyond a reasonable doubt “as a pure evidentiary matter.” But, partly due to this error, the Court of Appeal also held the trial court erred in instructing the jury on second degree felony murder. It found this error was prejudicial and reversed the murder conviction. It explained: “Second degree felony murder, the only express theory of second degree murder offered to the jury, was based on the underlying felony of shooting into an occupied vehicle. The merger doctrine prevents using an assaultive-type crime as the basis for felony murder unless the underlying crime is committed with an intent collateral to committing an injury that would cause death. Without the evidence of defendant’s statements about the shooting, there was no evidence from which a collateral intent or purpose could be found. Accordingly, it was error to instruct on second degree felony murder and the murder conviction must be reversed.”

Justice Nicholson dissented from the reversal of the murder conviction. Relying on People v. Hansen (1994) 9 Cal.4th 300 [36 Cal.Rptr.2d 609, 885 P.2d 1022] (Hansen), he argued that the underlying felony did not merge with the homicide for purposes of the second degree felony-murder rule and, accordingly, the trial court had properly instructed the jury on second degree felony murder.

We granted review. Later, we issued an order limiting review to the issues concerning whether the trial court prejudicially erred in instructing the jury on second degree felony murder.

II. Discussion

A. The Constitutionality of the Second Degree Felony-murder Rule

Defendant contends California’s second degree felony-murder rule is unconstitutional on separation of power grounds as a judicially created doctrine with no statutory basis. To explain the issue, we first describe how the doctrine fits in with the law of murder. Then we discuss defendant’s *1181contention. We will ultimately conclude that the doctrine is valid as an interpretation of broad statutory language.

Section 187, subdivision (a), defines murder as “the unlawful killing of a human being, or a fetus, with malice aforethought.” Except for the phrase “or a fetus,” which was added in 1970 in response to this court’s decision in Keeler v. Superior Court (1970) 2 Cal.3d 619 [87 Cal.Rptr. 481, 470 P.2d 617] (see People v. Davis (1994) 7 Cal.4th 797, 803 [30 Cal.Rptr.2d 50, 872 P.2d 591]), this definition has been unchanged since section 187 was first enacted as part of the Penal Code of 1872. Murder is divided into first and second degree murder. (§ 189.) “Second degree murder is the unlawful killing of a human being with malice, but without the additional elements (i.e., willfulness, premeditation, and deliberation) that would support a conviction of first degree murder. (§§ 187, subd. (a), 189; People v. Nieto Benitez (1992) 4 Cal.4th 91, 102 [13 Cal.Rptr.2d 864, 840 P.2d 969].)” (Hansen, supra, 9 Cal.4th at p. 307.)

Critical for our purposes is that the crime of murder, as defined in section 187, includes, as an element, malice. Section 188 defines malice. It may be either express or implied. It is express “when there is manifested a deliberate intention unlawfully to take away the life of a fellow creature.” (§ 188.) It is implied “when no considerable provocation appears, or when the circumstances attending the killing show an abandoned and malignant heart.” {Ibid.) This definition of implied malice is quite vague. Trial courts do not instruct the jury in the statutory language of an abandoned and malignant heart. Doing so would provide the jury with little guidance. “The statutory definition of implied malice has never proved of much assistance in defining the concept in concrete terms.” (People v. Dellinger (1989) 49 Cal.3d 1212, 1217 [264 Cal.Rptr. 841, 783 P.2d 200].) Accordingly, the statutory definition permits, even requires, judicial interpretation. We have interpreted implied malice as having “both a physical and a mental component. The physical component is satisfied by the performance of ‘an act, the natural consequences of which are dangerous to life.’ (People v. Watson (1981) 30 Cal.3d 290, 300 [179 Cal.Rptr. 43, 637 P.2d 279].) The mental component is the requirement that the defendant ‘knows that his conduct endangers the life of another and . . . acts with a conscious disregard for life.’ (Ibid., internal quotation marks omitted.)” (People v. Patterson (1989) 49 Cal.3d 615, 626 [262 Cal.Rptr. 195, 778 P.2d 549] (lead opn. of Kennard, J.) (Patterson).)2

*1182A defendant may also be found guilty of murder under the felony-murder rule. The felony-murder rule makes a killing while committing certain felonies murder without the necessity of further examining the defendant’s mental state. The rule has two applications: first degree felony murder and second degree felony murder. We have said that first degree felony murder is a “creation of statute” (i.e., § 189) but, because no statute specifically describes it, that second degree felony murder is a “common law doctrine.” (People v. Robertson (2004) 34 Cal.4th 156, 166 [17 Cal.Rptr.3d 604, 95 P.3d 872] (Robertson).) First degree felony murder is a killing during the course of a felony specified in section 189, such as rape, burglary, or robbery. Second degree felony murder is “an unlawful killing in the course of the commission of a felony that is inherently dangerous to human life but is not included among the felonies enumerated in section 189 ... .” (Robertson, supra, 34 Cal.4th at p. 164.)

In Patterson, Justice Kennard explained the reasoning behind and the justification for the second degree felony-murder rule: “The second degree felony-murder rule eliminates the need for the prosecution to establish the mental component [of conscious-disregard-for-life malice]. The justification therefor is that, when society has declared certain inherently dangerous conduct to be felonious, a defendant should not be allowed to excuse himself by saying he was unaware of the danger to life because, by declaring the conduct to be felonious, society has warned him of the risk involved. The physical requirement, however, remains the same; by committing a felony inherently dangerous to life, the defendant has committed ‘an act, the natural consequences of which are dangerous to life’ ([People v.] Watson, supra, 30 Cal.3d at p. 300), thus satisfying the physical component of implied malice.” (Patterson, supra, 49 Cal.3d at p. 626.)

The second degree felony-murder rule is venerable. It “has been a part of California’s criminal law for many decades. (See People v. Wright (1914) 167 Cal. 1, 5 [138 P. 349]; Pike, What Is Second Degree Murder in California (1936) 9 So.Cal. L.Rev. 112, 118-119.)” (Patterson, supra, 49 Cal.3d at p. 621; see also People v. Doyell (1874) 48 Cal. 85, 94.) Because of this, we declined to reconsider the rule in Patterson. (Patterson, supra, at p. 621.) Even earlier, in 1966, we rejected the argument that we should abandon the doctrine, explaining that “the concept lies imbedded in our law.” (People v. Phillips (1966) 64 Cal.2d 574, 582 [51 Cal.Rptr. 225, 414 P.2d 353]; see also People v. Mattison (1971) 4 Cal.3d 177, 184 [93 Cal.Rptr. 185, 481 P.2d 193] (Mattison) [describing the rule as “well-settled”].)

But some former and current members of this court have questioned the rule’s validity because no statute specifically addresses it. Chief Justice Bird argued for its abolition in her concurring opinion in People v. Burroughs *1183(1984) 35 Cal.3d 824, 836-854 [201 Cal.Rptr. 319, 678 P.2d 894], Justice Brown did so in dissent in Robertson, supra, 34 Cal.4th at pages 186-192, and again while concurring and dissenting in People v. Howard (2005) 34 Cal.4th 1129, 1140-1141 [23 Cal.Rptr.3d 306, 104 P.3d 107], Justices Werdegar and Moreno have viewed the rule as ripe for reconsideration in an appropriate case. (Robertson, supra, at pp. 174—177 (conc. opn. of Moreno, J.), 185-186 (dis. opn. of Werdegar, J.).) In Patterson, Justice Panelli questioned the rule’s constitutional validity. As he pointed out, “There are, or at least should be, no nonstatutory crimes in this state. (In re Brown (1973) 9 Cal.3d 612, 624 [108 Cal.Rptr. 465, 510 P.2d 1017]; see Pen. Code, § 6.)” (Patterson, supra, 49 Cal.3d at p. 641 (cone. & dis. opn. of Panelli, J.).) He was concerned that the second degree felony-murder rule is solely a judicial creation not derived from statute and was thus “not quite convinced” that it “stands on solid constitutional ground.” (Ibid.)

In line with these concerns, defendant argues that the second degree felony-murder rule is invalid on separation of powers grounds. As he points out, we have repeatedly said that “ ‘the power to define crimes and fix penalties is vested exclusively in the legislative branch.’ (Keeler v. Superior Court[, supra,] 2 Cal.3d 619, 631 ... ; [citations].)” (People v. Superior Court (Romero) (1996) 13 Cal.4th 497, 516 [53 Cal.Rptr.2d 789, 917 P.2d 628].) Defendant asks rhetorically, “How, then, in light of the statutory abrogation of common law crimes and the constitutional principle of separation of powers, does second degree felony murder continue to exist when this court has repeatedly acknowledged that the crime is a judicial creation?”

This court has never directly addressed these concerns and this argument, or explained the statutory basis of the second degree felony-murder rule. We do so now. We agree with Justice Panelli that there are no nonstatutory crimes in this state. Some statutory or regulatory provision must describe conduct as criminal in order for the courts to treat that conduct as criminal. (§ 6.)3 But, as we explain, the second degree felony-murder rule, although derived from the common law, is based on statute; it is simply another interpretation of section 188’s “abandoned and malignant heart” language.

Many provisions of the Penal Code were enacted using common law terms that must be interpreted in light of the common law. For example, section 484 defines theft as “feloniously” taking the property of another. The *1184term “feloniously”—which has little meaning by itself—incorporates the common law requirement that the perpetrator must intend to permanently deprive the owner of possession of the property. Accordingly, we have looked to the common law to determine the exact contours of that requirement. (People v. Avery (2002) 27 Cal.4th 49, 55 [115 Cal.Rptr.2d 403, 38 P.3d 1]; People v. Davis (1998) 19 Cal.4th 301, 304, fn. 1 [79 Cal.Rptr.2d 295, 965 P.2d 1165].) Thus, the intent-to-permanently-deprive requirement, although nonstatutory in the limited sense that no California statute uses those words, is based on statute. The murder statutes are similarly derived from the common law. (Keeler v. Superior Court, supra, 2 Cal.3d 619 [looking to the common law to determine the exact meaning of “human being” under § 187].) “It will be presumed . . . that in enacting a statute the Legislature was familiar with the relevant rules of the common law, and, when it couches its enactment in common law language, that its intent was to continue those rules in statutory form.” (Keeler v. Superior Court, supra, at p. 625.)

Even conscious-disregard-for-life malice is nonstatutory in the limited sense that no California statute specifically uses those words. But that form of implied malice is firmly based on statute; it is an interpretation of section 188’s “abandoned and malignant heart” language. Similarly, the second degree felony-murder rule is nonstatutory in the sense that no statute specifically spells it out, but it is also statutory as another interpretation of the same “abandoned and malignant heart” language. We have said that the “felony-murder rule eliminates the need for proof of malice in connection with a charge of murder, thereby rendering irrelevant the presence or absence of actual malice, both with regard to first degree felony murder and second degree felony murder.” (Robertson, supra, 34 Cal.4th at p. 165.) But analytically, this is not precisely correct. The felony-murder rule renders irrelevant conscious-disregard-for-life malice, but it does not render malice itself irrelevant. Instead, the felony-murder rule “acts as a substitute” for conscious-disregard-for-life malice. (Patterson, supra, 49 Cal.3d at p. 626.) It simply describes a different form of malice under section 188. “The felony-murder rule imputes the requisite malice for a murder conviction to those who commit a homicide during the perpetration of a felony inherently dangerous to human life.” (Hansen, supra, 9 Cal.4th at p. 308.)

A historical review confirms this view. California’s first penal law was the Crimes and Punishments Act of 1850 (Act of 1850). (Stats. 1850, ch. 99, p. 229.) Section 19 of that act defined murder as “the unlawful killing of a human being, with malice aforethought, either express or implied. The unlawful killing may be effected by any of the various means by which death may be occasioned.” (Stats. 1850, ch. 99, § 19, p. 231.) Sections 20 and 21 of the Act of 1850 defined express and implied malice, respectively. Section 21 stated, “Malice shall be implied when no considerable provocation appears, or when all the circumstances of the killing show an abandoned and *1185malignant heart.” (Stats. 1850, ch. 99, § 21, p. 231.) It also set the punishment for murder as death. At that time, murder was not divided into degrees. The division of murder into degrees “occurred in 1856, when the Legislature amended section 21 of the Act of 1850 to divide the crime of murder into two degrees: first degree murder was defined as that committed by certain listed means or in the perpetration of certain listed felonies, while all other murders were of the second degree.” (People v. Dillon (1983) 34 Cal.3d 441, 466 [194 Cal.Rptr. 390, 668 P.2d 697] (Dillon).)

Sections 22 to 25 of the Act of 1850 concerned voluntary and involuntary manslaughter. Section 25 provided, in its entirety, “Involuntary manslaughter shall consist in the killing of a human being, without any intent so to do; in the commission of an unlawful act, or a lawful act, which probably might produce such a consequence in an unlawful manner; Provided, that where such involuntary killing shall happen in the commission of an unlawful act, which in its consequences naturally tends to destroy the life of a human being, or is committed in the prosecution of a felonious intent, the offence shall be deemed and adjudged to be murder.'” (Stats. 1850, ch. 99, § 25, p. 231, italics of “Provided” in original, all other italics added.)

In 1872, the Legislature adopted the current Penal Code. Section 187 defined murder essentially the same as did the Act of 1850. (Keeler v. Superior Court, supra, 2 Cal.3d at p. 624.) As can readily be seen, section 188 also defined implied malice essentially the same as did the Act of 1850.

But the 1872 Penal Code did recast the definition of involuntary manslaughter. The new section 192 defined voluntary and involuntary manslaughter, as it still does today. (In the interim, vehicular manslaughter has been added as another form of manslaughter.) Subdivision 2 of that section defined and, now labeled subdivision (b), still defines, involuntary manslaughter as an unlawful killing without malice “in the commission of an unlawful act, not amounting to felony, or in the commission of a lawful act which might produce death, in an unlawful manner, or without due caution and circumspection.” (§ 192, subd. (b), italics added.) The proviso portion of section 25 of the Act of 1850 was deleted and essentially replaced with the italicized language “not amounting to [a] felony.”

In Dillon, supra, 34 Cal.3d 441, this court considered issues concerning the first degree felony-murder rule. As part of its discussion, Dillon stated that the proviso portion of section 25 of the Act of 1850 “codified the common law felony-murder rule in this state,” and that “the Legislature’s decision not to reenact the felony-murder provision of section 25 in the 1872 codification implied an intent to abrogate the common law felony-murder rule that the section had embodied since 1850.” (Dillon, supra, at pp. 465, 467.) If these *1186statements were correct, it would be difficult to conclude that second degree felony murder is based on statute today. But this language in Dillon was dicta because Dillon involved the first degree, not second degree, felony-murder rule. Now that the point is critical, we examine it further and, viewing the relevant 1850 and 1872 statutes in context, conclude that Dillon was not correct in this regard.

A codification of the felony-murder rule would logically be placed in the statutes defining murder, not in a statute defining involuntary manslaughter such as section 25 of the Act of 1850. Moreover, any reasonable felony-murder rule would apply to any killing during the course of a felony, not just an “involuntary killing” as stated in that same section 25. As Dillon noted, “It would have been absurd, of course, to punish as murder those killings [i.e., involuntary killings] but not ‘voluntary’ killings during a felony . . . .” (Dillon, supra, 34 Cal.3d at p. 465, fn. 12.) Dillon ascribed section 25’s apparent limitation of the felony-murder rule to involuntary killings to a “quirk of draftsmanship.” (Dillon, supra, at p. 465, fn. 12.) If that section’s proviso is viewed as a codification of the common law of felony murder, the draftsmanship would, indeed, be quirky. It would be doubly quirky: It would be unusual to codify a common law rule concerning murder in a statute defining involuntary manslaughter, and it would be quirky to include in the felony-murder rule only involuntary killings to the apparent exclusion of voluntary killings. But viewed instead as what it no doubt was-—a proviso merely limiting the scope of involuntary manslaughter—the draftsmanship makes sense.

Without the proviso, section 25 of the Act of 1850 would have meant, or at least would have been susceptible to the interpretation, that any killing “in the commission of an unlawful act”-—i.e., any unlawful act, whether misdemeanor or felony—is involuntary manslaughter. The proviso simply makes clear that involuntary manslaughter does not include killings in the course of a felony, which remain murder. As this court explained in a case in which the crime was committed before, but the opinion filed after, adoption of the 1872 Penal Code, “Whenever one, in doing an act with the design of committing a felony, takes the life of another, even accidentally, this is murder.” (People v. Doyell, supra, 48 Cal. at p. 94 [citing § 25 of the Act of 1850].) The new section 192 merely simplified the definition of involuntary manslaughter by replacing the earlier proviso with the new language, “not amounting to felony.” In this way, the Legislature avoided the awkwardness of having a broad definition of involuntary manslaughter followed by a proviso limiting that definition. So viewed, the language of section 25 of the Act of 1850 and 1872’s new section 192 all make sense; no need exists to ascribe any language to quirky draftsmanship or to view section 192’s simplified definition of involuntary manslaughter as abrogating a common law rule concerning murder.

*1187The notes of the California Code Commissioners accompanying the 1872 adoption of the Penal Code, which are entitled to substantial weight (Keeler v. Superior Court, supra, 2 Cal.3d at p. 630), provide no hint of an intent to abrogate the felony-murder rule. The note accompanying section 187, although not discussing this precise point, shows that the statutory term “malice aforethought” incorporated the term’s common law meaning. (Code commrs., note foil., Ann. Pen. Code, § 187 (1st ed. 1872, Haymond & Burch, commrs. annotators) pp. 80-81 (1872 Code commissioners note) [citing various common law sources in discussing the meaning of malice aforethought].) Similarly, nothing in the adoption of sections 188 and 189 suggests an intent to change the then existing law of murder, including, as relevant here, the definition of implied malice and its common law antecedents. The code commissioners note accompanying the 1872 adoption of section 192 states that “[t]his section embodies the material portions of Sections 22, 23, 24, and 25 of the Crimes and Punishment Act of 1850.” (1872 Code commrs. note, supra, p. 85, italics added.) This latter note strongly indicates that the language change from section 25 of the Act of 1850 to section 192 was not intended to change the law of manslaughter, much less to change the law of murder by abrogating the common law felony-murder rule. Any statute that “embodies the material portions” of predecessor statutes would not change the law in such a substantial manner.

We are unaware of any California case even remotely contemporaneous with the adoption of the 1872 Penal Code (i.e., any case before Dillon, supra, 34 Cal.3d 441) suggesting that the language change from section 25 of the Act of 1850 to section 192 abrogated the felony-murder rule or otherwise changed the law of murder. Indeed, cases postdating People v. Doyell, supra, 48 Cal. 85, and the adoption of the 1872 Penal Code, but still ancient from today’s perspective, cited Doyell in applying the second degree felony-murder rule without any hint that Doyell was obsolete because it had cited section 25 of the Act of 1850. (See People v. Olsen (1889) 80 Cal. 122, 126-127 [22 P. 125]; People v. Ferugia (1928) 95 Cal.App. 711, 718 [273 P. 99]; People v. Hubbard (1923) 64 Cal.App. 27, 33 [220 P. 315].)

For these reasons, we conclude that the Legislature’s replacement of the proviso language of section 25 of the Act of 1850 with the shorthand language “not amounting to felony” in section 192 did not imply an abrogation of the common law felony-murder rule. The “abandoned and malignant heart” language of both the original 1850 law and today’s section 188 contains within it the common law second degree felony-murder rule. The willingness to commit a felony inherently dangerous to life is a *1188circumstance showing an abandoned and malignant heart. The second degree felony-murder rule is based on statute and, accordingly, stands on firm constitutional ground.4

B. The Merger Doctrine and Second Degree Felony Murder

Although today we reaffirm the constitutional validity of the longstanding second degree felony-murder rule, we also recognize that the rule has often been criticized and, indeed, described as disfavored. (E.g., Patterson, supra, 49 Cal.3d at p. 621.) We have repeatedly stated, as recently as 2005, that the rule “ ‘ “deserves no extension beyond its required application.” ’ ” (People v. Howard, supra, 34 Cal.4th at p. 1135.) For these reasons, although the second degree felony-murder rule originally applied to all felonies (People v. Doyell, supra, 48 Cal. at pp. 94-95; Pike, What Is Second Degree Murder in California?, supra, 9 So.Cal. L.Rev. at pp. 118-119), this court has subsequently restricted its scope in at least two respects to ameliorate its perceived harshness.

First, “[i]n People v. Ford (1964) 60 Cal.2d 772, 795 [36 Cal.Rptr. 620, 388 P.2d 892], the court restricted the felonies that could support a conviction of second degree murder, based upon a felony-murder theory, to those felonies that are ‘inherently dangerous to human life.’ ” (Hansen, supra, 9 Cal.4th at p. 308.) Whether a felony is inherently dangerous is determined from the elements of the felony in the abstract, not the particular facts. (Patterson, supra, 49 Cal.3d at p. 621.) This restriction is not at issue here. Section 246 makes it a felony to “maliciously and willfully discharge a firearm at an . . . occupied motor vehicle . . . ,”5 In Hansen, supra, at pages 309-311, we held that shooting at an “inhabited dwelling house” under section 246 is inherently dangerous even though the inhabited dwelling house does not have to be actually occupied at the time of the shooting. That being the case, shooting at a vehicle that is actually occupied clearly is inherently dangerous.

*1189But the second restriction—the “merger doctrine”—is very much at issue. The merger doctrine developed due to the understanding that the underlying felony must be an independent crime and not merely the killing itself. Thus, certain underlying felonies “merge” with the homicide and cannot be used for purposes of felony murder. The specific question before us is how to apply the merger doctrine. The Court of Appeal divided on the question and on how to apply our precedents. But the majority and dissent agreed on one thing— that the current state of the law regarding merger is “muddled.” We agree that the scope and application of the merger doctrine as applied to second degree murder needs to be reconsidered. To explain this, we will first review the doctrine’s historical development. Then we will discuss what to do with the merger doctrine and, ultimately, will conclude that the trial court should not have instructed on felony murder.

1. Historical Review

The merger doctrine arose in the seminal case of Ireland, supra, 70 Cal.2d 522, and hence sometimes is called the “Ireland merger doctrine.” In Ireland, the defendant shot and killed his wife, and was convicted of second degree murder. The trial court instructed the jury on second degree felony murder with assault with a deadly weapon the underlying felony. We held the instruction improper, adopting the “so-called ‘merger’ doctrine” that had previously been developed in other jurisdictions. (Id. at p. 540.) We explained our reasons: “[T]he utilization of the felony-murder rule in circumstances such as those before us extends the operation of that rule ‘beyond any rational function that it is designed to serve.’ (People v. Washington (1965) 62 Cal.2d 777, 783 [44 Cal.Rptr. 442, 402 P.2d 130].) To allow such use of the felony-murder rule would effectively preclude the jury from considering the issue of malice aforethought in all cases wherein homicide has been committed as a result of a felonious assault—a category which includes the great majority of all homicides. This kind of bootstrapping finds support neither in logic nor in law. We therefore hold that a second degree felony-murder instruction may not properly be given when it is based upon a felony which is an integral part of the homicide and which the evidence produced by the prosecution shows to be an offense included in fact within the offense charged.” (Id. at p. 539.)6

We next confronted the merger doctrine in a second degree felony-murder case in Mattison, supra, 4 Cal.3d 177. As we later described Mattison’’ s facts, *1190“[i]n that case, the defendant and the victim both were inmates of a correctional institution. The defendant worked as a technician in the medical laboratory. He previously had offered to sell alcohol to inmates, leading the victim, an alcoholic, to seek alcohol from him. The defendant supplied the victim with methyl alcohol, resulting in the victim’s death by methyl alcohol poisoning. [][] At trial, the court instructed on felony murder based upon the felony of mixing poison with a beverage, an offense proscribed by the then current version of section 347 (‘ “Every person who wilfully mingles any poison with any food, drink or medicine, with intent that the same shall be taken by any human being to his injury, is guilty of a felony.” ’) (4 Cal.3d at p. 184.) The defendant was convicted of second degree murder.” (Hansen, supra, 9 Cal.4th at p. 313.)

The Mattison defendant argued “that the offense of administering poison with the intent to injure is an ‘integral part of’ and ‘included in fact within the offense’ of murder by poison” within the meaning of Ireland, supra, 70 Cal.2d 522. (Mattison, supra, 4 Cal.3d at p. 185.) We disagreed. “The instant case . . . presents an entirely different situation from the one that confronted us in Ireland. The facts before us are very similar to People v. Taylor (1970) 11 Cal.App.3d 57 [89 Cal.Rptr. 697], in which the victim died as a result of an overdose of heroin which had been furnished to her by the defendant. The defendant was convicted of second degree murder and the question presented was whether application of the felony-murder rule constituted error under Ireland. . . . [T]he Taylor court concluded that application of the felony-murder rule was proper because the underlying felony was committed with a ‘collateral and independent felonious design.’ (People v. Taylor, supra, 11 Cal.App.3d 57, 63.) In other words the felony was not done with the intent to commit injury which would cause death. Giving a felony-murder instruction in such a situation serves rather than subverts the purpose of the rule. ‘While the felony-murder rule can hardly be much of a deterrent to a defendant who has decided to assault his victim with a deadly weapon, it seems obvious that in the situation presented in the case at bar, it does serve a rational purpose: knowledge that the death of a person to whom heroin is furnished may result in a conviction for murder should have some effect on the defendant’s readiness to do the furnishing.’ (People v. Taylor, supra, 11 Cal.App.3d 57, 63.) The instant case is virtually indistinguishable from Taylor, and we hold that it was proper to instruct the jury on second degree felony murder.” (Mattison, supra, 4 Cal.3d at pp. 185-186.)

In People v. Smith (1984) 35 Cal.3d 798 [201 Cal.Rptr. 311, 678 P.2d 886], the defendant was convicted of the second degree murder of her two-year-old daughter. We had to decide whether the trial court correctly instructed the jury on second degree felony murder with felony child abuse (now § 273a, subd. (a)) the underlying felony. We reviewed some of the felonies that do not merge but found them distinguishable. (People v. Smith, supra, at p. 805.) *1191We explained that the crime at issue was “child abuse of the assaultive variety” for which we could “conceive of no independent purpose.” (Id. at p. 806.) Accordingly, we concluded that the offense merged with the resulting homicide, and that the trial court erred in instructing on felony murder.

Our merger jurisprudence took a different turn in Hansen, supra, 9 Cal.4th 300. In that case, the defendant was convicted of second degree murder for shooting at a house, killing one person. The trial court instmcted the jury on second degree felony murder, with discharging a firearm at an inhabited dwelling house (§ 246) the underlying felony. The majority concluded that the crime of discharging a firearm at an inhabited dwelling house “does not ‘merge’ with a resulting homicide so as to preclude application of the felony-murder doctrine.” (Hansen, supra, at p. 304.) We noted that this court “has not extended the Ireland doctrine beyond the context of assault, even under circumstances in which the underlying felony plausibly could be characterized as ‘an integral part of’ and ‘included in fact within’ the resulting homicide.” (Id. at p. 312.)

We discussed in detail Mattison, supra, 4 Cal.3d 177, and People v. Taylor, supra, 11 Cal.App.3d 57, the case Mattison relied on. We agreed with Taylor’s “rejection of the premise that Ireland’s ‘integral part of the homicide’ language constitutes the crucial test in determining the existence of merger. Such a test would be inconsistent with the underlying rule that only felonies ‘inherently dangerous to human life’ are sufficiently indicative of a defendant’s culpable mens rea to warrant application of the felony-murder rule. [Citation.] The more dangerous the felony, the more likely it is that a death may result directly from the commission of the felony, but resort to the ‘integral part of the homicide’ language would preclude application of the felony-murder rule for those felonies that are most likely to result in death and that are, consequently, the felonies as to which the felony-murder doctrine is most likely to act as a deterrent (because the perpetrator could foresee the great likelihood that death may result, negligently or accidentally).” (Hansen, supra, 9 Cal.4th at p. 314.)

But the Hansen majority also disagreed with People v. Taylor, supra, 11 Cal.App.3d 57, in an important respect. We declined “to adopt as the critical test determinative of merger in all cases” language in Taylor indicating “that the rationale for the merger doctrine does not encompass a felony ‘ “committed with a collateral and independent felonious design.” ’ (People v. Taylor, supra, 11 Cal.App.3d at p. 63; see also People v. Burton (1971) 6 Cal.3d 375, 387 [99 Cal.Rptr. 1, 491 P.2d 793].) Under such a test, a felon who acts with a purpose other than specifically to inflict injury upon someone—for example, with the intent to sell narcotics for financial gain, or to discharge a firearm at a building solely to intimidate the occupants—is subject to greater *1192criminal liability for an act resulting in death than a person who actually intends to injure the person of the victim. Rather than rely upon a somewhat artificial test that may lead to an anomalous result, we focus upon the principles and rationale underlying the foregoing language in Taylor, namely, that with respect to certain inherently dangerous felonies, their use as the predicate felony supporting application of the felony-murder rule will not elevate all felonious assaults to murder or otherwise subvert the legislative intent.” (Hansen, supra, 9 Cal.4th at p. 315.)

Hansen went on to explain that “application of the second degree felony-murder rule would not result in the subversion of legislative intent. Most homicides do not result from violations of section 246, and thus, unlike the situation in People v. Ireland, supra, 70 Cal.2d 522, application of the felony-murder doctrine in the present context will not have the effect of ‘precluding] the jury from considering the issue of malice aforethought . . . [in] the great majority of all homicides.’ {Id., at p. 539.) Similarly, application of the felony-murder doctrine in the case before us would not frustrate the Legislature’s deliberate calibration of punishment for assaultive conduct resulting in death, based upon the presence or absence of malice aforethought. . . . [T]his is not a situation in which the Legislature has demanded a showing of actual malice (apart from the statutory requirement that the firearm be discharged ‘maliciously and willfully’) in order to support a second degree murder conviction. Indeed, as discussed above, application of the felony-murder rule, when a violation of section 246 results in the death of a person, clearly is consistent with the traditionally recognized purpose of the second degree felony-murder doctrine—namely the deterrence of negligent or accidental killings that occur in the course of the commission of dangerous felonies.” (Hansen, supra, 9 Cal.4th at p. 315.)

Hansen generated three separate opinions in addition to the majority opinion. Justice Werdegar authored a concurring opinion arguing that the operative test for the merger doctrine is “whether the underlying felony was committed with a ‘collateral and independent felonious design.’ ” (Hansen, supra, 9 Cal.4th at p. 318.) She concurred in the judgment because “[t]he evidence in this case supports the conclusion defendant entertained a collateral and independent felonious design under Mattison and Taylor, namely to intimidate Behaves by firing shots into his house.” {Ibid.)

Justices Mosk and Kennard each authored separate concurring and dissenting opinions. They would have concluded that the underlying felony merged with the resulting homicide, thus precluding use of the felony-murder rule. Justice Kennard argued that “the prosecution’s evidence did not show that defendant had any independent felonious purpose for discharging the firearm at the Behaves residence. That conduct satisfies this court’s definition of an assault.” (Hansen, supra, 9 Cal.4th at p. 330.)

*1193People v. Tabios (1998) 67 Cal.App.4th 1 [78 Cal.Rptr.2d 753] involved the same issue as this case—whether shooting at an occupied vehicle under section 246 merges with the underlying homicide. Relying on Hansen, supra, 9 Cal.4th 300, the Court of Appeal found no merger. (People v. Tabios, supra, at p. 11.)

In Robertson, supra, 34 Cal.4th 156, the issue was whether the trial court properly instmcted the jury on felony murder based on discharging a firearm in a grossly negligent manner. (§ 246.3.) As we later summarized, “[t]he defendant in Robertson claimed he fired into the air, in order to frighten away several men who were burglarizing his car.” (People v. Randle (2005) 35 Cal.4th 987, 1005 [28 Cal.Rptr.3d 725, 111 P.3d 987] (Randle).) Robertson concluded that the merger doctrine did not bar a felony-murder instruction. (Robertson, supra, at p. 160.) Its reasons, however, were quite different than Hansen's reasons.

The Robertson majority reviewed some of the cases discussed above, then focused on Mattison, supra, 4 Cal.3d 177. We said that the Mattison court believed that finding no merger under its facts “was consistent with the deterrent purpose of the felony-murder rule, because we envisioned that application of the felony-murder rule would deter commission of the underlying inherently dangerous crime. (Id. at pp. 185-186.) Although a person who has decided to assault another would not be deterred by the felony-murder rule, we declared, a defendant with some collateral purpose may be deterred. The knowledge that a murder conviction may follow if an offense such as furnishing a controlled substance or tainted alcohol causes death 1 “should have some effect on the defendant’s readiness to do the furnishing.” ’ (Id. at p. 185.)” (Robertson, supra, 34 Cal.4th at pp. 170-171.)

We noted that Mattison, supra, 4 Cal.3d 177, focused on the fact that the underlying felony’s purpose “was independent of or collateral to an intent to cause injury that would result in death.” (Robertson, supra, 34 Cal.4th at p. 171.) Then we explained, “Although the collateral purpose rationale may have its drawbacks in some situations (Hansen, supra, 9 Cal.4th at p. 315), we believe it provides the most appropriate framework to determine whether, under the facts of the present case, the trial court properly instructed the jury. The defendant’s asserted underlying purpose was to frighten away the young men who were burglarizing his automobile. According to defendant’s own statements, the discharge of the firearm was undertaken with a purpose collateral to the resulting homicide, rendering the challenged instruction permissible. As Justice Werdegar pointed out in her concurring opinion in Hansen, a defendant who discharges a firearm at an inhabited dwelling house, for example, has a purpose independent from the commission of a resulting *1194homicide if the defendant claims he or she shot to intimidate, rather than to injure or kill the occupants. (Hansen, supra, 9 Cal.4th at p. 318 (conc. opn. of Werdegar, J.).)” (Ibid.)

In Robertson, the Court of Appeal had said “that application of the merger doctrine was necessary in order to avoid the absurd consequence that ‘[defendants who admit an intent to kill, but claim to have acted with provocation or in honest but unreasonable self-defense, would likely have a stronger chance [than defendants who claimed “I didn’t mean to do it”] of being convicted of the lesser offense of voluntary manslaughter.’ ” (Robertson, supra, 34 Cal.4th at pp. 172-173.) We responded: “The asserted anomaly identified by the Court of Appeal is characteristic of the second degree felony-murder rule in general and is inherent in the doctrine’s premise that it is reasonable to impute malice—or, more precisely, to eliminate consideration of the presence or absence of actual malice—because of the defendant’s commission of an underlying felony that is inherently and foreseeably dangerous. [Citations.] Reliance on section 246.3 as the predicate offense presents no greater anomaly in this regard than such reliance on any other inherently dangerous felony.” (Id. at p. 173.)

Thus, the Robertson majority abandoned the rationale of Hansen, supra, 9 Cal.4th 300, and resurrected the collateral purpose rationale of Mattison, supra, 4 Cal.3d 177, at least when the underlying felony is a violation of section 246.3.

Robertson generated four separate opinions in addition to the majority opinion. Justice Moreno’s concurring opinion agreed that the refusal to apply the merger doctrine was correct under the current state of the law, but he was concerned whether the court should continue to adhere to the second degree felony-murder doctrine at all. (Robertson, supra, at pp. 174-177.) Justice Brown argued in dissent that the second degree felony-murder rule should be abandoned entirely. (Robertson, supra, 34 Cal.4th at pp. 186-192.)

In a separate dissent, Justice Kennard disagreed that “defendant’s claimed objective to scare the victim” was “a felonious purpose that was independent of the killing.” (Robertson, supra, 34 Cal.4th at p. 178.) She noted with approval that “the majority, without explanation, abandoned] the rationale of the Hansen majority, and it returned] to the independent felonious purpose standard, which it had criticized in Hansen, supra, 9 Cal.4th 300.” (Id. at p. 180.) That was the test she had advocated in Hansen. (Ibid.) But she believed that the majority misapplied that test. “An intent to scare a person by shooting at the person is not independent of the homicide because it is, in essence, nothing more than the intent required for an assault, which is not considered an independent felonious purpose. [Citation.] Two examples of *1195independent felonious purpose come to mind: (1) When the felony underlying the homicide is manufacturing methamphetamine, the intent to manufacture this illegal drug is a felonious intent that is independent of the homicide, thus allowing the manufacturer to be convicted of murder if the methamphetamine laboratory explodes and kills an innocent bystander. (2) When the underlying felony is possession of a destructive device, the intent to possess that device is an independent felonious intent, allowing the possessor to be convicted of murder if the device accidentally explodes, killing an unintended victim. But when, as here, a defendant fires a gun to scare the victim, the intended harm—that of scaring the victim—is not independent of the greater harm that occurs when a shot fired with the intent to scare instead results in the victim’s death.” (Id. at p. 183.) “In sum, it makes no sense legally to treat defendant’s alleged intent to scare as ‘felonious’ when such an intent is legally irrelevant [to guilt of the underlying felony] and when the jury never decided whether he had that intent.” (Ibid..)

Justice Werdegar also dissented, arguing that the underlying felony merged with the resulting homicide. She said she “would like to join in the majority reasoning, which is consistent with my Hansen concurrence. But sometimes consistency must yield to a better understanding of the developing law. The anomalies created when assaultive conduct is used as the predicate for a second degree felony-murder theory (see dis. opn. of Kennard, J., ante, at pp. 180-182) are too stark and potentially too productive of injustice to be written off as ‘characteristic of the second degree felony-murder rule in general’ (maj. opn., ante, at p. 173). It simply cannot be the law that a defendant who shot the victim with the intent to kill or injure, but can show he or she acted in unreasonable self-defense, may be convicted of only voluntary manslaughter, whereas a defendant who shot only to scare the victim is precluded from raising that partial defense and is strictly liable as a murderer. The independent and collateral purposes referred to in Mattison must be understood as limited to nonassaultive conduct. In circumstances like the present, the merger doctrine should preclude presentation of a second degree felony-murder theory to the jury.” (Robertson, supra, 34 Cal.4th at p. 185 (dis. opn. of Werdegar, J.).)

In Randle, supra, 35 Cal.4th 987, the trial court, as in Robertson, instructed the jury on second degree felony murder, with discharging a firearm in a grossly negligent manner the underlying felony. (Randle, supra, at p. 1004.) We found the instruction erroneous under the facts. “Here, unlike Robertson, defendant admitted, in his pretrial statements to the police and to a deputy district attorney, he shot at Robinson [the homicide victim]. . . . [][] The fact that defendant admitted shooting at Robinson distinguishes Robertson and supports application of the merger rule here. Defendant’s claim that he shot Robinson in order to rescue [another person] simply provided a motive for the shooting; it was not a purpose independent of the shooting.” (Id. at p. 1005.)

*1196In People v. Bejarano (2007) 149 Cal.App.4th 975 [57 Cal.Rptr.3d 486], as in People v. Tabios, supra, 61 Cal.App.3d 1, and this case, the trial court instructed the jury on second degree felony murder, with shooting at an occupied vehicle under section 246 the underlying felony. The court concluded that the collateral purpose requirement of Robertson, supra, 34 Cal.4th 156, and Randle, supra, 35 Cal.4th 987, applied. “The facts of this case show that appellant discharged the firearm once, intending to shoot the motor vehicle’s occupants, rival gang members, and not intending merely to frighten them. The bullet, however, struck and killed an unintended victim, the driver of another vehicle.” (People v. Bejarano, supra, at p. 978.) Relying primarily on Randle, supra, 35 Cal.4th 987, the Court of Appeal concluded that the trial court erred in instructing on felony murder. “Thus, Randle controls this case, the predicate felony merged with the homicide, and the trial court erred in instructing the jury on second degree felony murder based on discharging a firearm at an occupied motor vehicle in violation of section 246.” (People v. Bejarano, supra, at p. 990.)

The most recent significant development is the Court of Appeal’s opinion in this case. The majority noted that People v. Tabios, supra, 61 Cal.App.4th 1, had relied on Hansen, supra, 9 Cal.4th 300, in finding no merger, but then it also noted that this court “returned to the Mattison collateral purpose rationale in” Robertson, supra, 34 Cal.4th 156. After reviewing other recent cases, it stated, “From this muddled state of the law, we discern the rule to be that second degree felony murder is applicable to an assaultive-type crime, such as when shooting at a person is involved, provided that the crime was committed with a purpose independent of and collateral to causing injury. Since the Supreme Court could have upheld instruction on felony murder in Randle on the basis that most homicides are not committed by negligently discharging a gun and did not, we conclude the collateral purpose rule is the proper test of merger in these types of cases.”

Regarding whether a collateral purpose exists in this case, the Court of Appeal majority noted that it had held defendant’s statement that he had fired the gun “ ‘to scare them’ ” should have been excluded. “Without defendant’s statements about firing the gun,” the majority concluded, “there was no admissible evidence of a collateral purpose by defendant or any of his companions. Indeed, the reasonable inference is that one who shoots another at close range intends to harm, if not to kill.” Thus it found the court erred, prejudicially, in instructing on second degree felony murder.

In dissent, Justice Nicholson agreed with the majority that the present state of the law is muddled. But he concluded that this court has not overruled Hansen, supra, 9 Cal.4th 300, and found that case, rather than Robertson, supra, 34 Cal.4th 156, or Randle, supra, 35 Cal.4th 987, to be on point. He *1197believed that “the only rule that can be gleaned from Robertson and Randle is that the collateral purpose rationale applies to cases involving a violation of section 246.3, which this case does not.” Accordingly, he would have held “that merger is inappropriate when the underlying offense is a violation of section 246.”

2. Analysis

The current state of the law regarding the Ireland merger doctrine is problematic in at least two respects.

First, two different approaches currently exist in determining whether a felony merges. Hansen, supra, 9 Cal.4th 300, which we have never expressly overruled, held that a violation of section 246, at least when predicated on shooting at an inhabited dwelling house, never merges. Robertson, supra, 34 Cal.4th 156, and Randle, supra, 35 Cal.4th 987, held that a violation of section 246.3 does merge unless it is done with a purpose collateral to the resulting homicide. If Hansen, on the one hand, and Robertson and Randle on the other hand, are all still valid authority, the question arises which approach applies here. People v. Tabios, supra, 67 Cal.App.4th 1, relied on Hansen to conclude that shooting at an occupied vehicle under section 246 never merges. People v. Bejarano, supra, 149 Cal.App.4th 975, relied on the more recent Robertson and Randle opinions to conclude that the same felony does merge unless accompanied by a collateral purpose. The Court of Appeal here, rather understandably, divided on the question. This court has never explained whether Hansen retains any viability after Robertson and Randle and, if so, how a court is to go about determining which approach to apply to a given underlying felony.

Second, Randle, when juxtaposed with Robertson, brings into sharp focus the anomaly that we noted in Robertson and accepted as inherent in the second degree felony-murder rule, and that we noted in Hansen and avoided by concluding that the merger doctrine never applies to shooting at an inhabited dwelling house. In combination, Robertson and Randle hold that, when the Hansen test does not apply (i.e., at least when the underlying felony is a violation of § 246.3), the underlying felony merges, and the felony-murder rule does not apply, if the defendant intended to shoot at the victim (Randle), but the underlying felony does not merge, and the felony-murder rule does apply, if the defendant merely intended to frighten, perhaps because he believed the victim was burglarizing his car (Robertson). This result is questionable for the reasons discussed in the separate opinions in Robertson. Moreover, as we discuss further below, the Robertson and Randle approach injected a factual component into the merger question that did not previously exist.

*1198In light of these problems, we conclude we need to reconsider our merger doctrine jurisprudence. As Justice Werdegar observed in her dissenting opinion in Robertson, “sometimes consistency must yield to a better understanding of the developing law.” (Robertson, supra, 34 Cal.4th at p. 185.) In considering this question, we must also keep in mind the purposes of the second degree felony-murder rule. We have identified two. The purpose we have most often identified “is to deter felons from killing negligently or accidentally by holding them strictly responsible for killings they commit.” (People v. Washington, supra, 62 Cal.2d at p. 781.) Another purpose is to deter commission of the inherently dangerous felony itself. (Robertson, supra, 34 Cal.4th at p. 171 [“the second degree felony-murder rule is intended to deter both carelessness in the commission of a crime and the commission of the inherently dangerous crime itself’]; Hansen, supra, 9 Cal.4th at pp. 310-311, 314.)

We first consider whether Hansen, supra, 9 Cal.4th 300, has any continuing vitality after Robertson, supra, 34 Cal.4th 156, and Randle, supra, 35 Cal.4th 987. In Robertson and Randle, we unanimously rejected the Hansen test, at least when the underlying felony is a violation of section 246.3. Although Hansen avoided the problems inherent in the Robertson approach by simply stating the felony at issue will never merge, we see no basis today to resurrect the Hansen approach for a violation of section 246.3. Indeed, doing so would arguably be inconsistent with Hansen’s reasoning. Hansen explained that most homicides do not involve violations of section 246, and thus holding that such homicides do not merge would not “subvert the legislative intent.” (Hansen, supra, at p. 315.) But most fatal shootings, and certainly those charged as murder, do involve discharging a firearm in at least a grossly negligent manner. Fatal shootings, in turn, are a high percentage of all homicides. Thus, holding that a violation of section 246.3 never merges would greatly expand the range of homicides subject to the second degree felony-murder rule. We adhere to Robertson and Randle to the extent they declined to extend the Hansen approach to a violation of section 246.3.

But if, as we conclude, the Hansen test does not apply to a violation of section 246.3, we must decide whether it still applies to any underlying felonies. The tests stated in Hansen and in Robertson and Randle cannot both apply at the same time. If Hansen governs, the underlying felony will never merge. If Robertson and Randle govern, the underlying felony will always merge unless the court can discern some independent felonious purpose. But we see no principled basis by which to hold that a violation of section 246 never merges, but a violation of section 246.3 does merge unless done with an independent purpose. We also see no principled test that another court could use to determine which approach applies to other possible underlying felonies. The court in People v. Bejarano, supra, 149 Cal.App.4th 975, implicitly concluded that Robertson and Randle now govern to the exclusion *1199of the Hansen test. We agree. The Robertson and Randle test and the Hansen test cannot coexist. Our analyses in Robertson and Randle implicitly overruled the Hansen test. We now expressly overrule People v. Hansen, supra, 9 Cal.4th 300, to the extent it stated a test different than the one of Robertson and Randle. Doing so also requires us to disapprove of People v. Tabios, supra, 67 Cal.App.4th 1.

But the test of Robertson, supra, 34 Cal.4th 156, and Randle, supra, 35 Cal.4th 987, has its own problems that were avoided in Hansen but resurfaced when we abandoned the Hansen test. Our holding in Randle made stark the anomalies that Justices Kennard and Werdegar identified in Robertson. On reflection, we do not believe that a person who claims he merely wanted to frighten the victim should be subject to the felony-murder rule (Robertson), but a person who says he intended to shoot at the victim is not subject to that rule (Randle). Additionally, Robertson said that the intent to frighten is a collateral purpose, but Randle said the intent to rescue another person is not an independent purpose but merely a motive. (Robertson, supra, at p. 171; Randle, supra, at p. 1005.) It is not clear how a future court should decide whether a given intent is a purpose or merely a motive.

The Robertson and Randle test presents yet another problem. In the past, we have treated the merger doctrine as a legal question with little or no factual content. Generally, we have held that an underlying felony either never or always merges (e.g., People v. Smith, supra, 35 Cal.3d at p. 805 [identifying certain underlying felonies that do not merge]), not that the question turns on the specific facts. Viewed as a legal question, the trial court properly decides whether to instruct the jury on the felony-murder rule, but if it does so instruct, it does not also instruct the jury on the merger doctrine. The Robertson and Randle test, however, turns on potentially disputed facts specific to the case. In Robertson, the defendant claimed he merely intended to frighten the victim, which caused this court to conclude the underlying felony did not merge. But the jury would not necessarily have to believe the defendant. Whether a defendant shot at someone intending to injure, or merely tried to frighten that someone, may often be a disputed factual question.

Defendant argues that the factual question whether the defendant had a collateral felonious purpose—and thus whether the felony-murder rule applies—involves an element of the crime and, accordingly, that the jury must decide that factual question. When the merger issue turns on potentially disputed factual questions, there is no obvious answer to this argument. Justice Kennard alluded to the problem in her dissent in Robertson when she observed that “the jury never decided whether he had that intent [to frighten].” (Robertson, supra, 34 Cal.4th at p. 183.) Because this factual *1200question determines whether the felony-murder rule applies under Robertson and Randle, and thus whether the prosecution would have to prove some other form of malice, it is not clear why the jury should not have to decide the factual question.

To avoid the anomaly of putting a person who merely intends to frighten the victim in a worse legal position than the person who actually intended to shoot at the victim, and the difficult question of whether and how the jury should decide questions of merger, we need to reconsider our holdings in Robertson, supra, 34 Cal.4th 156, and Randle, supra, 35 Cal.4th 987. When the underlying felony is assaultive in nature, such as a violation of section 246 or 246.3, we now conclude that the felony merges with the homicide and cannot be the basis of a felony-murder instruction. An “assaultive” felony is one that involves a threat of immediate violent injury. (See People v. Chance (2008) 44 Cal.4th 1164, 1167-1168 [81 Cal.Rptr.3d 723].) In determining whether a crime merges, the court looks to its elements and not the facts of the case. Accordingly, if the elements of the crime have an assaultive aspect, the crime merges with the underlying homicide even if the elements also include conduct that is not assaultive. For example, in People v. Smith, supra, 35 Cal.3d at page 806, the court noted that child abuse under section 273a “includes both active and passive conduct, i.e., child abuse by direct assault and child endangering by extreme neglect.” Looking to the facts before it, the court decided the offense was “of the assaultive variety,” and therefore merged. (Smith, supra, 35 Cal.3d at pp. 806-807.) It reserved the question whether the nonassaultive variety would merge. (Id. at p. 808, fn. 7.) Under the approach we now adopt, both varieties would merge. This approach both avoids the necessity of consulting facts that might be disputed and extends the protection of the merger doctrine to the potentially less culpable defendant whose conduct is not assaultive.

This conclusion is also consistent with our repeatedly stated view that the felony-murder rule should not be extended beyond its required application. (People v. Howard, supra, 34 Cal.4th at p. 1135.) We do not have to decide at this point exactly what felonies are assaultive in nature, and hence may not form the basis of a felony-murder instruction, and which are inherently collateral to the resulting homicide and do not merge. But shooting at an occupied vehicle under section 246 is assaultive in nature and hence cannot serve as the underlying felony for purposes of the felony-murder rule.7

*1201We overrule People v. Robertson, supra, 34 Cal.4th 156, and the reasoning, although not the result, of People v. Randle, supra, 35 Cal.4th 987. This conclusion means the trial court erred in this case in instructing the jury on the second degree felony-murder rule.8 We now turn to a consideration of whether this error was prejudicial.

C. Prejudice

California Constitution, article VI, section 13, prohibits a reviewing court from setting aside a judgment due to trial court error unless it finds the error prejudicial. Accordingly, we must decide whether the error in instructing on felony murder prejudiced defendant.

Instructional error regarding the elements of the offense requires reversal of the judgment unless the reviewing court concludes beyond a reasonable doubt that the error did not contribute to the verdict. (People v. Cross (2008) 45 Cal.4th 58, 69-71 [82 Cal.Rptr.3d 373, 190 P.3d 706] (conc. opn. of Baxter, J.); People v. Swain (1996) 12 Cal.4th 593, 607 [49 Cal.Rptr.2d 390, 909 P.2d 994]; People v. Calderon (2005) 129 Cal.App.4th 1301, 1306-1307 [29 Cal.Rptr.3d 277] [erroneous instruction on the second degree felony-murder rule]; see Hedgpeth v. Pulido (2008) 555 U.S._[172 L.Ed.2d 388, 129 S.Ct. 530] [reiterating that error of this nature is subject to harmless error analysis]; Neder v. United States (1999) 527 U.S. 1, 15 [144 L.Ed.2d 35, 119 S.Ct. 1827] [stating the reasonable doubt test].)

In finding prejudice, the Court of Appeal noted that the trial court “did not give CALJIC No. 8.30 on second degree express malice murder or CALJIC No. 8.31 on second degree implied malice murder.” It also stated, “While it is possible the jury selected second degree murder on another theory after finding no premeditation and deliberation, we cannot determine which theory the jury relied on, so if the second degree felony-murder instruction was legally flawed, the verdict must be reversed. (People v. Guiton (1993) 4 Cal.4th 1116, 1129 [17 Cal.Rptr.2d 365, 847 P.2d 45].)” Later, after it did find error, the court reiterated that the error was prejudicial: “Since ... the record does not show the murder conviction was based on a valid ground, we reverse the conviction for second degree murder. (People v. Guiton, supra, 4 Cal.4th 1116, 1129.)”

*1202Defendant argues that the trial court did not adequately instruct the jury on conscious-disregard-for-life malice as a theory of second degree murder, and therefore the jury could not have based its verdict on that theory. We disagree. Although the trial court did not give CALJIC Nos. 8.30 and 8.31, and hence did not instruct on implied (or express) malice murder precisely the way the authors of CALJIC intended, it did give CALJIC No. 8.11, which contains everything necessary to fully instruct the jury on this form of malice as a possible theory of second degree murder.

Specifically, the court instructed the jury that to prove murder, the prosecution had to prove an unlawful killing that “was done with malice aforethought or occurred during the commission or attempted commission of shooting at an occupied motor vehicle . . . .” (Italics added.) It also defined malice: “Malice may be either express or implied. Malice is express when there is manifested an intention unlawfully to kill a human being.

“Malice is implied when:

“1. The killing resulted from an intentional act;

“2. The natural consequences of the act are dangerous to human life; and

“3. The act was deliberately performed with knowledge of the danger to and with conscious disregard for human life.

“When it is shown that a killing resulted from the intentional doing of an act with express or implied malice, no other mental state need be shown to establish the mental state of malice aforethought.”

As the Attorney General notes, the only language from CALJIC No. 8.30 or No. 8.31 not included in CALJIC No. 8.11, which the court gave, is the last sentence of CALJIC No. 8.31: “When the killing is the direct result of such an act [an act committed with implied malice], it is not necessary to prove that the defendant intended that the act would result in the death of a human being.” But omission of this sentence, favorable to the prosecution, could neither have prejudiced defendant nor prevented the jury from finding implied malice.

Later, the court instructed the jury that a killing during the commission of shooting at an occupied motor vehicle is second degree murder “when the perpetrator had the specific intent to commit that crime.” The trial court did not reiterate at this point the conscious-disregard-for-life theory of second degree murder, but doing so was not necessary to adequately instruct the jury on that theory. The instructions permitted the jury to base a second degree *1203murder verdict on either malice or the felony-murder rule. Accordingly, the court did instruct the jury on conscious-disregard-for-life malice as a possible basis of murder.

Moreover, the prosecutor explained the applicable law to the jury. He explained that murder was an unlawful killing committed with malice or during the commission of a dangerous felony. He discussed what implied malice is and included examples. Defendant correctly notes that the prosecutor did not argue that defendant acted with implied malice. He argued for first degree, not second degree, murder. But the instructions, especially in light of the prosecutor’s explanation, permitted the jury to base a second degree murder verdict on a finding of malice separate from the felony-murder rule.

In this situation, to find the error harmless, a reviewing court must conclude, beyond a reasonable doubt, that the jury based its verdict on a legally valid theory, i.e., either express or conscious-disregard-for-life malice. Citing People v. Guiton, supra, 4 Cal.4th 1116, the Court of Appeal believed it could not do so. But Guitón does not dispose of this issue. In his concurring opinion in People v. Cross, supra, 45 Cal.4th at page 70, Justice Baxter discussed Guitón1 s significance in this context: “Although Guitón observed that reliance on other portions of the verdict is ‘[o]ne way’ of finding an instructional error harmless (Guiton, at p. 1130), we have never intimated that this was the only way to do so. Indeed, Guitón noted that we were not then presented with the situation of a jury having been instructed with a legally adequate and a legally inadequate theory and that we therefore ‘need not decide the exact standard of review’ in such circumstances— although we acknowledged that ‘[t]here may be additional ways by which a court can determine that error in [this] situation is harmless. We leave the question to future cases.’ (Id. at pp. 1130, 1131.) Because this case only now presents that issue, Guitón does not provide a dispositive answer to the question.” (See also People v. Harris (1994) 9 Cal.4th 407, 419, fn. 7 [37 Cal.Rptr.2d 200, 886 P.2d 1193].)

The Attorney General argues that the actual verdict does show that the jury did not base its murder verdict on the felony-murder rule but necessarily based it on a valid theory. He notes that the jury acquitted defendant of the separately charged underlying crime of shooting at an occupied vehicle. A jury that based a murder verdict solely on felony murder, the Attorney General argues, would not acquit a defendant of the underlying felony. Defendant counters with the argument that the verdict as a whole—finding defendant guilty of murder but not guilty of either shooting at or from a motor vehicle—is internally inconsistent. On these facts, it is hard to reconcile this verdict. If defendant did not commit this murder by firing at or from a vehicle, how did he commit it? There was no evidence the victims *1204were killed or injured by any method other than shooting from and at an occupied vehicle. The overall verdict had to have been either a compromise or an act of leniency.

Defendant recognizes that he may not argue that the murder conviction must be reversed due to this inconsistency. He may not argue that the acquittals imply that defendant could not have committed murder, and therefore the jury found he did not commit murder. Instead, courts necessarily tolerate, and give effect to all parts of, inconsistent verdicts. (See generally People v. Palmer (2001) 24 Cal.4th 856 [103 Cal.Rptr.2d 13, 15 P.3d 234].) But, defendant argues, this being the case, a reviewing court should not read more than is warranted into one part of an inconsistent verdict. Defendant posits the possibility that one or more jurors found him guilty of second degree murder on a felony-murder theory but then agreed to acquit him of the underlying felony either out of leniency or as a compromise, or perhaps simply out of confusion. In that event, defendant suggests, those jurors may simply have believed defendant was guilty of murder on the invalid felony-murder theory without ever considering a valid theory of malice.

Defendant’s argument has some force. The acquittal of the underlying felony strongly suggests the jury based its murder conviction on a valid theory of malice but, under the circumstances, we do not believe that it alone does so beyond a reasonable doubt. But for other reasons we find the error harmless. In his concurring opinion in California v. Roy (1996) 519 U.S. 2 [136 L.Ed.2d 266, 117 S.Ct. 337], Justice Scalia stated a test that fits the error of this case well. In Roy, the error was permitting a defendant to be convicted of a crime as an aider and abettor solely due to the defendant’s knowledge of the perpetrator’s intent without requiring a finding the aider and abettor shared that intent. That error is similar to the error of this case, which permitted defendant to be convicted of murder on a felony-murder theory without requiring a finding of a valid theory of malice. The high court held that the error was subject to harmless error analysis and remanded for the lower court to engage in that analysis.

California v. Roy, supra, 519 U.S. 2, involved collateral review of a state court judgment in a federal habeas corpus matter, a procedural posture in which the standard of review for prejudice is more deferential than the harmless-beyond-a-reasonable-doubt standard applicable to direct review. (Id. at pp. 4—5.) But Justice Scalia, in a concurring opinion, stated a test that is adaptable to the reasonable doubt standard of direct review: “The error in the present case can be harmless only if the jury verdict on other points effectively embraces this one or if it is impossible, upon the evidence, to have found what the verdict did find without finding this point as well.” (Id. at p. 7.) Without holding that this is the only way to find error harmless, we *1205think this test works well here, and we will use it. If other aspects of the verdict or the evidence leave no reasonable doubt that the jury made the findings necessary for conscious-disregard-for-life malice, the erroneous felony-murder instruction was harmless.

For felony murder, the court’s instructions required the jury to find that defendant had the specific intent to commit the underlying felony of shooting at an occupied vehicle. Later, it instructed that to find defendant committed that crime, the jury had to find these elements:

“l.A person discharged a firearm at an occupied motor vehicle; and

“2. The discharge of the firearm was willful and malicious.”

Thus any juror who relied on the felony-murder rule necessarily found that defendant willfully shot at an occupied vehicle. The undisputed evidence showed that the vehicle shot at was occupied by not one but three persons. The three were hit by multiple gunshots fired at close range from three different firearms. No juror could have found that defendant participated in this shooting, either as a shooter or as an aider and abettor, without also finding that defendant committed an act that is dangerous to life and did so knowing of the danger and with conscious disregard for life—which is a valid theory of malice. In other words, on this evidence, no juror could find felony murder without also finding conscious-disregard-for-life malice. The error in instmcting the jury on felony murder was, by itself, harmless beyond a reasonable doubt.

However, this instructional error is not the only error in the case. The Court of Appeal held that the jury should not have heard evidence that defendant admitted firing the gun, but said he did not point it at anyone and just wanted to scare them, and that this error was harmless “as a pure evidentiary matter.” Neither of these holdings is before us on review. The Court of Appeal also held that the error in instructing on felony murder was, by itself, prejudicial, a holding we are reversing. But the Court of Appeal never considered whether the two errors, in combination, were prejudicial. The parties have, understandably, not focused on this precise question. Under the circumstances, we think it prudent to remand the matter for the Court of Appeal to consider and decide whether the two errors, in combination, were prejudicial.

Ill, Conclusion

Although we agree with the Court of Appeal that the trial court erred in instructing the jury on second degree felony murder, we also conclude that *1206the error, alone, was harmless. Accordingly, we reverse the judgment of the Court of Appeal and remand the matter to that court for further proceedings consistent with this opinion.

George, C. J., Kennard, J., Werdegar, J., and Corrigan, J., concurred.

BAXTER, J., Concurring and Dissenting.

I concur in the majority’s decision to reaffirm the constitutional validity of the long-standing second degree felony-murder rule. (Maj. opn., ante, at pp. 1187-1188.) Ever since the Penal Code1 was enacted in 1872, and going back even before that, to California’s first penal law, the Crimes and Punishments Act of 1850 (Stats. 1850, ch. 99, p. 229), the second degree felony-murder rule has been recognized as a rule for imputing malice under the statutory definition of implied malice (§ 188)2 where the charge is second degree murder. (Maj. opn., ante, at pp. 1184-1188.) As the majority explains, “The willingness to commit a felony inherently dangerous to life is a circumstance showing an abandoned and malignant heart. The second degree felony-murder rule is based on statute and, accordingly, stands on firm constitutional ground.” (Maj. opn., ante, at pp. 1187-1188.)

Although the majority reaffirms the constitutional validity of the second degree felony-murder rule, it goes on to render the rule useless in this and future cases out of strict adherence to the so-called “merger doctrine” announced in People v. Ireland (1969) 70 Cal.2d 522 [75 Cal.Rptr. 188, 450 P.2d 580] (Ireland). Under the merger doctrine, no assaultive-type felony can be used as a basis for a second degree felony-murder conviction. The single rationale given in Ireland for the merger doctrine was that to allow assaultive-type felonies to serve as a basis for a second degree felony-murder conviction “would effectively preclude the jury from considering the issue of malice aforethought in all cases wherein homicide has been committed as a result of a felonious assault... a category which includes the great majority of all homicides. This kind of bootstrapping finds support neither in logic nor in law.” (Id. at p. 539.)

*1207In the 40 years since the Ireland court announced its sweeping “merger doctrine,” this court has struggled mightily with its fallout in an attempt to redefine the contours of the venerable second degree felony-murder rule. The history of our “ ‘muddled’ ” (maj. opn., ante, at p. 1189) case law on the subject is accurately recounted in painstaking detail in the majority opinion. (Id. at pp. 1188-1201.) Two decisions in particular are noteworthy here.

In People v. Hansen (1994) 9 Cal.4th 300 [36 Cal.Rptr.2d 609, 885 P.2d 1022] (Hansen), we concluded that maliciously and willfully shooting at an inhabited dwelling in violation of section 246, “involves a high probability that death will result and therefore is an inherently dangerous felony ... for purposes of the second degree felony-murder doctrine.” (Hansen, at p. 309.) Hansen explained that, “application of the second degree felony-murder rule to a homicide resulting from a violation of section 246 directly would serve the fundamental rationale of the felony-murder rule—the deterrence of negligent or accidental killings in the course of the commission of dangerous felonies. The tragic death of innocent and often random victims, both young and old, as the result of the discharge of firearms, has become an alarmingly common occurrence in our society—a phenomenon of enormous concern to the public. By providing notice to persons inclined to willfully discharge a firearm at an inhabited dwelling—even to those individuals who would do so merely to frighten or intimidate the occupants, or to ‘leave their calling card’—that such persons will be guilty of murder should their conduct result in the all-too-likely fatal injury of another, the felony-murder rule may serve to deter this type of reprehensible conduct, which has created a climate of fear for significant numbers of Californians even in the privacy of their own homes.” (Hansen, at pp. 310-311.)

I signed the majority opinion in Hansen, and continue to find that decision well reasoned and most directly on point in the matter now before us.3 I would follow Hansen and conclude the jury below was properly instructed on second degree felony murder based on defendant’s commission of the inherently dangerous felony of shooting at an occupied vehicle in violation of section 246 and the inference of malice that follows therefrom. The majority, *1208in contrast, rejects the analysis and holding in Hansen and expressly overrules it. (Maj. opn., ante, at pp. 1198-1199.)

In People v. Robertson (2004) 34 Cal.4th 156, 166 [17 Cal.Rptr.3d 604, 95 P.3d 872] (Robertson), we again considered whether the trial court had properly instructed the jury on second degree felony murder, this time based on the felony of discharging a firearm in a grossly negligent manner. (§ 246.3.) The defendant in Robertson claimed he fired his gun “upwards into the air” merely intending to “ ‘scare people away.’ ” (Robertson, supra, 34 Cal.4th at p. 162.) The Robertson majority rejected (although did not overrule) the rationale of Hansen, supra, 9 Cal.4th 300, and went on to resurrect and apply the so-called “collateral purpose” rule derived from two earlier decisions: People v. Mattison (1971) 4 Cal.3d 177 [93 Cal.Rptr. 185, 481 P.2d 193] (Mattison) and People v. Taylor (1970) 11 Cal.App.3d 57 [89 Cal.Rptr. 697]. Briefly, Robertson concluded that, under the collateral purpose rule, tire merger doctrine did not bar a second degree felony-murder instruction based on the violation of section 246.3. (Robertson, at p. 160.) The “collateral purpose” rule can be summarized as a test that reaches a compromise on the all-or-nothing approach taken in Ireland regarding assaultive-type felonies and their nonavailability as a basis for second degree felony-murder treatment. Under the collateral purpose rule or test, application of the second degree felony-murder rule is proper only where the underlying felony, although assaultive in nature, is nonetheless committed with a “ ‘collateral and independent felonious design.’ ” (Mattison, supra, 4 Cal.3d at p. 185; see Taylor, supra, 11 Cal.App.3d at p. 63.)

I signed the majority opinion in Robertson as well, but I have since come to appreciate that the collateral purpose rule on which it relied is unduly deferential to Ireland’s flawed merger doctrine. The majority itself points to several serious concerns raised in the wake of Robertson’s reliance on the collateral purpose rule in its effort to mitigate the harsh effects of Ireland’s all-or-nothing merger doctrine. (Maj. opn., ante, at pp. 1199-1200.) Nonetheless, it can fairly be observed that the decision in Robertson, right or wrong, did represent a compromise, for under its holding inherently dangerous felonies, though they be of the assaultive type, could still be used as a basis for second degree felony-murder rule treatment as long as a “collateral purpose” for the commission of such a felony could be demonstrated. (Robertson, supra, 34 Cal.4th at p. 160.)

The majority, in contrast, rejects the analysis and holding of Robertson and expressly overrules it along with our earlier decision in Hansen. (Maj. opn., ante, at p. 1200.) The majority, to put it bluntly, is unwilling to ameliorate the harsh effects of Ireland’s merger doctrine. The majority instead broadly holds that all felonies that are “assaultive in nature” (ibid.) henceforth may not *1209be used as a basis for a second degree felony-murder prosecution. In short, this court’s various attempts over the course of several decades to salvage the second degree felony-murder rule in the wake of Ireland’s merger doctrine, and to ameliorate the harsh effects of that all-or-nothing rule, have been wiped clean from the slate. The majority has effectively returned the law to where it stood 40 years ago, just after Ireland was decided. I cannot join in the majority’s wholesale capitulation to such a seriously flawed decision.

In the end, this case presented us with a clear opportunity to finally get this complex and difficult issue right. The majority’s recognition and unequivocal pronouncement, in part II.A. of its opinion—that the second degree felony-murder rule is simply a rule for imputing malice under section 188— furnishes the missing piece to this complex and confusing legal jigsaw puzzle. With that clear pronouncement of the second degree felony-murder rule’s true nature and function firmly in hand, I would go on to reach the following logical conclusions with regard to the long-standing tension between that rule and Ireland’s merger doctrine.

First, when a homicide has occurred during the perpetration of a felony inherently dangerous to human life, a jury’s finding that the perpetrator satisfied all the elements necessary for conviction of that offense, without legal justification or defense, is a finding that he or she acted with an “abandoned and malignant heart” (i.e., acted with malice) within the meaning of section 188. Put in terms of the modem definition of implied malice, where one commits a felony inherently dangerous to human life without legal justification or defense, then under operation of the second degree felony-murder rule, a homicide resulting therefrom is a killing “ ‘ “proximately resulting] from an act, the natural consequences of which are dangerous to life, which act was deliberately performed by a person who knows that his conduct endangers the life of another and who acts with conscious disregard for life.” ’ ” (Dellinger, supra, 49 Cal.3d at p. 1218.)

Once it is understood and accepted that the second degree felony-murder rale is simply a rule for imputing malice from the circumstances attending the commission of an inherently dangerous felony during which a homicide occurs, no grounds remain to support the sole rationale offered by the Ireland court for the merger doctrine—that use of an assaultive-type felony as the basis for a second degree felony-murder instruction “effectively precluded] the jury from considering the issue of malice aforethought in all cases wherein homicide has been committed as a result of a felonious assault.” (Ireland, supra, 70 Cal.2d at p. 539.) The majority’s holding in part II.A. of its opinion makes clear it understands and accepts that the second degree felony-murder rale is but a means by which juries impute malice under the *1210Legislature’s statutory definition of second degree implied malice murder. The majority’s holding in part II.B. of its opinion nonetheless fails to follow through and reach the logical conclusions to be drawn from the first premise, and instead simply rubberstamps the Ireland court’s misguided belief that the second degree felony-murder rule improperly removes consideration of malice from the jury’s purview.

Second, when a jury convicts of second degree murder under the second degree felony-murder rule, it has found the statutory element of malice necessary for conviction of murder. (§§ 187, 188.) Hence, there are no constitutional concerns with regard to whether the jury is finding all the elements of the charged murder, or is not finding all the “facts” that can increase punishment where the defendant is convicted of second degree murder in addition to being convicted of the underlying inherently dangerous felony. (See Apprendi v. New Jersey (2000) 530 U.S. 466 [147 L.Ed.2d 435, 120 S.Ct. 2348].)

Third, our recognition today that the second degree felony-murder rule is simply a rule under which the jury may impute malice from the defendant’s commission of inherently dangerous criminal acts, thereby undercutting the very rationale given by the Ireland court for the merger doctrine, should logically eliminate any impediment to the use of inherently dangerous felonies—such as the violation of section 246 (maliciously and willfully shooting at an occupied vehicle) at issue in this case—as the basis for an instruction on second degree felony murder.

The majority’s holding, in contrast, works just the opposite result. Prior to this court’s decision in Ireland, this court had already restricted the felonies that could support a second degree felony-murder conviction to those “inherently dangerous to human life.” (People v. Ford (1964) 60 Cal.2d 772, 795 [36 Cal.Rptr. 620, 388 P.2d 892].) The justification for the imputation of implied malice under these circumstances is that, “when society has declared certain inherently dangerous conduct to be felonious, a defendant should not be allowed to excuse himself by saying he was unaware of the danger to life.” (People v. Patterson (1989) 49 Cal.3d 615, 626 [262 Cal.Rptr. 195, 778 P.2d 549].) Hence, whatever felonies may remain available for use in connection with the second degree felony-murder rule after today’s holding will both have to qualify as inherently dangerous felonies (Ford, at p. 795), and not be “assaultive in nature” or contain any elements that have “an assaultive aspect.” (Maj. opn., ante, at p. 1200.) I fail to see how the second degree felony-murder rule, thus emasculated, will continue to serve its intended purposes of “ ‘deter[ring] felons from killing negligently or accidentally’ ” while “deter[ring] commission of the inherently dangerous felony itself.” (Maj. opn., ante, at p. 1198.)

*1211In sum, the majority has turned the second degree felony-murder rule on its head by excluding all felonies that are “assaultive in nature” (maj. opn., ante, at p. 1200), including a violation of section 246, in whatever form, from future use as a basis for second degree felony-murder treatment. In reaching its holding, the majority has rejected decades of sound felony-murder jurisprudence in deference to Ireland’s merger doctrine, a doctrine grounded on a single false premise, that use of the second degree felony-murder rule improperly insulates juries from the requirement of finding malice and thereby constitutes unfair “bootstrapping.” (Ireland, supra, 70 Cal.2d at p. 539.)

In concluding that Ireland’s merger doctrine trumps the second degree felony-murder rule in this and all future cases involving “assaultive-type” felonies (maj. opn., ante, at p. 1178), the majority professes to heed the concerns raised by some members of this court in past decisions that have addressed the tension between the second degree felony-murder rule and the merger doctrine. {Id. at pp. 1194-1195.) I do not believe those concerns justify the result reached by the majority in this case.

For example, in Robertson, supra, 34 Cal.4th 156, the issue was whether the trial court properly instructed the jury on second degree felony murder based on discharging a firearm in a grossly negligent manner. (§ 246.3.) In that case the defendant claimed he had heard a sound resembling “either a car backfire or the discharge of a firearm,” and merely “fired two warning shots” “upwards into the air” in order to “ ‘scare people away from my domain.’ ” (Robertson, at p. 162.) The physical evidence was otherwise; the defendant had fired at least three shots, two of which hit a car parked across the street “two feet above ground level.” (Ibid.) The homicide victim, found 50 yards from where the defendant was standing when he fired his weapon, died from a bullet wound to the back of his head. (Ibid.) The majority in Robertson concluded Ireland’s merger doctrine did not bar a second degree felony-murder instruction. (Robertson, at p. 160.)

As the majority observes, Justice Werdegar dissented in Robertson, arguing that the underlying felony merged with the resulting homicide. She wrote: “The anomalies created when assaultive conduct is used as the predicate for a second degree felony-murder theory [citation] are too stark and potentially too productive of injustice to be written off as ‘characteristic of the second degree felony-murder rule in general’ ([Robertson, supra, 34 Cal.4th] at p. 173). It simply cannot be the law that a defendant who shot the victim with the intent to kill or injure, but can show he or she acted in unreasonable self-defense, may be convicted of only voluntary manslaughter, whereas a defendant who shot only to scare the victim is precluded from raising that partial defense and is strictly liable as a murderer. The independent and *1212collateral purposes referred to in Mattison must be understood as limited to nonassaultive conduct. In circumstances like the present, the merger doctrine should preclude presentation of a second degree felony-murder theory to the jury.” (Robertson, supra, 34 Cal.4th at p. 185 (dis. opn. of Werdegar, J.).)

I appreciate and share the concerns voiced by Justice Werdegar in her dissent in Robertson. At the threshold, I fail to see why a bald claim by the defendant that he fired his gun “upwards into the air” intending merely to “ ‘scare people away’ ” (Robertson, supra, 34 Cal.4th at p. 162), a claim that was flatly contradicted by all the physical evidence in the case, including the dead victim who was found 50 yards away felled by a single shot to the back of his head, should be found controlling on the matter of what theory or theories of murder were rightfully available to the prosecution in trying the case. (In re Christian S. (1994) 7 Cal.4th 768, 783 [30 Cal.Rptr.2d 33, 872 P.2d 574] (Christian S.) [trial courts need only instruct on defenses supported by substantial evidence].)

The particular facts of Robertson aside, I agree with Justice Werdegar that the defendants are entitled to present all viable defenses supported by substantial evidence, like imperfect self-defense, in a second degree murder prosecution, whether it be tried on a theory of straight implied malice second degree murder or under the second degree felony-murder rule. But as we recognize today, the second degree felony-murder rule is simply a common law rule for imputing malice, a required element of murder under sections 187 and 188. Understood in that way, there is nothing in the rule, or in relevant murder statutes, to prevent a defendant from establishing that, even where the circumstances show he satisfied all the elements of an alleged inherently dangerous felony during which a homicide occurred, his actual state of mind nonetheless precludes drawing an inference of malice from those attending circumstances.

Under the modem construction of the statutory definition of implied malice (§ 188), “malice is presumed when ‘ “the killing proximately resulted from an act, the natural consequences of which are dangerous to life, which act was deliberately performed by a person who knows that his conduct endangers the life of another and who acts with conscious disregard for life.” ’ ” (Dellinger, supra, 49 Cal.3d at p. 1218, italics added; see also People v. Sedeno, supra, 10 Cal.3d at p. 719.) Notwithstanding a charge that a homicide occurred during the commission of an underlying inherently dangerous felony, a finding of second degree felony murder could still be negated by substantial evidence establishing unreasonable or imperfect self-defense, thereby reducing the murder to voluntary manslaughter (see Christian S., supra, 7 Cal.4th at p. 783), where the defendant, given his conduct and state of mind under the circumstances surrounding the crimes, is shown not to have actually harbored *1213a “ ‘conscious disregard for life.’ ” (Dellinger, at p. 1218.) Even a defendant who claims he “shot into the air” to scare away the homicide victim in an unreasonable or mistaken belief he had to do so in order to defend himself might successfully avoid an imputed inference of malice, and conviction under the second degree felony-murder rule, if substantial evidence bears out his claim and establishes he did not act with a conscious disregard for life.

One might reasonably speculate that if the Ireland court had had the benefit of our modem jurisprudence on second degree implied malice murder, including decisions like Christian S., supra, 7 Cal.4th 768, and People v. Flannel (1979) 25 Cal.3d 668 [160 Cal.Rptr. 84, 603 P.2d 1], which only firmly established the defense of unreasonable or imperfect self-defense years after Ireland was decided (see Flannel, at p. 683), the concerns that led the Ireland court to fashion its sweeping merger doctrine could have been alleviated.

In conclusion, I concur in the majority’s holding that the second degree felony-murder mle is a mle for imputing malice, and as such, withstands constitutional scrutiny. (Maj. opn., part HA., ante, at pp. 1180-1188.) I respectfully dissent from the analysis and conclusions reached by the majority with regard to Ireland’s merger doctrine. (Maj. opn., part H.B., ante, at pp. 1188-1201.) I would follow the well-reasoned decision in Hansen, supra, 9 Cal.4th 300, and conclude that the jury below was properly instructed on second degree felony murder based on defendant’s commission of the inherently dangerous felony of shooting at an occupied vehicle in violation of section 246.

MORENO, J., Concurring and Dissenting.

The second degree felony-murder mle is deeply flawed. The majority attempts once more to patch this judicially created mle and improves the state of the law considerably, but several years ago I expressed my willingness to “reassess[] the rule in an appropriate case.” (People v. Robertson (2004) 34 Cal.4th 156, 176 [17 Cal.Rptr.3d 604, 95 P.3d 872] (conc. opn. of Moreno, J.); see People v. Burroughs (1984) 35 Cal.3d 824, 829, fn. 3 [201 Cal.Rptr. 319, 678 P.2d 894] [“the time may be ripe to reconsider [the] continued vitality” of the second degree felony-murder mle].) This is that case. The time has come to abandon the second degree felony-murder mle.

“The felony-murder mle has been roundly criticized both by commentators and this court. As one commentator put it, ‘[t]he felony murder mle has an extensive history of thoughtful condemnation.’ [Citation.]” (People v. Robertson, supra, 34 Cal.4th 156, 174 (conc. opn. of Moreno, J.).) As the majority notes, “[t]he felony-murder mle makes a killing while committing certain felonies murder without the necessity of further examining the *1214defendant’s mental state.” (Maj. opn., ante, at p. 1182.) Regardless of this court’s view of the wisdom of doing so, it is within the Legislature’s prerogative to remove the necessity to prove malice when a death results from the commission of certain felonies, and the Legislature has done so by codifying the first degree felony-murder rule in Penal Code section 189. (People v. Dillon (1983) 34 Cal.3d 441, 472 [194 Cal.Rptr. 390, 668 P.2d 697].) Thus, we cannot abrogate the first degree felony-murder rule because it “is a creature of statute. . . . [T]his court does not sit as a super-legislature with the power to judicially abrogate a statute merely because it is unwise or outdated. [Citations.]” (Id. at p. 463.) We do, however, possess the authority to abrogate the second degree felony-murder doctrine because “ ‘the second degree felony-murder rule remains, as it has been since 1872, a judge-made doctrine without any express basis in the Penal Code.’ ” (People v. Robertson, supra, 34 Cal.4th at p. 174 (cone. opn. of Moreno, J.).)

My concerns about the felony-murder rule are neither new nor original. Nearly 45 years ago, this court acknowledged that “[t]he felony-murder rule has been criticized on the grounds that in almost all cases in which it is applied it is unnecessary and that it erodes the relation between criminal liability and moral culpability. [Citations.] Although it is the law in this state [citation], it should not be extended beyond any rational function that it is designed to serve.” (People v. Washington (1965) 62 Cal.2d 777, 783 [44 Cal.Rptr. 442, 402 P.2d 130], fn. omitted.) We have described the felony-murder rule as “ ‘a “highly artificial concept” ’ ” that this court long has held “in disfavor” (People v. Burroughs, supra, 35 Cal.3d 824, 829) “because it relieves the prosecution of the burden of proving one element of murder, malice aforethought” (People v. Henderson (1977) 19 Cal.3d 86, 92 [137 Cal.Rptr. 1, 560 P.2d 1180]). “The felony-murder doctrine has been censured not only because it artificially imposes malice as to one crime because of defendant’s commission of another but because it anachronistically resurrects from a bygone age a ‘barbaric’ concept that has been discarded in the place of its origin.” (People v. Phillips (1966) 64 Cal.2d 574, 583, fn. 6 [51 Cal.Rptr. 225, 414 P.2d 353], overruled on other grounds in People v. Flood (1998) 18 Cal.4th 470, 490, fn. 12 [76 Cal.Rptr.2d 180, 957 P.2d 869].)

The second degree felony-murder doctrine suffers from all the same infirmities as its first degree counterpart, and more. In People v. Satchell (1971) 6 Cal.3d 28, 33, footnote 11 [98 Cal.Rptr. 33, 489 P.2d 1361] (overruled on other grounds in People v. Flood, supra, 18 Cal.4th 470, 490, fn. 12) we observed that the second degree felony-murder rule is largely unnecessary and, in those unusual cases in which it would mandate a different result, may be unfair: “ ‘It may be that the rule is unnecessary in almost all cases in which it is applied, that is to say, that conviction in those cases can be predicated on the normal rules as to murder and as to accomplice liability. In the small residuum of cases, there may be a substantial question whether *1215the rule reaches a rational result or does not at least distract attention from more relevant criteria.’ (Fn. omitted.) [Citation.] [][] ‘If the defendant commits the felony in a highly reckless manner, he can be convicted of second degree murder independently of the shortcut of the felony-murder rule. Under California’s interpretation of the implied malice provision of the Penal Code [§ 188], proof of conduct evidencing extreme or wanton recklessness establishes the element of malice aforethought required for a second degree murder conviction. [Citation.] . . . The jury would decide whether the evidence, including the defendant’s conduct and inferences rising from it, established the requisite malice aforethought; they would not be bound by the conclusive presumption of malice which the felony murder rale compels.’ ”

The majority acknowledges the criticism heaped on the second degree felony-murder rule and describes this court’s halting and sometimes inconsistent attempts to circumscribe the scope of the rule, most notably by creating the Ireland merger doctrine (People v. Ireland 70 Cal.2d 522 [75 Cal.Rptr. 188, 450 P.2d 580]). The majority’s reformulation of the merger doctrine is an improvement, but it does not correct the basic flaw in the felony-murder rule; that it is largely unnecessary and, in those unusual instances in which it would produce a different result, may be unfair. “In most cases involving a felony-murder theory, prosecutors should have little difficulty proving second degree murder with implied malice. ‘[M]alice is implied “when the killing results from an intentional act, the natural consequences of which are dangerous to life, which act was deliberately performed by a person who knows that his conduct endangers the life of another and who acts with conscious disregard for life” [citation].’ [Citation.] Eliminating second degree felony murder from the prosecution’s arsenal would not have a detrimental effect on the prosecution’s ability to secure second degree murder convictions, but it would go a long way to restoring the proper balance between culpability and punishment.” (People v. Robertson, supra, 34 Cal.4th 156, 177 (conc. opn. of Moreno, J.).)

The lack of necessity for the second degree felony-murder rule is demonstrated by the majority’s conclusion that the error in instructing the jury on second degree felony murder in this case was harmless because no reasonable juror could have found that defendant participated in this shooting without also concluding that he harbored at least implied malice. I agree. This will be the rule, rather than the exception. In most instances, a juror who finds that the defendant killed the victim while committing a felony that is inherently dangerous to human life necessarily also will conclude that the defendant harbored either express or implied malice and thus committed second degree murder without relying upon the second degree felony-murder rule. Only in those rare cases in which it is not clear that the defendant acted in conscious disregard of fife will the second degree felony-murder rule make a difference, *1216but those are precisely the rare cases in which the rule might result in injustice. I would eliminate the second degree felony-murder rule and rely instead upon the wisdom of juries to recognize those situations in which a defendant commits second degree murder by killing the victim during the commission of a felony that is inherently dangerous to life.

Appellant’s petition for a rehearing was denied April 29, 2009.

The opinion of the court was delivered by

Larson, J.:

This is Sanexay Sophophone’s direct appeal of his felony-murder conviction for the death of his co-felon during flight from an aggravated burglary in which both men participated.

The facts are not in dispute. Sophophone and three other individuals conspired to and broke into a house in Emporia. The resident reported the break-in to the police.

Police officers responded to the call, saw four individuals leaving the back of the house, shined a light on the suspects, identified themselves as police officers, and ordered them to stop. The individuals, one being Sophophone, started to run away. One officer ran down Sophophone, hand-cuffed him, and placed him in a police car.

Other officers arrived to assist in apprehending the other individuals as they were running from the house. An officer chased one of the suspects later identified as Somphone Sysoumphone. Sysoumphone crossed railroad tracks, jumped a fence, and then stopped. The officer approached with his weapon drawn and ordered Sysoumphone to the ground and not to move. Sysoumphone was lying face down but raised up and fired at the officer, who returned fire and killed him. It is not disputed that Sysoumphone was one of the individuals observed by the officers leaving the house that had been burglarized.

Sophophone was charged with conspiracy to commit aggravated burglary, K.S.A. 21-3302; aggravated burglary, K.S.A. 21-3716; obstruction of official duty, K.S.A. 21-3808; and felony murder, K.S.A. 21-3401(b).

Sophophone moved to dismiss the felony-murder charges, contending the complaint was defective because it alleged that he and not the police officer had killed Sysoumphone and further because he was in custody and sitting in the police car when the deceased was killed and therefore not attempting to commit or even fleeing from an inherently dangerous felony. His motion to dismiss was denied by the trial court.

*705Sophophone was convicted by a jury of all counts. His motion for judgment of acquittal was denied. He was sentenced on all counts. He appeals only his conviction of felony murder.

Our jurisdiction is under K.S.A. 22-3601(b)(l) (maximum sentence of life imprisonment imposed).

Sophophone’s counsel contends (1) State v. Hoang, 243 Kan. 40, 755 P.2d 7 (1988), should be overruled insofar as it allows criminal responsibility for a co-felon’s death, (2) he cannot be convicted of felony murder of a co-felon caused by a police officer while he was in custody, and (3) there was not sufficient evidence to support his conviction.

Sophophone’s supplemental pro se brief raises contentions 1 and 2 above and further contends that the complaint/information was fatally or jurisdictionally defective.

We consider only the question of law, upon which our review is unlimited, of whether Sophophone can be convicted of felony murder for the killing of a co-felon not caused by his acts but by the lawful acts of a police officer acting in self-defense in the course and scope of his duties in apprehending the co-felon fleeing from an aggravated burglary.

The applicable provisions of K.S.A. 21-3401 read as follows:

“Murder in the first degree is the killing of a human being committed:

“(b) in the commission of, attempt to commit, or flight from an inherently dangerous felony as defined in K.S.A. 21-3436 and amendments thereto.”

Aggravated burglary is one of the inherently dangerous felonies as enumerated by K.S.A. 21-3436(10).

Sophophone does not dispute that aggravated burglary is an inherently dangerous felony which given the right circumstances would support a felony-murder charge. His principal argument centers on his being in custody at the time his co-felon was killed by the lawful act of the officer which he contends was a “break in circumstances” sufficient to insulate him from further criminal responsibility.

This “intervening cause” or “break in circumstances” argument has no merit under the facts of this case. We have held in numerous *706cases that “time, distance, and the causal relationship between the underlying felony and a killing are factors to be considered in determining whether the killing occurs in the commission of the underlying felony and the defendant is therefore subject to the felony-murder rule.” See, e.g. State v. Kaesontae, 260 Kan. 386, Syl. ¶ 1, 920 P.2d 959 (1996); State v. Rider, Edens & Lemons, 229 Kan. 394, Syl. ¶ 4, 625 P.2d 425 (1981); State v. Hearron, 228 Kan. 693, Syl. ¶ 2, 619 P.2d 1157 (1980). Based on the uncontroverted evidence in this case, the killing took place during flight from the aggravated burglary, and it is only because the act which resulted in the killing was a lawful one by a third party that a question of law exists as to whether Sophophone can be convicted of felony murder.

The history of felony murder in Kansas is well chronicled in Hoang, 243 Kan. at 43-44, and need not be repeated here. The changes that have taken place during the 13 years since Hoang was decided have mainly related to enumeration of the specific inherently dangerous felonies that are subject to felony murder (K.S.A. 21-3436) and the 1992 Legislative change in K.S.A. 21-3401 to state that murder in the first degree includes the killing of a human being committed “(b) in the commission of, attempt to commit, or flight from an inherently dangerous felony . . . .” L. 1992, ch. 298, § 3.

Prior to the enumeration of specific felonies that are statutorily deemed to be inherently dangerous, appeals often turned on the question of whether the collateral felony would support the application of the felony-murder rule. We said, however, in State v. Brantley 236 Kan. 379, 380-81, 691 P.2d 26 (1984):

“Although a literal application of K.S.A. 21-3401 would allow any felony to support a charge of felony murder so long as a causal relation exists, we have never allowed the doctrine to be applied so broadly. The purpose of the felony murder doctrine is to deter those engaged in felonies from killing negligently or accidentally, and it should not be extended beyond the rational function it was designed to serve.”

See State v. Vontress, 266 Kan. 248, 262, 970 P.2d 42 (1998); State v. Thompkins, 263 Kan. 602, 609, 952 P.2d 1332 (1998).

*707There was wording in Brantley concerning the killing being negligently or accidentally accomplished. This followed the result in State v. Branch & Bussey, 223 Kan. 381, 383-84, 573 P.2d 1041 (1978), where a killing was accidental but committed by Bussey during a robbery incidental to a drug deal, and we broadly stated:

“We conclude that any participant in a life-endangering felony is guilty of first degree murder when a life is taken in the course of committing or attempting to commit die felony, whether the deadi was intentional or accidental, or whether die participant directly caused it to occur. (See, State v. Bey, 217 Kan. 251, 535 P.2d 881; State v. Turner, [193 Kan. 189, 392 P.2d 863 (1964)]; State v. Bundy, 147 Kan. 4, 75 P.2d 236.)”

We have also stated that another purpose for the felony-murder doctrine “is to relieve the state of the burden of proving premeditation and malice when the victim’s death is caused by the killer while he is committing another felony.” State v. Clark, 204 Kan. 38, 43, 460 P.2d 586 (1969).

Our cases are legion in interpreting the felony-murder statute, but we have not previously decided a case where the killing was not by the direct acts of the felon but rather where a co-felon was killed during his flight from the scene of the felony by the lawful acts of a third party (in our case, a law enforcement officer).

A similar scenario took place in State v. Murrell, 224 Kan. 689, 585 P.2d 1017 (1978), where Murrell was charged with felony murder for the death of his co-felon who had been shot by the robbery victim who had returned gunfire from Murrell. However, Murrell was acquitted of felony murder and his appeal involved only issues relating to his other convictions. 224 Kan. at 690.

Although there were clearly different facts, we held in Hoang, 243 Kan. at 42-46 that felony murder may include the accidental death of a co-felon during the commission of arson. The decedents had conspired with Hoang to burn down a building housing a Wichita restaurant/club but died when they were trapped inside the building while starting the fire. Hoang was an active participant in the felony and present at the scene, although he remained outside the building while his three accomplices entered the building with containers of gasoline to start the fire.

*708We held, in a split decision, that the decedents were killed during the perpetration of a felony inherently dangerous to human life and there was nothing in the statute to exclude the killing of co-felons from its application. 243 Kan. at 46. It must be pointed out that the facts in Hoang involved the wrongful acts of a co-felon which were directly responsible for the deaths of his co-felons.

The dissent in Hoang noted that in previous cases the felony-murder rule had been applied only to the deaths of innocents and not to the deaths of co-felons. The result was deemed by the dissent to be contrary to legislative intent and the strict construction of criminal statutes that is required. 243 Kan. at 46-47.

With this brief background of our prior Kansas cases, we look to the prevailing views concerning the applicability of the felony-murder doctrine where the killing has been caused by the acts of a third party. The two different approaches applicable are succinctly set forth in Comment Kansas Felony Murder: Agency or Proximate Cause? 48 Kan. L. Rev. 1047, 1051-52 (2000), in the following manner:

“There are two basic approaches to application of the felony-murder doctrine: die agency and proximate cause theories. The agency approach, which is the majority view, limits application of the doctrine to those homicides committed by die felon or an agent of the felon. Under such an approach, ‘[t]he identity of the killer becomes the threshold requirement for finding liability under the felony-murder doctrine.’

“The proximate cause approach provides that liability attaches “for any death proximately resulting from the unlawful activity — even the death of a co-felon— notwithstanding the killing was by one resisting die crime.’ ” Under the proximate cause approach, felony murder may preclude consideration of the deceased’s identity, which would make a defendant liable for all deadis caused by others during the crime. Application of the proximate cause varies greatly by jurisdiction because the statutes differ substantially. The proximate cause approach becomes controversial when the homicide is committed by someone other than die felons, but only a minority of jurisdictions follow diis approach.”

In Dressier, Understanding Criminal Law, § 31.07[4] Killing by a Non-Felon, pp. 471-72 (1987), the question is posed of whether the felony-murder rule should apply when the fatal act is performed by a non-felon. Dressier states:

*709“This issue has perplexed courts. Two approaches to the question have been considered and applied by the courts.

“[b] The ‘Agency’ Approach

“The majority rule is that the felony-murder doctrine does not apply if the person who directly causes the death is a non-felon. . . .

“The reasoning of this approach stems from accomplice liability theory. Generally speaking, the acts of the primary party (the person who directly commits the offense) are imputed to an accomplice on the basis of the agency doctrine. It is as if the accomplice says to the primary party: ‘Your acts are my acts.’ It follows that [a co-felon] cannot be convicted of the homicides because the primary party was not die person widi whom she was an accomplice. It is not possible to impute die acts of the antagonistic party — [die non-felon or] the police officer — to [a co-felon] on the basis of agency.

“[c] The ‘Proximate Causation’ Approach

“An alternative dieory, followed by a few courts for awhile, holds that a felon may be held responsible under the felony-murder rule for a killing committed by a non-felon if the felon set in motion the acts which resulted in the victim’s deadi.

“Pursuant to diis rule, the issue becomes one of proximate causation: if an act by one felon is the proximate cause of the homicidal conduct by [die non-felon] or the police officer, murder liability is permitted.”

In 2 LaFave & Scott, Substantive Criminal Law, § 7.5(d), pp. 217-18 (1986), the author opines: “Although it is now generally accepted that there is no felony-murder liability when one of the felons is shot and killed by the victim, a police officer, or a bystander, it is not easy to explain why this is so.”

The author discusses forseeability and that it is not correct to say that a felon is never hable when the death is lawful because it is “justifiable” and goes on to state:

“A more plausible explanation, it is submitted, is the feeling that it is not justice (though it may be poetic justice) to hold the felon liable for murder on account of the death, which the felon did not intend, of a co-felon willingly participating in the risky venture. It is true that it is no defense to intentional homicide crimes that the victim voluntarily placed himself in danger of death at the hands of the defendant, or even that he consented to his own death: a mercy killing constitutes murder; and aiding suicide is murder unless special legislation reduces it to manslaughter. But with unintended killings it would seem proper to take the victim’s willing participation into account. . . .”

As we noted in Hoang, it is not very helpful to review case law from other states because of differences in statutory language; however, the high courts which have considered this precise question *710are divided between the agency approach and the proximate cause approach.

The leading case adopting the agency approach is Commonwealth v. Redline, 391 Pa. 486, 495, 137 A.2d 472 (1958), where the underlying principle of the agency theory is described as follows:

“In adjudging a felony-murder, it is to be remembered at all times that the thing which is imputed to a felon for a killing incidental to his felony is malice and not the act of killing. The mere coincidence of homicide and felony is not enough to satisfy the felony-murder doctrine.”

The following statement from Redline is more persuasive for Sophophone:

“In the present instance, the victim of the homicide was one of the robbers who, while resisting apprehension in his effort to escape, was shot and killed by a policeman in the performance of his duty. Thus, the homicide was justifiable and, obviously, could not be availed of, on any rational legal theory, to support a charge of murder. How can anyone, no matter how much of an outlaw he may be, have a criminal charge lodged against him for the consequences of the lawful conduct of another person? The mere question carries with it its own answer.” 391 Pa. at 509.

For a listing of states whose courts have adopted the agency theory, see Annot. Criminal Liability Where Act Of Killing Is Done By One Resisting Felony Or Other Unlawful Act Committed By Defendant, 56 A.L.jR. 3rd 239; Annot. Application of Felony-Murder Doctrine Where Person Killed was Co-Felon, 89 A.L.R. 4th 683, § 10[b], ll[b], 12[b].

The minority of the states whose courts have adopted the proximate cause theory believe their legislatures intended that any person, co-felon, or accomplice who commits an inherently dangerous felony should be held responsible for any death which is a direct and foreseeable consequence of the actions of those committing the felony. These courts apply the civil law concept of proximate cause to felony-murder situations. States which have so held are listed in Annot. 89 A.L.R. 4th 683 § 10[a], ll[a], 12[a].

It should be mentioned that some courts have been willing to impose felony-murder liability even where the shooting was by a person other than one of the felons in the so-called “shield” situ*711ations where it has been reasoned “that a felon’s act of using a victim as a shield in compelling a victim to occupy a place or position of danger constitutes a direct lethal act against the victim.” Campbell v. State, 293 Md. 438, 451 n. 3, 444 A.2d 1034 (1982).

It has previously been suggested that the wording in Kaesontae, 260 Kan. 386, Syl. ¶ 1, relating to “time, distance, and the causal relationship between the underlying felony and the killing” should be interpreted to mean we have adopted the proximate cause approach. That is not the case as this phrase is one which relates to the question of whether the killing occurs during the commission of or flight from the underlying felony. The Kansas cases which have used this wording involved situations where the court was determining if the killing was sufficiently related to the felony in terms of time, distance, and causation in order to constitute felony murder. None of those cases hold that we have adopted the proximate cause approach to the killing of a co-felon by a third party.

Nor have we adopted the proximate cause approach because of our holding and language in State v. Shaw, 260 Kan. 396, 405, 921 P.2d 779 (1990), where we held that a defendant who bound and gagged a 86-year-old robbery victim with duct tape was hable for the victim’s death when he died of a heart attack while so bound and gagged. Although we may speak of causation in such a case, our ruling in Shaw is better described by quoting syllabus ¶ 2: “The victim must be taken as the defendant finds him. Death resulting from a heart attack will support a felony-murder conviction if there is a causal connection between the heart attack and the felonious conduct of the defendant.” This is not the embracing of a proximate cause approach under the facts we face.

An additional argument has been made that when we approved the language of PIK Crim. 3d 56.02 relating to the causation required by the law for felony murder in State v. Lamae, 268 Kan. 544, 555, 998 P.2d 106 (2000), we recognized that the killing could be perpetrated by the defendant or another. (Emphasis added). The case involved the death of a participant in a metfiamphetamine fire. Our opinion did state: “It is true that there must be a direct causal connection between the commission of the felony and the homicide to invoke the felony-murder rule. See State v. Under*712wood, 228 Kan. 294, 302, 615 P.2d 153 (1980). However, the general rules of proximate cause used in civil actions do not apply.” (Emphasis added.) 268 Kan. at 555. This language, if taken in isolation, is much more favorable to Sophophone’s position. However, we believe that neither this statement nor the “or another” language in Lamae should be given undue consideration when we resolve the different question we face here.

There is language in K.S.A. 21-3205(2) that predicates criminal responsibility to an aider or abettor for “any other crime committed in pursuance of the intended crime if reasonably foreseeable by such person as a probable consequence of committing or attempting to commit the crime intended.” This wording does not assist us for the killing of the co-felon in our case where it was the lawful act by a law enforcement officer who was in no manner subject to these aider and abettor provisions.

The overriding fact which exists in our case is that neither Sophophone nor any of his accomplices “killed” anyone. The law enforcement officer acted lawfully in committing the act which resulted in the death of the co-felon. This does not fall within the language of K.S.A. 21-3205 since the officer committed no crime.

When the 1992 Kansas Legislature made extensive amendments to our criminal code, including K.S.A. 21-3401, one of the stated goals was to “conform the code to the majority of states’ laws when doing so was not clearly contrary to the Legislature’s intent.” See Comment 48 Kan. L. Rev. 1047, 1055 n. 68. But, this provides no insight to the problem we face and certainly does not require us to adopt the agency approach which the majority of the jurisdictions in the United States do follow.

Of more assistance to us is our long-time rule of statutory interpretation:

“[Cjriminal statutes must be strictly construed in favor of the accused. Any reasonable doubt about the meaning is decided in favor of anyone subjected to the criminal statute. The rule of strict construction, however, is subordinate to the rule that judicial interpretation must be reasonable and sensible to effect legislative design and intent.” State v. Vega-Fuentes, 264 Kan. 10, 14, 955 P.2d 1235 (1998).

It appears to the majority that to impute the act of killing to Sophophone when the act was the lawful and courageous one of a *713law enforcement officer acting in the line of his duties is contrary to the strict construction we are required to give criminal statutes. There is considerable doubt about the meaning of K.S.A. 21-3401(b) as applied to the facts of this case, and we believe that making one criminally responsible for the lawful acts of a law enforcement officer is not the intent of the felony-murder statute as it is currently written. Cf. State v. Murphy, 270 Kan. 804, 19 P.3d 80 (2001) (felon may not be convicted of felony murder for the killing of his co-felon caused not by his acts or actions but by the lawful acts of a victim of aggravated robbery and kidnapping acting in self-defense for the protection of his residence and the property thereof).

It does little good to suggest one construction over another would prevent the commission of dangerous felonies or that it would deter those who engage in dangerous felonies from killing purposely, negligently, or accidentally. Actually, innocent parties and victims of crimes appear to be those who are sought to be protected rather than co-felons.

We hold that under the facts of this case where the killing resulted from the lawful acts of a law enforcement officer in attempting to apprehend a co-felon, Sophophone is not criminally responsible for the resulting death of Somphone Sysoumphone, and his felony-murder conviction must be reversed.

This decision is in no manner inconsistent with our rulings in Hoang or Lamae, which are based on the direct acts of a co-felon and are simply factually different from our case. Sophophone’s request that State v. Hoang, 243 Kan. 40, 755 P.2d 7 (1988), be overruled is denied.

Reversed.

Abbott, J.,

dissenting: The issue facing the court in this case is

whether Sophophone may be legally convicted under the felony-murder statute when he did not pull the trigger and where the victim was one of the co-felons. The majority holds that Sophophone cannot be convicted of felony murder. I dissent.

An analysis of this issue must begin with an examination of the murder statute. K.S.A. 21-3401 provides:

*714“Murder in the first degree is the killing of a human being committed:

“(b) in the commission of, attempt to commit, or flight from an inherently dangerous felony as defined in K.S.A. 21-3436 and amendments thereto.”

When an issue requires statutory analysis and the statute is unambiguous, we are limited by the wording chosen by the legislature. We are not free to alter the statutory language, regardless of the result. In the present case, the felony-murder statute does not require us to adopt the “agency” theory favored by the majority. Indeed, there is nothing in the statute which establishes an agency approach. The statute does not address the issue at all. The requirements, according to the statute, are: (1) there must be a killing, and (2) the killing must be committed in the commission, attempt to commit, or flight from an inherently dangerous felony. The statute simply does not contain the limitations discussed by the majority. There is nothing in K.S.A. 21-3401 which requires us to adopt the agency approach or that requires Sophophone to be the shooter in this case. The facts in this case, in my opinion, satisfy all of the requirements set forth in K.S.A. 21-3401(b).

Moreover, there are sound reasons to adopt the proximate cause approach described in the majority opinion. In State v. Hoang, 243 Kan. 40, 755 P.2d 7 (1988), this court took such an approach, although never referring to it by name. In Hoang, Chief Justice McFarland, writing for the court, discussed at length the requirements of the felony-murder rule in Kansas and stated:

“In felony-murder cases, the elements of malice, deliberation, and premeditation which are required for murder in the first degree are deemed to be supplied by felonious conduct alone if a homicide results. To support a conviction for felony murder, all that is required is to prove that a felony was being committed, which felony was inherently dangerous to human life, and that the homicide which followed was a direct result of the commission of that felony. In a felony-murder case, evidence of who the triggerman is is irrelevant and all participants are principals. [Citations omitted.]

“The puipose of the felony-murder doctrine is to deter all those engaged in felonies from killing negligently or accidentally. . . .

“It is argued in the case before us that felony murder applies only to the deaths of Innocents’ rather than co-felons. There is nothing in our statute on which to base such a distinction. . . .

*715“Dung and Thuong, the decedents herein, were human beings who were killed in the perpetration of a felony. . . . Defendant was an active participant in die felony and present on the scene during all pertinent times. There is nothing in the statute excluding the killing of the co-felons herein from, its application. For this court to exclude the co-felons would constitute judicial amendment of a statute on philosophic rather than legal grounds. This would be highly improper. The legislature has defined felony murder. If this definition is to be amended to exclude the killing of co-felons therefrom under circumstances such as are before us, it is up to the legislature to make such an amendment.” 243 Kan. at 41-42, 45-46. (Emphasis added.)

It is not this court’s prerogative to decide what is and is not felony murder. That is a legislative function.

Hoang was decided in 1988, almost 13 years ago. The legislature has not amended or altered K.S.A. 21-3401 since that time to prevent the felony-murder statute from being used when the victim is a co-felon or where the defendant’s actions are not the immediate cause of the death. The majority states that the decision in this case is not inconsistent with the ruling in Hoang. I disagree. The language in Hoang warns of the dangers of judicial reconstruction and statutory revisionism; however, the majority has taken that approach regardless. Although the facts in Hoang are not identical to the facts in this case, the differences are inconsequential. In my opinion, Hoang is still good law and provides ample justification to apply the felony-murder rule to Sophophone.

The majority in this case points out that the majority of states have adopted the agency approach when faced with the death of a co-felon. They acknowledge, however, that because statutes vary significantly from state to state, reference to a “majority” rule and a “minority” rule is meaningless. Indeed, an in-depth analysis of the current case law in this area leads me to the following conclusions: (1) While a majority of states would agree with the majority opinion in this case, the margin is slim; (2) many of the states that have adopted the so-called “agency” approach have done so because the statutory language in their state requires them to do so; and (3) several of the states that have adopted the “proximate cause” approach have done so because their statutes are silent on the issue, like Kansas.

*716The following courts have used a proximate cause approach instead of following the agency theory adopted by the majority in this case. Several of the following cases also involve factual situations where the co-felon was killed by a police officer, as is the situation in the present case. See State v. Lopez, 173 Ariz. 552, 556, 845 P.2d 478 (Ct. App. 1993) (affirming felony-murder conviction where police officer shot co-felon while defendant was already under arrest by using the proximate cause approach as set forth by Arizona statute); State v. Wright, 379 So.2d 96, 96-97 (Fla. 1980) (holding that there was nothing in the Florida felony-murder statute which limited application to “innocent persons killed” by the defendant); People v. Dekens, 182 Ill. 2d 247, 252, 695 N.E.2d 474 (1998) (Illinois follows the proximate cause theory of felony murder); Jenkins v. State, 726 N.E.2d 268, 269-70 (Ind. 2000) (holding that Indiana felony-murder statute does not require the victim to be “innocent” and that defendant could be convicted of felony murder where robbery victim shot and killed co-felon); Palmer v. State, 704 N.E.2d 124, 125-26 (Ind. 1999) (affirming felony-murder conviction where co-felon was shot by police officer); State v. Baker, 607 S.W.2d 153, 156-57 (Mo. 1980) (affirming conviction where death was a proximate result of the acts of the defendant and his accomplices); State v. Blackmon, 587 S.W.2d 292, 293 (Mo. Ct. App. 1979) (affirming felony-murder conviction where victim was shot by an off-duty police officer); State v. Burton, 130 N.J. Super. 174, 178-79, 325 A.2d 856 (1974) (affirming felony-murder conviction where police killed co-felon during commission of a robbery and where statutory language indicated a preference for the proximate cause approach); In re Leon, 122 R.I. 548, 554-55, 410 A.2d 121 (1980) (taking a proximate cause approach and holding that defendant could be convicted of felony murder where the victim was a co-felon); State v. Oimen, 184 Wis. 2d 423, 435, 516 N.W.2d 399 (1994) (adopting proximate cause approach where co-felon was killed by burglary victim).

Some courts have been forced to take an agency approach because of the statutory language contained within their felony-murder statutes. See Weick v. State, 420 A.2d 159, 161-63 (Del. 1980) (reversing second-degree murder conviction where victim killed *717co-felon because statute requires that “he, with criminal negligence, causes the death of another person”); State v. Jones, 859 P.2d 514, 515 (Okla. Crim. 1993) (taking an agency approach because statute provides that the person committing the felony must “take the life of a human being”); State v. Hansen, 734 P.2d 421, 427 (Utah 1986) (holding that state law precluded second-degree felony-murder conviction where co-felon is killed because language in statute requires the death to be “other than a party” to the crime).

As noted in Hoang, references to cases from other jurisdictions, regardless of the “majority” or “minority” rule, is unnecessary because the statutory language, if unambiguous, should control the outcome. 243 Kan. at 44-46.

In my opinion, our statute is unambiguous and simply does not require the defendant to be the direct cause of the victim’s death, nor does it limit application of the felony-murder rule to the death of “innocents.”

In People v. Lowery, 178 Ill.2d 462, 687 N.E.2d 973 (1997), the Illinois Supreme Court discussed the public policy reasons justifying application of a proximate cause approach, stating:

“It is equally consistent with reason and sound public policy to hold that when a felon’s attempt to commit a forcible felony sets in motion a chain of events which were or should have been within his contemplation when the motion was initiated, he should be held responsible for any death which by direct and almost inevitable sequence results from the initial criminal act. Thus, there is no reason why the principle underlying the doctrine of proximate cause should not apply to criminal cases. Moreover, we believe that the intent behind the felony-murder doctrine would be thwarted if we did not hold felons responsible for the foreseeable consequences of their actions. [Citations omitted.]” 178 Ill. 2d at 467.

In Sheckles v. State, 684 N.E.2d 201 (Ind. Ct. App. 1997), the Indiana Court of Appeals opined:

“[A] person who commits or attempts to commit one of the offenses designated in the felony-murder statute is criminally responsible for a homicide which results from the act of one who was not a participant in the original criminal activity. Where the accused reasonably should have . . . foreseen that the commission of or attempt to commit the contemplated felony would likely create a situation which would expose another to the danger of death at the hands of a nonparticipant in the felony, and where death in fact occurs as was foreseeable, the creation *718of such a dangerous situation is an intermediary, secondary, or medium in effecting or bringing about the death of the victim.” 684 N.E.2d at 205.

Likewise, the Supreme Court of New Jersey discussed the historical justification for application of the proximate cause rule in felony-murder cases in State v. Martin, 119 N.J. 2, 573 A.2d 1359 (1990), stating:

“More recently, felony murder has been viewed not as a crime of transferred intent, but as one of absolute or strict liability. Whether the offense is viewed as a crime of transferred intent or as one of absolute liability, the continuing justification for the felony-murder rule is that in some circumstances one who commits a felony should be liable for a resulting, albeit unintended, death. Conversely, other deaths are so remotely related to the underlying felony that the actor should not be held culpable for them. Our task is to ascertain the circumstances in which the Legislature has decided that one who commits a felony should also be culpable for a resulting death.

“The historical justification for the rule is that it serves as a general deterrent against the commission of violent crimes. [Citation omitted.] The rationale is that if potential felons realize that they will be culpable as murderers for a death that occurs during the commission of a felony, they will be less likely to commit the felony. From this perspective, the imposition of strict liability without regard to the intent to kill serves to deter the commission of serious crimes.” 119 N.J. at 20.

Here, Sophophone set in motion acts which would have resulted in the death or serious injury of a law enforcement officer had it not been for the highly alert law enforcement officer. This set of events could have very easily resulted in the death of a law enforcement officer, and in my opinion this is exactly the type of case the legislature had in mind when it adopted the felony-murder rule.

The majority has opened a Pandora’s box and left the law grossly unsettled. It does not take much imagination to see a number of situations where a death is going to result from an inherently dangerous felony and the majority’s opinion is going to prevent the accused from being charged with felony murder.

If there is to be a change in the law, it should be by the legislature and not by this court adopting a statutory scheme set forth by the legislatures of other states. I would continue to follow the proximate cause theory of liability for felony murder which holds that criminal liability attaches for any death proximately resulting from *719the unlawful activity notwithstanding the fact that the killing was by one resisting the crime.

I would affirm the conviction based upon the statutory language found in K.S.A. 21-3401, the decision in Hoang, and the cases cited from other jurisdictions.

McFarland, C.J., and Davis, J., join in the foregoing dissenting opinion.