Opinion
ARABIAN, J.Pertinent provisions of the Penal Code and the Welfare and Institutions Code, combined with prior decisions of this court, seem to provide that a person under the age of 18 who commits first degree murder and is tried as an adult may be committed to the California Youth Authority (CYA), while the same person who attempts but fails to commit the same *63crime is not eligible for CYA, but must instead be sentenced to prison. The Court of Appeal in People v. Ladanio (1989) 211 Cal.App.3d 1114, 1121 [260 Cal.Rptr. 12], while finding the result “unquestionably anomalous,” felt compelled to hold precisely that. Here we face that question and the propriety of that result.
In addition, Penal Code section 12022.5 provides for a sentence enhancement when a person uses a firearm in the commission or attempted commission of a felony. In 1976, a four-to-three majority of this court, interpreting a substantially similar but earlier version of that section, held that even if there are multiple counts involving multiple victims of violent crime, the enhancement may be imposed only once “if all the charged offenses are incident to one objective and effectively comprise an indivisible transaction . . . .” (In re Culbreth (1976) 17 Cal.3d 330, 333 [130 Cal.Rptr. 719, 551 P.2d 23] (Culbreth).) This holding has been heavily criticized in the ensuing years, and Courts of Appeal have urged us to reconsider it. We do so here.
By way of preview, we hold that persons convicted of either attempted premeditated murder or first degree murder, committed when they are under the age of 18 years, are eligible for CYA commitment. We also overrule Culbreth, supra, 17 Cal.3d 330, but conclude that our decision cannot constitutionally be applied retroactively to this case.
I. Facts
Testimony at the preliminary hearing indicated that on the morning of October 8,1989, victims Nina Manlove and Marc Anthony Enciso arrived at work at the Thrifty Drug Store in Redlands. While they were preparing to open the store for business, defendant, 16 years old at the time, entered with a handgun. Defendant forced Enciso and Manlove into a back room at gunpoint; there he ordered them to get down on their knees. He demanded that Manlove give him her keys, some of which were for locked cases which held electronics merchandise. She complied. Defendant then forced the two to lie down on the floor.
Defendant “cocked the gun” and shot the prone Enciso in the head. He then pointed the gun at Manlove. She begged him not to shoot her. Nevertheless, he shot her in the arm. As he bent down and picked up the expended bullet casings, Manlove opened one eye to look at him. He then shot her again, this time in the face. Defendant walked out of the back room, leaving Manlove for dead. She was able, however, to call the police, and survived to testify against her assailant. Enciso died.
Defendant later told the police that he went to the store to commit a robbery. He shot the two ‘to hide his identity.” Afterwards, he stole a few items from the store and fled.
*64Defendant was ordered tried as an adult. (See Welf. & Inst. Code, § 707.) He pleaded guilty to one count of first degree murder, one count of attempted premeditated murder, and two counts of second degree robbery. He admitted that he personally used a firearm in the commission of the murder and attempted murder. Defendant asked to be committed to CYA. The court sent him there for an amenability study. The resulting report stated that defendant was amenable to CYA treatment. At sentencing, the court expressed doubts that defendant should be committed to CYA, but made no specific ruling on the question. Instead, it found that he was statutorily ineligible under Welfare and Institutions Code section 1731.5, and therefore stated that it need not decide whether it would commit him to CYA if he had been eligible.
The court sentenced defendant to state prison for 25 years to life for the murder plus 2 years for the firearm use enhancement. For the attempted murder, the court imposed a consecutive sentence of life in prison with possibility of parole plus two years for the firearm-use enhancement. Sentencing on the robbery counts was made concurrent. Due to defendant’s age, the court ordered that he be housed at CYA until that institution determined that he should be transferred to state prison to serve the balance of his term. (See Welf. & Inst. Code, § 1731.5, subd. (c).)
The Court of Appeal, relying on People v. Ladanio, supra, 211 Cal.App.3d 1114, affirmed the trial court’s finding that defendant was ineligible for a CYA commitment. However, it “reluctantly” struck the consecutive firearm-use enhancement for the attempted murder under compulsion of Culbreth, supra, 17 Cal.3d 330.
Defendant petitioned for review on the question of his eligibility for CYA. The Attorney General petitioned for review on the Culbreth question. We granted both petitions.
II. Discussion
A. Defendant’s Petition
1. The Problem
A person who was between the age of 16 and 18 years at the time of the crime may be tried as an adult if found unfit for juvenile court treatment. (Welf. & Inst. Code, § 707.) A person who was tried as an adult but was under the age of 21 at the time of apprehension may, with some exceptions, be committed to CYA rather than sentenced to state prison. (Welf. & Inst. *65Code, § 1731.5, subd. (a).) The issue here is whether defendant, who was 16 at the time of the crime and was tried as an adult, comes within one of the exceptions that makes him ineligible for CYA commitment.1
Among those not eligible for CYA commitment is a person who is “sentenced to death, [or] imprisonment for life . . . .” (Welf. & Inst. Code, § 1731.5, subd. (a).) The sentence for first degree murder is “confinement in tiie state prison for a term of 25 years to life.” (Pen. Code, § 190, subd. (a).) The sentence for attempted “willful, deliberate, and premeditated” murder is “imprisonment in the state prison for life with the possibility of parole.” (Pen. Code, § 664, subd. 1.) Defendant committed both of these crimes, and received both of these sentences. The precise question is therefore whether either or both of these sentences renders defendant ineligible for CYA, i.e., whether either of these is a sentence of “imprisonment for life.”
We have held that the punishment for first degree murder of “25 years to life” is not a life sentence under Welfare and Institutions Code section 1731.5. (In re Jeanice D. (1980) 28 Cal.3d 210 [168 Cal.Rptr. 455, 617 P.2d 1087] (Jeanice D.).) Under that decision, a juvenile convicted of first degree murder is eligible for CYA. If, conversely, we held that a sentence of imprisonment for “life with the possibility of parole” for attempted premeditated murder (Pen. Code, § 664, subd. 1) was a sentence to “imprisonment for life” under section 1731.5, that would render an attempted murderer ineligible for CYA. The anomaly would be obvious.
A review of the history of the relevant statutes helps explain how this problem arose. When Welfare and Institutions Code section 1731.5 was first enacted, California operated under the Indeterminate Sentence Law. Most sentences were indeterminate, and the California Adult Authority determined when the inmate would be released. Typical was the sentence for robbery, which was “not less then five years.” (Pen. Code, former § 213; People v. McNabb (1935) 3 Cal.2d 441, 444 [45 P.2d 334].) The question quickly arose whether such an indeterminate sentence was a life sentence for purposes of section 1731.5. We held that it was not. (People v. Ralph (1944) 24 Cal.2d 575 [150 P.2d 401].) However, a sentence for first degree murder, which used to be confinement “in state prison for life” (Pen. Code, former § 190; see Jeanice D., supra, 28 Cal.3d at p. 218 & fn. 6), did render the person ineligible under section 1731.5. (People v. Chi Ko Wong (1976) 18 Cal.3d 698, 724-725 [135 Cal.Rptr. 392, 557 P.2d 976].) As phrased in *66Jeanice D., supra, 28 Cal.3d at p. 218, such a sentence was a “ ‘straight’ life imprisonment sentence.”
An initiative measure passed on November 7, 1978, amended Penal Code section 190 to change the punishment for first degree murder from “state prison for life” to its current “25 years to life.” Although, in effect, this substantially increased the punishment for the crime, we held in Jeanice D., supra, 28 Cal.3d 210, that 25 years to life is an indeterminate term under People v. Ralph, supra, 24 Cal.2d 575. Therefore, the initiative made convicted first degree murderers eligible for CYA. When Jeanice D. was decided, the punishment for attempted murder was for a specified period of years, so the current question did not arise. (See People v. Ladanio, supra, 211 Cal.App.3d at p. 1118.)
The Legislature promptly responded to Jeanice D. Within a year, it passed emergency legislation making ineligible for CYA a person who is “convicted of first degree murder, committed when that person was 18 years of age or older . . . .” (Welf. & Inst. Code, § 1731.5, subd. (a)(2).) Also pertinent is Welfare and Institutions Code section 1732.5, adopted by initiative in 1982, which makes ineligible for CYA any person convicted of any serious felony, as defined in Penal Code section 1192.7, “committed when he or she was 18 years of age or older . . . .” The list of serious felonies is broad, and includes murder and attempted murder. Defendant was 16 at the time of these crimes, so neither of these pieces of legislation applies to him.
At this point, the current anomaly did not exist. In 1986, however, Penal Code section 664 was amended to provide, as it now does, that attempted premeditated murder is punishable by “life with the possiblity of parole.” This gave rise to the instant issue.
In People v. Ladanio, supra, 211 Cal.App.3d 1114, the court found that the sentence for attempted premeditated murder was “imprisonment for life” under Welfare and Institutions Code section 1731.5, thus making the defendant ineligible for CYA. It distinguished Jeanice D., finding that the sentence for attempted premeditated murder was like the earlier “ ‘ “straight” ’ life sentence” that had previously rendered juvenile murderers ineligible for CYA commitment, and not like the “25 years to life” term at issue in Jeanice D. (People v. Ladanio, supra, 211 Cal.App.3d at p. 1118.) Although it agreed that “This peculiar result may be, as defendant suggests, a legislative oversight,” the court found that “it is not the function of this court to define crimes and prescribe punishments.” (Id. at p. 1119.) “In the absence of statutory ambiguity or other constitutional infirmity, we cannot disregard the plain language of these statutes.” (Ibid.)
*67The Ladanio court went on to hold the statutes, as construed, constitutional, but concluded, “The disparity in sentencing treatment between juvenile defendants convicted of attempted murder and those convicted of the completed crime is unquestionably anomalous. Accordingly, we urge the Legislature to address this issue. Since neither section 1731.5 nor Penal Code section 664 is clearly unconstitutional, we must leave this task to the Legislature or the voters of this state.” (People v. Ladanio, supra, 211 Cal.App.3d at p. 1121.) The Legislature has not heeded the call.
2. Resolution
This issue is moot as to those 18 years of age or older. All such persons who commit any serious felony are ineligible for CYA commitment under Welfare and Institutions Code section 1732.5. As to 16- and 17-year-olds, we have 3 options: (1) find both successful and failed first degree murderers ineligible for CYA commitment; (2) find successful murderers eligible but failed murderers ineligible (the holding of People v. Ladanio, supra, 211 Cal.App.3d 1114); or (3) find both eligible.
The Attorney General urges us to adopt the first of these options, which would require us to overrule Jeanice D., supra, 28 Cal.3d 210, or, if not that, to adopt the second option. Defendant urges us to adopt the third option, which would require us to overrule People v. Ladanio, supra, 211 Cal.App.3d 1114. We find that the Legislature intended that both successful and intended first degree murderers under the age of 18 should be eligible for CYA, and thus agree with defendant.
Within a year of Jeanice D., supra, 28 Cal.3d 210, the Legislature overruled it, but only in part. (Stats. 1981, ch. 476, § 1, p. 1816.) As Welfare and Institutions Code section 1731.5 is now written, a person who is convicted of first degree murder is not eligible for CYA, but only if the crime was “committed when that person was 18 years of age or older.”2 This effectively overrules Jeanice D. as applied to persons between the ages of 18 and 21 years, but it also impliedly reaffirms the Jeanice D. rule as applied to 16- and 17-year-olds. The Legislature apparently intended to make adult, but not juvenile, murderers ineligible for CYA.
The history of the bill that was eventually enacted supports this view of the legislative intent. As originally introduced on December 2, 1980 (less *68than a month and a half after Jeanice D. was decided), the bill would have added Welfare and Institutions Code section 1732.1, providing that “No person convicted of first degree murder shall be committed to the Youth Authority.” (Assem. Bill No. 66 (1981-1982 Reg. Sess.) § 1.) The first amendment to the bill deleted the new proposed section, and instead amended Welfare and Institutions Code section 1731.5. It still would have made anyone “convicted of first degree murder” ineligible for CYA. (Assem. Amend, to Assem. Bill No. 66 (1981-1982 Reg. Sess.) Jan. 19, 1981.) A statement of legislative intent was added: “It is the intent of the Legislature in the enactment of this bill to restore the law concerning the sentencing of persons convicted of first degree murder to its status prior to the approval of an initiative measure on November 7, 1978, repealing and adding Section 190 of the Penal Code.” (Ibid.) Although the bill did not specifically mention Jeanice D., supra, 28 Cal.3d 210, the intent was clearly to overrule that decision.
The next amendment, the last one relevant to this issue (later amendments added the provision allowing a person under the age of 21 who is sentenced to state prison to be temporarily housed at CYA), is revealing. It added the qualifying language, “committed when such person was 18 years of age or older.” (Assem. Amend, to Assem. Bill No. 66 (1981-1982 Reg. Sess.) Apr. 20, 1981.) The statement of legislative intent added in the January 19, 1981, amendment was deleted. Instead, a statement of urgency requiring the act to go into immediate effect was added: “It is necessary in order to prevent adults convicted of first degree murder from being prematurely released.” (Ibid., italics added.) This statement was eventually enacted.
The report of the Assembly Committee on Criminal Justice on the bill as amended on January 19, 1981 (before the 18-year age qualification was added) states that the purpose of the bill is to “reverse” the Jeanice D. decision. It also states that the legislation “would apply to 16 year olds as well as 20 year olds.” After the April 20, 1981, amendment (adding the 18-year age qualification), the report of the Senate Committee on Judiciary explains that under Jeanice D., supra, 28 Cal.3d 210, “a person between the ages of 16 and 21 who has been convicted in adult court for first degree murder is eligible for a [CYA] commitment.” The report states that “The purpose of this bill is to overturn the Jeanice D. decision for those individuals who were 18 years or older when they committed the murder.” (Italics added.)
From this legislative history, the conclusion is inescapable that the Legislature intended to overrule Jeanice D. as applied to adults, i.e., those 18 years of age and older, but not as applied to juveniles, i.e., those 16 or 17 years old at the time of the crime. It did not intend to entirely restore the law *69to its pre-1978 status, but only as to adults. The Legislature effectively ratified Jeanice D. as applied to 16- and 17-year-olds. We may not now overturn that legislative action. A person who committed first degree murder while 16 or 17 years old is eligible for CYA.
Given this conclusion, are persons that age who merely attempt the crime similarly eligible? That is the only rational interpretation of the legislative intent. When the Legislature amended Penal Code section 664 five years after making first degree murderers under the age of eighteen eligible for a CYA commitment, surely it did not intend to make attempted premeditated murderers that age ineligible for the same commitment. It did not intend a lesser included offense to have potentially harsher penal consequences than the greater offense. Defendant should not be penalized because one of his victims survived; he should not be made to regret not applying the coup de gráce to that victim.
Settled rules of statutory construction support this interpretation. “The fundamental purpose of statutory construction is to ascertain the intent of the lawmakers so as to effectuate the purpose of the law. [Citations.] In order to determine this intent, we begin by examining the language of the statute. [Citations.] But ‘[i]t is a settled principle of statutory interpretation that language of a statute should not be given a literal meaning if doing so would result in absurd consequences which the Legislature did not intend.’ ” (People v. Pieters (1991) 52 Cal.3d 894, 898-899 [276 Cal.Rptr. 918, 802 P.2d 420].)
We are here confronted with the interplay of different statutory provisions. The clear legislative intent to make first degree murderers under the age of 18—and by extension those who attempt but fail to commit the crime—eligible for CYA should prevail over any irrational result caused by the amendment of different statutes in separate codes at different times for unrelated purposes. “[T]he ‘plain meaning’ rule does not prohibit a court from determining whether the literal meaning of a statute comports with its purpose or whether such a construction of one provision is consistent with other provisions of the statute. The meaning of a statute may not be determined from a single word or sentence; the words must be construed in context, and provisions relating to the same subject matter must be harmonized to the extent possible. [Citation.] Literal construction should not prevail if it is contrary to the legislative intent apparent in the statute. The intent prevails over the letter, and the letter will, if possible, be so read as to conform to the spirit of the act.” (Lungren v. Deukmejian (1988) 45 Cal.3d 727, 735 [248 Cal.Rptr. 115, 755 P.2d 299].)
Here, the specific reference to first degree murder in Welfare and Institutions Code section 1731.5, by reasonable implication, includes an *70attempt to commit that crime. The general reference to a life sentence in the same section must yield to the specific to avoid an absurd result. In such a case, the general provision is controlled by the specific, the latter being treated as an exception to the former. (San Francisco Taxpayers Assn. v. Board of Supervisors (1992) 2 Cal.4th 571, 577 [7 Cal.Rptr.2d 245, 828 P.2d 147].)
We thus hold that a person convicted of attempted premeditated murder, like the person convicted of the successful crime, is eligible for CYA commitment. We disapprove the contrary holding of People v. Ladanio, supra, 211 Cal.App.3d 1114.3 Since the trial court did not decide whether defendant should be committed to CYA if eligible, the matter must be remanded for that determination.
We stress, however, that simply because defendant is eligible for a CYA commitment does not mean he must actually receive it, but only that the trial court has that option. (See Jeanice D., supra, 28 Cal.3d at p. 221.) The trial court expressed doubts that defendant should be committed to CYA, which is understandable given the extraordinarily callous and serious nature of his crimes. Although the CYA report found defendant amenable to treatment, that is only part of what the trial court must consider. “The need to protect society, the nature and seriousness of the offense, the interests of justice, the suitability of the minor to the training and treatment offered by the Youth Authority, and the needs of the minor shall be the primary considerations in the court’s determination of the appropriate disposition for the minor.” (Welf. & Inst. Code, § 707.2; see People v. Jones (1988) 46 Cal.3d 585, 601 [250 Cal.Rptr. 635, 758 P.2d 1165].)
As noted in People v. Jones, supra, 46 Cal.3d at page 601, the CYA report “address[es] only the minor’s amenability to [CYA] training and treatment,” which is but one of the five statutory “ ‘primary considerations.’ ” On remand, the trial court will have to consider all of the statutory factors in deciding whether to actually commit defendant to CYA. (Id. at pp. 601-603.)
B. The Attorney General’s Petition
1. Culbreth Revisited
Penal Code section 12022.5, subdivision (a), provides in relevant part that “any person who personally uses a firearm in the commission or attempted *71commission of a felony shall, upon conviction of such felony or attempted felony, in addition and consecutive to the punishment prescribed for the felony or attempted felony of which he or she has been convicted, be punished” by a specified sentence enhancement. Culbreth, supra, 17 Cal.3d 330, construed a substantially similar predecessor statute.4
In Culbreth, supra, 17 Cal.3d 330, the defendant shot and killed his common law wife, his mother-in-law and his brother-in-law with a .30-.30 rifle. He was convicted of two counts of second degree murder and one of voluntary manslaughter. The question was whether the sentence could be enhanced for both of the murders (voluntary manslaughter was not one of the specified felonies at the time), or whether only one enhancement was allowed for the entire episode. We held the latter. “The legislative purpose of section 12022.5 has been described as deterrence, i.e., to deter the use of firearms on subsequent occasions. Thus it has been held that where there are consecutive robberies in several communities over a period of several hours, a defendant may not bootstrap himself into avoidance of additional penalties by claiming that the series of divisible acts, each of which had been committed with a separate identifiable intent and objective, composed an indivisible transaction. [Citations.] But if all the charged offenses are incident to one objective and effectively comprise an indivisible transaction, then section 12022.5 may be invoked only once and not in accordance with the number of victims. [Citation.]” (Id. at pp. 333-334.)
After discussing three Court of Appeal decisions, we continued our analysis. “It is clear that the term ‘uses’ was deliberately employed by the Legislature when it adopted section 12022.5. To ‘use’ means, among other things, ‘ “to carry out a purpose or action by means of,” to “make instrumental to an end or process,” and to “apply to advantage.” ’ [Citation.] The ‘end or process’ here was the commission of a single frenetic act of violence which, unfortunately, resulted in multiple victims.
“Under the People’s theory, no consideration of the facts is necessary; a body count of victims is sufficient to establish the number of separate transactions. This simplistic formula is untenable; an analysis of the events is essential to ascertain the apparent intent and objective of the defendant. Such an analysis here clearly indicates the homicides—the two second degree murders and the manslaughter—occurred in a matter of seconds, all *72part of a single melee. There was but one occasion, one intent, one objective, one indivisible transaction. Therefore section 12022.5 may be applied only once.” (Culbreth, supra, 17 Cal.3d at pp. 334-335.)
Justice Clark, joined by Justices McComb and Richardson, dissented. He argued that Penal Code section 654, which generally “prohibits multiple punishment when a single act or transaction is charged as multiple offenses,” does not apply to “crimes of violence against multiple victims.” (Culbreth, supra, 17 Cal.3d at p. 335 (dis. opn. of Clark, J.).) Thus, that section does not apply to the underlying murders, nor to Penal Code section 12022.5, which does not apply to the offense but “merely provides additional punishment for certain offenses in which a firearm is used.” (17 Cal.3d at p. 335.)
Justice Clark argued that the language of Penal Code section 12022.5 does not support the majority’s analysis. “Section 12022.5 speaks of ‘use’ of a firearm, not ‘occasions’ of firearm use. Neither ‘occasion’ nor any synonym associated with transactional analysis appears in the statute. [The dissent here quotes the same definition of “use” quoted by the majority.] Under section 12022.5, the ‘purpose or action,’ ‘end or process’ for which a firearm is used is the commission of one or more of the crimes specified therein. Accordingly, for the purposes of the section, petitioner used his firearm twice, committing two murders. By its terms the statute applies to ‘[a]ny person who uses a firearm in the commission of. . .murder. . . .’Nothing in the statute indicates that it is applicable to one of petitioner’s murders but not the other one.
“Nor is the . . . analysis supported by the fact that section 12022.5 is intended to deter use of a firearm in the commission of the offenses specified therein. The statutes directly sanctioning those crimes are also intended to have a deterrent effect. Nevertheless, a defendant committing two murders, e.g., on a ‘single occasion’ may be convicted and punished for both of them. [Citation.] So also may the punishment for each of the murders be enhanced under section 12022.5.” (Culbreth, supra, 17 Cal.3d at p. 336 (dis. opn. of Clark, J.), italics in original.)
Subsequent Court of Appeal opinions have not been kind to the Culbreth rule. In People v. Raby (1986) 179 Cal.App.3d 577 [224 Cal.Rptr. 576], the court, while yielding to the mandate of Culbreth, criticized it in a comprehensive analysis. There, the defendant was convicted of nine robberies against different victims committed in two stores on two dates. The court held that each series of robberies on different dates was a separate occasion permitting separate use enhancements, but that there could be only one enhancement per date despite the multiple victims. While so ruling, however, it “admit[ted] considerable unease with the current state of the law. *73After intensive study, we are not sure a coherent rule on the subject can be constructed on the foundation of Culbreth and its progeny. A test based in part on intent and objective is seriously flawed for several reasons: it favors those who harbor the graver criminal intent over those whose crimes are, in part, largely reactions to circumstances; worse, the test is so subjective that it approaches arbitrariness in its application.” (Id. at p. 583.)
The Raby court discussed the foundation of the rule, then continued its analysis: “Today the Culbreth rule remains at least as ambiguous as its heritage. Although a court may make as many findings of firearm use as there are victims, only one sentence enhancement may be imposed for each ‘occasion’ of gun use. But what is an ‘occasion’? That question has threatened judicial sanity and spawned distinct lines of authority.” (People v. Raby, supra, 179 Cal.App.3d at p. 585.) The court reviewed various Court of Appeal cases that found only one occasion of gun use. It then asked and sought to answer this question: “What then are the circumstances justifying the finding of multiple occasions of gun use? Frankly, we find little to distinguish the facts of the cases in this line of authority from those where but a single occasion of gun use was found.” (Id. at p. 587.)
After reviewing cases which found multiple occasions of gun use, the court summarized as follows. “Our efforts to identify and analyze the distinctions between these two lines of authority have proved exceedingly frustrating. In a broad sense, one might concede that the cases where multiple sentence enhancements have been imposed possess elements not present in Raby’s crime spree. What could be characterized as the original criminal plan in several of them was somehow altered in response to unanticipated events. ... In others, ‘the defendant [had] an opportunity to pause and reflect on the enhanced penal consequences of using his gun to achieve a newly-arising objective . . . .’ [Citations.] But do these thin distinctions justify the disparate sentencing treatment? And can a cogent rule be gleaned from the cases?
“The following example demonstrates the difficulty, if not futility, of the exercise: An armed defendant convicted of robbing seven solitary attendants at seven gas stations on the same street in the same evening may receive seven consecutive sentences and seven consecutive gun use enhancements. So might the armed crook who snares six successive drop-in customers while he is attempting to breach the safe at a gas station manned by a single employee. But the armed outlaw who robs a group of seven individuals at one gas station may receive seven consecutive robbery sentences and only one firearm use enhancement. On what basis is a more lenient sentence for the third felon justifiable? Are the ‘extra’ six victims any less terrorized *74because they were, from the outset, part of a group? Are one felon’s criminal actions less blameworthy than those of the others?
“There appears to be no easy or universal understanding of the Culbreth rule by those who must apply it, least of all us, perhaps. Clearly, neither multiple victims nor multiple motivations, without more, justify the finding of multiple occasions of gun use. Renouncing the chance to abandon a criminal scheme before encountering additional victims ... or seizing an opportunity to commit additional offenses on unexpected victims . . . , however, does appear to influence the determination as to the number of occasions of gun use. But this is just another way of saying that the more grandiose the perpetrator’s original plan, in terms of the number of victims, the less severe will be the punishment—a grotesque rule of law by any standard.” (People v. Raby, supra, 179 Cal.App.3d at pp. 589-590.)
The Raby court analyzed the facts before it, and concluded, “Raby obviously intended to rob everyone present each time; there were no unexpected victims; and there was no significant hiatus between the offenses committed in each store. Thus, the Culbreth rule, as we understand it, rewards him for the scope of his original criminal intent and permits the imposition of only two consecutive sentences for firearm use enhancements—one for each store. We believe the rule is ripe for reassessment but, of course, yield to its mandate.” (People v. Raby, supra, 179 Cal.App.3d at p. 591.)
In People v. Nguyen (1988) 204 Cal.App.3d 181, 194, footnote 11 [251 Cal.Rptr. 40], the same court reiterated its “frustration with Culbreth. It has generated considerable confusion among appellate courts as to the number of findings of firearm use that are permissible for a single criminal episode. [Citations.] Also, in our view it operates to punish those with a lesser criminal intent more severely than those who harbor more sinister objectives.”
The court in People v. Thomas (1990) 218 Cal.App.3d 1477 [267 Cal.Rptr. 865] applied Culbreth and limited the use enhancements to one for each set of crimes committed on a separate date. But it also questioned the rule. “Although the ‘single-occasion’ rule articulated in Culbreth has been criticized because it can result in punishing those with a lesser criminal intent more severely than those who harbor more sinister objectives (see People v. Raby (1986) 179 Cal.App.3d 577, 590 [224 Cal.Rptr. 576]; see also People v. Nguyen (1988) 204 Cal.App.3d 181,194, fn. 11 [251 Cal.Rptr. 40]), our Supreme Court has not yet undertaken to reconsider the underpinnings of its decision. To our mind the dissent in Culbreth contains the sounder and more logical reasoning. As an intermediate appellate court, *75however, we are compelled in this case to apply the rule of the majority opinion. [Citation.] We do respectfully urge the high court to reexamine Culbreth in light of the thoughtful criticisms voiced over the past decade.” (218 Cal.App.3d at p. 1491.)
As noted, the Court of Appeal in this case “reluctantly” struck the consecutive use enhancement under compulsion of Culbreth. But it added its voice to those urging reexamination of the rule. “We find no reason in logic why two separate acts of violence separately and properly punished with consecutive sentences cannot be enhanced with a use of a firearm as to each count. The cases are legion which have criticized Culbreth, and we count ourselves in that company. Many cases are tribute to the intellectual creativity of our brethren in circumventing its holding.”
With this background, the Attorney General asks us to overrule Culbreth, supra, 17 Cal.3d 330. Defendant argues that the decision was correct when decided, and that subsequently the Legislature has effectively approved the rule. We discuss defendant’s latter contention first.
When we examined the question of defendant’s eligibility for CYA, we found that the Legislature effectively ratified the decision of Jeanice D., supra, 28 Cal.3d 210, as applied to 16- and 17-year-olds. However, legislative inaction alone does not necessarily imply legislative approval. “The Legislature’s failure to act may indicate many things other than approval of a judicial construction of a statute: the sheer pressure of other and more important business, political considerations, or a tendency to trust to the courts to correct their own errors . . . .” (County of Los Angeles v. Workers’ Comp. Appeals Bd. (1981) 30 Cal.3d 391, 404 [179 Cal.Rptr. 214, 637 P.2d 681], internal quotation marks omitted; see also People v. Escobar (1992) 3 Cal.4th 740, 750-751 [12 Cal.Rptr.2d 586, 837 P.2d 1100].)
Defendant argues that past actions of the Legislature affirmatively show an intent to ratify Culbreth, supra, 17 Cal.3d 330.
First, in 1979, the Legislature added subdivision (h) (now (i)) to Penal Code section 1170.1, which provides, “For any violation” of specified sex offenses, ‘the number of enhancements which may be imposed shall not be limited, regardless of whether such enhancements are pursuant to this or some other section of law.” In People v. Cardenas (1982) 31 Cal.3d 897, 913-914, and footnote 9 [184 Cal.Rptr. 165, 647 P.2d 569], we recognized that the 1979 amendment to Penal Code section 1170.1 created a “possible” exception to the Culbreth rule. (Citing People v. Edwards (1981) 117 Cal.App.3d 436, 447 [172 Cal.Rptr. 652].) We rejected the argument that the *76amendment abrogated Culbreth for other purposes. “ ‘[T]o say that Culbreth no longer applies because of its apparent exclusion under Penal Code section 1170.1, subdivision (h), would ignore the maxim of statutory construction that when a statute expresses certain exceptions to a general rule, other exceptions are necessarily excluded.’ ” (People v. Cardenas, supra, at p. 914, quoting People v. Edwards, supra, at p. 448.)
The question here is not whether the amendment abolished Culbreth, supra, 17 Cal.3d 330, but whether it impliedly codified it. We believe it did not. Legislation adopting Culbreth, either expressly or impliedly, would logically be placed in Penal Code section 12022.5, the specific section at issue, not in Penal Code section 1170.1, which is a general statute implementing the Determinate Sentencing Act. Section 1170.1, subdivision (i), affects enhancements in general as applied to sex offenses. It contains no reference to the Culbreth rule or even section 12022.5. Even if that subdivision (or Penal Code section 12022.3, which the parties do not cite but which creates firearm enhancements applying specifically to sex offenses) contains an implied exception to the Culbreth rule (see People v. Ramirez (1987) 189 Cal.App.3d 603, 628-629 [233 Cal.Rptr. 645]), we see no evidence of an intent to endorse the rule in other cases. Any connection between what is now section 1170.1, subdivision (i), and the Culbreth rule is too oblique to signal an intent to codify the rule.
Second, in 1988, the Legislature added subdivision (f) to Penal Code section 12022.5, which provides, “For purposes of imposing an enhancement under Section 1170.1, the enhancements under this section shall count as one, single enhancement.” (Stats. 1988, ch. 1249, § 3, p. 4162.) Similarly, we do not believe that action endorsed Culbreth. It is not readily apparent exactly what was meant by subdivision (f) of section 12022.5. The Attorney General argues that it was the result of “poor drafting,” in that it was supposed to have been deleted from the bill when other portions to which it related were deleted. We need not resolve the point, nor need we decide what, if anything, subdivision (f) does do; it suffices to determine what it does not do—it does not codify Culbreth. If the Legislature had intended to adopt the Culbreth rule, surely it would have found a less obscure way to signal that intent.
The contrast between the legislation which partly overruled Jeanice D., supra, 28 Cal.3d 210, and that at issue here, could hardly be greater. By necessary implication, the former codified what it did not overrule. The latter acts, whatever they may have done, did not ratify Culbreth either expressly or impliedly. As we have often noted, legislative silence might support an arguable inference of acquiescence or passive approval, but *77something more than mere silence is needed to elevate the acquiescence to a species of implied legislation. In construing statutes, it is generally more fruitful to examine what the Legislature has done than what it has not done. (People v. Escobar, supra, 3 Cal.4th at p. 751.) “[L]egislative inaction is a weak reed upon which to lean . . . .” (Harris v. Capital Growth Investors XIV (1991) 52 Cal.3d 1142, 1156 [278 Cal.Rptr. 614, 805 P.2d 873], internal quotation marks omitted.)
The Legislature may have had many reasons for dealing with problems such as sentences for sex offenses but not addressing the Culbreth question, including the press of business and, not least, trusting to the courts to correct their own errors. (County of Los Angeles v. Workers’ Comp. Appeals Bd., supra, 30 Cal.3d at p. 404.) “Thus, although the Legislature has not affirmatively disapproved the court’s analysis in [Culbreth, supra, 17 Cal.3d 330] neither has it expressly or impliedly endorsed it.” (People v. Escobar, supra, 3 Cal.4th at p. 751; see also id. at p. 751, fn. 5.) In short, this court created the Culbreth rule; this court can reexamine it. (Ibid.)
Upon reexamination, we agree with the Courts of Appeal which have criticized the rule. As convincingly demonstrated in People v. Raby, supra, 179 Cal.App.3d 577, appellate courts (and undoubtedly trial courts) have found it very difficult, if not impossible, to consistently implement the rule. To the extent a consistent interpretation has emerged, it is illogical. Consider the situation of multiple victims. If all the victims were intended from the outset of the criminal scheme, multiple enhancements are prohibited; but if fewer victims were originally intended, and the defendant merely reacted to circumstances, the punishment may be more severe—“a grotesque rule of law by any standard.” (Id. at p. 590.)
The Culbreth rule finds no support in the statutory language. Penal Code section 12022.5, subdivision (a), states only that “any person who personally uses a firearm in the commission or attempted commission of a felony” shall receive the additional prescribed punishment. Nothing limits the enhancements to one for every separate occasion, whatever that might mean. It is ironic that both the majority and the dissent in Culbreth cite for their purposes the same definitions of the word “use.” (Culbreth, supra, 17 Cal.3d at pp. 334 (maj. opn.), 336 (dis. opn.).) But the dissent has the better of the argument. By any definition, defendant “use[d]” a firearm when he shot and killed Enciso; he used it again when he shot Manlove. Similarly, Culbreth used a firearm each time he killed one of his victims. As noted in People v. Raby, supra, 179 Cal.App.3d at page 584, the victims in Culbreth “were not killed with the same bullet.” The statutory language does not limit the number of enhancements in that situation.
*78We also find the majority’s deterrence analysis flawed. (Culbreth, supra, 17 Cal.3d at pp. 333-334.) We need not discuss whether the Legislature intended to punish as well as to deter (see Pen. Code, § 1170, subd. (a)(1)), for we doubt that the Legislature intended either deterrence or punishment to cease with the first victim. We think it far more likely, and consistent with the actual statutory language, that the Legislature intended to deter (and undoubtedly to punish) firearm use against multiple victims more strongly than firearm use against a single victim.
To the extent defendant contends that the Penal Code section 654 prohibition against multiple punishment mandates the Culbreth rule, we note, as did Justice Clark in dissent, that it was (and still is) settled that section 654 does not apply to “crimes of violence against multiple victims.” (Culbreth, supra, 17 Cal.3d at p. 335 (dis. opn. of Clark, J.); see also People v. McFarland (1989) 47 Cal.3d 798, 803 [254 Cal.Rptr. 331, 765 P.2d 493].) Here, there were multiple victims. Culbreth, supra, 17 Cal.3d 330, was solely an interpretation of Penal Code section 12022.5; it was not based on Penal Code section 654. For this reason, we need not decide whether section 654 applies at all to enhancements.5
Penal Code section 12022.5 simply enhances the term to be imposed for an offense; when multiple terms are imposed for multiple offenses, Penal Code section 654 exists to limit the number of terms which may be executed. However, as noted, when Culbreth was decided, and still today, the limitations of section 654 do not apply to crimes of violence against multiple victims. The Culbreth decision does not justify a conclusion that the Legislature intended that a term enhancement be treated more restrictively for multiple punishment purposes than the term for the underlying offense.
Because of the importance of the doctrine of stare decisis, we are reluctant to overturn prior opinions of this court. (Moradi-Shalal v. Fireman’s Fund Ins. Companies (1988) 46 Cal.3d 287, 296 [250 Cal.Rptr. 116, 758 P.2d 58].) Nevertheless, the doctrine is flexible, and “permits this court to reconsider, and ultimately to depart from, our own prior precedent in an appropriate case.” (Ibid.) Court-made error should not be shielded from correction. (Ibid.) One purpose behind the doctrine is to protect those who act in reliance upon existing law. “Moreover, the demands of the doctrine are ‘at their acme . . . where reliance interests are involved.’ ” (Quill Corp. *79v. North Dakota (1992) 504 U.S. _ [119 L.Ed.2d 91, 112, 112 S.Ct. 1904, 1923] (cone. opn. of Scalia, J.), quoting Payne v. Tennessee (1991) 501 U.S. _ [115 L.Ed.2d 720, 737, 111 S.Ct. 2597, 2610].) Such reliance interests can be minimized or even negated by making the new rule prospective only, as in Moradi Shalal v. Fireman’s Fund Ins. Companies, supra, 46 Cal.3d at page 305.
The Culbreth rule has proven to be almost impossible to implement in a nonarbitrary fashion. To the extent it has been interpreted consistently it has yielded illogical and “grotesque” results. It is contrary to the relevant statutory language, and is inconsistent with the very deterrence intent the Culbreth court found inherent in the statute.
For these reasons, we hereby overrule Culbreth, supra, 17 Cal.3d 330. Subject to Penal Code sections 654 and 1170.1, and any other applicable limitations, a firearm-use enhancement under section 12022.5 may be imposed for each separate offense for which the enhancement is found true. (We need not and do not decide here the propriety of multiple enhancements when the same act results in multiple victims, such as when one bullet hits two or more persons. We leave that question to a later day.)6
2. Application to This Case
Defendant argues that any judicial reinterpretation of Penal Code section 12022.5 cannot be applied to him without violating the due process and ex post facto clauses of the United States Constitution. We agree, and therefore do not apply the overruling of Culbreth to him.
As relevant here, any statute “ ‘which makes more burdensome the punishment for a crime, after its commission’ ” violates the ex post facto prohibition of the United States Constitution (Collins v. Youngblood (1990) 497 U.S. 37, 42 [111 L.Ed.2d 30, 38-39, 110 S.Ct. 2715], quoting Beazell v. Ohio (1925) 269 U.S. 167, 169-170 [70 L.Ed. 216, 217-218, 46 S.Ct. 68]), and its California counterpart. (Tapia v. Superior Court (1991) 53 Cal.3d 282, 294-297 [279 Cal.Rptr. 592, 807 P.2d 434].) While this limitation is specifically directed to the legislative, not judicial, branch, the same principle applies to judicial decisions. (People v. Escobar, supra, 3 Cal.4th at p. 752; People v. Wharton (1991) 53 Cal.3d 522, 586 [280 Cal.Rptr. 631, 809 *80P.2d 290].) Thus, “If a judicial construction of a criminal statute is ‘unexpected and indefensible by reference to the law which had been expressed prior to the conduct in issue,’ it must not be given retroactive effect. [Citation.]” (Bouie v. City of Columbia (1964) 378 U.S. 347, 354 [12 L.Ed.2d 894, 900, 84 S.Ct. 1697]; see also People v. Escobar, supra, 3 Cal.4th at p. 752 and In re Baert (1988) 205 Cal.App.3d 514, 518 [252 Cal.Rptr. 418].)
The Culbreth rule has been the law of this state since 1976. It was the law when defendant committed his crimes. Refusing to apply it here would make the punishment for his crimes more burdensome after he committed them. Defendant is therefore constitutionally entitled to its benefit.
In re Baert, supra, 205 Cal.App.3d 514, confronted the question whether our holding in People v. Anderson (1987) 43 Cal.3d 1104 [240 Cal.Rptr. 585, 742 P.2d 1306] “eliminating intent to kill as an element of the felony-murder special circumstance [citation] may be applied retroactively to crimes committed during the period when Carlos v. Superior Court (1983) 35 Cal.3d 131 [197 Cal.Rptr. 79, 672 P.2d 862], required proof of intentionality.” (In re Baert, supra, at p. 516.) The court reasoned that the “holding of Carlos was unambiguous and consistently applied in accord with its express language” (id. at p. 519), and that the “about-face in Anderson [was] ‘an unforeseeable judicial enlargement of a criminal statute. . . .’” (Id. at p. 520, quoting Bouie v. Columbia, supra, 378 U.S. at p. 353 [12 L.Ed.2d at p. 899].) Therefore, the court held, Anderson could not be applied to crimes committed after Carlos and before Anderson.
The same analysis applies here. Although the Culbreth rule has not been applied in a consistent fashion, it has consistently been applied. The Attorney General argues that the “plain words of the statute together with the controversy surrounding this decision constitute a fair warning to [defendant] that this Court might well reconsider” the rule. A similar argument that the defendant was on notice that Carlos v. Superior Court, supra, 35 Cal.3d 131, might be overruled was rejected in In re Baert, supra, 205 Cal.App.3d at page 520: “At best this argument is based on conjectural suppositions, inferences, and conclusions.” The mere possibility that this court might reconsider its own precedent is not the equivalent of actually overruling it. We thus conclude that our holding overruling Culbreth, supra, 17 Cal.3d 330, cannot be applied retroactively to crimes committed during the period between Culbreth and the finality of this decision.
Applying the Culbreth rule, we find that multiple enhancements were improper. Both victims were intended from the outset. Neither arrived *81unexpectedly after the other was shot. There was a time gap between the shootings, but it was brief, and the shootings occurred in the same location. We agree with the Court of Appeal that by any fair reading of the relevant cases, both shootings occurred on the same occasion within the meaning of Culbreth. Therefore, defendant must prevail under that decision. (Cf. People v. Pride (1992) 3 Cal.4th 195, 269 [10 Cal.Rptr.2d 636, 833 P.2d 643] [separate occasions found when “the stabbings occurred at least 30 minutes apart in different areas of the vast office building”]; People v. White (1981) 117 Cal.App.3d 270, 284 [172 Cal.Rptr. 612] [separate occasions found when defendant reloaded his gun and traveled from one side to the other of the city hall of San Francisco between two killings].)
As discussed in People v. Raby, supra, 179 Cal.App.3d 577, if the second victim had unexpectedly appeared on the scene, and defendant had reacted to the turn of events by shooting her, multiple firearm-use enhancements would have been permitted. The differing treatment is hard to justify. This observation, however, only illustrates the error of Culbreth, supra, 17 Cal.3d 330; it does not distinguish the decision.
III. Conclusion
The judgment of the Court of Appeal is reversed to the extent it holds that defendant is ineligible for a CYA commitment. In all other respects, it is affirmed.
Panelli, J., Baxter, J., and George, J., concurred.
A person under the age of 21 who is sentenced to state prison rather than committed to CYA may nonetheless be physically housed at CYA until such time as he is transferred to prison to serve the rest of the term. (Welf. & Inst. Code, § 1731.5, subd. (c).) The court here ordered defendant housed at CYA; that is not now at issue.
As it now reads in pertinent part, Welfare and Institutions Code section 1731.5, subdivision (a)(2), makes ineligible for a CYA commitment any person who has been “convicted of first degree murder, committed when that person was 18 years of age or older, or sentenced to death, [or] imprisonment for life ...”
The Legislature’s inaction since People v. Ladanio, supra, 211 Cal.App.3d 1114, was decided does not signal agreement with that decision, nor does it prevent us from overruling it. As discussed in part II. B. 1., post, legislative inaction may mean many things other than approval of a judicial ruling.
The predecessor statute applied only to specified felonies, not any felony, and did not contain the word “personally.” The length of the enhancement has been changed, and varies depending upon the crime and other factors. There have been other minor word changes. But the critical language “uses a firearm in the commission or attempted commission of’ has never been changed. (See Culbreth, supra, 17 Cal.3d at p. 332, fn. 1.)
In People v. Hernandez (1988) 46 Cal.3d 194, 205 [249 Cal.Rptr. 850, 757 P.2d 1013], a case dealing with the pleading and proof requirements for enhancements, we erroneously described People v. Cardenas, supra, 31 Cal.3d at pages 913-914, as “reaffirming that § 654 applies to enhancements.” The portion of Cardenas cited in Hernandez concerned the Culbreth rule. Neither Cardenas nor Culbreth even cited, much less relied on, Penal Code section 654.
As noted in the dissent, we cited Culbreth, supra, 17 Cal.3d 330, in People v. Harris (1989) 47 Cal.3d 1047, 1103 [255 Cal.Rptr. 352, 767 P.2d 619]. The issue in Harris was, however, entirely different than the one here. In Harris, convictions on two substantive counts were stayed under Penal Code section 654. (47 Cal.3d at p. 1102.) We merely noted that the trial court also properly stayed the section 12022.5 enhancements as to those counts. (47 Cal.3d at p. 1103.) Nothing in this opinion is inconsistent with Harris.